Encyclopedia of Cancer [Volume 1, 3 ed.] 9780128124840

Table of contents :
9780128124857v1_WEB
Cover
ENCYCLOPEDIA OF CANCER
ENCYCLOPEDIA OF CANCER
Copyright
EDITORS IN CHIEF
EDITORIAL BOARD
SECTION EDITORS
CONTRIBUTORS TO ALL VOLUMES
HOW TO USE THE ENCYCLOPEDIA
SUBJECT CLASSIFICATION
PREFACE
CONTENTS OF ALL VOLUMES
PERMISSION ACKNOWLEDGMENTS
9780128124857v2_WEB
Cover
ENCYCLOPEDIA OF CANCER
ENCYCLOPEDIA OF CANCER
Copyright
EDITORS IN CHIEF
EDITORIAL BOARD
SECTION EDITORS
CONTRIBUTORS TO ALL VOLUMES
HOW TO USE THE ENCYCLOPEDIA
SUBJECT CLASSIFICATION
PREFACE
CONTENTS OF ALL VOLUMES
PERMISSION ACKNOWLEDGMENTS
9780128124857v3_WEB
Cover
ENCYCLOPEDIA OF CANCER
ENCYCLOPEDIA OF CANCER
Copyright
EDITORS IN CHIEF
EDITORIAL BOARD
SECTION EDITORS
CONTRIBUTORS TO ALL VOLUMES
HOW TO USE THE ENCYCLOPEDIA
SUBJECT CLASSIFICATION
PREFACE
CONTENTS OF ALL VOLUMES
PERMISSION ACKNOWLEDGMENTS

Citation preview

ENCYCLOPEDIA OF CANCER THIRD EDITION EDITORS IN CHIEF

Paolo Boffetta Tisch Cancer Institute Icahn School of Medicine at Mount Sinai New York, NY, United States

Pierre Hainaut Cancer Biology, Faculty of Medicine, University Grenoble-Alpes Grenoble, France Institute for Advanced Biosciences, Inserm, CNRS, UGA Grenoble, France

VOLUME 1

Amsterdam • Boston • Heidelberg • London • New York • Oxford Paris • San Diego • San Francisco • Singapore • Sydney • Tokyo Academic Press is an imprint of Elsevier

Academic Press is an imprint of Elsevier The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK 50 Hampshire St, 5th Floor, Cambridge, MA 02139, USA Copyright Ó 2019 Elsevier Inc. unless otherwise stated. All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher’s permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions. This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein). Notices Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary. Practitioners and researchers may always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility. To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein. Library of Congress Cataloging-in-Publication Data A catalog record for this book is available from the Library of Congress British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library ISBN 978-0-12-812484-0

For information on all publications visit our website at http://store.elsevier.com

Publisher: Oliver Walter Acquisition Editor: Sam Crowe Content Project Manager: Kate Miklaszewska-Gorczyca Designer: Matthew Limbert

Printed and bound in the United States

EDITORS IN CHIEF Paolo Boffetta, MD, MPH, graduated in Medicine from the University of Turin and obtained a Master in Public Health from Columbia University. He worked at the American Cancer Society, the International Agency for Research on Cancer, a specialized agency of the World Health Organization located in Lyon, France, and at the German Cancer Research Center in Heidelberg, Germany. In 2010 he moved to Icahn School of Medicine at Mount Sinai, New York, where he is Professor of Medicine, Global Health, Oncological Sciences, and Preventive Medicine, and Associate Director for Global Oncology of the Tisch Cancer Institute, a NCI-designated Cancer Center. Dr. Boffetta also holds a full professorship at the University of Bologna and adjunct appointments at Harvard School of Public Health, Vanderbilt University, Catholic University of Rome, and University of South Carolina. Since 2017 he is Senior Advisor for Research at Vinmec Health System. His main fields of research are cancer epidemiology and cancer prevention, with emphasis on modifiable risk factors (environmental exposure and personal behaviors), gene–environment interactions, molecular epidemiology, and evidence integration. He is the initiator and coordinator of several large-scale international consortia of molecular cancer epidemiology studies, including ILCCO (lung cancer), INHANCE (head and neck cancer), PANC4 (pancreatic cancer), StoP (stomach cancer), and ILCEC (liver cancer). Dr. Boffetta is the editor or associate editor of 5 scientific journals and member of the editorial board of 10 additional journals; he is a member of review panels of NIH and several medical research agencies in Europe. He has edited 13 books and is Editor in Chief of the new edition of Elsevier’s Encyclopedia of Cancer. He has published over 1250 peer-reviewed publications; his publications have been quoted more than 90000 times; his h-index is 148.

Pierre Hainaut is Professor of Exceptional Class in Cancer Biology at University Grenoble-Alpes, France and Director of the Institute of Advanced Biosciences (IAB), a joint research center of Institut National de la Santé et de la Recherche Médicale (Inserm), Centre National de la Recherche Scientifique (CNRS), and University Grenoble-Alpes. He also heads the IAB research team on Molecular Biology and Biomarkers. Pierre Hainaut holds a PhD in Biology (Zoology) from University of Liège, Belgium (1987). After postdocs in Nice (France, 1988–90), Cambridge, and York (United Kingdom, 1990–94), he joined the International Agency for Research on Cancer (IARC, World Health Organization) in 1995, where he held the post of Head of Molecular Carcinogenesis from 1999 to 2011. In 2012, he joined the International Prevention Research Institute (Lyon, France) and became Professor at the Strathclyde Institute of Pharmacy and Biomedical Science (Glasgow, United Kingdom). In 2014, he was awarded a Chair of Excellence in Translational Research from University Grenoble-Alpes (Grenoble, France). His research focuses on TP53 mutations and p53 protein regulation in cancer and chronic diseases. From 1994 to 2011, he has led the development of the international IARC TP53 database, a source of information on the causes and consequences of mutations affecting the TP53 suppressor gene in cancer. His work addresses the mechanisms of TP53 mutagenesis as well as the prognostic and predictive significance of TP53 mutations in lung, liver, and oesophageal cancers. His studies on p53 regulation have focused on the role of environmental mutagens in TP53 mutagenesis, on the biochemical mechanisms of p53 control by oxidation-reduction and by metabolism, and on the identification of p53 isoforms as factors acting as dominant inhibitors of p53 functions in cancers without TP53 mutations. His current activities focus on germline TP53 mutation and on the diversity of genetic and nongenetic factors that modulate the penetrance of the Li–Fraumeni Syndrome, as well as on the mechanisms that maintain optimal p53 protein balance in cells and tissues over lifetime. He is the author of over 450 publications and 50 book chapters. He is Editor of the Cancer Biology section of Current Opinion in Oncology. He has co-edited books on p53 (25 Years of p53 Research, 2005, 2007, p53 in the Clinics, Springer), a textbook on molecular epidemiology (Molecular Epidemiology: Principle and Practice, IARC Press, 2011) and two-volume textbook on human biobanking (Human Biobanking, Principle and Practice, 2017, 2018).

v

EDITORIAL BOARD Fred T. Bosman Institute of Pathology University of Lausanne Medical Center (CHUV) Lausanne, Switzerland

Graham A. Colditz Alvin J. Siteman Cancer Center and Institute for Public Health Washington University School of Medicine St. Louis, MO, United States Barnes Jewish Hospital St. Louis, MO, United States Division of Public Health Sciences, Department of Surgery Washington University School of Medicine St. Louis, MO, United States

Carlo La Vecchia Department of Clinical Sciences and Community Health University of Milan Milan, Italy

Gerd P. Pfeifer Center for Epigenetics Van Andel Research Institute Grand Rapids, MI, United States

Marco Pierotti IFOM/Cogentech Milan, Italy

Thomas Tursz University of Paris-Sud Orsay, France Institut Gustave Roussy Villejuif, France

vii

SECTION EDITORS Fred T. Bosman MD PhD, studied Medicine at the University of Leiden (MD 1971), where he also earned his PhD degree (in cytogenetics, 1976), and trained as a pathologist. He was staff pathologist at the University of Leiden, Professor and Chair of Pathology at the Faculty of Medicine of the University of Maastricht, Faculty of Medicine of the Erasmus University in Rotterdam, Director of the University Institute of Pathology, and Professor of Pathology at the University Medical Center (CHUV) of Lausanne in Switzerland, now emeritus. He is Honorary Fellow of the Royal College of Pathologists (United Kingdom) and foreign correspondent of the Royal Netherlands Academy of Sciences. Fred Bosman’s research activities (combining diagnostic and experimental pathology) focused on the biology of digestive tract cancer, notably Barrett’s esophagus and colorectal cancer, with a strong emphasis on the development of molecular diagnostics. He has written over 350 original publications and over 50 book chapters. Fred Bosman was Series co-editor of the 4th edition of the WHO Series Classification of Human Tumours, the international standard for tumor classification, and co-editor of the Volume on Tumours of the Digestive Tract.

Graham A. Colditz, MD, DrPH, FAFPHM, is an internationally recognized leader in cancer prevention. As an epidemiologist and public health expert, he has a longstanding interest in the preventable causes of chronic disease, particularly among women. He focuses his research on early life and adolescent lifestyle, growth, and breast cancer risk. He is also interested in approaches to speed translation of research findings into prevention strategies that work. Dr. Colditz developed the award-winning Your Disease Risk website (www.yourdiseaserisk.wustl.edu) which communicates tailored prevention messages to the public. He has published over 1100 peer-reviewed publications, six books and six reports for the Institute of Medicine, National Academy of Sciences. His h-index is over 220. In October 2006, on the basis of professional achievement and commitment to public health, Dr. Colditz was elected to membership of the National Academy of Medicine, an independent body that advises the US government on issues affecting public health. In 2011, he was awarded the American Cancer Society Medal of Honor for cancer control research. In 2012 he received the AACR-American Cancer Society Award for Research Excellence in Cancer Epidemiology and Prevention. He also received awards in 2014 for cancer prevention research from ASCO and from AACR. During 2016 he served on the Implementation Science Work Group of the Blue-Ribbon Panel to advise the National Cancer Moonshot. He received the 2018 Daniel P. Schuster Award for Distinguished Work in Clinical and Translational Science, Washington University School of Medicine. He was also elected as a Fellow, American Association for the Advancement of Science

Carlo La Vecchia received his MD from the University of Milan and a Master of Science degree in Medicine (epidemiology) from Oxford University. Presently, he is Professor of Medical Statistics and Epidemiology at the Faculty of Medicine at the University of Milan. Dr. La Vecchia serves as an editor for numerous clinical and epidemiologic journals. He is among the most renowned and productive epidemiologists in the field with over 2040 peer-reviewed papers in the literature and is among the most highly cited medical researchers in the world, according to ISI HighlyCited.com, the developer and publisher of the Science Citation Index (2003, 2017, H index 153, H10 index 1543, second Italian in Clinical Medicine). Dr. La Vecchia is Adjunct Professor of Medicine at Vanderbilt Medical Center and the Vanderbilt-Ingram Cancer Center (2002-18).

ix

x

Section Editors Gerd. P. Pfeifer received a PhD degree in biochemistry from Goethe University in Frankfurt, Germany. After postdoctoral training, he became a faculty member at the Beckman Research Institute of the City of Hope in Duarte, California, where he spent much of his career working on cancer research. In 2014, Dr. Pfeifer joined the new Center for Epigenetics at the Van Andel Research Institute in Grand Rapids, MI, United States, as a Professor of Epigenetics. Dr. Pfeifer has authored more than 300 publications, has held an NIH MERIT award, and was elected Fellow of the American Association for the Advancement of Science in 2015. Research in Dr. Pfeifer’s laboratory has been concerned with genetic and epigenetic mechanisms of human carcinogenesis, with emphasis on DNA methylation and genetic toxicology.

Marco Alessandro Pierotti graduated in 1973 in Biological Sciences at the University of Milan, Italy, and started working at the Fondazione IRCCS Istituto Nazionale dei Tumori (INT) in Milan. From 1978 to 1980 he was Visiting Investigator at the Laboratory of Chemical Cancerogenesis of the NCINIH Bethesda (MD, United States) and Postdoctoral Research Fellow at the Laboratory of Viral Oncology of the Memorial Sloan-Kettering Institute in New York. In 2006, Dr. Pierotti was appointed Scientific Director of the Fondazione IRCCS Istituto Nazionale dei Tumori in Milan, where, since 1970, he had already held various positions, including Director of the Department of Experimental Oncology. Since 1988, he has been Professor of Molecular Genetics of Cancer at the Postgraduate School of Oncology, University of Milan Medical School and co-director of the Laboratory of Molecular Diagnosis at the INT. In September 2014 he resigned from his position at INT to take the position of President and CEO of Nerviano Medical Sciences (NMS) srl, one of the biggest oncological pharma companies in Europe. In April 2015 he left the company and took the position of Scientific Coordinator of the Institute of Pediatric Researches (IRP) in Padua, Italy, devoted to study molecular aspects of the main pediatric diseases with particular focus on pediatric onco-hematology. The Institute was created by a private Charity, The City of Hope Foundation of Monte Malo (Vi). Since 2000 he is Senior Group leader of the Molecular Genetics of Cancer group at the Institute FIRC of Molecular Oncology (IFOM, Milan). Past President (2006–08) of the Italian Cancer Society, Dr. Pierotti is a member of the American Association for Cancer Research and of its Advisory Board and the Laboratory Research Awards Selection Committee. He has also been President (2006–08) of the European Association for Cancer Research (EACR) and in this role was among the founders of the European CanCer Organisation (ECCO) where he was appointed as member of the Policy Committee. In recent years, he was the Italian Representative at the Scientific Committee of the International Agency for Research on Cancer (IARC), Lyon. From 2006 to 2014 he was Scientific Secretary of Alleanza Contro il Cancro (ACC), promoted by the Italian Ministry of Health. In 2008 he was Member of the Evaluation of the Research Program Functional and Structural Genomics for DKFZ. He was an expert for the Oncology Research Projects of the European Community and was a consultant in oncology research for the Ministries of Research of different Countries. From 2006 to 2014 he was Chair of the regional Project, The Region Lombardy Oncological Network (ROL), selected in 2014 by the EC as one of the best examples of oncological network. His appointments in the Organisation of European Cancer Institutes (OECI) started in 2007 when he took the position of Vice-President. From 2008 he was the President-elect and then the President of the Organization. Finally, in 2014 he was appointed OECI Executive Secretary. Over the years, Dr. Pierotti has been Principal Investigator or Head of several national and international research grants, funded by both private and public bodies. His authorship includes over 470 publications that deal with various aspect of experimental oncology including studies on immunology, biochemistry, and molecular biology using both experimental and human tumors. In addition, since its fifth edition he is the first author of the chapter on “Oncogenes” in the most reputed textbook Cancer Medicine (Holland-Frei). The metrics of his scientific activity is summarized by an H index of 96 and total citations of 37.256 (Google scholar June 2018).

Section Editors

xi

Professor Thomas Tursz, born in Kraków, Poland, in 1946, died in Paris, France, on April 27 2018. He was Professor of Oncology at the Faculty of Medicine Paris-Sud since 1986 and General Director of the Institut Gustave Roussy (1994–2010). He was the leader of the French Doctoral School of Oncology which he founded in 1999, and President of the French Federation of Comprehensive Anticancer Centres (FNCLCC) from 2004 to 2010. He was highly involved in the European Organization for the Research and Treatment of Cancer (EORTC) as both Chairman of the Scientific Advisory Committee (2003–06) and Vice President of the Board (2006–09). His experience as President of the FNCLCC was crucial for the Organization of European Cancer Institutes (OECI) when he acted as President from 2002 to 2005. His scientific interests included the biology of virus-induced tumors, as well as immunological responses including the role of thioredoxin in lymphocytes infected by Epstein–Barr virus. In the clinical research area, he conducted a number of important clinical trials in breast cancer, lung cancer, and soft-tissue sarcoma. He had a particular interest in cytokines and gene therapy, and his clinical research activities were further disseminated to the European level when he was the Chairman of the Sarcoma Group of the EORTC (1993–96). Prof. Tursz received several prestigious awards, such as the Prix de Cancérologie from the French National League Against Cancer (1979), the Bernard Halpern Immunology Award (1983), the Rosen Oncology Award (1989), the Grand Prix in Oncology from the Academy of Medicine (1992), the Hamilton Fairley Award for clinical research (1998), and the Prix de Rayonnement Français (2001). He was the author of 350 international scientific publications. He was also an esteemed member of the Editorial Board of Molecular Oncology ever since its creation in 2007. Modified from Ulrik Ringborg and Julio E. Celis. Thomas Tursz (1946–2018) in: Molecular Oncology (2018). Published by FEBS Press and John Wiley & Sons Ltd. https://doi.org/10.1002/1878-0261.12361

CONTRIBUTORS TO ALL VOLUMES Rachel Abbotts University of Maryland School of Medicine, Baltimore, MD, United States

Christina Bamia National and Kapodistrian University of Athens, Athens, Greece

Ludmil B Alexandrov University of California San Diego, San Diego, CA, United States

Thomas D Bannister The Scripps Research Institute, Jupiter, FL, United States

Geneviève Almouzni Institut Curie, PSL Research University, CNRS, Equipe Labellisée Ligue contre le Cancer, Paris, France; and Sorbonne University, UPMC Univ Paris 06, CNRS, Paris, France Fatima Ameer University of the Punjab, Lahore, Pakistan Volker M Arlt MRC-PHE Centre for Environment and Health, King’s College London, London, United Kingdom; and NIHR Health Protection Research Unit in Health Impact of Environmental Hazards at King’s College London in partnership with Public Health England, London, United Kingdom Elena Arrigoni University of Pisa, Pisa, Italy Sylvia L Asa University Health Network and University of Toronto, Toronto, ON, Canada Caroline Aspord Institute for Advanced Biosciences, Immunobiology and Immunotherapy in Chronic Diseases Team, Inserm U 1209, CNRS UMR 5309, University Grenoble Alpes, Grenoble, France Livia S A Augustin National Cancer Institute IRCCS G. Pascale Foundation, Naples, Italy; and Clinical Nutrition and Risk Factor Modification Centre, St Michael’s Hospital, Toronto, On, Canada Rameshwar N K Bamezai National Centre of Applied Human Genetics, School of Life Sciences, Jawaharlal Nehru University, New Delhi, India

Michael Baumann German Cancer Research Center (DKFZ), Heidelberg, Germany; and Technical University of Dresden, Dresden, Germany Daniel Baumhoer University Hospital Basel, Basel, Switzerland Aditya Bele University of Florida, Gainesville, FL, United States Richard L Bennett University of Florida, Gainesville, FL, United States Anton Berns Oncode Institute, Division of Molecular Genetics, The Netherlands Cancer Institute, Amsterdam, The Netherlands Venkaiah Betapudi US Army Medical Research Institute of Chemical Defense, Aberdeen Proving Ground, MD, United States Joydeep Bhadury The University of Tokyo, Tokyo, Japan Justin A Bishop University of Texas Southwestern Medical Center, Dallas, TX, United States Stefania Boccia Institute of Public Health, Catholic University of the Sacred Heart, Fondazione Policlinico “Agostino Gemelli”, Rome, Italy Stefan K Bohlander University of Auckland, Auckland, New Zealand Lubor Borsig University of Zurich and Zurich Center for Integrative Human Physiology, Zurich, Switzerland

xiii

xiv

Contributors to All Volumes

Fred T Bosman Institute of Pathology, University of Lausanne Medical Center (CHUV), Lausanne, Switzerland Ekaterina Bourova-Flin University of Grenoble Alpes, Grenoble, France

Dhyan Chandra Pharmacology and Therapeutics, Roswell Park Comprehensive Cancer Center, Buffalo, NY, United States

Mariana Brandão University of Porto, Porto, Portugal; and Institute Jules Bordet, Brussels, Belgium

Laurence Chaperot Institute for Advanced Biosciences, Immunobiology and Immunotherapy in Chronic Diseases Team, Inserm U 1209, CNRS UMR 5309, University Grenoble Alpes, Grenoble, France

Kristen D Brantley Harvard T.H. Chan School of Public Health, Boston, MA, United States

Adrian K Charles Sidra Medicine, Doha, Qatar

Thomas Brenn University of Calgary, Calgary, AB, Canada; and Alberta Health Services, Calgary, AB, Canada

Harvey Checkoway University of California San Diego, La Jolla, CA, United States

Jan Buckner Mayo Clinic, Rochester, MN, United States

Sai-Juan Chen State Key Laboratory of Medical Genomics, Shanghai Institute of Hematology, Rui Jin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China

Januario B Cabral-Neto Federal University of Rio de Janeiro, Rio de Janeiro, Brazil Jean Cadet University of Sherbrooke, Sherbrooke, QC, Canada Laia Caja Uppsala University, Uppsala, Sweden Ilaria Calabrese University of Naples Federico II, Naples, Italy Elías Campo University of Barcelona, Barcelona, Spain Kaixiang Cao Northwestern University, Chicago, IL, United States Fátima Carneiro São João Hospital, Porto, Portugal; and University of Porto (FMUP), Porto, Portugal Francesca Carozzi ISPO, Cancer Prevention and Research Institute, Florence, Italy

Zhu Chen State Key Laboratory of Medical Genomics, Shanghai Institute of Hematology, Rui Jin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China Nai-Kong V Cheung Memorial Sloan-Kettering Cancer Center, New York, NY, United States Jennifer Choe Duke University School of Medicine, Durham, NC, United States Edward Chow Odette Cancer Centre, Sunnybrook Health Sciences Centre, Toronto, ON, Canada; and University of Toronto, Toronto, ON, Canada Ying-Hsia Chu The University of Wisconsin School of Medicine and Public Health, Madison, WI, United States

France Carrier University of Maryland, Baltimore, MD, United States

Fabio Ciccarone University of Rome “Tor Vergata”, Rome, Italy

Carla Carrilho Hospital Central of Maputo, Maputo, Mozambique; and University Eduardo Mondlane, Maputo, Mozambique

Maria Rosa Ciriolo University of Rome “Tor Vergata”, Rome, Italy

Nathalie Cassoux Institut Curie, Paris, France

Graham A Colditz Alvin J. Siteman Cancer Center and Institute for Public Health, Washington University School of Medicine, St. Louis, MO, United States

JoAnn C Castelli University of California San Francisco, San Francisco, CA, United States

Laura C Collopy University College London, London, United Kingdom

Contributors to All Volumes

Antonio Cossu University Hospital Health Unit/Azienda OspedalieroUniversitaria (AOU), Sassari, Italy Sarah E Coupland Royal Liverpool University Hospital, Liverpool, United Kingdom; and University of Liverpool, Liverpool, United Kingdom Alix Coysh University of Auckland, Auckland, New Zealand Ana Cuenda Centro Nacional de Biotecnología/CSIC (CNB-CSIC), Madrid, Spain Romano Danesi University of Pisa, Pisa, Italy Rachel Dankner Sheba Medical Center, Tel Hashomer, Israel; Sackler Faculty of Medicine, School of Public Health, Tel Aviv University, Tel Aviv, Israel; and The Feinstein Institute for Medical Research, Manhasset, NY, United States Ben Davidson Oslo University Hospital, Oslo, Norway; and University of Oslo, Oslo, Norway Carlos E de Andrea University of Navarra, Pamplona, Spain Carlo DeAngelis Odette Cancer Centre, Sunnybrook Health Sciences Centre, Toronto, ON, Canada; and University of Toronto, Toronto, ON, Canada Vincenzo De Giorgi University of Florence, Florence, Italy Vincenzo De Iuliis Gabriele D’Annunzio University of Chieti, Chieti, Italy Olivier Delattre Institute Curie Hospital, Paris, France; and Inserm U830, SIREDO Oncology Center, Paris Sciences and Letters (PSL) Research University, Paris, France Laurence de Leval Lausanne University Hospital (CHUV), Lausanne, Switzerland

Marzia Del Re University of Pisa, Pisa, Italy Marco Demaria European Institute for the Biology of Aging (ERIBA), University Medical Center Groningen (UMCG), Groningen, The Netherlands; and University of Groningen, Groningen, The Netherlands Catherine de Martel International Agency for Research on Cancer, Lyon, France Evangelina López de Maturana Spanish National Cancer Research Center (CNIO), Madrid, Spain; and Biomedical Research Networking Centre in Oncology, Madrid, Spain Laurence Desjardins Institut Curie, Paris, France Elizabeth E Devore Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, United States Luca Di Leo University of Rome “Tor Vergata”, Rome, Italy Antonello Di Paolo University of Pisa, Pisa, Italy Emanuela D’Angelo   University of LAquila, LAquila, Italy Nadja Ebert German Cancer Research Center (DKFZ), Heidelberg, Germany; and Technical University of Dresden, Dresden, Germany A Heather Eliassen Harvard T.H. Chan School of Public Health, Boston, MA, United States; Brigham and Women’s Hospital; and Harvard Medical School, Boston, MA, United States Iman El Sayed Medical Research Institute, Alexandria University, Alexandria, Egypt Meira Epplein Duke University, Durham, NC, United States

Julio Delgado University of Barcelona, Barcelona, Spain

Iñigo Espinosa Autonomous University of Barcelona, Barcelona, Spain

Stefan Delorme German Cancer Research Center (DKFZ), Heidelberg, Germany

Irene Esposito Heinrich Heine University, University Hospital, Duesseldorf, Germany

xv

xvi

Contributors to All Volumes

Manel Esteller Bellvitge Biomedical Biomedical Research Institute (IDIBELL), Barcelona, Catalonia, Spain; University of Barcelona, Barcelona, Catalonia, Spain; and Catalan Institution for Research and Advanced Studies (ICREA), Barcelona, Catalonia, Spain Shereen Ezzat University Health Network and University of Toronto, Toronto, ON, Canada Chuannan Fan Oncode Institute, Leiden University Medical Center, Leiden, Netherlands

Preethi S George Regional Cancer Centre, Trivandrum, India Katherine A Giles UNSW Sydney, Sydney, NSW, Australia Thomas J Giordano University of Michigan Health System, Ann Arbor, MI, United States Nicolas Girard Curie Institute, Paris, France; and Lyon University, Lyon, France

Lynnette Fernandez-Cuesta International Agency for Research on Cancer, Lyon, France

Abhiram Gokhale Pharmacology and Therapeutics, Roswell Park Comprehensive Cancer Center, Buffalo, NY, United States

Giustina Ferone Oncode Institute, Division of Molecular Genetics, The Netherlands Cancer Institute, Amsterdam, The Netherlands

Tyler Golato National Institute on Aging, Intramural Research Program, National Institutes of Health, Baltimore, MD, United States

Stefano Fiori European Institute of Oncology, Milan, Italy

Paulina Gomez-Rubio Spanish National Cancer Research Center (CNIO), Madrid, Spain; and Biomedical Research Networking Centre in Oncology, Madrid, Spain

Martin Fischer Leibniz Institute on Aging e Fritz Lipmann Institute (FLI), Jena, Germany Jean-François Fléjou Saint-Antoine Hospital, AP-HP, Faculty of Medicine, Sorbonne University, Paris, France Filipa Fontes University of Porto, Porto, Portugal; and Portuguese Institute of Oncology Francisco Gentil, Porto, Portugal Simon J Furney Royal College of Surgeons in Ireland, Dublin, Ireland Nina Gale Faculty of Medicine, University of Ljubljana, Ljubljana, Slovenia Sara Gandini European Institute of Oncology, Milan, Italy Vithusha Ganesh Odette Cancer Centre, Sunnybrook Health Sciences Centre, Toronto, ON, Canada; and University of Toronto, Toronto, ON, Canada Mengqing Gao Shanghai Jiao Tong University School of Medicine, Shanghai, China

M M Gottesman Laboratory of Cell Biology, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA Bernardo H L Goulart Hutchinson Institute of Cancer Outcomes Research (HICOR), Seattle, WA, United States; Fred Hutchinson Cancer Research Center, Seattle, WA, United States; and University of Washington, Seattle, WA, United States William J Graham V Ludwig Institute for Cancer Research, San Diego Branch, La Jolla, CA, United States Emmanuel Griessinger INSERM, C3M, Nice, France; and University of Nice Sophia, Nice, France John D Groopman Johns Hopkins University, Baltimore, MD, United States Irene Gullo São João Hospital, Porto, Portugal; and University of Porto (FMUP), Porto, Portugal Meiyun Guo University of British Columbia, Vancouver, BC, Canada

Contributors to All Volumes

Lena Haeberle Heinrich Heine University, University Hospital, Duesseldorf, Germany Pierre Hainaut Cancer Biology, Faculty of Medicine, University Grenoble-Alpes, Grenoble, France; and Institute for Advanced Biosciences, Inserm, CNRS, UGA, Grenoble, France Susan E Hankinson University of Massachusetts Amherst, Amherst, MA, United States Leonie Hartl University of Gothenburg, Gothenburg, Sweden Christian Hartmann Medical School Hannover, Hannover, Germany Dana Hashim International Agency for Research on Cancer, Lyon, France; and Icahn School of Medicine at Mount Sinai, New York, NY, United States Dan He Drexel University, Philadelphia, PA, United States

xvii

Janusz A Jankowski University Hospitals of Morecambe Bay NHS Foundation Trust, Lancaster, United Kingdom; National Institute of Health and Care Excellence, Manchester, United Kingdom; and Royal College of Surgeons in Ireland, Dublin, Ireland Daniel Jeffery Institut Curie, PSL Research University, CNRS, Equipe Labellisée Ligue contre le Cancer, Paris, France; and Sorbonne University, UPMC Univ Paris 06, CNRS, Paris, France Albert Jeltsch Stuttgart University, Stuttgart, Germany Seung-Gi Jin Center for Epigenetics, Van Andel Research Institute, Grand Rapids, MI, United States Ivo Julião University of Porto, Porto, Portugal; and Portuguese Institute of Oncology Francisco Gentil, Porto, Portugal Kunal C Kadakia Levine Cancer Institute, Carolinas HealthCare System, Charlotte, NC, United States

Andrew D Higham University Hospitals of Morecambe Bay NHS Foundation Trust, Lancaster, United Kingdom

Santosh U Kafle Kathmandu University Birat Medical College and Teaching Hospital, Morang, Nepal; and Biratnagar Eye Hospital, Biratnagar, Nepal

Pancras C W Hogendoorn Leiden University Medical Center, Leiden, the Netherlands

Russel Kahmke Duke University School of Medicine, Durham, NC, United States

Wan Hua Oncode Institute, Leiden University Medical Center, Leiden, Netherlands

Purvi M Kakadiya University of Auckland, Auckland, New Zealand

Tao Huang Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China Kent Hunter National Institutes of Health, Bethesda, MD, United States Joseph Inigo Pharmacology and Therapeutics, Roswell Park Comprehensive Cancer Center, Buffalo, NY, United States

Sudhakar Kalakonda University of Maryland School of Medicine, and Greenebaum Cancer Center, Baltimore, MD, United States Emarene Kalaw The University of Queensland, Brisbane, QLD, Australia Dhananjaya V Kalvakolanu University of Maryland School of Medicine, and Greenebaum Cancer Center, Baltimore, MD, United States

Thergiory Irrazábal University of Toronto, Toronto, ON, Canada

Heidrun Karlic Ludwig Boltzmann Cluster Oncology, Hanusch Hospital, Vienna, Austria

Corbin D Jacobs Duke University School of Medicine, Durham, NC, United States

Stefan Kempa Berlin Institute for Medical Systems Biology at the MDC Berlin-Buch, Berlin, Germany

xviii

Contributors to All Volumes

Saadi Khochbin University of Grenoble Alpes, Grenoble, France Jakub Khzouz King Hussein Cancer Center, Amman, Jordan Edward S Kim Levine Cancer Institute, Carolinas HealthCare System, Charlotte, NC, United States Peter S Klein University of Pennsylvania, Philadelphia, PA, United States Richard D Kolodner Ludwig Institute for Cancer Research, San Diego Branch, La Jolla, CA, United States Annette M Krais Lund University, Lund, Sweden Paul Krimpenfort Oncode Institute, Division of Molecular Genetics, The Netherlands Cancer Institute, Amsterdam, The Netherlands Ivana Kulhánová Section of Cancer Surveillance, International Agency for Research on Cancer, Lyon, France Rahul Kumar Pharmacology and Therapeutics, Roswell Park Comprehensive Cancer Center, Buffalo, NY, United States Razelle Kurzrock UC San Diego Moores Cancer Center, San Diego, CA, United States Sunil R Lakhani The University of Queensland, Brisbane, QLD, Australia; and Pathology Queensland, Herston, QLD, Australia Sylvie Lantuejoul Léon Bérard Centre, Lyon, France; and Grenoble University, La Tronche, France Stefano La Rosa Institute of Pathology, Lausanne University Hospital, Lausanne, Switzerland Heinz Läubli University Hospital Basel, Basel, Switzerland Matteo Lazzeroni European Institute of Oncology, Milan, Italy Christopher J Lengner University of Pennsylvania, Philadelphia, PA, United States

Jay A Levy University of California San Francisco, San Francisco, CA, United States Jonathan D Licht University of Florida, Gainesville, FL, United States Ricardo V Lloyd The University of Wisconsin School of Medicine and Public Health, Madison, WI, United States Leendert H J Looijenga Erasmus University Medical Center, Rotterdam, The Netherlands Estibaliz Lopez-Rodrigo Memorial Sloan-Kettering Cancer Center, New York, NY, United States Fernando López Álvarez Asturias Central University Hospital, University of Oviedo, University Institute of Oncology of Asturias (IUOPA), ISPA, CIBERONC, Oviedo, Spain Nuno Lunet University of Porto, Porto, Portugal Eamonn R Maher University of Cambridge, Cambridge, United Kingdom Sayantan Maji University of Florida, Gainesville, FL, United States Sharan Malagobadan University of Malaya, Kuala Lumpur, Malaysia Leila Malek Odette Cancer Centre, Sunnybrook Health Sciences Centre, Toronto, ON, Canada; and University of Toronto, Toronto, ON, Canada David Malkin The Hospital for Sick Children, Toronto, ON, Canada; and University of Toronto, Toronto, ON, Canada Olivier Manches Institute for Advanced Biosciences, Immunobiology and Immunotherapy in Chronic Diseases Team, Inserm U 1209, CNRS UMR 5309, University Grenoble Alpes, Grenoble, France Mario Mandalà Papa Giovanni XXIII Hospital, Bergamo, Italy Salomon Manier University of Lille, INSERM UMR-S 1172, Lille, France; University of Lille, CHU Lille, Lille, France; and Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, United States

Contributors to All Volumes

Alberto Mantovani Humanitas Clinical and Research Center, Milan, Italy; Humanitas University, Milan, Italy; and Queen Mary University of London, London, United Kingdom Alberto Martin University of Toronto, Toronto, ON, Canada Daniela Massi University of Florence, Florence, Italy Alexandre Matet Curie Institute, Paris, France Aleyamma Mathew Regional Cancer Centre, Trivandrum, India Ashley G Matusz-Fisher Levine Cancer Institute, Carolinas HealthCare System, Charlotte, NC, United States Christopher A Maxwell University of British Columbia, Vancouver, BC, Canada Amy E McCart Reed The University of Queensland, Brisbane, QLD, Australia Simon S McDade Queen’s University Belfast, Belfast, United Kingdom Lin Mei University of British Columbia, Vancouver, BC, Canada Carlos F M Menck University of São Paulo, São Paulo, Brazil

Rimsha Munir University of the Punjab, Lahore, Pakistan Matthew Murtha Bellvitge Biomedical Biomedical Research Institute (IDIBELL), Barcelona, Catalonia, Spain Rebecca L Myers University of Pennsylvania, Philadelphia, PA, United States Ferran Nadeu IDIBAPS, Barcelona, Spain Noor Hasima Nagoor University of Malaya, Kuala Lumpur, Malaysia Jean Nakhle IRMB, INSERM, CNRS, University of Montpellier, Montpellier, France Shreeram C Nallar University of Maryland School of Medicine, and Greenebaum Cancer Center, Baltimore, MD, United States Charlotte K Y Ng University Hospital Basel, Basel, Switzerland Mina Nikanjam UC San Diego Moores Cancer Center, San Diego, CA, United States

Jian-Qing Mi Shanghai Jiao Tong University School of Medicine, Shanghai, China

Stephanie Nishi National Cancer Institute IRCCS G. Pascale Foundation, Naples, Italy; and University of Toronto, Toronto, ON, Canada

Orli Michaeli The Hospital for Sick Children, Toronto, ON, Canada; and University of Toronto, Toronto, ON, Canada

J W Oosterhuis Erasmus University Medical Center, Rotterdam, The Netherlands

Concetta Montagnese National Cancer Institute IRCCS G. Pascale Foundation, Naples, Italy

Peter O’Donovan Royal College of Surgeons in Ireland, Dublin, Ireland

Michael J Moravan Duke University School of Medicine, Durham, NC, United States Yvonne Mowery Duke University School of Medicine, Durham, NC, United States

xix

Jordan O’Malley Pharmacology and Therapeutics, Roswell Park Comprehensive Cancer Center, Buffalo, NY, United States Giuseppe Palmieri National Research Council (CNR), Sassari, Italy

Maciej M Mrugala Mayo Clinic, Scottsdale, AZ, United States

Elvira Palumbo National Cancer Institute IRCCS G. Pascale Foundation, Naples, Italy

Karl Munger Tufts University School of Medicine, Boston, MA, United States

Thomas G Papathomas University of Birmingham, Birmingham, United Kingdom

xx

Contributors to All Volumes

Palak R Parekh University of Maryland, Baltimore, MD, United States Sophie Park Grenoble-Alpes University, Grenoble, France; and Institute for Advanced Biosciences, Grenoble, France Yikyung Park Washington University School of Medicine, St. Louis, MO, United States Roberta Pastorino Institute of Public Health, Catholic University of the Sacred Heart, Rome, Italy Alpa V Patel American Cancer Society, Atlanta, GA, United States Päivi Peltomäki University of Helsinki, Helsinki, Finland Francesco Pezzella University of Oxford, Oxford, United Kingdom Gerd P Pfeifer Center for Epigenetics, Van Andel Research Institute, Grand Rapids, MI, United States Thierry Philip Curie Institute, Paris, France Stefano A Pileri European Institute of Oncology, Milan, Italy; and Bologna University School of Medicine, Bologna, Italy Scott W Piraino Royal College of Surgeons in Ireland, Dublin, Ireland Salvatore Piscuoglio University Hospital Basel, Basel, Switzerland Fiona J Pixley The University of Western Australia, Crawley, WA, Australia Martyn Plummer International Agency for Research on Cancer, Lyon, France Katrina Podsypanina Curie Institute, PSL Research University, CNRS, Equipe Labellisée Ligue contre le Cancer, Paris, France; and Sorbonne University, UPMC Univ Paris 06, CNRS, Paris, France

Giuseppe Porciello National Cancer Institute IRCCS G. Pascale Foundation, Naples, Italy Gopinath Prakasam National Centre of Applied Human Genetics, School of Life Sciences, Jawaharlal Nehru University, New Delhi, India Jaime Prat Autonomous University of Barcelona, Barcelona, Spain Christopher D Putnam Ludwig Institute for Cancer Research, San Diego Branch, La Jolla, CA, United States Shuo Qie Medical University of South Carolina, Charleston, SC, United States Kavitha J Ramchandran Stanford University, Stanford, CA, United States Elizabeth G Ratcliffe University Hospitals of Morecambe Bay NHS Foundation Trust, Lancaster, United Kingdom Tibor A Rauch Rush University Medical Center, Chicago, IL, United States Itedale N Redwan University of Gothenburg, Gothenburg, Sweden Erika Rees-Punia University of Georgia, Athens, GA, United States Andrew Reid Pathology Queensland, Herston, QLD, Australia Giuliana Restante University of Pisa, Pisa, Italy J Richard Wagner University of Sherbrooke, Sherbrooke, QC, Canada Juan Pablo Rodrigo Asturias Central University Hospital, University of Oviedo, University Institute of Oncology of Asturias (IUOPA), ISPA, CIBERONC, Oviedo, Spain Isabelle Romieu National Institute of Public Health, Cuernavaca, Mexico; and Emory University, Atlanta, GA, United States

Igor P Pogribny National Center for Toxicological Research, Jefferson, AR, United States

Christina Ross National Institutes of Health, Bethesda, MD, United States

Sergey D Popov University Hospital of Wales, Cardiff, United Kingdom

Paolo Giorgi Rossi AUSL Reggio Emilia, IRCCS, Reggio Emilia, Italy

Contributors to All Volumes

Sophie Rousseaux University of Grenoble Alpes, Grenoble, France Nadine Royla Berlin Institute for Medical Systems Biology at the MDC Berlin-Buch, Berlin, Germany Joseph K Salama Duke University School of Medicine, Durham, NC, United States Lorenzo Salvati University of Florence, Florence, Italy Nianli Sang Drexel University, Philadelphia, PA, United States; and Sidney Kimmel Cancer Center of Philadelphia, Philadelphia, PA, United States Alain Sarasin Paris-Sud University, Villejuif, France Eva S Schernhammer Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, United States; and Medical University of Vienna, Wien, Austria Gudrun Schleiermacher Curie Institute, Paris, France Neil J Sebire Great Ormond Hospital for Sick Children, London, United Kingdom Veena Shankaran Hutchinson Institute of Cancer Outcomes Research (HICOR), Seattle, WA, United States; Fred Hutchinson Cancer Research Center, Seattle, WA, United States; and University of Washington, Seattle, WA, United States Niva Shapira Ashkelon Academic College, Ashkelon, Israel Akanksha Sharma Mayo Clinic, Scottsdale, AZ, United States Ossie Sharon Ashkelon Academic College, Ashkelon, Israel Ali Shilatifard Northwestern University, Chicago, IL, United States Masahiro Shuda University of Pittsburgh Medical Center (UPMC) Hillman Cancer Center, Pittsburgh, PA, United States; and University of Pittsburgh, Pittsburgh, PA, United States

xxi

C Simon Herrington University of Edinburgh, Edinburgh Cancer Research Centre, Institute of Genetics and Molecular Medicine, Western General Hospital, Edinburgh, United Kingdom Stina Simonsson University of Gothenburg, Gothenburg, Sweden Rajinder Singh University of Leicester, Leicester, United Kingdom Pieter J Slootweg Radboud University Medical Center, Nijmegen, The Netherlands Egbert F Smit Netherlands Cancer Institute, Amsterdam, The Netherlands Joshua W Smith Johns Hopkins University, Baltimore, MD, United States Eric Solary Gustave Roussy, Villejuif, France; and Paris-Sud University, Le Kremlin-Bicêtre, France Mingyang Song Harvard Medical School, Boston, MA, United States; Massachusetts General Hospital, Boston, MA, United States; and Harvard T.H. Chan School of Public Health, Boston, MA, United States Claus Storgaard Sørensen University of Copenhagen, Copenhagen, Denmark Xiao-Jian Sun State Key Laboratory of Medical Genomics, Shanghai Institute of Hematology, Rui Jin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China Katarzyna Szyma nska Science to the Point, Archamps Technopole, Archamps, France Valentina Tabanelli European Institute of Oncology, Milan, Italy Phillippa C Taberlay University of Tasmania, Hobart, TAS, Australia Hideyuki Takeshima National Cancer Center Research Institute, Tokyo, Japan E-Jean Tan Uppsala University, Uppsala, Sweden

xxii

Contributors to All Volumes

Peter ten Dijke Oncode Institute, Leiden University Medical Center, Leiden, Netherlands Luigi M Terracciano University Hospital Basel, Basel, Switzerland Valentina Thomas Royal College of Surgeons in Ireland, Dublin, Ireland Mangesh A Thorat Centre for Cancer Prevention, Wolfson Institute of Preventive Medicine, Queen Mary University of London, London, United Kingdom; and Breast Services, Division of Surgery and Interventional Science, Whittington Hospital, London, United Kingdom Andrew Thorburn University of Colorado School of Medicine, Aurora, CO, United States Erik Thunnissen VU University Medical Center, Amsterdam, The Netherlands Falk Tillner Technical University of Dresden, Dresden, Germany Arthur S Tischler Tufts University School of Medicine, Boston, MA, United States Alessia Tognetto Institute of Public Health, Catholic University of the Sacred Heart, Rome, Italy Kazunori Tomita University College London, London, United Kingdom Christina G Towers University of Colorado School of Medicine, Aurora, CO, United States Lawrence D True University of Washington, Seattle, WA, United States Silvia Uccella University of Insubria, Varese, Italy Toshikazu Ushijima National Cancer Center Research Institute, Tokyo, Japan

Funda Vakar-Lopez University of Washington, Seattle, WA, United States Franz Varga Hanusch Hospital of WGKK and AUVA Trauma Centre Meidling, Vienna, Austria Matthew G Varga University of North Carolina, Chapel Hill, NC, United States Michael C Velarde University of the Philippines Diliman, Quezon City, Philippines Marie-Luce Vignais IRMB, INSERM, CNRS, University of Montpellier, Montpellier, France Neus Villamor University of Barcelona, Barcelona, Spain Rocío C Viñuelas University of Gothenburg, Gothenburg, Sweden Sara Vitale National Cancer Institute IRCCS G. Pascale Foundation, Naples, Italy Jamie Von Roenn American Society of Clinical Oncology, Alexandria, VA, United States Gordan M Vujanic Sidra Medicine, Doha, Qatar Jean Y J Wang University of California San Diego, La Jolla, CA, United States Jin Wang Shanghai Jiao Tong University School of Medicine, Shanghai, China Sarah Watson Institute Curie Hospital, Paris, France; and Inserm U830, SIREDO Oncology Center, Paris Sciences and Letters (PSL) Research University, Paris, France

Salvatore Vaccarella Section of Infections, International Agency for Research on Cancer, Lyon, France

Elisabete Weiderpass Cancer Registry of Norway, Oslo, Norway; Karolinska Institute, Stockholm, Sweden; UiT The Arctic University of Norway, Tromsø, Norway; and Folkhälsan Research Center, Helsinki, Finland

William Vainchenker Gustave Roussy, Villejuif, France

Sara Weirich Stuttgart University, Stuttgart, Germany

Contributors to All Volumes

Pieter Wesseling VU University Medical Center/Brain Tumor Center Amsterdam, Amsterdam, The Netherlands; Princess Máxima Center for Pediatric Oncology, Utrecht, The Netherlands; and University Medical Center Utrecht, Utrecht, The Netherlands Katharina Wiedemeyer University of Heidelberg, Heidelberg, Germany Sean R Williamson Henry Ford Cancer Institute, Henry Ford Health System, Detroit, MI, United States; and Wayne State University School of Medicine, Detroit, MI, United States David M Wilson, III National Institute on Aging, Intramural Research Program, National Institutes of Health, Baltimore, MD, United States Frank Winkler University Hospital Heidelberg, Heidelberg, Germany; and German Cancer Consortium (DKTK), German Cancer Research Center (DKFZ), Heidelberg, Germany Matthias Wirth Heinrich Heine University, University Hospital, Duesseldorf, Germany Tejas Yadav Curie Institute, PSL Research University, CNRS, Equipe Labellisée Ligue contre le Cancer, Paris, France; and Sorbonne University, UPMC Univ Paris 06, CNRS, Paris, France

xxiii

Marie Yammine University of Lille, INSERM UMR-S 1172, Lille, France Maryam Yousefi University of Pennsylvania, Philadelphia, PA, United States Nousheen Zaidi University of the Punjab, Lahore, Pakistan Jing Zhang Oncode Institute, Leiden University Medical Center, Leiden, Netherlands Yin Zhang Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, United States Maria Zhivagui International Agency for Research on Cancer (WHO), Lyon, France Nina Zidar Faculty of Medicine, University of Ljubljana, Ljubljana, Slovenia

HOW TO USE THE ENCYCLOPEDIA Structure of the Encyclopedia All articles in the encyclopedia are arranged alphabetically as a series of entries. There are four features to help you easily find the topic you are interested in: an alphabetical contents list, cross references, a full subject index, and contributors. 1. Alphabetical contents list: The alphabetical contents list, which appears at the front of each volume, lists the entries in the order that they appear in the encyclopedia. So that they can be easily located, entry titles generally begin with the key word or phrase indicating the topic, with any generic terms following. For example, “Multiple Myeloma: Pathology and Genetics” is the entry title rather than “Pathology and Genetics of Multiple Myeloma”. 2. Cross references: Virtually all the entries in the encyclopedia have been extensively cross-referenced. The cross references which appear at the end of an entry, serve three different functions: i. To draw the reader’s attention to related material on other entries ii. To indicate material that broadens and extends the scope of the article iii. To indicate material that covers a topic in more depth Example The following list of cross-references appears at the end of the entry “CarcinogendDNA Adducts”. See also: Cancer Risk Reduction Through Lifestyle Changes. Cell Responses to DNA Damage. Genetic Instability. Hereditary Risk of Breast and Ovarian Cancer: BRCA1 and BRCA2. Molecular Epidemiology and Cancer Risk. Role of DNA Repair in Carcinogenesis and Cancer Therapeutics.

3. Index: The index appears at the end of volume 3 and includes page numbers for quick reference to the information you are looking for. The index entries differentiate between references to a whole entry, a part of an entry, and a table or figure. 4. Contributors: At the start of each volume there is a list of the authors who contributed to all volumes.

xxv

SUBJECT CLASSIFICATION Causes of Cancer Aflatoxins Aging and Cancer Cancers as Ecosystems: From Cells to Population Diabetes and Cancer Dietary Factors and Cancer Helicobacter Pylori-Mediated Carcinogenesis HIV (Human Immunodeficiency Virus) Obesity and Cancer: Epidemiological Evidence Opisthorchis Viverrini, Clonorchis Sinensis, and Cholangiocarcinoma Papillomaviruses Physical Inactivity and Cancer Radiation Therapy-Induced Metastasis and Secondary Malignancy Sleep Disturbances and Misalignment in Cancer

Diagnosis and Therapy Acute Lymphocytic Leukemia: Diagnosis and Treatment Acute Myelogeneous Leukemia: Diagnosis and Treatment Bladder Cancer: Pathology, Genetics, Diagnosis, and Treatment Bone and Soft Tissue Sarcoma: From Molecular Features to Clinical Applications Cancer Vaccines: Dendritic Cell-Based Vaccines and Related Approaches Cholangiocarcinoma: Diagnosis and Treatment Chromatin Dynamics and Cancer: Epigenetic Parameters and Cellular Fate Chronic Myelogenous Leukemia: Pathology, Genetics, Diagnosis, and Treatment Colorectal Cancer: Diagnosis and Treatment End of Life Support Esophageal Cancer: Diagnosis and Treatment Glioblastoma: Biology, Diagnosis, and Treatment Interferons: Cellular and Molecular Biology of Their Actions Kidney Cancer: Diagnosis and Treatment Laryngeal Cancer: Diagnosis and Treatment Malignant Tumors of the Eye, Conjunctiva, and Orbit: Diagnosis and Therapy Myelodysplastic Syndromes: Mechanisms, Diagnosis, and Treatment Nasopharyngeal Carcinoma: Diagnosis and Treatment Neuroblastoma: Diagnosis and Treatment New Rationales and Designs for Clinical Trials in the Era of Precision Medicine Non-Hodgkin Lymphoma: Diagnosis and Treatment Oncology Imaging Oral Cavity Cancer: Diagnosis and Treatment Ovarian Cancer: Diagnosis and Treatment Pancreatic Cancer: Diagnosis and Treatment P-Glycoprotein-Mediated Multidrug Resistance Pituitary Tumors: Diagnosis and Treatment

xxvii

xxviii

Subject Classification

Prostate Cancer: Diagnosis and Treatment Radiation Oncology Squamous Cell and Basal Cell Carcinoma of the Skin: Diagnosis and Treatment Symptom Control Thyroid Cancer: Pathology, Genetics, Diagnosis, and Treatment Uterine Cervix Cancer: Diagnosis and Treatment

Hallmarks of Cancer Anoikis Autophagy and Cancer Cancer-Related Inflammation in Tumor Progression Cell Adhesion During Tumorigenesis and Metastasis Cell Responses to DNA Damage DNA Mismatch Repair: Mechanisms and Cancer Genetics Epithelium to Mesenchyme Transition Genetic Instability Glutamine Metabolism and Cancer Induced Pluripotent Stem Cells and Yamanaka factors Inhibitors of Lactate Transport: A Promising Approach in Cancer Drug Discovery Lipid Metabolism Metastatic SignaturesdThe Tell-Tale Signs of Metastasis Mevalonate Pathway Mitogen-Activated Protein Kinases (MAPK) in Cancer Pyruvate Kinase Role of DNA Repair in Carcinogenesis and Cancer Therapeutics Senescence and Cellular Immortality Telomeres, Telomerase, and Cancer TGF-b in Cancer Progression: From Tumor Suppressor to Tumor Promotor TP53 Tumor-Associated Macrophages Tumors and Blood Vessel Interactions: A Changing Hallmark of Cancer Tunneling Nanotubes (TNTs): Intratumoral Cell-to-Cell Communication

Mechanisms Animal Models of Cancer: What We Can Learn From Mice Ataxia Telangiectasia Syndrome CarcinogendDNA Adducts Carcinogenesis: Role of Reactive Oxygen and Nitrogen Species Chromosome Rearrangements and Translocations Copy Number Variations in Tumors Defective 5-Methylcytosine Oxidation in Tumorigenesis DNA Methylation Changes in Cancer: Cataloguing DNA Methylation Changes in Cancer: Mechanisms Enhancers in Cancer: Genetic and Epigenetic Deregulation Environmental Exposures and Epigenetic Perturbations Epigenetic Therapy Genome Wide Association Studies (GWAS) Hereditary Risk of Breast and Ovarian Cancer: BRCA1 and BRCA2 Hormones and Cancer Li–Fraumeni Syndrome Lynch Syndrome Microbiota and Colon Cancer: Orchestrating Neoplasia Through DNA Damage and Immune Dysregulation Mod Squad: Altered Histone Modifications in Cancer Mutational Signatures and the Etiology of Human Cancers Mutations in Chromatin Remodeling Factors

Subject Classification Mutations in DNA Methyltransferases and Demethylases Mutations in Histone Lysine Methyltransferases and Demethylases Mutations: Driver Versus Passenger Polyomaviruses in Human Cancer Systems Biology Approach to Study Cancer Metabolism TCA Cycle Aberrations and Cancer Unprogrammed Gene Activation: A Critical Evaluation of Cancer Testis Genes Wnt Signaling in Intestinal Stem Cells and Cancer Xeroderma Pigmentosum: When the Sun Is the Enemy

Pathology and Genetics of Specific Cancers Adrenal Glands Tumors: Pathology and Genetics Anal Cancer: Pathology and Genetics Bladder Cancer: Pathology, Genetics, Diagnosis, and Treatment Bones and Joints Cancer: Pathology and Genetics Breast Cancer: Pathology and Genetics Chronic Lymphocytic Leukemia: Pathology and Genetics Chronic Myelogenous Leukemia: Pathology, Genetics, Diagnosis, and Treatment Colorectal Cancer: Pathology and Genetics Endometrial Cancer: Pathology and Genetics Esophageal Cancer: Pathology and Genetics Eye and Orbit Cancer: Pathology and Genetics Gastric Cancer: Pathology and Genetics Germ Cell Tumors: Pathology and Genetics Glioblastoma: Pathology and Genetics Hepatocellular Carcinoma: Pathology and Genetics Hodgkin Lymphoma: Pathology and Genetics Integrative Molecular Tumor Classification: A Pathologist’s View Jaws Cancer: Pathology and Genetics Larynx Cancer: Pathology and Genetics Malignant Skin Adnexal Tumors: Pathology and Genetics Melanoma: Pathology and Genetics Multiple Myeloma: Pathology and Genetics Neuroblastoma: Pathology and Genetics Non-Hodgkin Lymphoma: Pathology and Genetics Nonsmall-Cell Cancers of the Lung: Pathology and Genetics Oral and Oropharyngeal Cancer: Pathology and Genetics Ovarian Cancer: Pathology and Genetics Pancreatic Cancer: Pathology and Genetics Parathyroid Cancer: Pathology and Genetics Pituitary Tumors: Pathology and Genetics Prostate Cancer: Pathology and Genetics Renal Cell Cancer: Pathology and Genetics Small-Cell Cancer of the Lung: Pathology and Genetics Thyroid Cancer: Pathology, Genetics, Diagnosis, and Treatment Uterine Cervix Cancer: Pathology and Genetics Wilms Tumor: Pathology and Genetics

Prevention and Control Aspirin and Cancer Cancer Disparities Cancer in Populations in Transition Cancer in Sub-Saharan Africa Cancer in the Middle East Cancer Risk Reduction Through Lifestyle Changes

xxix

xxx

Subject Classification

Cancer Survival and Survivorship Cervical Cancer: Screening, Vaccination, and Preventive Strategies Chemoprevention of Cancer: An Overview of Promising Agents and Current Research Chemoprevention Trials Diet and Cancer Environmental and Occupational Exposures Financial Burden of Cancer Care Hereditary Cancer Syndromes: Identification and Management Metformin Molecular Epidemiology and Cancer Risk Prevention and Control: Nutrition, Obesity, and Metabolism

PREFACE Cancer holds a special status in biology, medicine, and society. It offers formidable challenges to basic, clinical, and population science; it ranks at the top of medical research priorities and healthcare costs in most countries. The general public is continuously exposed to news of discoveries promising to defeat the disease in the near future. Indeed, ground-breaking advances are being made, from preventive vaccines to genome-guided personalized medicine, sophisticated imaging and surgical technologies, and systemic therapies aimed at awakening natural immune responses against cancer. These novel therapeutic approaches make cure a real possibility for a growing number of patients. They also launch cancer care into a new era of maintaining the disease under control for an indefinite period of time, turning it into a form of chronic disease. At the same time, evidence-based prevention and early detection strategies have opened a new window on the natural history of the disease, enabling effective intervention well ahead of diagnosis. As a result, the first two decades of this millennium have witnessed a marked decrease in the mortality and, in some instances, the incidence of cancers that have dominated the death toll in more developed countries during the second half of the 20th century. A turning point in our understanding of cancer is the deciphering of the human genome and its spin-off endeavors aimed at exploring the genomic landscape and architecture of human cancers. These discoveries are causing a major overhaul of our vision of cancer as a dynamic, rapidly evolving, and heterogeneous disease at the individual level. Harnessing this complexity requires mastering increasingly complex sources of data at molecular, cellular, systemic, personal, environmental, and societal level, heralding the emergence of big-data science in cancer diagnosis and treatment. However, this exceptional acceleration in knowledge and solutions cannot hide the fact that cancer remains a global scourge that exerts a massive burden on humankind and societies worldwide, in particular in societies in transition and in low-resource contexts. Today, cancer crystallizes many of the major societal challenges pertaining to lifestyles, sustainable development and environmental policies, demography and population aging, access to education and healthcare, sharing of resources and knowledge, and protection of persons and personal information. The information on cancer available at a fingertip is overwhelming in volume, complexity, veracity, and velocity. We worked on the Third Edition of Elsevier’s Encyclopedia of Cancer with this rapidly changing background. Rather than aiming at developing a comprehensive framework encompassing all aspects, we attempted to address the literal meaning of the greek terms ἐgkύklιo2 paιdείa, which means “general education”. While we retained some articles from the previous edition, which, at the time of the publication, represented an exceptional achievement of Dr. J. Bertino, we largely modified the structure and the list of chapters, and the possibility of continuous update of the articles has been a great incentive for us and for the authors of the chapters. This new edition of the Encyclopedia consists of six major parts: (i) mechanisms of cancer, (ii) hallmarks of cancer, (iii) causes of cancer, (iv) cancer prevention and control, (v) diagnosis and treatment of specific cancers, and (vi) pathology and genetics of specific cancers. This repartition is necessarily artificial and is complemented by the extensive cross-references between articles. The repartition, however, reflects our effort to identify discrete topics that would best address the needs of a wide community of readers. The primary target readership of the Encyclopedia comprises medical and other health science students, as well as non-specialized physicians and other health practitioners. Cancer researchers, oncologists, and other cancer professionals may find the articles pertaining to their specific field to be too short, over-simplistic, and perhaps obsolete; they too, however, may benefit from articles on topics other than their own. The Encyclopedia also offers an easy way to navigate across concepts and topics that should be appealing to readers from other communities, including social sciences or stakeholders in public decision-making.

xxxi

xxxii

Preface

We were fortunate to work with a formidable team of section editors, including Fred Bosman, Graham Colditz, Carlo La Vecchia, Gerd Pfeifer, and Marco Pierotti. An additional section editor was Professor Thomas Tursz, who passed away prematurely during the preparation of the Encyclopedia. Thomas was a great colleague and mentor, and a major figure in oncology in France and internationally. We had the privilege to involve him in the last project of his long career, and we want to dedicate this work to him. We wish to thank the many article authors, who agreed to contribute to the success of this international endeavor, and in particular Dr. Katarzyna Szyma nska, who drafted several cancer-specific articles. Finally, we want to thank the staff at Elsevier, whose patience and perseverance helped us bringing the project to the final stage. All these individuals are responsible for the many strengths of the new edition of the Encyclopedia of Cancer, while weaknesses are mainly ours. Paolo Boffetta Pierre Hainaut

CONTENTS OF ALL VOLUMES

VOLUME 1 Acute Lymphocytic Leukemia: Diagnosis and Treatment Katarzyna Szyma nska and Sophie Park

1

Acute Myelogeneous Leukemia: Diagnosis and Treatment Katarzyna Szyma nska and Sophie Park

9

Adrenal Glands Tumors: Pathology and Genetics Thomas G Papathomas, Thomas J Giordano, Eamonn R Maher, and Arthur S Tischler

18

Aflatoxins Joshua W Smith and John D Groopman

30

Aging and Cancer Mingyang Song

44

Anal Cancer: Pathology and Genetics Fred T Bosman

53

Animal Models of Cancer: What We Can Learn From Mice Giustina Ferone, Paul Krimpenfort, and Anton Berns

60

Anoikis Sharan Malagobadan and Noor Hasima Nagoor

75

Aspirin and Cancer Mangesh A Thorat

85

Ataxia Telangiectasia Syndrome Palak R Parekh and France Carrier

101

Autophagy and Cancer Christina G Towers and Andrew Thorburn

112

Bladder Cancer: Pathology, Genetics, Diagnosis, and Treatment Katarzyna Szyma nska, Fred T Bosman, and Pierre Hainaut

122

Bone and Soft Tissue Sarcoma: From Molecular Features to Clinical Applications Sarah Watson and Olivier Delattre

134

xxxiii

xxxiv

Contents of All Volumes

Bones and Joints Cancer: Pathology and Genetics Pancras C W Hogendoorn and Carlos E de Andrea

153

Breast Cancer: Pathology and Genetics Amy E McCart Reed, Emarene Kalaw, Andrew Reid, and Sunil R Lakhani

170

Cancer Disparities Ivana Kulhánová and Salvatore Vaccarella

184

Cancer in Populations in Transition Aleyamma Mathew and Preethi S George

191

Cancer in Sub-Saharan Africa Mariana Brandão, Ivo Julião, Carla Carrilho, Filipa Fontes, and Nuno Lunet

212

Cancer in the Middle East Iman El Sayed

225

Cancer Risk Reduction Through Lifestyle Changes Christina Bamia

243

Cancer Survival and Survivorship Dana Hashim and Elisabete Weiderpass

250

Cancer Vaccines: Dendritic Cell-Based Vaccines and Related Approaches Laurence Chaperot, Olivier Manches, and Caroline Aspord

260

Cancer-Related Inflammation in Tumor Progression Alberto Mantovani

275

Cancers as Ecosystems: From Cells to Population Graham A Colditz

278

CarcinogendDNA Adducts Annette M Krais, Rajinder Singh, and Volker M Arlt

282

Carcinogenesis: Role of Reactive Oxygen and Nitrogen Species Jean Cadet and J Richard Wagner

296

Cell Adhesion During Tumorigenesis and Metastasis Lubor Borsig and Heinz Läubli

307

Cell Responses to DNA Damage Jean Y J Wang

315

Cervical Cancer: Screening, Vaccination, and Preventive Strategies Paolo Giorgi Rossi and Francesca Carozzi

326

Chemoprevention of Cancer: An Overview of Promising Agents and Current Research Elizabeth G Ratcliffe, Andrew D Higham, and Janusz A Jankowski

345

Chemoprevention Trials Kunal C Kadakia, Ashley G Matusz-Fisher, and Edward S Kim

352

Cholangiocarcinoma: Diagnosis and Treatment Katarzyna Szyma nska

367

Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate Daniel Jeffery, Katrina Podsypanina, Tejas Yadav, and Geneviève Almouzni

372

Contents of All Volumes

xxxv

Chromosome Rearrangements and Translocations Stefan K Bohlander, Purvi M Kakadiya, and Alix Coysh

389

Chronic Lymphocytic Leukemia: Pathology and Genetics Julio Delgado, Neus Villamor, Ferran Nadeu, and Elías Campo

405

Chronic Myelogenous Leukemia: Pathology, Genetics, Diagnosis, and Treatment Katarzyna Szyma nska and Sophie Park

418

Colorectal Cancer: Diagnosis and Treatment Katarzyna Szyma nska

425

Colorectal Cancer: Pathology and Genetics Fred T Bosman

428

Copy Number Variations in Tumors Tao Huang

444

Defective 5-Methylcytosine Oxidation in Tumorigenesis Gerd P Pfeifer and Seung-Gi Jin

452

Diabetes and Cancer Rachel Dankner

459

Diet and Cancer Livia S A Augustin, Concetta Montagnese, Ilaria Calabrese, Giuseppe Porciello, Elvira Palumbo, Sara Vitale, and Stephanie Nishi

471

Dietary Factors and Cancer Isabelle Romieu

501

DNA Methylation Changes in Cancer: Cataloguing Tibor A Rauch and Gerd P Pfeifer

512

DNA Methylation Changes in Cancer: Mechanisms Hideyuki Takeshima and Toshikazu Ushijima

520

DNA Mismatch Repair: Mechanisms and Cancer Genetics William J Graham V, Christopher D Putnam, and Richard D Kolodner

530

End of Life Support Kavitha J Ramchandran and Jamie Von Roenn

539

Endometrial Cancer: Pathology and Genetics Ben Davidson

549

Enhancers in Cancer: Genetic and Epigenetic Deregulation Kaixiang Cao and Ali Shilatifard

559

Environmental and Occupational Exposures Harvey Checkoway

569

Environmental Exposures and Epigenetic Perturbations Igor P Pogribny

574

VOLUME 2 Epigenetic Therapy Richard L Bennett, Aditya Bele, Sayantan Maji, and Jonathan D Licht

1

xxxvi

Contents of All Volumes

Epithelium to Mesenchyme Transition Laia Caja and E-Jean Tan

14

Esophageal Cancer: Diagnosis and Treatment Katarzyna Szyma nska

24

Esophageal Cancer: Pathology and Genetics Jean-François Fléjou

32

Eye and Orbit Cancer: Pathology and Genetics Sarah E Coupland, Santosh U Kafle, and Jakub Khzouz

47

Financial Burden of Cancer Care Bernardo H L Goulart and Veena Shankaran

61

Gastric Cancer: Pathology and Genetics Irene Gullo and Fátima Carneiro

77

Genetic Instability Meiyun Guo, Lin Mei, and Christopher A Maxwell

99

Genome Wide Association Studies (GWAS) Giuliana Restante, Elena Arrigoni, Marzia Del Re, Antonello Di Paolo, and Romano Danesi

110

Germ Cell Tumors: Pathology and Genetics J W Oosterhuis and Leendert H J Looijenga

121

Glioblastoma: Biology, Diagnosis, and Treatment Akanksha Sharma, Maciej M Mrugala, and Jan Buckner

154

Glioblastoma: Pathology and Genetics Christian Hartmann and Pieter Wesseling

165

Glutamine Metabolism and Cancer Nianli Sang, Dan He, and Shuo Qie

179

Helicobacter Pylori-Mediated Carcinogenesis Matthew G Varga and Meira Epplein

187

Hepatocellular Carcinoma: Pathology and Genetics Luigi M Terracciano, Salvatore Piscuoglio, and Charlotte K Y Ng

198

Hereditary Cancer Syndromes: Identification and Management Roberta Pastorino, Alessia Tognetto, and Stefania Boccia

211

Hereditary Risk of Breast and Ovarian Cancer: BRCA1 and BRCA2 Claus Storgaard Sørensen

214

HIV (Human Immunodeficiency Virus) Jay A Levy and JoAnn C Castelli

218

Hodgkin Lymphoma: Pathology and Genetics Stefano A Pileri, Stefano Fiori, Vincenzo De Iuliis, and Valentina Tabanelli

227

Hormones and Cancer Kristen D Brantley, Susan E Hankinson, and A Heather Eliassen

237

Induced Pluripotent Stem Cells and Yamanaka factors Stina Simonsson, Rocío C Viñuelas, Leonie Hartl, Itedale N Redwan, and Joydeep Bhadury

254

Contents of All Volumes

xxxvii

Inhibitors of Lactate Transport: A Promising Approach in Cancer Drug Discovery Thomas D Bannister

266

Integrative Molecular Tumor Classification: A Pathologist’s View Fred T Bosman

279

Interferons: Cellular and Molecular Biology of Their Actions Dhananjaya V Kalvakolanu, Shreeram C Nallar, and Sudhakar Kalakonda

286

Jaws Cancer: Pathology and Genetics Pieter J Slootweg and Daniel Baumhoer

313

Kidney Cancer: Diagnosis and Treatment Katarzyna Szyma nska

325

Laryngeal Cancer: Diagnosis and Treatment Fernando López Álvarez and Juan Pablo Rodrigo

332

Larynx Cancer: Pathology and Genetics Nina Zidar and Nina Gale

346

Li–Fraumeni Syndrome Orli Michaeli and David Malkin

356

Lipid Metabolism Fatima Ameer, Rimsha Munir, and Nousheen Zaidi

369

Lynch Syndrome Päivi Peltomäki

374

Malignant Skin Adnexal Tumors: Pathology and Genetics Katharina Wiedemeyer and Thomas Brenn

389

Malignant Tumors of the Eye, Conjunctiva, and Orbit: Diagnosis and Therapy Laurence Desjardins, Nathalie Cassoux, and Alexandre Matet

402

Melanoma: Pathology and Genetics Lorenzo Salvati, Giuseppe Palmieri, Vincenzo De Giorgi, Antonio Cossu, Mario Mandalà, and Daniela Massi

414

Metastatic SignaturesdThe Tell-Tale Signs of Metastasis Christina Ross and Kent Hunter

426

Metformin Matteo Lazzeroni and Sara Gandini

435

Mevalonate Pathway Heidrun Karlic and Franz Varga

445

Microbiota and Colon Cancer: Orchestrating Neoplasia Through DNA Damage and Immune Dysregulation Alberto Martin and Thergiory Irrazábal

458

Mitogen-Activated Protein Kinases (MAPK) in Cancer Ana Cuenda

472

Mod Squad: Altered Histone Modifications in Cancer Matthew Murtha and Manel Esteller

481

Molecular Epidemiology and Cancer Risk Paulina Gomez-Rubio and Evangelina López de Maturana

487

xxxviii

Contents of All Volumes

Multiple Myeloma: Pathology and Genetics Marie Yammine and Salomon Manier

494

Mutational Signatures and the Etiology of Human Cancers Ludmil B Alexandrov and Maria Zhivagui

499

Mutations in Chromatin Remodeling Factors Katherine A Giles and Phillippa C Taberlay

511

Mutations in DNA Methyltransferases and Demethylases Xiao-Jian Sun, Zhu Chen, and Sai-Juan Chen

528

Mutations in Histone Lysine Methyltransferases and Demethylases Sara Weirich and Albert Jeltsch

538

Mutations: Driver Versus Passenger Scott W Piraino, Valentina Thomas, Peter O’Donovan, and Simon J Furney

551

Myelodysplastic Syndromes: Mechanisms, Diagnosis, and Treatment Eric Solary and William Vainchenker

563

Nasopharyngeal Carcinoma: Diagnosis and Treatment Katarzyna Szyma nska

571

VOLUME 3 Neuroblastoma: Diagnosis and Treatment Gudrun Schleiermacher and Thierry Philip

1

Neuroblastoma: Pathology and Genetics Nai-Kong V Cheung and Estibaliz Lopez-Rodrigo

15

New Rationales and Designs for Clinical Trials in the Era of Precision Medicine Mina Nikanjam and Razelle Kurzrock

30

Non-Hodgkin Lymphoma: Diagnosis and Treatment Katarzyna Szyma nska and Sophie Park

44

Non-Hodgkin Lymphomas: Pathology and Genetics Laurence de Leval

48

Nonsmall-Cell Cancers of the Lung: Pathology and Genetics Erik Thunnissen and Egbert F Smit

76

Obesity and Cancer: Epidemiological Evidence Yikyung Park

88

Oncology Imaging Stefan Delorme and Michael Baumann

98

Opisthorchis viverrini, Clonorchis sinensis, and Cholangiocarcinoma Catherine de Martel and Martyn Plummer

119

Oral and Oropharyngeal Cancer: Pathology and Genetics Pieter J Slootweg and Justin A Bishop

124

Oral Cavity Cancer: Diagnosis and Treatment Corbin D Jacobs, Michael J Moravan, Jennifer Choe, Russel Kahmke, Yvonne Mowery, and Joseph K Salama

131

Contents of All Volumes

xxxix

Ovarian Cancer: Diagnosis and Treatment Katarzyna Szyma nska

169

Ovarian Cancer: Pathology and Genetics Jaime Prat, Iñigo Espinosa, and Emanuela D’Angelo

173

Pancreatic Cancer: Diagnosis and Treatment Katarzyna Szyma nska

193

Pancreatic Cancer: Pathology and Genetics Irene Esposito, Lena Haeberle, and Matthias Wirth

199

Papillomaviruses Karl Munger

212

Parathyroid Cancer: Pathology and Genetics Ying-Hsia Chu and Ricardo V Lloyd

219

P-Glycoprotein-Mediated Multidrug Resistance M M Gottesman

228

Physical Inactivity and Cancer Alpa V Patel and Erika Rees-Punia

235

Pituitary Tumors: Pathology and Genetics Stefano La Rosa and Silvia Uccella

241

Pituitary Tumors: Diagnosis and Treatment Sylvia L Asa and Shereen Ezzat

257

Polyomaviruses in Human Cancer Masahiro Shuda

266

Prevention and Control: Nutrition, Obesity, and Metabolism Niva Shapira and Ossie Sharon

278

Prostate Cancer: Diagnosis and Treatment Katarzyna Szyma nska and Pierre Hainaut

292

Prostate Cancer: Pathology and Genetics Funda Vakar-Lopez and Lawrence D True

299

Pyruvate Kinase Gopinath Prakasam and Rameshwar N K Bamezai

311

Radiation Oncology Nadja Ebert, Falk Tillner, and Michael Baumann

321

Radiation Therapy-Induced Metastasis and Secondary Malignancy Jordan O’Malley, Joseph Inigo, Abhiram Gokhale, Venkaiah Betapudi, Rahul Kumar, and Dhyan Chandra

337

Renal Cell Cancer: Pathology and Genetics Sean R Williamson

347

Role of DNA Repair in Carcinogenesis and Cancer Therapeutics Rachel Abbotts, Tyler Golato, and David M Wilson, III

363

Senescence and Cellular Immortality Marco Demaria and Michael C Velarde

386

xl

Contents of All Volumes

Sleep Disturbances and Misalignment in Cancer Eva S Schernhammer, Yin Zhang, and Elizabeth E Devore

395

Small-Cell Cancer of the Lung: Pathology and Genetics Lynnette Fernandez-Cuesta, Sylvie Lantuejoul, and Nicolas Girard

403

Squamous Cell and Basal Cell Carcinoma of the Skin: Diagnosis and Treatment Katarzyna Szyma nska

412

Symptom Control Vithusha Ganesh, Leila Malek, Carlo DeAngelis, and Edward Chow

417

Systems Biology Approach to Study Cancer Metabolism Stefan Kempa and Nadine Royla

426

TCA Cycle Aberrations and Cancer Fabio Ciccarone, Luca Di Leo, and Maria Rosa Ciriolo

429

Telomeres, Telomerase, and Cancer Kazunori Tomita and Laura C Collopy

437

TGF-b in Cancer Progression: From Tumor Suppressor to Tumor Promotor Chuannan Fan, Jing Zhang, Wan Hua, and Peter ten Dijke

455

Thyroid Cancer: Pathology, Genetics, Diagnosis, and Treatment Katarzyna Szyma nska and Fred T Bosman

471

TP53 Simon S McDade and Martin Fischer

483

Tumor-Associated Macrophages Fiona J Pixley

496

Tumors and Blood Vessel Interactions: A Changing Hallmark of Cancer Francesco Pezzella and Frank Winkler

504

Tunneling Nanotubes (TNTs): Intratumoral Cell-to-Cell Communication Marie-Luce Vignais, Jean Nakhle, and Emmanuel Griessinger

513

Unprogrammed Gene Activation: A Critical Evaluation of Cancer Testis Genes Sophie Rousseaux, Ekaterina Bourova-Flin, Mengqing Gao, Jin Wang, Jian-Qing Mi, and Saadi Khochbin

523

Uterine Cervix Cancer: Diagnosis and Treatment Katarzyna Szyma nska

531

Uterine Cervix Cancer: Pathology and Genetics C Simon Herrington

535

Wilms Tumor: Pathology and Genetics Gordan M Vujanic, Adrian K Charles, Sergey D Popov, and Neil J Sebire

542

Wnt Signaling in Intestinal Stem Cells and Cancer Rebecca L Myers, Maryam Yousefi, Christopher J Lengner, and Peter S Klein

553

Xeroderma Pigmentosum: When the Sun Is the Enemy Alain Sarasin, Carlos F M Menck, and Januario B Cabral-Neto

562

PERMISSION ACKNOWLEDGMENTS The following material is reproduced with kind permission of American Association for the Advancement of Science Figure 3 Neuroblastoma; Pathology and Genetics www.aaas.org The following material is reproduced with kind permission of Oxford University Press Table 2 Environmental and Occupational Exposures Figure 1 Driver Versus Passenger Mutations in Tumors Figure 3 Driver Versus Passenger Mutations in Tumors Figure 4 Driver Versus Passenger Mutations in Tumors Figure 5 Driver Versus Passenger Mutations in Tumors Figure 1 Financial Burden of Cancer – Therapies www.oup.com The following material is reproduced with kind permission of Nature Publishing Group Figure 5 Neuroblastoma; Pathology and Genetics Table 2 Neuroblastoma; Pathology and Genetics Figure 13 Neuroblastoma; Pathology and Genetics Table 1 Neuroblastoma; Pathology and Genetics Table 2 Neuroblastoma; Pathology and Genetics Table 3 Neuroblastoma; Pathology and Genetics Table 4 Neuroblastoma; Pathology and Genetics Figure 2 Oesophageal Cancer; Diagnosis and Treatment Figure 15 Germ Cell Tumors: Pathology and Genetics Figure 2 Chemoprevention of Cancer: an Overview of Promising Agents and Current Research http://www.nature.com

i

Acute Lymphocytic Leukemia: Diagnosis and Treatment ska, Science to the Point, Archamps Technopole, Archamps, France Katarzyna Szyman Sophie Park, Grenoble-Alpes University, Grenoble, France; and Institute for Advanced Biosciences, Grenoble, France © 2019 Elsevier Inc. All rights reserved.

Abbreviations 6-MP 6-Mercaptopurine ALL Acute lymphocytic (lymphoblastic) leukemia AYA Adolescents and young adults B-ALL B-lymphoblastic leukemia B-LBL B-lymphoblastic lymphoma CAR-T Chimeric antigen receptor T-cells CBC Complete blood count CD Cluster of differentiation CGH Comparative genomic hybridization CNS Central nervous system COG The Children’s Oncology Group CR Complete remission CT Computed tomography DIC Disseminated intravascular coagulation FAB French–American–British (histopathologic classification of acute leukemia) FDA US Food and Drug Administration FISH Fluorescence in situ hybridization HSCT Hematopoietic stem cell transplantation ICD-O International Classification of Diseases for Oncology IGH Immunoglobulin heavy locus LDH Lactate dehydrogenase MRD Minimal residual disease NCCN US National Comprehensive Cancer Network NK Natural killer PET Positron emission tomography Ph Philadelphia chromosome RT-PCR Reverse-transcriptase polymerase chain reaction SNP Single nucleotide polymorphism T-ALL T-lymphoblastic leukemia TCR T-cell receptor TdT Terminal deoxynucleotidyl transferase TK Tyrosine kinase TKI Tyrosine kinase inhibitor TMPT Thiopurine methyltransferase WBC White blood cells WHO World Health Organization

Definition and Classification Acute lymphocytic leukemia (also called acute lymphoblastic leukemia; ALL) is a neoplasm of immature B- or T-cells (lymphoblasts). In the most recent World Health Organization (WHO) classification of hematopoietic and lymphoid tumors, it is classified under “precursor B-cell lymphoblastic leukemia/lymphoma” and “precursor T-cell lymphoblastic leukemia/lymphoma.” This

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.11132-8

1

2

Acute Lymphocytic Leukemia: Diagnosis and Treatment

classification has replaced the French–American–British (FAB) histopathologic classification of acute leukemia, originally proposed in 1976, which classified ALL into three entities: adult, childhood, and Burkitt type ALL. The new WHO classification system takes into account not only morphological and cytochemical characteristics of the disease but also its cytogenetic and clinical diversity. All in all, the widely used term “ALL” encompasses several entities as classified by WHO, defined below. B-lymphoblastic leukemia (B-ALL) not otherwise specified (NOS; ICD-O code: 9811/3) is a neoplasm of precursor lymphoid cells committed to the B-cell lineage, typically composed of small to medium-size blast cells with scant cytoplasm, moderately condensed to dispersed chromatin and inconspicuous nucleoli, involving bone marrow and blood. By convention, the term Blymphoblastic lymphoma (B-LBL), which is included under the same ICD-O code, is used when a process is confined to a mass lesion with no or minimal evidence of blood and bone marrow involvement (generally defined as less than 20% of lymphoblasts in the marrow). The term B-ALL does not encompass Burkitt leukemia/lymphoma. B-ALLs with specific recurrent genetic abnormalities associated with particular clinical and phenotypic features, and/or of prognostic significance are classified as separate entities. These are as follows: B lymphoblastic leukemia with t(9;22)(q34;q11.2); BCR-ABL1 (ICD-O code: 9812/3) B lymphoblastic leukemia with t(v;11q23); MLL (also called KMT2A or MLL1) rearranged (9813/3) B lymphoblastic leukemia with t(12;21)(p13;q22); TEL-AML1 (ETV6-RUNX1) (9814/3) B lymphoblastic leukemia with hyperdiploidy (9815/3) B lymphoblastic leukemia with hypodiploidy (Hypodiploid ALL; 9816/3) B lymphoblastic leukemia with t(5;14)(q31;q32); IL3-IGH (9817/3) B lymphoblastic leukemia with t(1;19)(q23;p13.3); E2A-PBX1 (TCF3-PBX1) (9818/3) B lymphoblastic leukemia, BCR-ABL1-like (9819/3) Additionally, B lymphoblastic leukemia with iAMP21 has been depicted as a provisional entity with no separate ICD-O code (it is coded under B-ALL, NOS; ICD-O code: 9811/3). T-lymphoblastic leukemia (T-ALL) is a neoplasm of precursor lymphoid cells with the same cellular characteristics as B-ALL but concerning precursor cells committed to the T-cell lineage.

Presentation and Diagnosis The symptoms of ALL are usually nonspecific and may appear only weeks or even days before diagnosis. Fatigue and weakness are among the most common nonspecific symptoms. Fever, with or without night sweats, and weight loss are frequent. The cause of fever is often not found, although granulocytopenia may lead to rapidly progressing and potentially life-threatening bacterial infections. Many patients present with thrombocytopenia and/or neutropenia, easy bruising and/or bleeding (bleeding gums, epistaxis, purplish patches in the skin or petechiae), and anemia as a result of bone marrow failure and disrupted hematopoiesis. Dyspnea, chest pain, and dizziness may also occur. Bone and joint pain due to bone marrow and periosteal infiltration is frequent in children in whom it may be the only presenting symptom. Headaches, vomiting, irritability, cranial nerve palsies, seizures, papilledema, and blurred vision are manifestations of the central nervous system (CNS) involvement and/or leukemic meningitis but initial CNS involvement is uncommon. A sensation of abdominal fullness and/or discomfort may appear due to hepatomegaly and/or splenomegaly. Hepatomegaly, splenomegaly, and lymphadenopathy at physical examination are found in about 20% of all patients and in approximately 50% of adult ALL presentations. Extramedullary infiltration may also lead to leukemia cutis (a raised, nonpruritic rash). Oliguria may occur as a result of dehydration, uric acid nephropathy, or disseminated intravascular coagulation (DIC). The initial workup includes complete blood count (CBC) with peripheral blood smear, blood chemistry profile, liver and kidney function tests, coagulation analysis (including measurement of prothrombin time, partial thromboplastin time, and fibrinogen), and a tumor lysis syndrome panel (including measurement of serum lactate dehydrogenase (LDH), potassium, uricemia, calcium, and phosphorus). Computed tomography (CT) scans of the neck, chest, abdomen, and pelvis are recommended for detecting possible organ involvement, lymphadenopathy, and organomegaly. If any extramedullary involvement is suspected, a positron emission tomography (PET)/CT is used for diagnosis. The leukocyte count in ALL patients may be elevated (majority of cases), normal, or decreased. T-ALL patients typically present with a high leukocyte count and often with a large mediastinal mass or other tissue mass. Circulating blast cells are present in virtually all cases. However, the percentage may be very low in some patients. ALL diagnosis is based on a hematopathological evaluation of bone marrow aspirate and biopsy material, with a demonstration of at least 20% of bone marrow lymphoblasts confirming a definitive ALL diagnosis. According to the guidelines of the US National Comprehensive Cancer Network (NCCN), the diagnostic workup should include a morphologic assessment of Wright/Giemsa-stained bone marrow aspirate smears and of hematoxylin-and-eosin-stained bone marrow core biopsy and clot sections, a comprehensive immunophenotyping using flow cytometry, and a baseline characterization of the leukemic clone(s) (see “Pathology and Genetics” section for more) to facilitate subsequent analysis of the minimal residual disease (MRD). For the latter, cytogenetic testing by fluorescence in situ hybridization (FISH) with a panel of the acute leukemia probes and reverse-transcriptase polymerase chain reaction (RT-PCR) for BCR-ABL1 fusions (and other fusions in case of BCR-ABL1 negativity), including determining the transcript size, are commonly used. In cases of aneuploidy, array comparative genomic hybridization (CGH) may also be useful.

Acute Lymphocytic Leukemia: Diagnosis and Treatment

3

Epidemiology and Risk Factors Burden ALL is mainly a childhood disease, with 75% of cases diagnosed in children under 6 years of age. It is also the most common form of childhood leukemia, accounting for 75%–80% of all childhood leukemia cases, while it accounts for only 20% of cases in adults. The estimated annual incidence worldwide is 1–4.8 cases per 100,000 population. In the United States, the age-adjusted incidence rate of all ALL is 1.58 per 100,000 population per year, with approximately 5970 new cases and 1440 deaths estimated in 2017. The prevalence of specific B-ALL types and of T-ALL is summarized in Table 1.

Etiology and Risk Factors The etiology of both B- and T-ALL is not well known. However, some environmental exposures have been associated with an increased risk. Radiation is a well-documented leukemogenic factor in humans, with a direct relationship between cumulative radiation dose and the incidence of some leukemia types. ALL may develop following exposure to ionizing radiation as well as radiotherapy. Also chemotherapy has been associated with an increased risk of developing the disease. Moreover, some chemicals, like benzene and toluene, contribute to the risk of acute leukemias, including ALL, likely by mutating or ablating the bone marrow stem cells. Acute leukemia usually develops 1–5 years after exposure and is often preceded by bone marrow hypoplasia, dysplasia, and pancytopenia. Age over 70 years is also a known risk factor for ALL. Moreover, several hereditary conditions are associated with an increased risk of ALL, in particular Down syndrome but also less frequent disorders: neurofibromatosis type 1, Bloom syndrome, Klinefelter syndrome, Schwachman–Diamond syndrome, Fanconi anemia, and ataxia telangiectasia. Genome-wide association studies have identified several single nucleotide polymorphisms (SNPs) associated with an increased risk of B-ALL, including in GATA2, ARID5B, IKZF1, CEBPE, and CDKN2A. In true familial ALL cases, even though rare, mutations in PAX5, ETV6, and TP53 have been described.

Pathology and Genetics In smear preparations, B-ALL lymphoblasts present variably, from small blasts with scant cytoplasm, condensed nuclear chromatin and indistinct nucleoli to larger cells with moderate amounts of light-blue to bluish-grey cytoplasm, sometimes vacuolated, dispersed chromatin and multiple variably prominent nucleoli. Nuclei are round or convoluted. Coarse azurophilic granules are present in some lymphoblasts in about 10% of cases. In bone marrow specimens, B-ALL lymphoblasts have a relatively uniform appearance, with round or oval, indented or convoluted nuclei. Nuclei range from inconspicuous to prominent, the chromatin is finely dispersed and the number of mitotic figures varies. Lymphoblasts in B-ALLs with specific genetic abnormalities usually do not have any specific morphological or cytochemical features that would allow a more specific diagnosis. Also T-ALL lymphoblasts are morphologically indistinguishable from B-ALL lymphoblasts. B-ALL lymphoblasts are nearly always positive for the B-cell markers CD19, cCD79a, and cCD22. None of these molecules on its own is a specific diagnostic marker but their combination or high intensity strongly supports the B-ALL diagnosis. In most cases, the expression of CD10, surface CD22, CD24, PAX5, and terminal deoxynucleotidyl transferase (TdT) is also found. The myeloidassociated markers CD13 and CD33 may be expressed and this does not exclude the B-ALL diagnosis. The PAX5 expression is

Table 1

The prevalence of different acute lymphoblastic leukemia (ALL) entities

Entity

Epidemiology

B-ALL with t(9;22)(q34;q11.2); BCR-ABL1 B-ALL with t(v;11q23); MLL rearranged

More common in adults (25% of all adult ALLs) than in children (2%–4%) The most common leukemia in infants under 1 year of age; less common in older children; incidence increases with age into adulthood Common in children (25% of all B-ALLs in children) but not seen in infants; incidence decreases with age (this entity is uncommon in adults) Common in children (25% of all B-ALLs in children) but not seen in infants; incidence decreases with age; accounts for 7%–8% of adult B-ALL cases 1%–5% of cases, both adults and children (nearly-haploid ALL likely in children only) Less than 1% of all ALL cases, both adults and children About 6% of childhood B-ALL cases, uncommon in adults 10%–25% of all ALL cases; incidence increases with age About 2% of ALL cases in children; no data on the incidence in adults About 15% of childhood ALL cases (more common in adolescents than in younger children) and 25% of adult ALL cases; more common in males than in females

B-ALL with t(12;21)(p13;q22); TEL-AML1 (ETV6-RUNX1) B-ALL with hyperdiploidy B-ALL with hypodiploidy B-ALL with t(5;14)(q31;q32); IL3-IGH B-ALL with t(1;19)(q23;p13.3); E2A-PBX1 (TCF3-PBX1) B-ALL, BCR-ABL1-like B-ALL with iAMP21 T-ALL

B-ALL, B-lymphoblastic leukemia; T-ALL, T-lymphoblastic leukemia.

4

Acute Lymphocytic Leukemia: Diagnosis and Treatment

considered to be the most specific and sensitive marker for the B-cell lineage in tissue sections but it is also expressed in some variants of acute myeloid leukemia. The expression profile of cell surface markers within the B-cell lineage varies with the B-cell differentiation (maturation stage). The immunophenotypes characterizing B-ALLs with specific recurrent genetic aberrations are summarized in Table 2. T-ALL lymphoblasts are usually positive for TdT and variably express CD1a, CD2–CD5, CD7, and CD8, with CD3 (cytoplasmic) and CD7 being positive most frequently. Only CD3, however, is specific for the T-lineage. CD4/CD8 coexpression is common but it is not specific to T-ALL. TdT expression combined with that of CD34, CD1a, and CD99 may help to establish the precursor nature of T-lymphoblasts. CD52 may be expressed in 30%–50% of adult T-ALL cases. One or both of the myeloid markers may also be expressed but, like for B-ALL, this does not exclude the T-ALL diagnosis. Moreover, many markers characteristic for premature Tcells (e.g., CD2 or CD7) may also be expressed by immature natural killer (NK) cells. Therefore, it may be very difficult to distinguish the true NK-cell ALL (which is rare) and T-ALL associated with the expression of immature T-cell markers. Clonal rearrangements of the immunoglobulin heavy locus (IGH) gene and rearrangements of the T-cell receptor (TCR) gene are found in both B-ALL and T-ALL, albeit at different frequencies. In B-ALL, IGH clonal rearrangements are found in nearly all cases and TCR rearrangements in up to 70% of cases, whereas in T-ALL, clonal TCR rearrangements are seen in most cases and simultaneous IGH rearrangements in about 20%. Moreover, cytogenetic abnormalities are seen in a majority of B-ALL cases, many of them defining specific entities with unique phenotypic and prognostic features (Table 2). In addition to those “classifying” aberrations, deletions of chromosome 6q, 9p, and 12p are frequent in B-ALL but they do not seem to impact prognosis. Some other abnormalities observed in B-ALL may emerge as having some clinical utility. For example, the rare (17;19)(q22;p13.3) translocation is associated with a very poor prognosis but there are too few cases to validate this alteration as a prognostic marker and classify these B-ALL cases as a separate entity. In T-ALL, abnormal karyotype is found in 50%–70% of cases. The most common recurrent cytogenetic abnormality involves the alpha and delta TCR loci (14q11.2) as well as the beta (7q34) and gamma locus (7p14.1), with a multitude of partner genes, most common of them being the TLX1 (also called HOX11; 10q24.3) and TLX3 (HOX11L2; 5q35.1) transcription factors. The most prevalent deletions are those of chromosome 9p, encompassing the locus of the CDKN2A gene. Alterations at a gene level are also found in ALL. In particular, a large number of recurrent genetic alterations (copy number alterations and specific intragenic mutations) are found in B-ALL. Many of them, such as those in the PAX5 gene, are found in most of the subtypes, suggesting that they play a fundamental role in leukemogenesis. Others, like those in RAS and IKFZ1, tend to be associated with specific subtypes (Table 2). In T-ALL, the most frequently mutated gene is NOTCH1 which encodes a protein critical for early T-cell development (mutated in 50% of cases). In 30% of cases, mutations in FBXW7, a negative regulator of NOTCH1, are found. According to data in the Cosmic database, the genes that are most frequently mutated in ALL are the following: IKFZ1 (35% of cases), ABL1 (21%), NRAS (10%), TP53 (10%), CDKN2A, PAX5, and KMT2D (9% each). However, the data do not distinguish between B-ALL and T-ALL, whichdgiven remarkable differences between the subtypesdis a serious limitation.

Management and Therapy The introduction of new therapies has resulted in a dramatic improvement of cure rates, in particular in B-ALL patients. ALL has a relatively good prognosis in children (5-year survival rates up to 89%). However, the prognosis for adults remains less favorable, especially for those of older age. Risk assessment in ALL is an essential part of treatment planning and the criteria, widely debated, have been evolving. Patient age, initial white blood cells (WBC) count, disease subtype, presence of CNS disease, and rapidity of response to induction therapy are recognized as important factors in assessing prognosis. For years, clinicians used to stratify children, adolescents, and young adults (AYA) with ALL into two groups: standard risk and high risk, based mainly on age, WBC count, and disease subtype. The Children’s Oncology Group (COG) has refined risk assessment criteria, creating four groups: very high, high, standard, and low risk. The very high risk category comprises all B-ALL patients with the t(9;22) translocation (Ph-positive ALL) and/or presence of the BCR-ABL1 fusion protein, B-ALL patients with hypodiploidy (defined as less than 44 chromosomes), BCR-ABL1-like, or iAMP21 subtype, and those with KMT2A alterations or failure to achieve remission with induction therapy, regardless of age and WBC count. About 4% of nonadult B-ALL patients fall into this category. Risk stratification for T-ALL has been more difficult and this entity is frequently categorized as very high risk. However, newer treatments have resulted in improved outcomes also for T-ALL patients. According to the current NCCN guidelines, the initial risk stratification of all ALL patients should be based on the presence of the t(9;22) chromosomal translocation and/or BCR-ABL fusion protein (Ph-positive versus Ph-negative). Further stratification may be based on patient age, WBC count, comorbidities, and the presence of minimal residual disease (MRD). The treatment of ALL is extremely complex, with treatment regimens, selection of drugs, doses, and schedules which differ according to the age group of patients (children, AYA, or older adults), risk stratification, and ALL subtype. However, all treatment protocols follow the same basic principles, with three main treatment phases: induction, consolidation, and maintenance. CNS prophylactic and/or therapeutic treatment is also a mandatory element of ALL management. The goal of the induction therapy, the first phase of the ALL treatment, is to reduce tumor burden by clearing as many leukemic cells as possible from the bone marrow, that is to induce remission. Most inductions regimens are based on a combination of vincristine, anthracycline (e.g., daunorubicin or doxorubicin, in particular for higher risk patients), and a corticosteroid (e.g.,

Table 2

The genetic and immunophenotypic profiles of different B-lymphoblastic leukemia (B-ALL) entities as well as their clinical significance Genetic profile

Immunophenotype

Prognosis and predictive factors

B-ALL with t(9;22)(q34;q11.2); BCR-ABL1

The t(9;22) translocation resulting from fusion of BCR at 22q11.2 and ABL1 at 9q34.1, with production of a BCR-ABL1 fusion protein (p190 BCR-ABL1 in most childhood cases and in about half of adult cases); the resulting aberrant chromosome 22 harboring the fusion BCR-ALB1 gene locus is called Philadelphia (Ph) chromosome Fusions of the KMT2A (MLL) gene with multiple genes (over 100 identified fusion partners), the most common partner being AFF1 (AFF4; 4q21), followed by MLLT1 (19p13.3), and MLLT3 (9p21.3); none of these fusions is specific to this entity; frequent overexpression of FTL3; very few additional mutations in infants, usually in the RAS pathway The ETV6-RUNX1 translocation with production of the fusion protein (a putative negative regulator of the transcription factor RUNX1), likely an early event in the leukemogenesis Increased chromosome number (usually without structural aberrations), most commonly extra copies of chromosome 21, X, 14, and 4; the least common: chromosome 2 and 3

Typically positive for CD10, CD19, and TdT, and negative for CD117 (KIT); CD25 expression highly associated with this entity, especially in adults; myeloidassociated markers CD13 and CD33 frequently expressed; T-cell precursor phenotype in rare cases

The worst prognosis of all types, particularly in adults; patients potentially responsive to tyrosine kinase inhibitors (e.g. imatinib), so prognosis may change with the introduction of targeted therapies

Typically positive for CD19 and CD15, and negative for CD10 and CD24; the NG2 homolog usually expressed and relatively specific

KMT2A-FFA1 translocation associated with a very poor prognosis, particularly in infants under the age of 6 months

Positive for CD10 and CD19, and usually also for CD34; characteristic nearly complete absence of CD9, CD20, and CD66c (relatively specific); myeloid-associated markers, especially CD13, frequently expressed Positive for CD10 and CD19, and other typical B-ALL markers; most cases positive for CD34 and negative for CD45

Very favorable prognosis (cure in >90% of childhood cases); relapses later than for other B-ALL types

B-ALL with t(v;11q23); MLL rearranged

B-ALL with t(12;21)(p13;q22); TEL-AML1 (ETV6-RUNX1) B-ALL with hyperdiploidy

B-ALL with hypodiploidy B-ALL with t(5;14)(q31;q32); IL3-IGH

Loss of one or more chromosomes and nonspecific structural aberrations in the other chromosomes (rare in near-haploid cases) A functional rearrangement between IL3 (chromosome 5) and IDH (chromosome 14) leading to constitutional overexpression of IL3 and eosinophilia

Typically positive for CD10 and CD19 but no other distinct features Typically positive for CD10 and CD19; even small numbers of blasts expressing these two markers in a patient with eosinophilia strongly suggests this diagnosis

Very favorable prognosis in children (cure in >90% of cases); not enough data for adults; trisomies may have more significance as prognostic factors than the actual number of chromosomes; simultaneous trisomy of chromosome 4 and 10 being associated with the best prognosis Poor prognosis (likely even without MRD); the worst prognosis for nearly-haploid cases suggested by some studies No data

(Continued)

Acute Lymphocytic Leukemia: Diagnosis and Treatment

Entity

5

6

The genetic and immunophenotypic profiles of different B-lymphoblastic leukemia (B-ALL) entities as well as their clinical significancedcont'd

Entity

Genetic profile

Immunophenotype

Prognosis and predictive factors

B-ALL with t(1;19)(q23;p13.3); E2A-PBX1 (TCF3-PBX1)

TCF3-PBX translocation with production of the fusion protein (oncogenic); the fusion gene is located on chromosome 19, may be associated with loss of chromosome 1 (unbalanced translocation); an alternative TCF3 translocation involving the HLF gene (chromosome 17) in rare cases Various chromosomal rearrangements of multiple genes with various partners; CRLF2 rearrangements in approximately half of the cases; the tyrosine kinasetype translocations often involve ABL1 with partners other than BCR, as well as other kinases; additionally frequent deletions and point mutations in other genes, particularly IKZF1 (not specific to this entity) and CDKNA2A/B; mutations in JAK1 or JAK2 found in about a half of cases with CRLF2 rearrangements Intrachromosomal amplification of chromosome 21 (iAMP21) detectable with FISH probes recognizing the ETV6-RUNX1 translocation; multiple other chromosomal abnormalities (most commonly involving chromosome X and 7) in 80% of cases; deletions of RB1 and ETV6 as well as CRLF2 rearrangements more frequent than in other ALLs

Typically positive for CD10, CD19, and cytoplasmic mu heavy chain (pre-B phenotype); typical strong expression of CD9 with lack or very limited expression of CD34 (with these features, this diagnosis may be suspected even if cytoplasmic mu is undetermined)

The TCF3 translocation associated with very poor prognosis; possibly an increased risk of CNS relapses in patients with this entity

Typically positive for CD10 and CD19; very high levels of the translocated CRLF2 gene product (detectable by flow cytometry) in a subset of cases with these translocations

Overall poor prognosis; this may be due to an increased risk of MRD (controversial); children resistant to induction therapy usually have a translocation involving PDGFRB (most often with EBF1) and are very responsive to ABL-class tyrosine kinase inhibitors (e.g. imatinib or dasatinib); patients with JAK mutations or translocations may be candidates for treatment with JAK inhibitors (to be tested)

Unknown

Unclear

B-ALL, BCR-ABL1-like

B-ALL with iAMP21

CNS, central nervous system; FISH, fluorescence in situ hybridization; MRD, minimal residual disease; t, translocation; TdT, terminal deoxynucleotidyl transferase.

Acute Lymphocytic Leukemia: Diagnosis and Treatment

Table 2

Acute Lymphocytic Leukemia: Diagnosis and Treatment

7

dexamethasone or prednisone) with or without L-asparaginase and/or cyclophosphamide, given over the course of 4–6 weeks. An alternative is the so-called Hyper-CVAD regimen which alternates cycles of hyperfractionated cyclophosphamide, vincristine, doxorubicin (adriamycin), and dexamethasone with those of high-dose methotrexate and cytarabine. Most children achieve complete remission (CR) within 4 weeks of the induction therapy. If CR is not achieved within 6 weeks, alternative treatments are started. In adults, a regimen based on a combination vincristine/prednisone/anthracycline is associated with 75% CR rates. Dexamethasone has been shown to give better outcomes compared to prednisone. It has a better penetration into the CNS, which explains its efficacy in preventing CNS relapse. However, it is associated with severe toxicities and its advantage for the overall survival has yet to be demonstrated. CNS prophylaxis after induction chemotherapy aims at preventing CNS disease or early CNS relapse. The CNS is the initial site of ALL relapse in more than half of children unless prophylaxis is given and it is also a frequent site of relapse in adults. The goal of CNS prophylaxis and/or treatment is to clear leukemic cells within sites that cannot be easily accessed with systemic chemotherapy because of the blood–brain barrier. Different regimens of CNS-directed therapy exist. Many authorities recommend intrathecal methotrexate, often combined with cranial irradiation. Triple intrathecal therapy (methotrexate, hydrocortisone, cytarabine), with or without high-dose systemic chemotherapy (e.g., methotrexate or cytarabine), may substitute for cranial irradiation which can lead to late intellectual disabilities when used in children. For adults, prophylactic intrathecal chemotherapy alone is considered sufficient. CNS prophylaxis is typically administered to all patients throughout the entire course of ALL therapy. The consolidation treatment aims at eliminating any leukemic cells remaining after induction therapy, further eradicating residual disease. The postremission induction phase of treatment may also be referred to as intensification therapy. It typically involves an intensive multidrug regimen. A variety of regimens are used in different centers and trials, although frequently containing similar drug combinations to those that were used during the induction treatment. High-dose methotrexate, cytarabine, 6mercaptopurine (6-MP), cyclophosphamide, vincristine, corticosteroids, and L-asparaginase are frequently incorporated into consolidation/intensification regimens. For treatment of younger children, adjusted doses and frequencies are used. Maintenance therapy aims to prevent disease relapse after postremission induction and consolidation therapy. Most regimens are based on daily 6-MP and weekly methotrexate, typically with a periodic addition of vincristine and corticosteroids, for 2–3 years. Of note, the efficacy of the maintenance therapy is determined by the metabolism of 6-MP. In particular, the efficacy of the active metabolite production is modified by polymorphisms in thiopurine methyltransferase (TMPT). Genotyping TMPT allows to anticipate decreased bioavailability or increased toxicity of 6-MP and adjust the dose. The optimal duration of the maintenance therapy is not clearly defined and the therapy is frequently stopped due to associated toxic effects. The emergence of targeted drugs offers hope for more effective treatments of ALL. In particular, adding tyrosine kinase inhibitors (TKIs) to the induction treatment regimens of Ph-positive ALL patients has significantly improved their prognosis. Imatinib (also called imatinib mesylate; brand name: GleevecÒ, Novartis Pharmaceuticals), an inhibitor of the BCR-ABL tyrosine kinase, has been approved by the US Food and Drug Administration (FDA) for treatment of relapsed or refractory Ph-positive ALL adult patients and of all untreated Ph-positive pediatric patients. Dasatanib, a second-generation TKI which inhibits both the BCR-ABL tyrosine kinase and kinases of the Src family, seems to be even more effective than imatinib due to its capacity to inhibit several TK-related signaling pathways. It has also been shown to maintain activity against cells harboring several (but not all) ABL domain mutations which confer resistance to imatinib. Furthermore, it has a better penetration of the blood–brain barrier and so it may prove useful in the CNS prophylaxis. Dasatinib (under the brand name of SPRYCEL, Bristol-Myers Squibb Co) has been approved by FDA for treatment of pediatric patients with Ph-positive chronic myeloid leukemia and it is being extensively trialed (phase II and III) for treatment of ALL. Single-agent TKI therapy in Ph-positive ALL patients has been shown to yield improved response to induction therapy as compared to chemotherapy. However, the remission-free periods are relatively short for both imatinib and dasatinib alone. TKIs have been shown to provide most benefit, with significantly improved disease-free and overall survival rates, when combined with corticosteroids. Patients with CD20-positive B-ALL benefit from the addition of rituximab, an anti-CD20 monoclonal antibody, to standard ALL chemotherapy regimens. All these agents may be incorporated as part of the induction, consolidation, and/or maintenance therapy in the initial treatment of ALL, or in regimens for treating relapsed and/or refractory disease. Allogeneic hematopoietic stem cell transplantation (HSCT) used to be considered the standard of care in AYA and adults with Ph-positive ALL, and for relapsing ALL patients. However, with the advent of BCR-ABL-targeted TKIs, its role has been questioned. Still, it may offer some benefit to specific groups of Ph-positive patients who relapse after initial remission. Treatment of relapsed ALL remains a challenge. Combining standard chemotherapy regimens with emerging targeted drugs offers much hope to this effect. In particular, mitoxantrone (NovantroneÒ, Immunex Corporation; an anthracycline derivative), cytarabine, fludarabine/cytarabine/granulocyte colony-stimulating factor/idarubicin (FLAG-IDA), methotrexate/asparaginase (CAPIZZI), and blinatumomab (a monoclonal antibody against CD19 and CD3) have been shown to provide benefit to patients with CD19-positive ALL, while inotuzumab ozagamicin (FDA-approved since August 2017) has been shown to be useful in patients with CD22-postiive disease. Several first-, second-, and third-generation TKIs are being currently tested in clinical trials and their approval for treatment of relapsed/refractory ALL, and in particular for BCR-ABL1-like ALL, is likely just a matter of time. However, as of now, bone marrow transplant remains the only cure for relapsed/refractory ALL. For patients who are not eligible for transplant, generation of chimeric antigen receptor T-cells (CAR-T) targeting specific molecules on the surface of cancer cells may be a solution. In August 2017, FDA approved the anti-CD10 CAR-T therapeutic agent tisagencleucel (KYMRIAHÒ, Novartis Pharmaceuticals) for treatment of refractory or relapsed B-ALL patients (at least second relapse) below 25 years of age. However, CAR-T therapy is associated with severe potentially fatal toxicities (like cytokine release syndrome, B-cell aplasia, and cerebral edema) and its use should be preceded by a specific risk evaluation. FDA also requires a special

8

Acute Lymphocytic Leukemia: Diagnosis and Treatment

certification for centers administering this therapy. For relapsed or refractory T-ALL, nelarabine, a purine nucleoside analog, is currently an approved treatment. Assessing the response to treatment also remains a challenging issue. Complete remission (CR) is defined as the absence of circulating blasts and extramedullary disease. However, patients who achieved CR according to morphological assessment can potentially harbor leukemic cells in the bone marrow, giving rise to a low-level disease which is not detectable by conventional cytomorphological techniques (minimal residual disease, MRD). Many trials have confirmed that MRD is the strongest prognostic factor in both children and adults. Therefore, applying reliable sensitive methods to detect MRD is of utmost importance for further treatment planning and monitoring of treatment response at all stages following the induction therapy. MRD can be detected by multicolor flow cytometry or sensitive molecular techniques targeting characteristic genetic or chromosomal aberrations, like RTPCR or next-generation sequencing.

Further Reading Chmielowski, B., Territo, M., 2017. Manual of clinical oncology, 8th edn. Wolters Kluwer, Philadelphia. International Agency for Research on Cancer, 2017. In: Swerdlow, S.H., Campo, E., Harris, N.L., Jaffe, E.S., Pileri, S.A., Stein, H., Thiele, J. (Eds.), WHO classification of tumours of hematopoietic and lymphoid tissues, Revised 4th edn. International Agency for Research on Cancer, Lyon. Kotrova, M., Brüggemann, M., 2017. Minimal residual disease in adult ALL: Technical aspects and implications for correct clinical interpretation. Blood Advances 1 (25), 2456–2466. National Comprehensive Cancer Network (NCCN), 2017. NCCN clinical practice guidelines in oncology: Acute lymphoblastic leukemia, Version 5.2017. October 27, 2017. https:// www.nccn.org/professionals/physician_gls/default.aspx#site.

Relevant Website https://clinicaltrials.gov/dClinical trial information at the US National Library of Medicine, National Health Institutes.

Acute Myelogeneous Leukemia: Diagnosis and Treatment ska, Science to the Point, Archamps Technopole, Archamps, France Katarzyna Szyman Sophie Park, Grenoble-Alpes University, Grenoble, France; and Institute for Advanced Biosciences, Grenoble, France © 2019 Elsevier Inc. All rights reserved.

Abbreviations AML Acute myeloid (myelogeneous) leukemia APL Acute promyelocytic leukemia APLDS APL differentiation syndrome ATO Arsenic trioxide ATRA All-trans-retinoic acid CD Cluster of differentiation CNS Central nervous system CR Complete remission CT Computed tomography DIC Disseminated intravascular coagulation FAB French–American–British [histopathologic classification of acute leukemia] FDA US Food and Drug Administration FISH Fluorescence in situ hybridization GO Gemtuzumab ozogamicin HiDAC High-dose cytarabine HSCT Hematopoietic stem cell transplantation ICD-O International Classification of Diseases for Oncology IDH Isocitrate dehydrogenase ITD Internal tandem duplication MDS Myelodysplastic syndrome MDS/MPN Myelodysplastic/myeloproliferative neoplasm MRD Minimal residual disease MRI Magnetic resonance imaging NCCN US National Comprehensive Cancer Network PET Positron emission tomography t-AML Therapy-related AML WBC White blood cells WHO World Health Organization

Definition and Classification Acute myelogenous leukemia (AML), also called acute myeloid leukemia and acute nonlymphocytic leukemia, is a heterogeneous hematological malignancy characterized by clonal expansion of myeloid blasts in the peripheral blood, bone marrow, and/or other tissues. The term encompasses a number of entities whose basic characteristics are summarized in Table 1. Of note, the current classification of acute leukemia is the 2016 classification by the World Health Organization (WHO). It has replaced the French–American–British (FAB) histopathologic classification of acute leukemia, originally proposed in 1976, and it takes into account not only morphological and cytochemical characteristics of the disease but also its cytogenetic and clinical diversity. However, the “historical” names of some leukemia types referring to FAB categories have persisted and they are frequently used in clinical practice and related publications. These names are listed among synonyms in Table 1. Moreover, the WHO classification has lowered the diagnostic threshold of blasts which defines AML to 20% (compared to 30% in the FAB classification) and it also allows the AML diagnosis regardless of the blast count in patients with abnormal hematopoiesis and specific cytogenetic abnormalities.

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65299-6

9

10

Classification and defining characteristics of acute myeloid leukemia (AML) entities

Entity name [synonyms]

ICD-O code

AML with balanced translocations/inversions AML with t(8;21)(q22;q22); RUNX1-RUNX1T1 [FAB M2, t(8;21)(q22;q22); FAB M2, AML1(CBF-alpha)/ETO]

9869/3

AML with abnormal marrow eosinophils [AML, t(16;16)(p 13;q 11); AML, CBF-beta/MYH11; acute myelomonocytic leukemia with abnormal eosinophils; FAB M4Eo; AML, inv(16)(p13;q22)] Acute promyelocytic leukemia (APL) with PML-RAR-alpha [acute promyelocytic leukemia, t(15;17)(q22;q11-12); acute promyelocytic leukemia, NOS; FAB M3; AML, PML/RAR-alpha; AML with t(15:17)(g22;q1112)] AML with t(9;11) (p22;q23); MLLT3-MLL

9871/3

AML with t(6;9)(p23;q34); DEK-NUP214

9865/3

AML with inv(3)(q21q26.2) or t(3;3)(q21;q26.2); RPN1-EVI1 [AML with inv(3)(q21q26.2) or t(3;3)(q21;q26.2); GATA2-MECOM]

9869/3

AML (megakaryoblastic) with t(1;22)(p13;q13); RBM15-MKL1

9911/3

AML with BCR-ABL1

9912/3 (provisional entity)

AML with gene mutations AML with mutated NPM1 AML with biallelic mutations of CEBPA

9877/3 9878/3

AML with mutated RUNX1

9879/3 (provisional entity)

Other AMLs AML with myelodysplasia-related changes [AML with multilineage dysplasia; AML with prior myelodysplastic syndrome; AML without prior myelodysplastic syndrome]

9866/3

9897/3

9895/3

Defining characteristics other than cytogenetic/genetic abnormalities reflected in the name of the entity Generally shows maturation in the neutrophil lineage; Bone marrow and peripheral blood show large myeloblasts with abundant basophilic cytoplasm, often containing azurophilic granules; Classified as AML regardless of the blast count Usually shows monocytic and granulocytic differentiation and characteristically abnormal eosinophil component in the bone marrow; Classified as AML regardless of the blast count Predomination of abnormal promyelocytes; Two types exist: hypergranular and microgranular (hypogranular); Classified as AML regardless of the blast count Usually associated with monocytic features; It is controversial whether cases with blast count > PCC > PGL/PCC PGL > PCC PGL > PCC PGL > PCC

Multiple Multiple Multiple

SDHAF2 FH

0.1 1

PGL > PCC PCC > PGL

Multiple Single

MDH2 TMEM127

g)

1 PGL > PCC/PGL PCC > PGL PCC?

Retinal and CNS hemangioblastoma, clear cell RCC and renal cysts, pancreatic neuroendocrine tumor and pancreatic cyst, endolymphatic sac tumors, papillary cystadenoma of the epididymis/broad ligament and mesosalpinx Medullary thyroid carcinoma, multiglandular parathyroid disease, marfanoid habitus, mucocutaneous neuromas, gastrointestinal ganglioneuromatosis, cutaneous lichen amyloidosis Neurofibroma, malignant peripheral nerve sheath tumor, gastrointestinal stromal tumor, optic nerve and other brain stem glioma, neuroendocrine tumor (somatostatinoma) of the periampullary duodenum, cafe-au-lait spots and axillary or inguinal freckling, Lisch nodules, dysplasia of sphenoid bone or long bones, kyphoscoliosis, cerebral arteriopathy, pulmonary artery stenosis, learning disabilities Gastrointestinal stromal tumor, pituitary adenoma, SDH-deficient RCC, melanoma? Gastrointestinal stromal tumor, SDH-deficient RCC, pituitary adenoma Gastrointestinal stromal tumor, SDH-deficient RCC, pituitary adenoma Gastrointestinal stromal tumor, SDH-deficient RCC, pituitary adenoma, pancreatic neuroendocrine tumor? None reported Hereditary leiomyomatosis and RCC (HLRCC) phenotypic manifestations including cutaneous and uterine leiomyomas and/or leiomyosarcomas? and FH-deficient RCC encompassing papillary RCC Type 2, tubulo-papillary RCC and/or tubulocystic RCC with poorly differentiated foci, and collecting duct RCC; and leydig cell tumor of the testis? ?c Clear cell RCC

EPAS1/HIF2ab (s and s/ m >> g)d

1

PGL > PCC >> PGL/PCC

Multiple > single

H3F3A (m) PHD1/EGLN2 PHD2/EGLN1 BAP1

somatotrophinoma > corticotrophinoma/ nonfunctioning), adrenal cortical tumor, brochopulmonary neuroendocrine tumor, thymic neuroendocrine tumor, gastric neuroendocrine tumor, angiofibromas, collagenomas, lipomas, meningiomas Germline MET mutations are associated with Hereditary Papillary RCCe Medullary thyroid carcinoma?

Abbreviations: CNS, central nervous system; EPO, erythropoietin; g, germline mutation; PGL, paraganglioma; PCC, pheochromocytoma; RCC, renal cell carcinoma; s, somatic mutation; s/m, somatic mosaicism. a Combined data for PPGL prevalence (Favier et al., Nature Review Endocrinology, 2014; Bausch et al., JAMA Oncology, 2017; Dahia PL, Nature Review. Cancer, 2014). b Gain-of-function mutations; loss-of-function mutations in RET result in intestinal aganglionosis and Hirschprung disease. c One truncating MDH2 mutation is recorded in The Catalogue of Somatic Mutations in Cancer (COSMIC; http://cancer.sanger.ac.uk/cancergenome/projects/cosmic/). d Indicating low level constitutional mosaicism. e Paraganglionic tumors are not features of this disorder.

22

Adrenal Glands Tumors: Pathology and Genetics Table 2

Tumor combinations of pheochromocytoma and paraganglioma (PPGL) raising the suspicion of occult hereditary disease

Tumor combinations

Syndromes and/or germline mutations

PPGL and entero-pancreatic neuroendocrine tumor

Duodenal somatostatinoma Polycythemia-PPGL-somatostatinoma syndrome (or Pacak-Zhuang syndrome) Neurofibromatosis type 1 (NF1) Pancreatic neuroendocrine tumor Von Hippel-Lindau (VHL) syndrome Multiple endocrine neoplasia type 1 (MEN1) PGL1 (or SDHD-related PPGL syndrome) Multiple endocrine neoplasia type 1 (MEN1) SDH- and MAX-related PPGL syndromes Multiple endocrine neoplasia type 1 (MEN1) Carney Triad SDH-related PPGL syndromes Carney-Stratakis syndrome Carney Triad Neurofibromatosis type 1 (NF1) Von Hippel-Lindau (VHL) syndrome SDH-related PPGL syndromes MAX- and TMEM127-related PPGL syndromes Hereditary leiomyomatosis and renal cell cancer (HLRCC) KIF1b, SDHB, SDHC Composite tumor NF1, VHL, MEN2A, MEN2, SDHB

PPGL and pituitary adenoma PPGL and adrenal cortical adenoma PPGL and gastrointestinal stromal tumor

PPGL and renal tumor

PPGL and peripheral neuroblastic tumor

Table 3 Stage I Stage II Stage III

Stage IV

Adrenocortical tumor staging criteria as proposed in children Completely resected tumors 5 years) and hormonal secretion pattern of pediatric adrenocortical tumors might be tissue (fetal or transient embryonic adrenal cortex)- and/or genotype (germline TP53 mutation)-dependent, possibly reflecting perturbations of sex- and age-specific developmental processes in the adrenal cortex.

Risk Factors Hereditary susceptibility is the main causative factor (see Genetic Susceptibility). Subsets of ACC exhibit mutation signature 1, featuring C > T substitution in a CG context and resembling the age- and DNA-mismatch repair-deficiency signatures. They also exhibit mutation signature 2, featuring A > T substitution in a CG context and resembling the smoking signature. The latter is in accordance with data derived from the 1986 National Mortality Followback Survey suggesting heavy cigarette smoking as a risk factor in men. The same study supported a role of using oral contraceptives before the age of 25 as risk factor in women. This is consistent with clinical data indicating that proliferation and/or secretion of adrenocortical neoplasms might be affected by the

Adrenal Glands Tumors: Pathology and Genetics

25

hormonal context of pregnancy. It is also supported by experimental evidence: a growth-inhibitory effect of selective estrogen receptor modulators on human adrenocortical cell line H295R and on human tumors xenografted in athymic nude mice.

Pathology For a diagnosis of malignancy (i.e., the ability to invade adjacent structures and metastasize) of an adrenal cortical neoplasm, multiparametric histopathological scoring systems are used, including parameters such as cell type, tumor architecture, tumor cell invasion, and cellular proliferation. These systems include the Weiss scoring system, Hough scoring system, van Slooten scoring system, Weiss revisited index and/or diagnostic algorithms, including the stepwise discriminate diagnostic system and the “reticulin” diagnostic algorithm. The latter defines malignancy in a two-step process, the presence of a disrupted reticulin framework constituting the first step of this diagnostic approach, and subsequent identification of at least 1 out of 3 parameters: a high mitotic rate (> 5/50 HPFs), necrosis, and venous invasion. Both the “reticulin” algorithm and the Helsinki score, a novel model for prediction of metastases in ACCs based on mitotic index (> 5/50 HPFs), necrosis and Ki67 labeling index, have been subsequently validated. The Weiss system is the most popular among all multifactorial scoring systems, comprising nine histologic parameters, that is, high nuclear grade, mitotic rate (> 5 mitotic figures/50 HPFs), atypical mitoses, < 25% clear cells, diffuse architecture, tumor necrosis, capsular invasion, venous invasion, and sinusoidal invasion, with presence of  3 indicating malignant behaviour. Nevertheless, no particular system is currently endorsed in the 2017 WHO classification. Various limitations of the Weiss system include lack of reproducibility of several criteria, borderline tumors with a Weiss score of 2 or 3, oncocytic and myxoid variants of adrenocortical neoplasms, and pediatric adrenocortical tumors, which are frequently overdiagnosed as malignant given the presence of impressively atypical histologic features. With regard to the latter, a specialized classification system has been proposed by Wieneke et al., while the Lin-Weiss-Bisceglia system has been established for oncocytic adrenocortical neoplasms given the inherent presence of three definitional criteria and overdiagnosis of malignancy if Weiss system is applied. On a large ACC series, assessment of the most frequent diagnostic pitfalls resulted in a disagreement level as high as 9% of referral cases. Misclassification of metastatic and/or primary intra-adrenal tumors (PCC, ACA, and soft tissue neoplasm) as ACCs and vice versa accounted for such discrepancies, which significantly modified clinical management. An immunohistochemically determined Ki67 proliferation index > 5% and/or over-expression of IGF2 with a characteristic juxtanuclear (Golgi) dot-like pattern are useful for accurate classification of adrenocortical neoplasms. From a differential diagnostic perspective, ACCs usually exhibit reactivity for steroidogenic factor-1 (SF-1), Melan-A, inhibin-a, calretinin and synaptophysin, while they are negative for cytokeratins, EMA, CEA, PAX-8, HMB-45 and chromogranin-A. Hence, this immunohistochemical panel can exclude intraadrenal tumors, for example, PCC and primary soft-tissue neoplasms, metastatic tumors, for example, lung/breast/colorectal adenocarcinomas, urothelial carcinomas and melanomas, and adrenal gland involvement by other malignancies, for example, renal cell carcinoma, hepatocellular carcinoma and retroperitoneal sarcomas. Despite these advances, some adrenocortical neoplasms defy classification into benign and malignant categories. In such cases, the diagnosis germ of adrenocortical neoplasm of uncertain/undetermined malignant potential can be used.

Molecular Pathology and Genetics Genetic Profile Driver-gene mutations and activation of key cellular signaling pathways they control appear to suffice for adrenal tumorigenesis. By adapting methods from molecular evolution to the study of cancer genomes, it has been shown that positive selection outweighs negative selection during cancer development, while the number of driver mutations increases, but not linearly, with increasing mutation burden. Multistep tumorigenesis within ACCs was further corroborated by data emerging from pan-genomic analysis of TCGA based on variant allele fractions, genotype and LOH patterns and from a comprehensive genomic analysis of pediatric adrenocortical tumors based on timing of copy-neutral-LOH patterns and the temporal order of somatic single nucleotide variations (SNV). The former displayed widespread chromosomal loss with whole genome doubling (WGD) as important genomic event in ACC progression, while the latter highlighted copy-neutral-LOH of chromosomes 11/17 as early driver events with subsequent SNV acquisition in these regions. Whether ACCs arise de novo or as result of an adenoma-to-carcinoma sequence is still not settled, although clinicopathologic and molecular evidence in support of a progression model is accumulating. Molecular evidence in support of stepwise progression derives from isolated reports, single nucleotide polymorphism (SNP) array profiling, comparative genomic hybridization, Xchromosome inactivation pattern analysis and microsatellite allelotyping. Nevertheless, this sequence is currently regarded as exceptionally rare with an extremely low progression rate given molecular and/or epidemiological data. In the pediatric context, a similar model was also proposed for TP53-associated adrenocortical tumors, arising from a simpler genomic background and progressing to acquire complex genomic aberrations. Given shared molecular attributes in pediatric adrenal cortical tumors, that is, germline TP53 mutations, chromosome 11p15 abnormalities, and IGF2 overexpression, it has been recently proposed that adrenocortical adenomas, tumors of undetermined histology and ACCs represent a spectrum of the same disease. The comprehensive TCGA study has recently provided detailed insight in molecular events in ACC oncogenesis. ACC displayed a heterogeneous somatic mutation density with a median of 0.9 per Mb, which is higher than all genitourinary malignancies, with

26

Adrenal Glands Tumors: Pathology and Genetics

the exception of urothelial carcinoma of the bladder. Significantly mutated genes were TP53, CTNNB1, MEN1, PRKAR1A, and RPL22. The catalogue of known ACC driver genes was expanded to include TERF2, CCNE1, and NF1. A low frequency of gene fusions was found without recurrent events. Copy number alterations were significant with WGD as hallmark mechanism for disease progression. Expression level of telomerase reverse transcriptase (TERT) was high but telomere length decreased, highlighting the role of telomerase reactivation and telomere maintenance. Finally, p53 apoptosis/Rb1 cell-cycle pathway and Wnt/b-catenin pathway as well as PKA subnetwork were significantly modified. Despite the relative lack of targetable hotspot mutations, a subset of ACCs demonstrated approximately 50 potentially actionable alterations, encompassing mutations and copy-number changes. The TCGA study was based on a near-global ACC sample and in continuity with previous efforts on adult and pediatric adrenocortical cancer from Europe, North America and Brazil. This confirmed (i) highly prevalent IGF2 over-expression; (ii) high frequency of TP53 mutations, which are dominant in the pediatric setting; (iii) high frequency of diverse defects resulting in deregulation of the Wnt/b-catenin pathway, that is, CTNNB1 point mutations, homozygous deletion of Wnt repressors ZNRF3/KREMEN1 and/or deactivating APC/MEN1 alterations; (iv) evidence for epigenetic deregulation, for example, alterations of chromatin remodeling (ATRX/DAXX) and histone modification (MLL/MLL2/MLL4) genes; and (v) significant copy number changes with recurrent focal involvement of the ZNRF3 and/or TERT loci. Along these lines, TERT promoter mutations were also detected in a small subset of ACCs in accordance with previously published data. In addition, the TCGA study expanded the transcriptomics classification and refined the multi-omics classification of two distinct molecular entities C1A/C1B driven by different oncogenic alterations into three molecular Cluster of Cluster (CoC) subtypes I/II/III with distinct clinical outcomes based on integrated analysis of DNA copy number, mRNA expression, DNA methylation and miRNA expression. The occurrence of genome near-haploidisation with or without subsequent endo-reduplication has been demonstrated in thyroid, parathyroid and adrenocortical tumors with oncocytic features and without any associations with mitochondrial (mt) DNA mutations. In another study, the 4977 bp mtDNA “common deletion” was detected in approximately 50% of oncocytic adrenocortical neoplasms and with a relatively high degree of intercellular and intracellular heteroplasmy. A subset of sarcomatoid ACC was shown to be monoclonal in origin with dysregulation of Wnt/b-catenin signaling pathway and mutational TP53 inactivation as frequent genetic events. In this context, SF1 immunonegativity and/or loss of expression of steroidogenic enzymes provided further evidence of a dedifferentiation process in this ACC variant. Molecular diagnostic markers that differentiate ACCs from adenomas have been stemming from gene expression, microRNA and DNA methylation profiling analyses. Further studies are warranted in order to validate and define the level of diagnostic accuracy for such molecular tests especially in borderline adrenocortical tumors.

Genetic Susceptibility Hereditary transmission of ACCs is uncommon: most tumors are sporadic. Nonetheless, the spectrum of genes associated with familial cancer susceptibility syndromes encompasses TP53 (Li-Fraumeni syndrome), IGF2/CDKN1C/H19 locus alterations on 11p15 (Beckwith-Wiedemann syndrome), mismatch repair (MMR) genes, that is, MSH2, MSH6, MLH1, PMS2 (Lynch syndrome), APC (Familial Adenomatous Polyposis syndrome), NF1 (Neurofibromatosis type 1), MEN1 (Multiple Endocrine Neoplasia type 1), and PRKAR1A (Carney Complex). Pathogenic MUTYH mutations were also present in ACC patients of two series analyzed by whole-exome sequencing and with the corresponding tumors displaying loss of the wild-type MUTYH allele. Germline MUTYH mutations are associated with MUTYH-associated polyposis and of interest is that mostly benign adrenal lesions are relatively frequent in patients with polyposis, that is, Familial Adenomatous Polyposis, Attenuated Familial Adenomatous Polyposis and/ or MUTYH-Associated Polyposis, who undergo abdominal imaging. Exceedingly rare examples of ACCs have been reported arising in patients with germline SDHC/SDHA mutations (PGL3/PGL5), FH (Hereditary Leiomyomatosis and Renal Cell Cancer), BRCA2 (BRCA2 hereditary breast and ovarian cancer syndrome) and in the context of Werner syndrome, a rare genetic disease caused by pathogenic WRN mutations and often accompanied by various types of malignancy. However, ACC is not considered component of the tumor spectrum of the aforementioned syndromes, while evidence for biallelic inactivation in the tumor was documented only in two cases, that is, SDHC and BRCA2. Adrenal cortical carcinoma is a core malignancy of the classical tumor spectrum of Li-Fraumeni syndrome with germline TP53 mutations accounting for 50%–80% of ACC cases in the pediatric population. Nonetheless, only 4%–6% of carriers develop adrenal cortical tumors, indicative of co-operating genetic alterations in the tumorigenic process. Hotspot TP53 p.R337H mutation exerts founder effect and was found in one out of 375 individuals within a total population of  100 million in Southern and Southeastern Brazil. Those carriers exhibit variable tumor susceptibility ranging from isolated pediatric ACC cases to Li-Fraumeni or LiFraumeni-like syndromes. Genotype-phenotype correlations were reported in the setting of this pleiotropic hereditary cancer syndrome with a proposed clinical gradient of germline TP53 mutations: (i) dominant-negative missense mutations as the most severe aberrations significantly associated with early tumor onset and representing the main germline alterations in carriers who develop childhood cancers apart from ACC; (ii) nondominant-negative missense mutations which define an intermediate class with regard to clinical severity; and (iii) loss of function mutations, that is, nonsense mutations, frameshift mutations, or genomic rearrangements, as the less severe aberrations associated with later tumor onset, and mainly detected in pedigrees characterized by cancers occurring in adults. Gene alterations other than dominant-negative missense mutations were detected in the majority of pediatric ACCs; with the low-penetrant TP53 p.R158H mutation as the predominant mutation in childhood ACC and adult tumors.

Adrenal Glands Tumors: Pathology and Genetics

27

Fig. 3 Immunohistochemistry stains of a mismatch repair (MMR)-deficient adrenocortical carcinoma (A; H/E stain) including the proliferation marker Ki67 (B), p53 (C), beta-catenin (D), and the MMR panel of four markers, that is, MLH1 (E), MSH2 (F), MSH6 (G) and PMS2 (H). This tumor displays nuclear p53 and membranous beta-catenin immunoreactivity with Ki67 labeling index estimated at 22% by manual counting. As in this case a germline MSH6 mutation had been identified. MSH6 nuclear staining is absent, while nuclear immunoreactivity for MLH1, MSH2 and PMS2 is retained.

Analysis of the International Agency for Research on Cancer (IARC) TP53 database to re-evaluate age and variant-dependent tumor patterns showed that pathogenic TP53 variants in the germline have different consequences according to cell, tissue, context and age. This indicates variable penetrance according to age, sex and mutation type and temporal tumor patterns displaying distinct phases. At 0–15 years, 22% of all malignancies includes ACC, rhabdomyosarcoma, choroid plexus carcinoma, and medulloblastoma. At 16–50 years this is 51%, encompassing breast cancer, osteosarcoma, soft tissue sarcomas, leukemia, astrocytoma and glioblastoma, colorectal and lung cancer. At 51–80 years this is 27%, comprising prostate and pancreatic cancer. In the adult population, germline TP53 mutations and MMR gene mutations account for 3%–7% and 3% of ACC cases respectively. Of note is that the prevalence of Lynch syndrome among patients with ACC is comparable to that in colorectal and endometrial cancer. Irrespective of family history and/or restriction of age at diagnosis, it is recommended that all patients with newly diagnosed ACC shall be offered genetic counseling and assessed for inherited predisposition, i.e. germline testing for TP53 mutations and/or immunohistochemical analysis for MMR deficiency (Fig. 3) and subsequent molecular genetic testing for germline MMR gene mutations and/or EPCAM deletion. Given that MMR-deficient ACCs were found to be microsatellite stable, microsatellite instability analysis might not be needed as a complementary test to the IHC screening.

Microenvironment Including Immune Response Adrenocortical carcinomas are regarded as relatively pure tumors with low fractions of tumor infiltrated stromal cells and lymphocytes in cortisol-secreting tumors; the latter mainly attributable to suppressive action exerted by glucocorticosteroids.

Staging and Grading Staging TNM staging of adrenal cortical carcinoma, as introduced in the 8th edition of AJCC staging manual, is consistent with the European Network for the Study of Adrenal Tumors (ENSAT) staging system. Accordingly, stage 1 and stage 2 are defined as strictly localized tumors with a size of  5 cm or > 5 cm, respectively. Stage 3 tumors are characterized by infiltration into surrounding tissue or invading adjacent organs, that is, kidney, diaphragm, pancreas, liver or great vessels (renal vein or vena cava), or positive regional lymph nodes. Stage 4 is restricted to patients with distant metastasis. In the pediatric population, the staging system for ACCs (Table 3) is based on a modification of the work of Sandrini et al. by investigators from Children’s Oncology Group (COG) and relies on the clinical data from the International Pediatric Adrenocortical Tumour Registry.

Grading The mitotic rate has been incorporated in all multiparametric systems and/or diagnostic algorithms for evaluating the biological behaviour and also included in the grading of ACCs, further adding to the prognostic value of proliferative activity (see Prognostic

28

Adrenal Glands Tumors: Pathology and Genetics

and Predictive Biomarkers). Proliferation grading of ACCs is based on mitotic rate, that is, low grade  20 mitoses per 50 high-power fields and high grade > 20 mitoses per 50 high-power fields. The TCGA study measured adrenal cortical differentiation using a single metric, namely Adrenocortical Differentiation Score (ADS), based on 25 genes, which are of importance for adrenal function, i.e. cholesterol transporters, steroidogenic enzymes and their transcriptional regulator SF1, and with very high expression levels in the adult adrenal cortex. Of interest, higher ADS values were noted in functional tumors, but without any association with histopathological parameters, that is, Weiss score.

Prognostic and Predictive Biomarkers The prognosis of ACC is poor, but heterogeneous, with a 5-year overall survival rate varying between 37% and 47%. The resection status and the proliferation marker Ki67 constitute the most relevant prognostic parameters in both localized and advanced ACCs. Modified ENSAT stages (stages IVa, IVb, or IVc) along with G R A S parameters [Grade (Weiss > 6 and/or Ki67  20%), resection status, age  50 years and tumor- or hormone-related symptoms] correlated with overall survival in stage III–IV ACCs. Clinically relevant hypercortisolism has an adverse effect on survival in patients with completely resected ACC. Surgical margin status has an impact on long-term outcome, roughly two-thirds of ACC patients experience disease recurrence. Patients with combined locoregional and distant recurrence have worse survival than those with distant-only or locoregional-only recurrence. Some studies confirmed Ki67 as a powerful tool for prognostic stratification, and it has been integrated into a model for prediction of metastatic disease. However, preanalytical variation, lack of standardized assessment methods and significant levels of interobserver variation pose important challenges. This is reflected in the variable threshold (10%–30%) used to incorporate the parameter into treatment flow charts to guide therapy decision making. In the pediatric setting, Ki67 proliferation marker has recently emerged as a strong prognostic indicator with Ki67 LI  15% independently associated with poor prognosis. In an effort to refine prognostic prediction in ACC, it has been recently shown that combined assessment of Ki67 LI and VAV2 expression improved patient stratification to low-risk and high-risk groups. VAV2 is a guanine nucleotide exchange factor for small GTPases that control the cytoskeleton, and that is driven by increased expression of the gene encoding SF-1 (also known as NR5A1) in ACC. SF1 is a marker of adrenal cortical derivation, but also of biological aggressiveness as reflected in immunohistochemical marker staining patterns. Other proliferation-based (immunohistochemical) scoring methods and/or routine histology, that is, assessment of mitotic activity, angioinvasion, and Weiss score, can provide valuable aid in determining prognosis. In addition to the aforementioned methods, -omics derived parameters seem very promising to recognize aggressive behaviour in adrenocortical cancer. Two distinct molecular groups have been identified: the C1A subgroup with poor outcome, characterized by numerous mutations as well as alterations of DNA methylation, and the C1B subgroup with good prognosis, exhibiting specific deregulation of two microRNA clusters. Furthermore, three ACC subtypes have been identified with distinct clinical outcome, based solely on integrated multi-platform molecular profiling: Cluster of Cluster (CoC) I/II/III. CoC I ACCs showed better outcome, significantly unregulated genes in immune-mediated pathways and lower Ki67 expression levels, whereas CoC III tumors showed dismal outcome, significant disregulation of genes in mitotic pathways and higher Ki67 defined proliferative activity. CoC II ACCs had more heterogeneous outcome. A recent TCGA study showed that high mutation density and a specific copy-number phenotype, characterized by a high number of chromosomal breaks and frequent loss of 1p (with 1q intact), are associated with aggressive disease. Whole genome doubling, associated with TERT expression along with telomere shortening, predict dismal outcome in ACC. This is consistent with recent immunohistochemical data suggesting a better prognosis of ACCs with loss of expression of the telomere regulator DAXX. Methylation profiles support a correlation between CpG island hypermethylation and poor survival. Tumor DNA methylation has recently emerged as independently prognostic for disease-free survival (along with stage) and overall survival (along with stage and Ki67 proliferative index) in ACC. miRNA expression signatures (high levels of miR-483-5p and/or low levels of miR-195) have been reported to predict worse overall survival. Expression of TOP2A, EZH2 and BARD1, identified through gene expression studies, might provide additional prognostic information. Overexpression of histone methyltransferase EZH2, a result of deregulated P53/ RB/E2F pathway activity, has been associated with increased proliferation and poor prognosis. With regard to pediatric adrenocortical cancer, recent National Cancer Data Base (NCDB) and European Cooperative Study Group on Pediatric Rare Tumors (EXPeRT) studies showed that age  4 years, tumor size  10.0 cm in diameter, extension of primary disease into adjacent structures, metastatic disease, and margin status were significantly associated with poor long-term survival. Furthermore, distant metastatic disease and large tumor volume > 200 cm3 were the main unfavorable prognostic factors. Of note, a cut-off  10.0 cm in tumor diameter, corresponding to a tumor volume of 524 cm3, was found to be of prognostic relevance, consistent with data generated by the Surveillance, Epidemiology, and End Results (SEER) program study on pediatric ACC. This challenges the 200 cm3 threshold currently used in the COG modification of the Sanrini et al. staging system. In pediatric cases, concomitant germline TP53 and somatic ATRX mutations as well as associated genomic aberrations, such as massive structural variations and/or frequent background mutations, are associated with a dismal outcome. Such cases showed high tumor weight and advanced disease (COG stage III/IV). Glucose transporter 1 (GLUT1) was found to be more strongly expressed in clinically malignant pediatric adrenocortical tumors and associated with shorter overall and disease-free survival. Of note, metabolic reprogramming towards a hyperglycolytic and acid-resistant phenotype is also found in adult cases, with GLUT1 emerging as a stage-independent predictor of outcome. Accordingly, ACC patients with a high metabolic tumor volume, total lesion glycolysis and maximum standardized uptake value by 18F-FDG PET/CT prior to treatment, had worse overall survival.

Adrenal Glands Tumors: Pathology and Genetics

29

Human cytochrome P450 2W1 (CYP2W1), ribonucleotide reductase large subunit 1 (RRM1) and sterol-O-acyl-transferase 1 (SOAT1) might represent predictive markers for the response to mitotane. This agent confers adrenal-specific cytotoxicity and downregulates steroidogenesis. However, mitotane is significantly toxic. Topoisomerase II alpha (TOP2A) and thymidylate synthase (TS) were not found predictive of mitotane efficacy in ACC patients. These markers await further validation in prospective studies.

Further Reading Assié, G., Letouzé, E., Fassnacht, M., Jouinot, A., Luscap, W., Barreau, O., et al., 2014. Integrated genomic characterization of adrenocortical carcinoma. Nature Genetics 46 (6), 607–612. Bausch, B., Schiavi, F., Ni, Y., Welander, J., Patocs, A., Ngeow, J., et al., 2017. Clinical characterization of the pheochromocytoma and paraganglioma susceptibility genes SDHA, TMEM127, MAX, and SDHAF2 for gene-informed prevention. JAMA Oncology 3 (9), 1204–1212. Carney, J.A., 2013. Carney triad. Frontiers of Hormone Research 41, 92–110. Castro-Vega, L.J., Lepoutre-Lussey, C., Gimenez-Roqueplo, A.P., Favier, J., 2016. Rethinking pheochromocytomas and paragangliomas from a genomic perspective. Oncogene 35 (9), 1080–1089. Castro-Vega, L.J., Letouzé, E., Burnichon, N., Buffet, A., Disderot, P.H., Khalifa, E., et al., 2015. Multi-omics analysis defines core genomic alterations in pheochromocytomas and paragangliomas. Nature Communications 6, 6044. Dahia, P.L., 2014. Pheochromocytoma and paraganglioma pathogenesis: Learning from genetic heterogeneity. Nature Reviews Cancer 14 (2), 108–119. de Krijger, R.R., Papathomas, T.G., 2012. Adrenocortical neoplasia: Evolving concepts in tumorigenesis with an emphasis on adrenal cortical carcinoma variants. Virchows Archiv 460 (1), 9–18. de Reyniès, A., Assié, G., Rickman, D.S., Tissier, F., Groussin, L., René-Corail, F., et al., 2009. Gene expression profiling reveals a new classification of adrenocortical tumors and identifies molecular predictors of malignancy and survival. Journal of Clinical Oncology 27 (7), 1108–1115. Else, T., Kim, A.C., Sabolch, A., Raymond, V.M., Kandathil, A., Caoili, E.M., et al., 2014. Adrenocortical carcinoma. Endocrine Reviews 35 (2), 282–326. Favier, J., Amar, L., Gimenez-Roqueplo, A.P., 2015. Paraganglioma and phaeochromocytoma: From genetics to personalized medicine. Nature Reviews Endocrinology 11 (2), 101–111. Fishbein, L., Leshchiner, I., Walter, V., Danilova, L., Robertson, A.G., Johnson, A.R., et al., 2017. Comprehensive molecular characterization of pheochromocytoma and paraganglioma. Cancer Cell 31 (2), 181–193. Giordano, T.J., Kuick, R., Else, T., Gauger, P.G., Vinco, M., Bauersfeld, J., et al., 2009. Molecular classification and prognostication of adrenocortical tumors by transcriptome profiling. Clinical Cancer Research 15 (2), 668–676. Huang KL, Mashl RJ, Wu Y, Ritter DI, Wang J, Oh C, et al. (2018) Pathogenic germline variants in 10,389 adult cancers. Cell 173(2): 355–370.e14. Jimenez, C., Libutti, S.K., Landry, C.S., et al., 2017. AdrenaldNeuroendocrine tumours. In: AJCC cancer staging manual, 8th edn. Springer International Publishing. Lloyd, R.V., Osamura, R.Y., Kloppel, G., Rosai, J. (Eds.), 2017. WHO classification of tumours of endocrine organs, 4th edn. IARC, Lyon. Papathomas, T.G., de Krijger, R.R., Tischler, A.S., 2013. Paragangliomas: Update on differential diagnostic considerations, composite tumors, and recent genetic developments. Seminars in Diagnostic Pathology 30 (3), 207–223. Phan, A.T., Grogan, R.H., Rohren, E., Perrier, N.D., 2017. Adrenal cortical carcinoma. In: AJCC cancer staging manual, 8th edn. Springer International Publishing. Pinto, E.M., Chen, X., Easton, J., Finkelstein, D., Liu, Z., Pounds, S., et al., 2015. Genomic landscape of pediatric adrenocortical tumors. Nature Communications 6, 6302. Zheng, S., Cherniack, A.D., Dewal, N., Moffitt, R.A., Danilova, L., Murray, B.A., et al., 2016. Comprehensive pan-genomic characterization of adrenocortical carcinoma. Cancer Cell 29 (5), 723–736.

Aflatoxins Joshua W Smith and John D Groopman, Johns Hopkins University, Baltimore, MD, United States © 2019 Elsevier Inc. All rights reserved.

Introduction As one of the most potent naturally occurring known carcinogens, aflatoxins represent a special case in the study of cancer etiology (Squire, 1981). Research spanning more than five decades has created a detailed corpus of histological, biochemical, and molecular toxicology, which comprehensively and convincingly characterizes the carcinogenic mechanisms following aflatoxin exposure (Kensler et al., 2011; Wogan et al., 2012). These efforts have produced a suite of non- and minimally-invasive biomarkers, the use of which have allowed the mechanistic relation of aflatoxin exposures to internal dose, DNA adduct formation, and a signature carcinogenic mutation in the tumor suppressor p53. Harnessing these tools, epidemiological studies have consistently demonstrated a significantly increased risk of hepatocellular carcinoma (HCC) from dietary aflatoxin exposure (Liu et al., 2012) and, most dramatically, a synergistic increase in risk for those also chronically infected with hepatitis B virus (HBV) (Wild et al., 2009). Consequently, out of total global HCC incidence (nearly 800,000 new cases per year (Ferlay et al., 2015)), it is estimated that approximately one quarter is attributable to aflatoxin exposure, either alone or in combination with chronic viral hepatitis (Liu et al., 2010, 2012). The International Agency for Research on Cancer (IARC) declared aflatoxins, as a group, to be human carcinogens in 1987 (International Agency for Research on Cancer, 1987), a classification that has been confirmed in each of three subsequent reviews (International Agency for Research on Cancer, 1993, 2002, 2012). Concordantly, cancer prevention efforts have benefitted from the same half-century of molecular toxicology that enabled the determination of aflatoxins’ carcinogenicity. Biomarkers of aflatoxin exposure, with their mechanistically defined relationships to HCC development, unshackled scientists from the necessity to directly link exposures to cancer outcomes, greatly reducing the time and resources required to test novel preventative measures. As a result, a wide array of promising primary and secondary prevention strategies have been identified (Groopman et al., 2008; International Agency for Research on Cancer, 2015), some of which have now been associated with reductions in HCC mortality (Chen et al., 2013). The goals of this chapter are not only to describe the carcinogenic effects of aflatoxin exposure and to recount major discoveries in aflatoxin research, but also to highlight throughout the strength and breadth of this body of literature.

Sources of Aflatoxin Exposure Aflatoxins are secondary metabolites produced by members of the fungal genus Aspergillus, which contains approximately 250 known species (Klich, 2007). However, less than a dozen are known to produce aflatoxins (International Agency for Research on Cancer, 2002; Klich, 2007; Bennett et al., 2003) and only A. flavus and A. parasiticus are truly relevant to human health through contamination of the food supply. A. flavus is found in soils across the globe, but is more common in tropical and sub-tropical climates between latitudes 35 degrees north and south of the equator (Klich, 2007; Abbas et al., 2009; Council for Agricultural Science and Technology, 2003). Regardless, due to globalized trade, dietary aflatoxins are a worldwide concern. A major distinction between these two aflatoxigenic species are their profiles of toxin productiondwhile A. parasiticus produces both the B and G series of aflatoxins, A. flavus produces only B toxins (Dorner et al., 1984; Klich et al., 1988). Many factors can influence the crop colonization, growth, and toxin production of Aspergillus species, notably heat, humidity, pest or environmental host stressors, and post-harvest practices (Abbas et al., 2009; Pitt et al., 2013; Diener et al., 1987). Maize, groundnuts, cottonseed, and tree nuts represent the most common sources of food supply contamination worldwide, as these crops are preferred by Aspergillus for colonization pre-harvest, while also being susceptible to contamination due to improper drying and storage conditions post-harvest. However, although other crops like wheat, sorghum, and rice are much less susceptible to preharvest colonization, poor post-harvest handling can also encourage aflatoxin contamination in these stored commodities and levels of contamination vary widely (International Agency for Research on Cancer, 2002; Council for Agricultural Science and Technology, 2003).

Aflatoxin Discovery and Structural Elucidation The aflatoxins were first discovered as a result of their role in the acute intoxication and death of many thousand British poultry in 1960 (Blount, 1961). The toxic outbreaks were linked to moldy groundnut meal used in animal feed (Blount, 1961) and crude extracts revealed the presence of a potent substance produced by A. flavus, which was able to recapitulate the lethal and histological effects of the original toxic feed (Sargeant et al., 1961). The carcinogenic potential of the as-yet-unidentified agent was quickly suspected, however, as liver tumors were found in rats fed the toxic groundnut meal (Lancaster et al., 1961). Crude extracts were revealed to contain at least two primary constituents (van der Zijden et al., 1962; De Iongh et al., 1962), which, given their mycological source and intense blue or green fluorescence, were designated as aflatoxin (A. flavus toxin) B and G, respectively (Nesbitt et al., 1962). Büchi, Wogan, and colleagues proposed the structures of AFB1 and AFG1 in 1963 (Asao et al., 1963), followed by

30

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65030-4

Aflatoxins

31

a longer report 2 years later (Asao et al., 1965), in which they identified the structures of AFB2 and AFG2. In 1967, Büchi’s group produced the first total synthesis of AFB1 (Büchi et al., 1967). The structures of the naturally occurring aflatoxins are shown in Fig. 1.

Aflatoxin Metabolism and Molecular Mechanisms of Carcinogenicity Human exposure to aflatoxins occurs through the diet, however, these parent compounds are not carcinogenic per se. Rather, as procarcinogens, they must first be transformed in vivo to their carcinogenic derivative (Smith et al., 2016; Rendic et al., 2012). The ultimate carcinogenic metabolite of AFB1, AFB1-exo-8,9-epoxide, is produced in the liver through oxidation of AFB1 by cytochrome P450 (CYP) enzymes. If not conjugated by GSTs, the AFB1-exo-8,9-epoxide intercalates with DNA, facilitating electrophilic attack on guanine residues at the N7 position, leading to mutagenesis.

Structure–Activity Relationships Driving Aflatoxin Carcinogenicity Mixtures of the four naturally occurring aflatoxins (AFB1, AFG1, AFB2, and AFG2) represent the relevant form of aflatoxin exposure for humans and have been classified by IARC as a Group 1 carcinogen, indicating sufficient evidence for carcinogenicity in humans (Ostry et al., 2017). However, while carcinogenic as mixed species (Carnaghan, 1967; Norred et al., 1983), in their purified forms, these four aflatoxins are not equally potent. In bacterial mutagenicity assays, with AFB1 as the reference, AFG1 is 3% as potent, and both AFB2 and AFG2 retain < 1% activity (Wong et al., 1976). In general, carcinogenesis studies agree both with this order and the relative decrements in potency, to the point where AFG2 is essentially non-toxic (Wogan et al., 1971, 1974a; Bailey et al., 1988; Ayres et al., 1971; Butler et al., 1969). AFB1, by far the most carcinogenic of the group, displays a dose–response in carcinogenicity in rats down to 1 ppb in the diet (
0.2 mg/kg bw) with long-term dietary exposure (Wogan et al., 1974a), while very large studies in trout suggest that it may retain carcinogenic potency well below the 1 ppb level (Williams et al., 2009). Single large doses of purified AFB1 cause substantial rates of hepatocarcinogenesis in rats (Carnaghan, 1967), while even a single 15 minute exposure to 0.5 ppm AFB1 is carcinogenic in trout (Bailey et al., 1988). Appreciation of critical structural differences between these similar compounds reveals structure–activity relationships responsible for their differing carcinogenicities. As shown in Fig. 1, all four naturally occurring aflatoxins share a common structural featuredthe methoxylated difuranocoumarin backbone. Very strong biochemical evidence supports the 8,9-exo-epoxide (Fig. 2) as the ultimately carcinogenic DNA-adducting metabolite of AFB1 (Swenson et al., 1973, 1974, 1977) and its generation depends on the oxidation potential of the 8,9 carbon–carbon bond of the outermost furan ring (Guengerich et al., 1998) (or the 9,10 bond in G series aflatoxins). A blunt illustration of this requirement is provided by synthetic analogues entirely lacking the furofuran moiety (e.g., 5,7-dimethoxycyclopentenone coumarins)dthese agents are completely inactive as liver carcinogens in the trout (Ayres et al., 1971) or at high doses in the rat (Wogan et al., 1971). Comparison of AFB2 and AFB1 provides a more relevant perspective. Saturation of the 8,9 double bond in AFB1 significantly reduces the carcinogenicity of AFB2: doses of AFB2 115-fold greater than that of AFB1 induced fewer tumors in rats (Wogan et al., 1971), while equivalent doses of AFB1 and AFB2 in trout induced liver tumors at drastically different rates (78% vs. 5%) (Ayres et al., 1971). Molecular evidence supports these studies, showing that administration of radiolabeled AFB2 to rats results in hepatic DNA adduct formation at levels approximately 1% that of an equivalent dose of radiolabeled AFB1 (Swenson et al., 1977).

Fig. 1

Structures of the naturally occurring aflatoxins.

32

Aflatoxins

Fig. 2

AFB1 metabolism.

Oxidation of AFB1 by cytochrome P450s produces two stereoisomers: the AFB1-8,9-exo- and AFB1-8,9-endo-epoxides (Guengerich et al., 1998; Dohnal et al., 2014). However, the exo epoxide is
103 times as mutagenic as the endo (Iyer et al., 1994; Ueng et al., 1995), which exhibits negligible reactivity towards double-stranded DNA (Iyer et al., 1994). Intercalation of the 8,9-epoxide into DNA dramatically enhances adductiondsingle-stranded DNA is 6.5-fold less reactive with the epoxide than is double-stranded DNA (Johnson et al., 1997a), while adduction to free deoxyguanosine is 2000-fold less efficient (Brown et al., 2009). While both the exo- and endo-epoxides intercalate into double-stranded DNA (Gopalakrishnan et al., 1989, 1990), structural differences result in differing rates of adduct formation. After intercalation, the oxirane moiety of the exo-epoxide faces away from the guanine N7, providing optimal orientation for SN2 attack (Iyer et al., 1994). In contrast, the position of the endo-epoxide precludes the reaction, and consequently, adduction. Secondly, calculations from molecular modeling suggest that steric hindrance from solvating water molecules during the reaction with guanine is systematically greater in the endo configuration than the exo (Okajima et al., 2000; Bren et al., 2007). These mechanisms likely explain why the exo-epoxide is a highly potent carcinogen, while the endo isomer is non-toxic. Finally, additional structural differences between aflatoxin species affects their intercalation into DNA and, thus, carcinogenic potency. Compared to the cyclopentenone ring of AFB1 and AFB2, the lactone substituent of the G series aflatoxins is less planar (Cheung et al., 1964); as a result, AFG1 and AFG2 intercalate into DNA with an affinity approximately one order of magnitude lower than AFB1 and AFB2 (Raney et al., 1990). AFB1-8,9-epoxide forms more guanine adducts than its equivalent AFG1 epoxide at equivalent concentrations, and, accordingly, AFG1 has been shown to be less carcinogenic than AFB1 in both rats (Wogan et al., 1971) and trout (Ayres et al., 1971). To summarize, saturation of the furofuran ring is the largest determinant of carcinogenic activity and separates the relatively nontoxic AFB2 and AFG2 from the carcinogenic AFB1 and AFG1. Further refinement in potency is provided by substitution of the cyclopentenone ring for a lactone ring, resulting in the reduced intercalation efficiency and DNA adduction of AFG1 compared to AFB1.

Aflatoxin Metabolism Phase I activation As with procarcinogen metabolism more generally (Rendic et al., 2012), the P450s are the major players in aflatoxin oxidation (Guengerich et al., 1998; Dohnal et al., 2014). Multiple CYPs have been shown to activate AFB1 (Pelkonen et al., 1997; Aoyama et al., 1990), but CYP3A4 appears to be the enzyme of greatest relevance for AFB1 epoxidation in humans (Guengerich et al., 1998; Dohnal et al., 2014; Ueng et al., 1995; Shimada et al., 1989); one report in human liver microsomes suggests that CYP3A4 accounts for 79%–95% of total AFB1-8,9-epoxide formation (Kamdem et al., 2006). This has been somewhat controversial, though, as some epidemiological data suggests that polymorphisms in CYP1A2 modify HCC risk (Chen et al., 2006), and experimental reports suggested CYP1A2 as the dominant enzymatic source of AFB1 epoxidation, rather than 3A4 (Gallagher et al., 1994, 1996). However, CYP1A2 hepatic protein content is 10–25 times lower than CYP3A4 and, even granting the former an advantage in reaction rate, 1A2 is likely far outstripped by 3A4 in total AFB1 epoxidation capacity (Kamdem et al., 2006). Moreover, while CYP3A4 produces only the mutagenic AFB1-8,9-exo-epoxide but not the non-toxic endo-epoxide, CYP1A2-derived AFB1 epoxides are split roughly equally between the exo and endo stereoisomers (Ueng et al., 1995). Finally, higher contribution of other CYP enzymes to

Aflatoxins

33

AFB1 epoxidation has been shown to depend on concurrently low CYP3A4 activity (Kamdem et al., 2006; Wojnowski et al., 2004). Genetic differences may impact AFB1 epoxidation and risk of carcinogenesis. Several reports have demonstrated very high interindividual variation in hepatic P450 protein contentdup to 57-fold in CYP3A4 (Kamdem et al., 2006; Doi et al., 2002; Kirby et al., 1993). Accordingly, microsomal AFB1 epoxide formation exhibits high variability (Kamdem et al., 2006) and correlates with CYP3A4 expression (68). Male rats fed AFB1 demonstrated two- to threefold higher hepatic microsomal P450 content and AFB1DNA adduct formation than AFB1-fed female rats, but castration of males lowered these values to those in females, while testosterone administration in castrated males rescued values to those of intact males (Gurtoo et al., 1976). A similar result was shown with hypophysectomy (Swenson et al., 1977; Goodall et al., 1969), while testosterone administration to female rats resulted in a phenotype very similar to that of intact males (Gurtoo et al., 1976). In the analysis of human hepatic microsomes by Kamdem et al., males exhibited a median rate of aflatoxin epoxidation nearly 50% higher than females (Kamdem et al., 2006). These mechanisms may explain, at least in part, the epidemiological evidence showing that men are at two- to threefold higher risk of HCC incidence than women (Petrick et al., 2016). Once formed, the 8,9-exo-epoxide is highly unstable in solution (t½ z 1 s), rapidly undergoing hydrolysis to the 8,9-dihydrodiol (Johnson et al., 1996) (Fig. 2). Hydrolysis of the carcinogenic exo-epoxide is catalyzed enzymatically by epoxide hydrolase, but in vitro evidence suggests that only supra-physiological expression of the enzyme would be able to provide added protection from genotoxicity (Johnson et al., 1997b). Despite this, epidemiological studies have revealed elevated AFB1 adduct levels (McGlynn et al., 1995; Dash et al., 2007) or an increased risk of HCC in individuals carrying mutant alleles of the enzyme (McGlynn et al., 1995; Tiemersma et al., 2001). An alternate possibility is that epoxide hydrolase status may impact HCC risk independently from aflatoxin metabolism (Tiemersma et al., 2001; Rahat et al., 2012).

Major metabolites of AFB1 AFB1 is also detoxified through conversion to less-potent metabolites (Fig. 2). CYP3A4 indeed produces the AFB1-8,9-exo-epoxide, but also catalyzes hydroxylation of the cyclopentenone ring to yield aflatoxin Q1 (AFQ1). The yield for each of these products is vastly different, with production of AFQ1 being up to 8-fold greater than the epoxide (Ueng et al., 1995; Wang et al., 1998). Similarly, another major metabolite, AFM1, is produced though 9a-hydroxylation by CYP1A2, at a ratio of up to 5:1 AFM1:epoxide (Ueng et al., 1995). Additionally, AFB1 can be demethylated to aflatoxin P1 (AFP1) (Dalezios et al., 1971, 1973) or subjected to ketoreduction at the cyclopentenone ring to form aflatoxicol. Notably, these major detoxification products (AFM1, AFQ1, AFP1, and aflatoxicol) all retain the crucial 8,9 double bond and thus themselves can be epoxidized at this site, capable of DNA adduct formation (Bujons et al., 1995; Neal et al., 1998; Eaton et al., 1988; Essigmann et al., 1980). However, it appears that very little epoxidation occurs relative to AFB1 (Neal et al., 1998; Eaton et al., 1988), limiting their mutagenicity. Indeed, while AFM1 is carcinogenic in its own right (International Agency for Research on Cancer, 2002; Hsieh et al., 1984; Cullen et al., 1987; Wogan et al., 1974b), AFM1, AFQ1, and AFP1 are all at least 10 times less potent than AFB1 (Wong et al., 1976). An exception is aflatoxicol, which retains
20%–50% the potency of its parent (Wong et al., 1976; Garner et al., 1972). Interestingly, although the epoxide of aflatoxicol can form DNA adducts in vitro, aflatoxicol-DNA adducts are undetectable after dietary aflatoxicol treatment in vivo, while AFB1 adducts are readily observed (Bailey et al., 1994). This, as well as the high potency of aflatoxicol relative to the other major AFB1 metabolites, can be explained by the in vivo conversion of aflatoxicol back to AFB1 (Salhab et al., 1976). In total, nearly 90% of cytochrome P450 oxidation products may be diverted away from epoxidation during initial metabolism of AFB1 (Kamdem et al., 2006).

Phase II conjugation Phase II detoxification involves the conjugation of phase I-activated electrophiles to endogenous nucleophiles, to create polar metabolites that may be safely passed in the urine or excreted through biliary secretion into feces. In general, most of this work is performed through hepatic glucuronidation by UDP-glucuronosyl transferases (UGT) or adduction to glutathione by glutathione S-transferase (GST) enzymes (Rendic et al., 2012). As discussed above, the bulk of ingested AFB1 is converted to its hydroxylated metabolites (AFM1, AFQ1, AFP1), which are subsequently glucuronidated or sulfonated and excreted (Dalezios et al., 1973; Wei et al., 1985a, b; Wogan et al., 1967; Raj et al., 1984). However, as for the carcinogenic epoxide, GST-dependent conjugation to glutathione (Fig. 2) appears to be the most critical phase II detoxification pathway (Appleton et al., 1982; Monroe et al., 1988; Lotlikar et al., 1980; Raj et al., 1986), as well as a significant source of variation for between- and within-species aflatoxin sensitivity (Dohnal et al., 2014; Hayes et al., 1991). With the exceptions of GSTM1 and GST1, most human GST isoforms do not conjugate the AFB1 epoxide well, or at all (Raney et al., 1992; Johnson et al., 1997c). However, primary hepatocytes from GSTM1-positive subjectsdbut not GSTM1-nulldhave been shown to be competent in glutathione conjugation of AFB1-8,9-epoxide (Langouët et al., 1995; GrossSteinmeyer et al., 2010), GSTM1-null individuals were more likely than GSTM1-positive to have elevated AFB1 adduct levels (McGlynn et al., 1995), and GSTM1 expression and GST conjugating activity have been shown to be lower in liver tumors than neighboring normal tissue (Kirby et al., 1993). Accordingly, GST polymorphismsdespecially in GSTM1dhave been shown to increase HCC risk in numerous epidemiological studies, particularly in the context of high aflatoxin exposure (Sun et al., 2001; Kirk et al., 2005a; Long et al., 2006; Shen et al., 2014). Due to metabolic differences throughout the aflatoxin detoxification pathway (Dohnal et al., 2014), animal species exhibit substantial variability in sensitivity to aflatoxin (Newberne et al., 1969). However, the capacity to detoxify the AFB1-epoxide through conjugation to glutathione appears to be the largest determinant of a species’ sensitivity to aflatoxins, and thus the

34

Aflatoxins

most crucial of the metabolic idiosyncrasies between clades. Beginning with the paper from Degen et al. (1981), many reports have characterized the divergent susceptibilities of adult rats (sensitive) and mice (resistant) to be the result of a striking difference in AFB1-epoxide detoxification capacity (Monroe et al., 1987, 1988; O’Brien et al., 1983). Subsequent interspecies comparisons revealed that the significant kinetic advantage of murine hepatic cytosolic fractions over those from rats (7–50-fold) and humans (300–3500-fold) (Raney et al., 1992; Slone et al., 1995) was not simply due to greater enzymatic abundance, but that reaction rates for purified mouse GSTs were orders of magnitude greater than their rat and human homologues (Raney et al., 1992; Johnson et al., 1997c; Buetler et al., 1992). However, differences in GST expression may impact aflatoxin sensitivity within a speciesdrelative to male rats, female rats are more resistant to aflatoxin-induced DNA adduction (Gurtoo et al., 1976), apparently due to 10-fold higher expression of GSTA5 (Hayes et al., 1994) and greater epoxide detoxification (Degen et al., 1981). Additionally, temporal shifts in aflatoxin metabolism may dramatically impact the sensitivity of an organism throughout its development. While adult mice are resistant to aflatoxin-induced carcinogenesis, infant mice have long been known to be much more sensitive (Vesselinovitch et al., 1972). Recent work has shown that prenatal/neonatal AFB1 exposure induced 20 times more mutations than exposures during adulthood (Chen et al., 2010; Chawanthayatham et al., 2015). Moreover, Shupe et al. have shown that murine hepatic GST content rose fivefold from 4 days postnatal to the end of the first year of life, while DNA adducts from AFB1 gavage decreased 13-fold over the same period (Shupe et al., 2004).

Targeting Nrf2 for chemoprotective modulation of aflatoxin metabolism Nuclear factor erythroid 2 like 2 (Nrf2) is a crucial regulator of the endogenous antioxidant response, which protects cells from oxidative damage by reactive oxygen species and electrophiles (Suzuki et al., 2015). This is accomplished through binding of Nrf2 to the antioxidant response element (ARE) motif within the promoter regions of phase II detoxifying enzymes, including GSTs (Itoh et al., 1997; Nguyen et al., 2000; Kensler et al., 2007). Indeed, loss of Nrf2 in rodents increases sensitivity to chemical carcinogenesis (Ramos-Gomez et al., 2001; Taguchi et al., 2016), while stabilization of Nrf2 through disruption of its negative regulator, Keap1, results in increased expression of cytoprotective Nrf2-target genes (Thimmulappa et al., 2002; McWalter et al., 2004; Yates et al., 2007). Mice lacking the Nrf2 target GSTA3 (Jowsey et al., 2003) are rendered hypersensitive to dietary AFB1 treatments, with AFB1-DNA adducts accumulating to levels > 100-fold greater than in wild-type mice (Kensler et al., 2014; Ilic et al., 2010). However, neither simultaneous deletion of Keap1 nor pharmacological activation of Nrf2 was able to rescue adduct levels in GSTA3/ mice back to control levels, despite the presence of an active Nrf2 signaling pathway (Kensler et al., 2014). Similar approaches have been used to identify chemopreventative strategies for translation to at-risk human populations. Studies with oltipraz, a dithiolthione previously studied as an antischistosomal agent, have demonstrated it to be an inducer of GST expression and activity (Davidson et al., 1990; Kensler et al., 1985, 1987; Primiano et al., 1995), an inhibitor of AFB1-DNA adduct formation (Kensler et al., 1985, 1987), and protective against AFB1-induced hepatocarcinogenesis (Kensler et al., 1997). While oltipraz does appear to be a modulator of cytochrome P450-dependent AFB1 oxidation (Langouët et al., 1995; Kensler et al., 1985, 1987), DNA adduction by AFB1 was inversely and highly correlated with induction of GST activity (Kensler et al., 1985, 1987), suggesting this to be the predominant effect on AFB1 metabolism. Another inducer of carcinogen detoxification, 1[2-cyano-3-,12-dioxooleana-1,9(11)-dien-28-oyl]imidazole (CDDO-Im), has been shown to exhibit exceptional chemoprotective effects in animals exposed to AFB1. Demonstrating that CDDO-Im acts through Nrf2, Taguchi et al. reported that in wild-type ratsdbut not Nrf2-knockoutsdCDDO-Im potently induced Nrf2, GSTA3, and GSTA5, while reducing hepatic AFB1-DNA adducts (Taguchi et al., 2016). Others have shown that CDDO-Im dose-dependently induced GST expression and inhibited hepatic AFB1DNA adduct formation (Yates et al., 2006), while another report revealed complete protection against AFB1-induced hepatocarcinogenesis with simultaneous administration of CDDO-Im (Johnson et al., 2014). Finally, natural agents, such as the isothiocyanate sulforaphane, are able to provide protection against aflatoxin exposure through the same Nrf2-dependent mechanism (Kensler et al., 2013; Wakabayashi et al., 2004). Although not as potently as CDDO-Im, sulforaphane induces phase II detoxification genes (Hu et al., 2004) in an Nrf2-dependent manner (Thimmulappa et al., 2002) and inhibits the formation of mutagenic AFB1-DNA adducts (Fiala et al., 2011).

DNA Adduct Formation Prior to DNA adduction, the AFB1-8,9-exo-epoxide must remain stable long enough to enter the nucleus and interact with DNA after formation in the endoplasmic reticulum by P450s. Given the rapid solvolysis of the AFB1-exo-epoxide in aqueous solution (Johnson et al., 1996), a pertinent consideration is a comparison of the rate of epoxide hydrolysis with its anticipated rate of encounter with DNA. Based on estimates of diffusion rate and distance within the cell, Johnson et al. have calculated the epoxide’s rate of interaction with DNA to be roughly 1011-times greater than its spontaneous degradation (Johnson et al., 1997a). As described above in the section “Structure–activity relationships driving aflatoxin carcinogenicity”, intercalation of the AFB1-8, 9-exo-epoxide into doublestranded DNA facilitates SN2 attack on the N7 position of guanine residues; specifically, Kobertz et al. have shown that this intercalation must occur to the 5’ side of guanine (Kobertz et al., 1997). The AFB1-N7-guanine adduct (N7-guanine; Fig. 3) was first identified in vitro by Wogan and colleagues (Essigmann et al., 1977), followed by their demonstration of its dose-dependent formation in vivo (Croy et al., 1978). Due to its positively charged imidazole ring, the N7-guanine DNA adduct is unstable and susceptible to either spontaneous depurination or base-catalyzed hydrolysis and imidazole ring-opening (Fig. 3). In the former case, this produces an apurinic (AP) site and the free AFB1-N7-guanine adduct, which is excreted and can be detected in the urine (Bennett

Aflatoxins

Fig. 3

35

Formation of major AFB1 adducts.

et al., 1981). Spontaneous depurination is not uncommon in mammalian cells in general (Loeb et al., 1986) and dose–response experiments in rats suggest that approximately 30% of the initial AFB1-N7-guanine DNA adducts may be depurinated and excreted within 48 hours of initial dosing (Bennett et al., 1981). Regarding base-catalyzed ring-opening, hydrolysis of the AFB1-N7-guanine imidazole ring results in formation of an AFB1-formamidopyrimidine (FAPyr) adduct that is much more stable than the initial N7guanine lesion. This accounts for the observed temporal shifts in adduct profiles in AFB1-perfused livers or single AFB1 doses in vivo. Two hours after dosing, N7-guanine adducts represent > 80% of total AFB1 adducts (at a rate of
10 AFB1-N7-guanine per 1011 nucleotides), while FAPyr adducts (
1 per 1011) account for less than 10% (Essigmann et al., 1980; Croy et al., 1981a, b). However, the chemical instability of the N7-guanine DNA adduct and the persistence of the FAPyr adduct results in the predominance of FAPyr at 24 hours and a reversal of the initial profile by 48–72 hours (Croy et al., 1981a; Hertzog et al., 1980). This shift is delayed with repeated AFB1 exposuredfor example, dosing to rats over 14 days (Croy et al., 1981a)dbut eventually results in an accumulation of FAPyr at
80% of total DNA adducts.

36

Aflatoxins

Adduct Repair in DNA Enzymatic repair of AFB1 lesions in DNA is carried out through both nucleotide excision repair (NER) and base excision repair (BER) mechanisms. Typically, NER is considered to act on a variety of bulky, helix-distorting lesions, with BER utilized for small lesions, such as AP sites (Bedard et al., 2006). However, both NER and BER have been shown to contribute to N7-guanine and FAPyr adduct repair. In human fibroblasts treated with AFB1, deficiency of the NER gene XPA resulted in lower efficiency of N7guanine adduct repair (Leadon et al., 1981) and higher rates of mutation frequency (Levy et al., 1992), while AFB1-dosed Xpa/ mice had greater hepatic carcinoma incidence and multiplicity than similarly treated wild-type controls (Takahashi et al., 2002). While the FAPyr adduct has been shown to be repaired by NER mechanisms in bacteria (Alekseyev et al., 2004), this adduct is persistent in mammals (Croy et al., 1981a; Hertzog et al., 1980), independently from NER status (Leadon et al., 1981). However, the FAPyr adduct appears to be less helix-distorting than the N7-guanine adduct and prokaryotic NER mechanisms are less sensitive to helix distortion than their mammalian homologues (Bedard et al., 2006). Thus, poor recognition of FAPyr adducts by mammalian NER mechanisms may increase its persistence and explain differences between assays performed in prokaryotic and eukaryotic systems. Several BER enzymes are capable of DNA glycosylase activity at guanine formamidopyrimidine adducts, allowing for specific excision of these mutagenic lesions (Dizdaroglu et al., 2008). One of these, Neil1, has recently been shown to catalyze excision of the AFB1-FAPyr adduct (Vartanian et al., 2017), while AFB1-treated newborn mice deficient in Neil1 exhibit more FAPyr adducts 48 hours post-dosing and are more susceptible to HCC than wild-type animals (Vartanian et al., 2017). In humans, epidemiological evidence suggests that defects in adduct repair may increase AFB1-related HCC risk. Polymorphisms of several DNA repair genes (XRCC1, XRCC3, XRCC4, XRCC7, ERCC2, XPC) have been shown to increase HCC risk (Kirk et al., 2005a; Long et al., 2006, 2008, 2013; Yao et al., 2014), especially in individuals with high AFB1 exposure (Long et al., 2006, 2008, 2013; Yao et al., 2014). In particular, ERCC2 and XPC are involved in NER, while XRCC1 is involved in BER.

Mutagenesis AFB1 exposure induces G / T transversion mutations in humans and animal models (Hsu et al., 1991; Bressac et al., 1991a; Smela et al., 2001; Chawanthayatham et al., 2017). While the AP, N7-guanine, and FAPyr adducts all result from AFB1 exposure, they may not contribute equally to the resulting mutational spectra. AP sites are well-known to be mutagenic, as they induce incorporation of non-complementary bases opposite to the adducted base (Loeb et al., 1986). Given the origination of AFB1-induced AP sites from adducted guanine residues and a predominance of AP/T mutations (Bailey et al., 1996), AP sites may logically contribute to AFB1 mutations. However, although most AFB1-induced mutations occur at the site of adduction, a minor fraction (7%–13%) occur at the residue 50 to the adducted guanine (Bailey et al., 1996). Although these mutations can be caused by intercalation of the bulky N7-guanine or FAPyr adducts, they cannot result from AP sites, which only form mutations at the lesioned base (Smela et al., 2001; Bailey et al., 1996). Furthermore, while N7-guanine and FAPyr adducts exhibit the same mutagenic profiles (> 80% G/T, 10% G/A, < 5% G/C, < 5% deletion), the FAPyr is at least twice as mutagenic as the N7-guanine adduct (Lin et al., 2014a, b; Smela et al., 2002). Thus, combined with observed differences in stability and repair, the FAPyr adduct is likely responsible for the majority of AFB1 mutagenesis. In 1991, two reports in the same issue of Nature reported the presence of a G/T transversion hotspot at the third position of codon 249 (50 -CGG AGG CCC-30 ) of the human TP53 gene (Hsu et al., 1991; Bressac et al., 1991a), resulting in the nonsynonymous substitution of arginine with serine (R249S) within the p53 protein (Li et al., 1993). Experiments in vitro have shown that AFB1 causally induces this R249S mutation (Aguilar et al., 1993) and epidemiological evidence strongly supports a role of aflatoxin exposure in its occurrence in human HCC (Chawanthayatham et al., 2017; Villar et al., 2011; Stern et al., 2001; Qi et al., 2015; Zhang et al., 2006; Lunn et al., 1997; Xie et al., 1994; Unsal et al., 1994; Szymañska et al., 2009; Kirk et al., 2005b; Shimizu et al., 1999). While the R249S mutation is found in only 3%–7% of HCCs in general (TCGA Research Network, 2017; Wellcome Trust Sanger Institute, 2017), its prevalence rises to 30%–60% in HCC cases with documented or likely AFB1 exposure (Villar et al., 2011; Stern et al., 2001; Qi et al., 2015; Xie et al., 1994; Unsal et al., 1994; Szymañska et al., 2004, 2009; Kirk et al., 2005b; Shimizu et al., 1999; Jackson et al., 2001; Coursaget et al., 1993). Indeed, it is rarely seen in liver tumors diagnosed in Europe or North America, where aflatoxin exposure is typically very low (Unsal et al., 1994; Wong et al., 2000; Bressac et al., 1991b); exceptions are often migrants from high-exposure regions (Amaddeo et al., 2015). The emergence of this mutation as a prototypical AFB1 hotspot can be linked to two factors: the contextual preference of AFB1 adduction within nucleotide sequences and disruption of p53 protein function. First, a study which systematically investigated the sequence context of AFB1 adduction found a preference for guanines in GC-rich regions, particularly when flanked by other guanines (50 -GGG-30 ) or with guanine or cytosine at the 50 positiondfor example, GGX, CGX (Benasutti et al., 1988). These in vitro predictions align with the observed hotspot mutation in codon 249 of human TP53 (GGC), as well as the predominating genome-wide mutational spectra (CGC; G/T) seen in AFB1-treated rodents (Chawanthayatham et al., 2017; Woo et al., 2011; Wattanawaraporn et al., 2012) and humans with R249S-mutant HCC (Chawanthayatham et al., 2017). Secondly, the arginine / serine missense alteration of mutant R249S p53 substantially disrupts its DNA-binding and promoter transactivation activity, leading to loss of cell cycle and apoptotic control (Gouas et al., 2009). However, its mechanism of selection in vivo (e.g., dominant negative) remains unclear and a subject of investigation.

Aflatoxins

37

Carcinogenicity of Aflatoxins: Evidence in Humans IARC has conducted several systematic reviews of the available data, which most recently in 2012 reaffirmed the carcinogenicity of aflatoxins to humans (International Agency for Research on Cancer, 1987, 1993, 2002, 2012). This section will not replicate those reports, but rather highlight the validation and implementation of molecular biomarkers to establish aflatoxins’ carcinogenicity. Before the availability of biomarkers of internal and biologically effective dose, epidemiological studies relied on estimates of dietary AFB1 exposure, acquired through food contamination surveys. However, substantial imprecision in such estimates can arise from heterogeneity in the spatial distribution of contamination within sampled food items (on the order of 103), errors in actual vs. assumed food consumption, and unappreciated temporal or geographical variation in contamination, due to unaccounted ecological factors. Nonetheless, such studies have provided circumstantial support for an association between AFB1 intake and HCC risk. In a cohort of nearly 8000 men in Guangxi, China, HCC rates among four communities correlated with estimates of dietary aflatoxin contamination (Yeh et al., 1989). In an ecological study in high- and low-incidence regions of China, although 85% of corn samples in the high-incidence region were contaminated with AFB1, only 5% of samples in the low-incidence community had detectable levels (Li et al., 2001). While these and other data were suggestive of the carcinogenic effects that had long been observed in animal models, the limitations of the ecological approach persist. This necessitated the development, validation, and implementation of minimally-invasive biomarkers of aflatoxin exposure, internal dose, and biologically effective dose.

Aflatoxins and Cancer Risk: Assessment with Biomarkers of Exposure Measurement of AFB1 or its metabolites in the urine has long been an approach for the subject-level assessment of aflatoxin exposure, continuing into the present day. Analytical techniques have advanced since early studies with radiolabeled tracers in animals (Dalezios et al., 1973; Wogan et al., 1967), and a breakthrough was the development of high-affinity monoclonal antibodies recognizing aflatoxins (Groopman et al., 1984), which could be implemented in immunoassays (Hatch et al., 1993; Zhu et al., 1987) or affinity columns for aflatoxin purification prior to analytical HPLC (Groopman et al., 1985, 1992a). Using these approaches, studies have shown that 24-hour urinary AFM1 excretion is highly correlated with AFB1 dose in animals (Groopman et al., 1992a) and AFB1 intake in humans (Zhu et al., 1987; Groopman et al., 1992b). In contrast, AFB1 exposure does not show a linear relationship with urinary AFP1 excretion (Groopman et al., 1992a, b) and correlations with total urinary aflatoxin excretion are inconsistent, likely due to variation in AFP1 production (Groopman et al., 1992b, c). Thus, urinary AFM1 is a valid biomarker of short-term (
24 hour) dietary AFB1 exposure. With this validated tool in hand, Ross et al. were the first to report an increased risk of HCC with subject-level estimates of aflatoxin exposure (Ross et al., 1992), a finding which has subsequently been replicated several times (Hatch et al., 1993; Qian et al., 1994; Sun et al., 1999; Wu et al., 2009; Wang et al., 1996). Finally, in settings of pre-existing aflatoxin exposure, modulation of urinary AFM1 excretion has been used as a biomarker of efficacy in chemoprevention trials (Scholl et al., 1996; Wang et al., 1999), or interventions utilizing enteric adsorbents to reduce AFB1 uptake in the intestine (Mitchell et al., 2013).

Aflatoxins and Cancer Risk: Assessment with Biomarkers of Internal Dose Similarly to urinary metabolites, levels of urinary AFB1-mercapturic acid (AFB1-NAC)dthe ultimate product of AFB1-glutathione conjugation and excretiondreflects recent aflatoxin exposures. However, unlike AFM1 or other oxidative metabolites (AFP1, AFQ1), AFB1-NAC links aflatoxin exposure with the production and detoxification of the toxic AFB1-8,9-exo-epoxide (Degen et al., 1978; Moss et al., 1983, 1985). Similarly, the AFB1-albumin (AFB1-alb) adduct (Fig. 3) is formed downstream from the AFB1-epoxide: solvolysis of the epoxide results in the AFB1-8,9-dihydrodiol, which exists in equilibrium with AFB1-dialdehyde (Johnson et al., 1996). This dialdehyde does not adduct DNA or participate in GST-mediated conjugation, but can react with lysine residues within proteins (Guengerich et al., 2002), notably in serum albumin (Sabbioni et al., 1987, 1990). Enhanced reduction of the dialdehyde by aldo-ketoreductases (Guengerich et al., 2001) results in lower AFB1-alb adduct levels, but has no impact on carcinogenesis (Roebuck et al., 2009). Thus, like AFB1-NAC, AFB1-alb is a biomarker of internal dose, as it integrates dietary exposure with metabolic production of the mutagenic epoxide; unlike AFB1-NAC, it does not reflect activity of the predominant phase II detoxification pathway (GST conjugation). Despite this difference, both are highly correlated with AFB1 exposure (Sabbioni et al., 1990; Scholl et al., 1997, 2006; Wild et al., 1990, 1992) and hepatic DNA adduct formation (Scholl et al., 1997; Wild et al., 1986, 1996). Rather, the most substantial difference lies in the temporal window each provides to investigators. While urinary biomarkers reflect only the most recent exposure/dose, the half-life of albumin in circulation is
3 days in the rat,
21 days in humans, and is unaffected by AFB1 adduction (Sabbioni et al., 2017)dmeaning that previous exposures may be quantifiable for several months in humans. Additionally, this long-term indicator of internal dose represents a “smoothed” estimate of habitual aflatoxin exposure (Wild et al., 1992), while levels of dietary contamination may vary significantly (e.g., daily or between wet and dry seasons (Wild et al., 2000; Gong et al., 2004)). Both of these biomarkers of internal dose have been used to interrogate cancer risk in humans. AFB1-NAC was detectable in 24 of 27 HCC cases from Guangxi, China (Wang et al., 2001), while oltipraz (Wang et al., 1999) and green tea polyphenols (Tang et al., 2008) have been shown to increase urinary AFB1-NAC excretion in secondary prevention trials. Several reports from a Taiwanese case-control study (Hatch et al., 1993) have demonstrated a significantly elevated risk of HCC with detectable (Chen et al.,

38

Aflatoxins

1996) or elevated (Wu et al., 2009) AFB1-alb levels, while secondary prevention trials with enterosorbent clay (Pollock et al., 2016) and post-harvest crop processing and storage (Turner et al., 2005) have shown benefits in mitigating existing AFB1 exposure.

Aflatoxins and Cancer Risk: Assessment with Biomarkers of Biologically Effective Dose Spontaneous depurination of the unstable N7-guanine adduct leads to its dose-dependent appearance in the urine of rodents (Bennett et al., 1981; Groopman et al., 1992a; Walton et al., 2001), in proportion to the initial mutagenic N7-guanine adduct burden in the liver (Bennett et al., 1981; Groopman et al., 1992a). In humans, although direct validation of urinary N7-guanine to hepatic adducts is not possible, urinary N7-guanine excretion in humans very closely reflects dietary AFB1 intake (Groopman et al., 1992b, 1992c). While also present in the urine, the urinary N7-guanine adduct does not reflect the same 24-hour window of aflatoxin exposure as do AFM1 and AFB1-NAC. A single dose of AFB1 produces an initial burden of N7-guanine adducts, of which 30%– 50% may depurinate in a biphasic decay and appear as urinary N7-guanine over the first 48 hours (Bennett et al., 1981; Hertzog et al., 1980). Of course, AFB1 exposure in humans is rarely in the form of a single bolus and multiple-dosing regimens in animals demonstrate that repeated exposure can result in a sustained release of N7-guanine adducts for many days (Croy et al., 1981a). Thus, although the N7-guanine biomarker appears in the urine and certainly reflects a narrower window of exposure (days) than does AFB1-alb (weeks-months), it exhibits different kinetics than the other aflatoxin biomarkers measured in the urine compartment. Ross et al. reported associations between HCC risk and subject-level aflatoxin exposure in two papers from a case-control study in China in the late 1980s (Ross et al., 1992; Qian et al., 1994). In addition to finding elevated risk with detectable urinary AFM1, detectable urinary N7-guanine was also associated with increased HCC risk, and to an even stronger degree: nearly fivefold in the initial report (95% CI 1.5–16.3) and more than ninefold in their follow-up paper (95% CI 2.1–11.8). Additionally, in the second report (Qian et al., 1994), the HCC risk estimates associated with detectable urinary AFM1, AFP1, AFB1, or AFQ1 all increased when detected simultaneously with the N7-guanine adduct. Similar patterns have been reported in a Taiwanese case-control study (Yu et al., 1997). As with other aflatoxin biomarkers, N7-guanine levels have been used in intervention and prevention trials, revealing associations with broccoli-sprout glucosinolates (Kensler et al., 2005) and reduced N7-guanine adduct formation with the use of chlorophyllin as an orally administered enterosorbent (Egner et al., 2001).

Aflatoxins and Cancer Risk: Assessment with Biomarkers of Early Biologic Effect The use of circulating tumor DNA as a biomarker of early carcinogenic events or tumor emergence has been of great interest for some time (Christie et al., 2016; Jahr et al., 2001). Given the high penetrance of the R249S mutation in AFB1-induced HCC, as well as the great importance of p53 as a tumor suppressor, detection of this mutation in circulating DNA from pre-neoplastic lesions or early carcinomas became an attractive biomarker target. Kirk et al. (2000) tested plasma from Gambian controls or patients diagnosed with HCC or cirrhosis; using restriction endonuclease digestion, they found the R249S mutation in only 3/53 controls and 2/13 cirrhotic patients, but in 19/53 HCCs (36%; OR 16.4, 95% CI 3.0–90.5). Moreover, they were not able to detect the R249S biomarker in any of the 10 European controls or 50 European HCC cases. Other publications followed, with most utilizing the short oligonucleotide mass analysis (SOMA) technique (Laken et al., 1998; Qian et al., 2002). Jackson and colleagues detected the R249S mutation first in 10/25 HCC tumors from China (validated with DNA sequencing) and subsequently in 11/20 tumors from a second Chinese cohort (Jackson et al., 2001). In the second cohort, paired plasma for 6/11 mutant tumors tested positive for the circulating R249S mutation, while 0/10 plasma samples from healthy American controls did. A study in Gambian HCCs reported similar rates of tumoral R249S mutation, with greater concordance (89%) between paired tumor-plasma dyads (Szymañska et al., 2004). A later study of HCC in Qidong, initiated after aflatoxin exposure had likely begin to decrease (Chen et al., 2013), found
60% prevalence of both tumoral and plasma R249S mutations, despite low levels of serum AFB1-alb adducts (Szymañska et al., 2009)dthus demonstrating the ability to reveal significant past exposures not captured temporally with other aflatoxin biomarkers. Others have demonstrated relative quantitation of plasma R249S levels, enabling its use as a continuous HCC risk variable (Lleonart et al., 2005); for example, while only 2% of controls had R249S levels > 103 copies per mL, 26% of HCC cases did (OR 62, 95% CI 4.7–820). Subsequent studies have found increased HCC risk with either higher relative levels of plasma R249s (Kirk et al., 2005b; Villar et al., 2012) or R249S detection (Kirk et al., 2005b; Jackson et al., 2003).

Chemical–Viral Interactions A crucial and consistent factor impacting the role of AFB1 in HCC risk has been the interaction between concurrent chronic HBV infection and AFB1 exposure. This relationship has been noted since the earliest AFB1 biomarker studiesdwhile Ross and colleagues reported a substantially increased HCC risk overall with any detectable urinary N7-guanine adduct (RR 4.9, 95% CI 1.5–16.3), stratification by HBV surface antigen (HBsAg) status revealed a crucial interaction (Ross et al., 1992). Detectable urinary N7-guanine conferred no significant HCC risk in HBsAg-negative individuals (RR 1.9, 95% CI 0.5–7.5), but a 60-fold increased risk (95% CI, 6.4–561.8)dgreater than multiplicativedwith concurrent HBsAg-positive status. This remarkable finding was replicated in their follow-up report: while detectable urinary aflatoxin biomarkers now significantly elevated HCC risk in HBsAg-negative individuals (likely due to improved power from additional cases and controls), aflatoxin exposure in HBsAg-positive participants was still associated with a nearly 60-fold increased risk (Qian et al., 1994). While some recent studies have not returned the same magnitude of

Aflatoxins

39

synergy as earlier reports, this may be partly due to improved HBV vaccination in some regions (Chang et al., 2009; Sun et al., 2013; Qu et al., 2014), declining aflatoxin exposure in others (Chen et al., 2013; Sun et al., 2013), and may explain some findings of additive, rather than synergistic, interaction (Wu et al., 2009). Regardless, the existence of a chemical-viral interaction appears to hold a firm consensus, regardless of the biomarker used for estimating aflatoxin exposure (Lunn et al., 1997; Kirk et al., 2005b; Yeh et al., 1989; Wang et al., 1996). Indeed, much of the impact of aflatoxin exposure on HCC incidence must be viewed in context with HBV infection. Reduced aflatoxin exposure, rather than HBV vaccination, likely accounts for the attenuation of HCC mortality in Qidong, China over the past 30 years; however, the great majority of this effect (83%) was attributable to reducing aflatoxin exposure in HBV-infected individuals (Chen et al., 2013). Similarly, while a recent systematic review and meta-analysis reported that 17% of worldwide HCC risk could be attributed to aflatoxin, this was higher in HBV-infected (21%) than HBV-uninfected individuals (9%), with odds ratios of 73.0 (95% CI 36.0–148.3) for concurrent exposure and 6.4 (95% CI 3.7–10.9) for aflatoxin alone (Liu et al., 2012). Several mechanisms have been proposed to explain this striking interaction between HBV and aflatoxin exposure. First, mutations in the HBV HBx gene are predictive of HCC development (Kuang et al., 2004; Muñoz et al., 2011) and HBx proteindparticularly mutant HBxdphysically interacts with (Gouas et al., 2010; Iyer et al., 2011) and inhibits p53 (Iyer et al., 2011), disrupting genome surveillance. Additionally, coexpression of HBx and R249S mutants has been shown to increase proliferation and aneuploidy in immortalized hepatocytes (Jiang et al., 2010). Second, HBV infection may exacerbate AFB1-induced mutagenesis and inhibit adduct repair. HBx protein and the HBV large envelope protein have been shown to induce CYP3A4 expression, potentially leading to greater AFB1 epoxidation and adduct formation (Niu et al., 2013; Kirby et al., 1994). Moreover, HBx appears to reduce NER through both p53-dependent and -independent mechanisms (Jia et al., 1999), with one study demonstrating specific impairment of AFB1-adduct repair (Groisman et al., 1999). Unsurprisingly, epidemiological evidence suggests that TP53 R249S mutations co-occur with HBx genomic integration and mutation in HCC (Ortiz-Cuaran et al., 2013; Kuang et al., 2005; Gouas et al., 2012).

Summary and Outlook In closing, the half-century of investigation into the toxicology and molecular biology of aflatoxin exposure now represents a classic and voluminous case study in chemical carcinogenesis. Few agents have been so thoroughly characterized through basic research; fewer still provide molecular epidemiologists with as wide of an array of tools with which to investigate. While much has been revealed, work in many subfields remains ongoingdfrom characterization of the epigenetic impacts of AFB1-induced hepatocarcinogenesis (Livingstone et al., 2017), to the role of AFB1 in carcinogenesis at other organ sites (Noguier et al., 2015; Koshiol et al., 2017; Carvajal et al., 2012), to predictions of elevated and shifting aflatoxin exposures as a result of impending climate change (Medina et al., 2014; Battilani et al., 2016). Work from many investigators has identified effective, culturally competent, and cost-effective strategies for the primary and secondary prevention of aflatoxin exposure (Groopman et al., 2008). Given that the burden of aflatoxin exposure will continue to fall on many of the world’s most vulnerable populations (International Agency for Research on Cancer, 2015; Wu et al., 2012), efforts to implement these strategies will likely (and rightly) become a priority.

See also: Aspirin and Cancer. Environmental Exposures and Epigenetic Perturbations. Epigenetic Therapy. Genetic and Epigenetic Deregulation of Enhancers in Cancer. Hepatocellular Carcinoma: Pathology and Genetics.

References Abbas, H., et al., 2009. Toxin Reviews 28 (2–3), 142–153. Aguilar, F., et al., 1993. Proceedings of the National Academy of Sciences of the United States of America 90 (18), 8586–8590. Alekseyev, Y.O., et al., 2004. Carcinogenesis 25 (6), 1045–1051. Amaddeo, G., et al., 2015. Gut 64 (5), 820–829. Aoyama, T., et al., 1990. Proceedings of the National Academy of Sciences of the United States of America 87 (12), 4790–4793. Appleton, B.S., et al., 1982. Cancer Research 42 (9), 3659–3662. Asao, T., et al., 1963. Journal of the American Chemical Society 85 (7), 1706–1707. Asao, T., et al., 1965. Journal of the American Chemical Society 87 (4), 882–886. Ayres, J.L., et al., 1971. Journal of the National Cancer Institute 46, 561–564. Bailey, E.A., et al., 1996. Proceedings of the National Academy of Sciences of the United States of America 93 (4), 1535–1539. Bailey, G.S., et al., 1988. Carcinogenesis 9 (11), 1919–1926. Bailey, G.S., et al., 1994. Mutation Research: Environmental Mutagenesis and Related Subjects 313 (1), 25–38. Battilani, P., et al., 2016. Scientific Reports 6, 24328. Bedard, L.L., et al., 2006. Cancer Letters 241 (2), 174–183. Benasutti, M., et al., 1988. Biochemistry 27, 472–481. Bennett, J.W., et al., 2003. Clinical Microbiology Reviews 16 (3), 497–516. Bennett, R.A., et al., 1981. Cancer Research 41 (2), 650–654. Blount, W.P., 1961. Turkeys 9, 52–61. Bren, U., et al., 2007. Chemical Research in Toxicology 20 (8), 1134–1140. Bressac, B., et al., 1991a. Nature 350, 429–431. Bressac, B., et al., 1991b. Lancet 338 (8779), 1356–1359.

40

Aflatoxins

Brown, K.L., et al., 2009. Chemical Research in Toxicology 22 (5), 913–917. Büchi, G., et al., 1967. Journal of the American Chemical Society 89 (25), 6745–6753. Buetler, T.M., et al., 1992. Biochemical and Biophysical Research Communications 188 (2), 597–603. Bujons, J., et al., 1995. Chemical Research in Toxicology 8 (3), 328–332. Butler, W.H., et al., 1969. Cancer Research 29, 2206–2211. Carnaghan, R.B., 1967. British Journal of Cancer 21 (4), 811–814. Carvajal, M., et al., 2012. Food Additives & Contaminants: Part A 29 (2), 258–268. Chang, M.H., et al., 2009. Journal of the National Cancer Institute 101 (19), 1348–1355. Chawanthayatham, S., et al., 2015. International Journal of Cancer 136, 1254–1262. Chawanthayatham, S., et al., 2017. Proceedings of the National Academy of Sciences of the United States of America 114 (15), E3101–E3109. Chen, C.-J., et al., 1996. Hepatology 24 (1), 38–42. Chen, J.-G., et al., 2013. Cancer Prevention Research 6 (10), 1038–1045. Chen, T., et al., 2010. Environmental and Molecular Mutagenesis 51 (3), 156–163. Chen, X., et al., 2006. Pharmacogenetics and Genomics 16, 219–227. Cheung, K.K., et al., 1964. Nature 201 (4925), 1185–1188. Christie, E.L., et al., 2016. Cancer Research 76 (19), 5590–5591. Council for Agricultural Science and Technology (2003) Mycotoxins: Risks in plant, animal, and human systems council for agricultural science and technology. Coursaget, P., et al., 1993. British Journal of Cancer 67 (6), 1395–1397. Croy, R.G., et al., 1978. Proceedings of the National Academy of Sciences of the United States of America 75 (4), 1745–1749. Croy, R.G., et al., 1981a. Cancer Research 41 (1), 197–203. Croy, R.G., et al., 1981b. Journal of the National Cancer Institute 66 (4), 761–768. Cullen, J.M., et al., 1987. Cancer Research 47, 1913–1917. Dalezios, J.I., et al., 1971. Science 171 (3971), 584–585. Dalezios, J.I., et al., 1973. Food and Cosmetics Toxicology 11 (4), 605–616. Dash, B., et al., 2007. Journal of Toxicology and Environmental Health. Part A 70 (1), 58–66. Davidson, N.E., et al., 1990. Cancer Research 50 (8), 2251–2255. De Iongh, H., et al., 1962. Biochimica et Biophysica Acta 65, 548–551. Degen, G., et al., 1978. Chemico-Biological Interactions 22, 239–255. Degen, G.H., et al., 1981. Carcinogenesis 2 (4), 299–306. Diener, U.L., et al., 1987. Annual Review of Phytopathology 25, 249–270. Dizdaroglu, M., et al., 2008. Free Radical Biology and Medicine 45 (12), 1610–1621. Dohnal, V., et al., 2014. Archives of Toxicology 88 (9), 1635–1644. Doi, A.M., et al., 2002. Toxicology and Applied Pharmacology 181 (1), 48–59. Dorner, J.W., et al., 1984. Mycopathologia 87 (1–2), 13–15. Eaton, D.L., et al., 1988. Chemical Research in Toxicology 1 (2), 108–114. Egner, P.A., et al., 2001. Proceedings of the National Academy of Sciences of the United States of America 98 (25), 14601–14606. Essigmann, J.M., et al., 1977. Proceedings of the National Academy of Sciences of the United States of America 74 (5), 1870–1874. Essigmann, J.M., et al., 1980. Cancer Research 40, 4085–4091. Ferlay, J., et al., 2015. International Journal of Cancer 136 (5), E359–E386. Fiala, J.L.A., et al., 2011. Toxicological Sciences 121 (1), 57–62. Gallagher, E.P., et al., 1994. Cancer Research 54, 101–108. Gallagher, E.P., et al., 1996. Toxicology and Applied Pharmacology 141, 595–606. Garner, R.C., et al., 1972. Cancer Research 32, 2058–2066. Gong, Y., et al., 2004. Environmental Health Perspectives 112 (13), 1334–1338. Goodall, C.M., et al., 1969. International Journal of Cancer 4, 422–429. Gopalakrishnan, S., et al., 1989. Journal of the American Chemical Society 111 (18), 7232–7239. Gopalakrishnan, S., et al., 1990. Biochemistry 29 (46), 10438–10448. Gouas, D.A., et al., 2009. Cancer Letters 286 (1), 29–37. Gouas, D.A., et al., 2010. Carcinogenesis 31 (8), 1475–1482. Gouas, D.A., et al., 2012. Carcinogenesis 33 (6), 1219–1224. Groisman, I.J., et al., 1999. Carcinogenesis 20 (3), 479–483. Groopman, J.D., et al., 1984. Proceedings of the National Academy of Sciences of the United States of America 81 (24), 7728–7731. Groopman, J.D., et al., 1985. Proceedings of the National Academy of Sciences of the United States of America 82 (19), 6492–6496. Groopman, J.D., et al., 1992a. Cancer Research 52 (2), 267–274. Groopman, J.D., et al., 1992b. Cancer Research 52 (1), 45–52. Groopman, J.D., et al., 1992c. Cancer Epidemiology, Biomarkers & Prevention 1 (3), 221–227. Groopman, J.D., et al., 2008. Annual Review of Public Health 29, 187–203. Gross-Steinmeyer, K., et al., 2010. Toxicological Sciences 116 (2), 422–432. Guengerich, F.P., et al., 1998. Mutation Research 402 (1–2), 121–128. Guengerich, F.P., et al., 2001. Chemical Research in Toxicology 14 (6), 727–737. Guengerich, F.P., et al., 2002. Chemical Research in Toxicology 15 (6), 780–792. Gurtoo, H.L., et al., 1976. Cancer Research 36 (12), 4663–4671. Hatch, M.C., et al., 1993. International Journal of Cancer 54 (6), 931–934. Hayes, J.D., et al., 1991. Pharmacology & Therapeutics 50, 443–472. Hayes, J.D., et al., 1994. The Journal of Biological Chemistry 269 (32), 20707–20717. Hertzog, P.J., et al., 1980. Carcinogenesis 1 (9), 787–793. Hsieh, D.P.H., et al., 1984. Food and Chemical Toxicology 22 (12), 1027–1028. Hsu, I.C., et al., 1991. Nature 350 (6317), 427–428. Hu, R., et al., 2004. The Journal of Pharmacology and Experimental Therapeutics 310 (1), 263–271. Ilic, Z., et al., 2010. Toxicology and Applied Pharmacology 242 (3), 241–246. International Agency for Research on Cancer (1987) IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, 7:1-440. International Agency for Research on Cancer (1993) Some naturally occurring substances: Food items and constituents, heterocyclic aromatic amines and mycotoxins, 56. International Agency for Research on Cancer (2002) Some traditional herbal medicines, some mycotoxins, naphthalene and styrene, 82.

Aflatoxins International Agency for Research on Cancer (2012) Chemical agents and related occupants, 100F. International Agency for Research on Cancer (2015) Mycotoxin control in low- and middle-income countries, 9. Itoh, K., et al., 1997. Biochemical and Biophysical Research Communications 236 (2), 313–322. Iyer, R.S., et al., 1994. Journal of the American Chemical Society 116 (5), 1603–1609. Iyer, S., et al., 2011. Molecular Carcinogenesis 50 (12), 972–980. Jackson, P.E., et al., 2001. Cancer Research 61, 33–35. Jackson, P.E., et al., 2003. Carcinogenesis 24 (10), 1657–1663. Jahr, S., et al., 2001. Cancer Research 61 (4), 1659–1665. Jia, L., et al., 1999. International Journal of Cancer 80 (6), 875–879. Jiang, W., et al., 2010. International Journal of Cancer 127 (5), 1011–1020. Johnson, N.M., et al., 2014. Cancer Prevention Research 7 (7), 658–665. Johnson, W.W. et al. (1996) Journal of the American Chemical Society, 118 (35), 8213–8220. Johnson, W.W., et al., 1997a. Proceedings of the National Academy of Sciences of the United States of America 94 (12), 6121–6125. Johnson, W.W., et al., 1997b. Chemical Research in Toxicology 10 (6), 672–676. Johnson, W.W., et al., 1997c. Biochemistry 36 (11), 3056–3060. Jowsey, I.R., et al., 2003. Molecular Pharmacology 64 (5), 1018–1028. Kamdem, L.K., et al., 2006. Chemical Research in Toxicology 19 (4), 577–586. Kensler, K.H., et al., 2014. Toxicological Sciences 139 (2), 293–300. Kensler, T.W., et al., 1985. Carcinogenesis 6 (5), 759–763. Kensler, T.W., et al., 1987. Cancer Research 47 (16), 4271–4277. Kensler, T.W., et al., 1997. Cancer Epidemiology, Biomarkers & Prevention 6, 603–610. Kensler, T.W., et al., 2005. Cancer Epidemiology, Biomarkers & Prevention 14, 2605–2613. Kensler, T.W., et al., 2007. Annual Review of Pharmacology and Toxicology 47, 89–116. Kensler, T.W., et al., 2011. Toxicological Sciences 120 (S1), S28–S48. Kensler, T.W., et al., 2013. Current Topics in Medicinal Chemistry 329, 163–177. Kirby, G.M., et al., 1993. Carcinogenesis 14 (12), 2613–2620. Kirby, G.M., et al., 1994. Molecular Carcinogenesis 11 (2), 74–80. Kirk, G.D., et al., 2000. Journal of the National Cancer Institute 92 (2), 2–7. Kirk, G.D., et al., 2005a. Cancer Epidemiology, Biomarkers & Prevention 14 (2), 373–379. Kirk, G.D., et al., 2005b. Oncogene 24 (38), 5858–5867. Klich, M.A., 2007. Molecular Plant Pathology 8 (6), 713–722. Klich, M.A., et al., 1988. Transactions of the British Mycological Society 99 (1), 99–108. Kobertz, W.R., et al., 1997. Proceedings of the National Academy of Sciences of the United States of America 94 (18), 9579–9584. Koshiol, J., et al., 2017. Gastroenterology 153 (2), 488–494. Kuang, S., et al., 2004. Proceedings of the National Academy of Sciences of the United States of America 101 (10), 3575–3580. Kuang, S.-Y., et al., 2005. Cancer Epidemiology, Biomarkers & Prevention 14, 380–384. Laken, S.J., et al., 1998. Nature Biotechnology 16 (13), 1352–1356. Lancaster, M.C., et al., 1961. Nature 192 (4807), 1095–1096. Langouët, S., et al., 1995. Cancer Research 55, 5574–5579. Leadon, S.A., et al., 1981. Cancer Research 41, 5125–5129. Levy, D.D., et al., 1992. Cancer Research 52 (20), 5668–5673. Li, D., et al., 1993. Carcinogenesis 14 (2), 169–173. Li, F.Q., et al., 2001. Journal of Agricultural and Food Chemistry 49 (8), 4122–4126. Lin, Y.C., et al., 2014a. Carcinogenesis 35 (7), 1461–1468. Lin, Y.C., et al., 2014b. Journal of Biological Chemistry 289 (26), 18497–18506. Liu, Y., et al., 2010. Environmental Health Perspectives 118 (6), 818–824. Liu, Y., et al., 2012. European Journal of Cancer 48 (14), 2125–2136. Livingstone, M.C., et al., 2017. Molecular Carcinogenesis 9, 2382–2390. Lleonart, M.E., et al., 2005. Cancer Epidemiology, Biomarkers & Prevention 14 (12), 2956–2962. Loeb, L.A., et al., 1986. Annual Review of Genetics 20, 201–230. Long, X.D., et al., 2006. Hepatology Research 36 (1), 48–55. Long, X.D., et al., 2008. Annals of Epidemiology 18 (7), 572–578. Long, X.-D., et al., 2013. Epidemiology 24 (5), 671–681. Lotlikar, P.D., et al., 1980. Cancer Letters 9 (2), 143–149. Lunn, R.M., et al., 1997. Cancer Research 57, 3471–3477. McGlynn, K.A., et al., 1995. Proceedings of the National Academy of Sciences of the United States of America 92 (6), 2384–2387. McWalter, G.K., et al., 2004. Journal of Nutrition 134 (12 Suppl), 3499S–3506S. Medina, A., et al., 2014. Frontiers in Microbiology 5, 348. Mitchell, N.J., et al., 2013. Biomarkers 18 (5), 391–398. Monroe, D.H., et al., 1987. Toxicology and Applied Pharmacology 90, 401–409. Monroe, D.H., et al., 1988. Toxicology and Applied Pharmacology 94 (1), 118–127. Moss, E.J., et al., 1983. Biochemical Journal 210 (1), 227–233. Moss, E.J., et al., 1985. Chemico-Biological Interactions 55, 139–155. Muñoz, A., et al., 2011. Carcinogenesis 32 (6), 860–865. Neal, G.E., et al., 1998. Toxicology and Applied Pharmacology 151 (1), 152–158. Nesbitt, B.F., et al., 1962. Nature 195 (4846), 1062–1063. Newberne, P.M., et al., 1969. Cancer Research 29, 236–251. Nguyen, T., et al., 2000. The Journal of Biological Chemistry 275 (20), 15466–15473. Niu, Y., et al., 2013. Digestive and Liver Disease 45 (12), 1041–1048. Noguier, L., et al., 2015. Journal of the American Medical Association 313 (20), 26–28. Norred, W.P., et al., 1983. Toxicology and Applied Pharmacology 70 (1), 96–104. O’Brien, K., et al., 1983. Biochemical and Biophysical Research Communications 114 (2), 813–821. Okajima, T., et al., 2000. Journal of Molecular Structure 532, 205–212.

41

42

Aflatoxins

Ortiz-Cuaran, S., et al., 2013. Cancer Letters 331 (1), 46–51. Ostry, V., et al., 2017. Mycotoxin Research 33, 65–73. Pelkonen, P., et al., 1997. Chemical Research in Toxicology 10 (1), 85–90. Petrick, J.L., et al., 2016. International Journal of Cancer 139 (7), 1534–1545. Pitt, J.I., et al., 2013. Food Control 32 (1), 205–215. Pollock, B.H., et al., 2016. Food Additives & Contaminants: Part A 33 (8), 1–9. Primiano, T., et al., 1995. Cancer Research 55 (19), 4319–4324. Qi, L.-N., et al., 2015. Liver International 35 (3), 999–1009. Qian, G.-S., et al., 1994. Cancer Epidemiology, Biomarkers & Prevention 3, 3–10. Qian, G.-S., et al., 2002. Cancer Epidemiology, Biomarkers & Prevention 11, 1126–1129. Qu, C., et al., 2014. PLOS Medicine 11 (12) e1001774. Rahat, B., et al., 2012. Journal of Clinical and Experimental Hepatology 2 (2), 104–111. Raj, H.G., et al., 1984. Cancer Letters 22 (2), 125–133. Raj, H.G., et al., 1986. Cancer Letters 33 (1), 1–9. Ramos-Gomez, M., et al., 2001. Proceedings of the National Academy of Sciences of the United States of America 98 (6), 3410–3415. Raney, K.D., et al., 1990. Chemical Research in Toxicology 3 (3), 254–261. Raney, K.D., et al., 1992. Chemical Research in Toxicology 5 (4), 470–478. Rendic, S., et al., 2012. Chemical Research in Toxicology 25 (7), 1316–1383. Roebuck, B.D., et al., 2009. Toxicological Sciences 109 (1), 41–49. Ross, R.K., et al., 1992. Lancet 339 (8799), 943–946. Sabbioni, G., et al., 1987. Carcinogenesis 8 (6), 819–824. Sabbioni, G., et al., 1990. Carcinogenesis 11 (11), 2063–2066. Sabbioni, G., et al., 2017. Chemical Research in Toxicology 30, 332–366. Salhab, A.S., et al., 1976. Cancer Research 37 (4), 1016–1021. Sargeant, K., et al., 1961. Nature 192 (4807), 1096–1097. Scholl, P.F., et al., 1996. Carcinogenesis 17 (6), 1385–1388. Scholl, P.F., et al., 1997. Chemical Research in Toxicology 10 (10), 1144–1151. Scholl, P.F., et al., 2006. Chemical Research in Toxicology 19 (1), 44–49. Shen, Y.-H., et al., 2014. Tumor Biology 35 (5), 4007–4015. Shimada, T., et al., 1989. Proceedings of the National Academy of Sciences of the United States of America 86 (2), 462–465. Shimizu, Y., et al., 1999. International Journal of Cancer 82 (2), 187–190. Shupe, T., et al., 2004. Toxicology Letters 148 (1–2), 1–9. Slone, D.H., et al., 1995. Pharmacogenetics 5, 224–233. Smela, M.E., et al., 2001. Carcinogenesis 22 (4), 535–545. Smela, M.E., et al., 2002. Proceedings of the National Academy of Sciences of the United States of America 99 (10), 6655–6660. Smith, M.T., et al., 2016. Environmental Health Perspectives 124 (6), 713–721. Squire, R.A., 1981. Science 214 (4523), 877–880. Stern, M.C., et al., 2001. Cancer Epidemiology, Biomarkers & Prevention 10, 617–625. Sun, C.-A., et al., 2001. Carcinogenesis 22 (8), 1289–1294. Sun, Z., et al., 1999. Hepatology 30 (2), 379–383. Sun, Z., et al., 2013. Carcinogenesis 34 (8), 1800–1805. Suzuki, T., et al., 2015. Free Radical Biology & Medicine 88, 93–100. Swenson, D.H., et al., 1973. Biochemical and Biophysical Research Communications 53 (4), 1260–1267. Swenson, D.H., et al., 1974. Biochemical and Biophysical Research Communications 60 (3), 1036–1043. Swenson, D.H., et al., 1977. Cancer Research 37, 172–181. Szymañska, K., et al., 2004. International Journal of Cancer 110 (3), 374–379. Szymañska, K., et al., 2009. Cancer Epidemiology, Biomarkers & Prevention 18 (5), 1638–1643. Taguchi, K., et al., 2016. Toxicological Sciences 152 (1), 40–52. Takahashi, Y., et al., 2002. Carcinogenesis 23 (4), 627–633. Tang, L., et al., 2008. Carcinogenesis 29 (2), 411–417. TCGA Research Network, 2017 [Online]. Available. http://cancergenome.nih.gov. Accessed: 10 October 2017. Thimmulappa, R.K., et al., 2002. Cancer Research 62, 5196–5203. Tiemersma, E.W., et al., 2001. Cancer Epidemiology, Biomarkers & Prevention 10, 785–791. Turner, P.C., et al., 2005. Lancet 365 (9475), 1950–1956. Ueng, Y.-F., et al., 1995. Chemical Research in Toxicology 8 (2), 218–225. Unsal, H., et al., 1994. Proceedings of the National Academy of Sciences of the United States of America 91 (2), 822–826. Vartanian, V., et al., 2017. Proceedings of the National Academy of Sciences of the United States of America 114 (16), 4207–4212. Vesselinovitch, S.D., et al., 1972. Cancer Research 32, 2289–2291. Villar, S., et al., 2011. Environmental Health Perspectives 119 (11), 1635–1640. Villar, S., et al., 2012. PLoS One 7 (6), e37707. Wakabayashi, N., et al., 2004. Proceedings of the National Academy of Sciences of the United States of America 101 (7), 2040–2045. Walton, M., et al., 2001. Chemical Research in Toxicology 14, 919–926. Wang, H., et al., 1998. Biochemistry 37 (36), 12536–12545. Wang, J.-S., et al., 1999. Journal of the National Cancer Institute 91, 347–354. Wang, J.-S., et al., 2001. Cancer Epidemiology, Biomarkers & Prevention 10, 143–146. Wang, L.-Y., et al., 1996. International Journal of Cancer 67 (5), 620–625. Wattanawaraporn, R., et al., 2012. Toxicological Sciences 128 (2), 326–333. Wei, C.I., et al., 1985a. Food and Chemical Toxicology 23 (9), 809–819. Wei, C.I., et al., 1985b. Food and Chemical Toxicology 23 (9), 821–825. Wellcome Trust Sanger Institute, 2017 [Online]. Available. www.sanger.ac.uk. Accessed: 10 October 2017. Wild, C.P., et al., 1986. Carcinogenesis 7 (6), 853–858. Wild, C.P., et al., 1990. Cancer Research 50 (16), 245–251. Wild, C.P., et al., 1992. Cancer Epidemiology, Biomarkers & Prevention 1 (3), 229–234.

Aflatoxins Wild, C.P., et al., 1996. Cancer Epidemiology, Biomarkers & Prevention 5, 179–189. Wild, C.P., et al., 2000. International Journal of Cancer 86 (1), 1–7. Wild, C.P., et al., 2009. Cancer Letters 286 (1), 22–28. Williams, D.E., et al., 2009. Comparative Biochemistry and Physiology, Part C: Toxicology & Pharmacology 149 (2), 175–181. Wogan, G.N., et al., 1967. Cancer Research 27 (10), 1729–1736. Wogan, G.N., et al., 1971. Cancer Research 31, 1936–1942. Wogan, G.N., et al., 1974a. Food and Cosmetics Toxicology 12 (5), 681–685. Wogan, G.N., et al., 1974b. Food and Cosmetics Toxicology 12, 381–384. Wogan, G.N., et al., 2012. Food Additives & Contaminants: Part A 29 (2), 249–257. Wojnowski, L., et al., 2004. Pharmacogenetics 14 (10), 691–700. Wong, J.J., et al., 1976. Proceedings of the National Academy of Sciences of the United States of America 73 (7), 2241–2244. Wong, N., et al., 2000. Clinical Cancer Research 6, 4000–4009. Woo, L.L., et al., 2011. Toxicological Sciences 122 (1), 38–44. Wu, F., et al., 2012. PLoS One 7 (9) e45151. Wu, H.-C., et al., 2009. Cancer Epidemiology, Biomarkers & Prevention 18 (3), 846–853. Xie, J.P., et al., 1994. Cancer Research 54 (1), 281–285. Yao, J.-G., et al., 2014. International Journal of Clinical and Experimental Pathology 7 (9), 6231–6244. Yates, M.S., et al., 2006. Cancer Research 66 (4), 2488–2494. Yates, M.S., et al., 2007. Molecular Cancer Therapeutics 6 (1), 154–162. Yeh, F.-S., et al., 1989. Cancer Research 49, 2506–2509. Yu, M.W., et al., 1997. Journal of Hepatology 27 (2), 320–330. Zhang, Y.J., et al., 2006. International Journal of Cancer 119 (5), 985–991. Zhu, J., et al., 1987. Cancer Research 47, 1848–1852. van der Zijden, A.S.M., et al., 1962. Nature 195 (4846), 1060–1062.

43

Aging and Cancer Mingyang Song, Harvard Medical School, Boston, MA, United States; Massachusetts General Hospital, Boston, MA, United States; and Harvard T.H. Chan School of Public Health, Boston, MA, United States © 2019 Elsevier Inc. All rights reserved.

Glossary Aerobic glycolysis Refers to a metabolic phenomenon commonly observed in cancer cells that convert glucose to lactate in the presence of oxygen. Autophagy Is a major intracellular self-degradative process that delivers cytoplasmic materials to the lysosome for degradation. Cellular senescence Also known as replicative senescence, refers to the essentially irreversible arrest of cell proliferation (growth) that occurs when cells experience potentially oncogenic stress. Clonal hematopoiesis Refers to a common aging-related phenomenon in which hematopoietic stem cells (HSCs) or other early blood cell progenitors carrying recurrent somatic mutations contribute to the formation of a genetically distinct subpopulation of blood cells. Inflammasomes Are innate immune system receptors and sensors that regulate the activation of caspase-1 and induce inflammation in response to infectious microbes and molecules derived from host proteins. Loss of heterozygosity Is a gross chromosomal event that results in loss of the entire gene and the surrounding chromosomal region. Oxidative stress Refers to an imbalance between the production of reactive oxygen species and the ability of the body to counteract or detoxify their harmful effects through neutralization by antioxidants. Progeroid syndromes Refer to a group of rare genetic disorders which mimic physiological aging, such as Werner syndrome, Hutchinson–Gilford progeria syndrome, trichothiodystrophy syndrome, and Cockayne syndrome. Reactive oxygen species (ROS) Refer to a number of reactive molecules and free radicals derived from incomplete reaction of oxygen in mitochondria. Warburg effect Is the enhanced conversion of glucose to lactate observed in tumor cells, even in the presence of normal levels of oxygen.

Incidence of most cancers increases over age (Fig. 1). In the United States, 50% of all malignancies occur in individuals over the age of 65, which represent 12% of the population. With the current aging rate, it is estimated that 70% of all cancers will occur in the elderly by 2030. Despite the evident age-dependent pattern of cancer incidence, the relationship between the biology of aging and cancer is far from clear. According to the widely accepted multi-step model of cancer, cancer is the end-result of various sequential mutations and successive cellular changes. Therefore, it has been debated whether aging has any causal effect on cancer independent of the prolonged exposure time to carcinogens due to aging. However, emerging evidence from studies of genetic and dietary manipulations indicates that cancer may be reduced by chronic downregulation of pro-aging pathways. On the other hand, although several mechanisms are well known to be shared between aging and cancer, the molecular and cellular hypotheses posited to

(B)

2500 2000 1500 1000 500 0

500 450 400 350 300 250 200 150 100 50 0

Major cancers Breast Lung Colorectum Pancreas Bladder Non-Hodgkin Lymphoma

75%

Yes

DK

DK

–

Dibouti Egypt

No Yes

Six to seven Three to five

Yes Yes

Yes Yes

Yes Yes

26%–50% 51%–75%

No No

No No

No No

– –

Iran

Yes

Yes þ national quit line

Yes

Yes

25%

Yes

Yes

Yes

–

Iraq

Yes

All public places completely smoke-free Up to two

No

Yes

25

Yes

Yes

DK

–

Birth

66%

Israel

No

Up to two public places

Yes þ national quit line Yes þ national quit line

Yes

Complete absence of ban

>75%

No

No

Yes

Birth, þ 1, þ 6 months

98%

Jordan

Yes

Three to five

Yes

Yes

>75%

Yes

Yes

Yes

No

Six to seven

Yes

Yes

25%

No

No

No

–

3, þ 4, þ 5 months Birth

98%

Kwait

99%

Lebanon

Yes

all

Yes þ national quit line Yes þ national quit line Yes

13 years, þ 2 months, þ6 months –

Yes

Yes

26%–50%

DK

–

Birth

81%

Warning labels (no/yes)

Bans on advertising, Tobacco taxes promotion and (% of tax in retail sponsorship a price)

Presence of an operational policy, strategy or action plan

HPV vaccine schedule

Hepatitis B vaccine schedule – Birth, þ 1 month 2, þ 4, þ 8 to 12 months Birth months, þ 4 months,þ 6 months Birth, þ 2, þ 6 months

Hepatitis B vaccine coverage, infants 71% 99% 96% 82% 97% 99%

Libya

No

all

Yes

Morocco

Yes

Three to five

Three to five

Oman

Yes

Yes

Pakistan

No

Data not recognized all

No or small size No or small size Yes

Yes

Yes

Qatar Saudi Arabia Somalia

Yes Yes

Up to two Six to seven

Yes Yes

Yes Yes

No

Up to two

None

Sudan

Yes

Up to two

None

Syria

Yes

Six to seven

Yes

Tunisia

Yes

Up to two

Yes

Turkey

Yes

All

UAE

No

Three to five

Yemen

No

Up to two

Yes þ national quit line Yes þ national quit line Yes

No or small size No or small size No or small size No or small size Yes

Yes

25%

No

No

No

Birth

98%

No

15 years (3) –

Yes

51%–75%

No

No

Birth

99%

absence

25%

Yes

Yes

No

–

Birth

97%

absence

51%–75%

No

No

No

–

–

72%

absence

25% 25%

Yes Yes

Yes Yes

Yes No

– –

Birth Birth

99% 98%

absence

25%

No

No

No

–

–

34%

51%–75%

Yes

Yes

No

–

–

98%

Yes

51%–75%

No

No

No

–

71%

Yes

>75%

No

No

No

–

Birth, þ 2, þ 6 months Birth

Yes

>75%

Yes

Yes

No

–

Yes

Yes

25%

No

No

No

–

Yes

Yes

51%–75%

No

No

No

–

Complete of ban Complete of ban Yes Complete of ban Complete of ban Yes

Birth, þ 1, þ 6 months Birth, þ 2, þ 6 months –

98% 97% 94% 88%

DK: Country responded “don’t know”. – Data not available. a Ban on national television, radio and print media. Source: WHO Cancer country profiles (http://www.who.int/cancer/country-profiles/en/).

Cancer in the Middle East 235

236

Cancer in the Middle East

Viral Hepatitis (NCCVH) was established to provide a novel model of care (MOC). The mission of MOC is to provide national treatment programs supplying antiviral therapy either at very minimum cost or totally free of charge. Egypt has succeeded to negotiate in reducing the price of sofosbuvir with previous commitment in treatment by Peg-interferon/RBV. Preventive strategies were provided to raise public awareness especially the high-risk groups to further minimize the risk of HCV transmission, improve infection control, ensure safe injection practices, and adopt safe blood products. The MOC activity also establishes the practice of adapted national guidelines for chronic HCV treatment as supplied by national expert hepatologists and training of health care professionals for efficient counseling of chronic HCV patients. The national guidelines were adapted from guidelines for screening and management of patients with Hepatitis C virus. It involves stepwise approach for screening, diagnosis, and treatment of HCV infection and follow-up of patients to prevent hepatocellular carcinoma.

•

Screening for HCV: Patients on haemodialysis, blood/tissue donors, and health care professional who are exposed to procedures should be tested for HCV. Screening also includes patients with persistently elevated alanine aminotransferase, injecting drug users, patients diagnosed with HIV infection, patients with medical and dental treatment or those making tattoo or piercing in ME countries known to be endemic with HCV, children born to HCV-infected mothers and finally, household contact with HCVinfected patients. • HCV diagnostic testing: It should be tested on serum or plasma where possible and HCV genotyping should be undertaken if antiviral therapy is being considered. If health care professionals are suspected of percutaneous exposure to infection, they should be tested for HCV ribonucleic acid (RNA) at 6, 12, and 24 weeks and anti-HCV testing should be considered at 12 and 24 weeks. • Treatment of acute infection: Patients require close monitoring, both clinical and laboratory for the first 3 months so that the infection may be resolved spontaneously. They should be offered IFN therapy either pegylated IFN or nonpegylated IFN for 24 weeks regardless the genotype. • Treatment of chronic HCV infection: All patients should be treated with antiviral therapy and sustained virological response should be a marker for viral clearance. The selection of antiviral therapy is based on the genotype: – Genotype 1 Treatment: Naïve and treatment-experienced patients should be treated with pegylated IFN, weight-based ribavirin, and a protease inhibitor as triple therapy. The duration of therapy varies depending on early viral response at 12 weeks. The treatment should continue until 48 weeks; those with still positive virus RNA at 24 weeks, should discontinue the treatment. – Genotype 4 Patients with HCV genotype 4 infections: Standard treatment should be 48 weeks of pegylated IFN and weight-based ribavirin. Caution should be taken to avoid adverse effects of interferon (IFN) and ribavirin therapy. They should not be prescribed in pregnant women and in patients with psychiatric illnesses. Hepatitis B virus Hepatocellular carcinoma is caused by HBV in most ME countries with the exception of Egypt and Pakistan. Prevention of HBV involves three levels: primary prevention by effective HBV vaccination for healthy population, secondary prevention through hepatitis B antiviral agents, and tertiary prevention to prevent infection recurrence. Hepatitis B vaccine is safe and very costeffective as licensed and incorporated by WHO into expanded program immunization (EPI) worldwide with wide coverage in the low-middle income countries (> 90% response rate in nonimmunocompromized subjects). It is available in three doses that significantly reduce the prevalence of HBsAg in children and young adults (Table 4). WHO recommends a birth dose vaccine given to all newborns within the 24 h after birth. Although the birth dose vaccine significantly reduced HBV infection by about 70% in infants born to highly infectious mothers, not all African countries adopt it; even these counties face difficulties in implementing the vaccine. Some Countries didn’t not yet integrate it. This may be due to the minimal estimated perinatal transmission and lack of sufficient evidence and clinical trials conduction for it in pregnant women in the African countries (Table 4). The following strategies are needed for better control of both HCV and HBV infections and decrease HCC incidence:

•

•

Operational research programs: Epidemiological surveys and population-based screening programs are needed to assess research priorities about risk factors of transmission and to determine cost-effective approaches for screening and treatment. Unfortunately, most of the screening programs in Africa are expensive and are not of considerable quality. The PROLIFICA (Prevention of Liver Fibrosis and Cancer in Africa) program is a screening regimen that was conducted in Gambia, not in the ME. It is concluded that only 1% of positive HBs antigen in the adult previously tested and aware of their diagnosis. Though the PROLIFICA program estimated around 50% of HBV infection in the Gambia was attributed to vertical mother to infant transmission, only 3% of infants received the birth dose HBV vaccine due to recognized social and financial barriers that are similar to the situation in many ME countries. Improving screening and early treatment of infected patients: WHO facilitates the selection criteria for infected patients who need early treatment in resources constrained countries: (a) A patient who is HBsAg positive and has cirrhosis and (b) patients with HBsAg who are over 30 and have persistently raised alanine aminotransferase (ALT) levels. Policy makers should focus on resource allocation for treatment, negotiation with generic drugs production companies to supply drugs at lower prices. Treatment of

Cancer in the Middle East

•

237

HBV is challenging in the ME. Proper screening and treatment programs are lacking in many ME countries because the cost of the programs are the responsibility of patients and due to their availability in private sector only. Tenovir, the drug of choice for HBV treatment, is supplied by the global funds at a lower price of $50 per patient per year but not for HBV monoinfection and even not licensed by some African countries. Surveillance for HCC in cirrhotic patients: Cirrhosis is the main causative agent for occurrence of HCC. Regular surveillance interval at 6 months is recommended than annual surveillance to improve the survival of cirrhotic patients and minimize the incidence of HCC. The best-known surveillance option is the liver ultrasound that is available with good accuracy in most of ME countries, but not in the rural areas and it is highly dependent on operator experience. It is imperative to train health care providers on ultrasound as an easy and applicable surveillance system tool.

H. pylori infection WHO recommends all countries consider screening for H. pylori for primary prevention of gastric cancer. The International Agency for Research on Cancer (IARC) Working Group of international experts suggested population-based screening for eradication of H. pylori which reduce the incidence of gastric cancer by 30%–40%. However, low-middle income countries recognized with high burden of gastric cancer do not properly implement this screening program. The screening program consists of three steps: First: culture, histology, rapid urease test (RUT), histopathology, PCR, and a serology test to screen for H. pylori in healthy individuals. Second step involves the eradication treatment with antibiotics from a primary care doctor if required, then the final step is retest by urea breath test (UBT), stool antigen test (SAT), and second-line treatment if required. H. pylori screening programs are costeffective. A cost-effectiveness of a H. pylori serology-based screening program in New Zealand, a country with relatively high gastric cancer rates, illustrated that the incremental cost for national screening program was US$196 million (95% uncertainty interval [95% UI]: $182-$211 million) with health gains of 14,200 quality adjusted life years (QALYs) (95% UI: 5,100–26,300) and the screening program was more cost-effective by US$8,000 ($3,800-$18,500) per QALY. H. pylori eradication therapy was proved to protect against the occurrence of gastric carcinoma and reduce its incidence (risk ratio (RR) 0.66; 95% CI: 0.46–0.95). However, in patients with preneoplastic lesions, H. pylori eradication measures do not play a significant role for gastric cancer prevention. It is available in many forms: (1) proton pump inhibitor (PPI) dual therapy (PPI plus either amoxicillin or clarithromycin); (2) PPI triple therapy (PPI plus any two of the following: amoxicillin, macrolide, 5-nitroimidazole); (3) histamine 2-receptor antagonist (H2RA) triple therapy (H2RA plus any two of the following: amoxicillin, macrolide, 5-nitro imidazole); (4) bismuth triple therapy (bismuth salt and 5-nitroimidazole with either amoxicillin or tetracycline); (5) bismuth quadruple therapy (as bismuth triple therapy, but with the addition of a PPI); (6) ranitidine bismuth citrate (RBC) dual therapy (RBC plus either amoxicillin or clarithromycin); (7) RBC triple therapy (RBC plus any two of the following: amoxicillin, macrolide, 5-nitroimidazole); (8) clarithromycin monotherapy. Antibiotic resistance is a possible cause of H. pylori treatment failure in the ME. The combined levofloxacin, and clarithromycin and esomeprazole-based regimen as first-line triple therapy for H. pylori eradication can give better eradication rate with same safety when compared with standard triple therapy (clarithromycin 500 mg twice daily, Amoxicillin 1000 mg twice daily, plus esomeprazole 20 mg twice daily for 7 days). The rapid urease test (RUT) is a popular, easy, and cheap screening test based on the presence of Urease in the H. pylori-infected gastric mucosa. It could be easily used in lowmiddle income countries of the ME. However, the former intake of proton pump inhibitors (PPI) should be taken into account prior to testing to avoid false negative results. Also, More than 10,000 bacterial cells are required to achieve positive test results. In Iranian population, Gemifloxacin-containing quadruple therapy (amoxicillin 1 g, bismuth 240 mg, and omeprazole 20 mg for 14 days) provides high H. pylori eradication rate ( 90% per protocol cure rate) after the failure of first-line standard quadruple therapy (clarithromycin–amoxicillin–bismuth–omeprazole). Early screening and prevention of EBV could prevent gastric carcinoma, lymphoma, and nasopharyngeal carcinoma. IgA antibody titres to EBV viral capsid antigen (EBV-IgA-VCA) and EBV early antigen (EBV-EA) can be used for the serologic screening of nasopharyngeal carcinoma. However, even after screening, attempts to develop vaccines continue and still under investigations because the success of this vaccine depends on cancer pattern in each country and the cost versus benefit.Table 4 illustrates the summary of cancer primary prevention policies in all ME countries as adapted from WHO country profiles report.

Cancer Screening and Early Detection The focus of cancer control is to diagnose cancer in the earliest possible curable stage, thus improving the quality of life and prognosis of cancer patients. WHO summarized the major elements for cancer early diagnosis into three main domains: 1. Step 1: Awareness of cancer symptoms and accessing care 2. Step 2: Clinical evaluation, diagnosis, and staging 3. Step 3: Access to treatment, including pain relief.

Step 1: Awareness of cancer symptoms and accessing care The patient should be aware of cancer symptoms (symptom appraisal interval). He or she should be encouraged to overcome fear and cancer stigma, and access primary care. Thus, awareness of symptoms will be translated into health-seeking interval.

238

Cancer in the Middle East

Barriers (a) Poor health literacy (b) Cancer stigma (c) Poor access to primary care Solutions (a) Empower and engage people and communities (b) Improve health literacy and reduce cancer stigma (c) Leadership and governance to improve access to care

Step 2: Clinical evaluation, diagnosis, and staging This step is known as diagnosis interval. It involves three stages: First, the patient should be evaluated by health care provider to ensure if cancer may be present (diagnostic evaluation); second, the patient receives diagnostic testing for confirmation including clinical, laboratory, and pathological confirmation testing and patients undergo staging to determine cancer spread (diagnostic testing and staging). Finally, if patient confirmed by cancer presence, he or she should be referred for treatment to specialized centers providing safe and effective treatment across ranges of modalities needed (treatment interval). Barriers (a) Inaccurate clinical assessment and delays in clinical diagnosis (b) Inaccessible diagnostic testing, pathology, and staging (c) Poor coordination and loss to follow-up Solutions (a) (b) (c) (d)

Improve provider capacity at first contact point Strengthen diagnostic and pathology services Develop referral mechanisms and integrated care Provide supportive counseling and people-centered care

Step 3: Access to treatment Many patients from LMIC recently diagnosed with cancer do not immediately initiate treatment due to financial, geographic, logistical, and sociocultural barriers.Table 5 shows how ME countries use cervical, breast, and colorectal cancer screening and early detection strategies at the public primary health care level for early diagnosis of cancer types. An example of early detection program is the EUROMED CANCER Network project. It aimed to support the non-EU Mediterranean countries about the appropriate cancer screening programs. It illustrates the importance of public health surveillance through a structured questionnaire survey applied to 15 countries (Albania, Algeria, Bosnia and Herzegovina (BiH), Croatia, Egypt, Jordan, UN Interim Administration Mission in Kosovo, Lebanon, Montenegro, Morocco, Palestinian National Authority, Serbia, Syria, Tunisia, and Turkey). Many of them are ME countries. It is concluded that applying a uniform strengthening system for cancer control for all these countries may not be feasible. However, tailored surveillance of early detection programs considering each country’s resources should start early with appropriate uniform health information system. One of the project’s recommendations is breast cancer (BC) and cervical cancer (CC) screening as the commonest female cancers in the region. The most frequent breast cancer screening test is mammography, either alone or combined with clinical breast examination (CBE), breast self-examination (BSE), or ultrasound. It is feasible and applied in all project’s studied countries (Table 5). The starting age in BC organized screening is 40–45 years in most countries but in others, screening starts even earlier. The age upper limit is 69 years in most countries, but in Lebanon, it goes up to 75. An interview by structured questionnaire in a multicenter survey among the Qatari women showed that fewer than half of the participants were aware of breast cancer screening recommendations; less than one third performed this screening. Around one-quarter of them stated that they were counseled by their doctors about appropriate BC screening. Most of the participants stated that they were ready for a mammogram appointment based on a recommendation by their health care professional, hence, notifying the important role of heath care providers for women counseling about different cancer screening programs. Many ME countries lack the National Organized Cervical Cancer Screening (NOCCS) program due to the insufficient political support of public health programs, lack of resources, little participation due to cultural factors, and insufficient women knowledge. This subsequently contributes to late detection of cervical cancer. Screening by Pap test could be possible in many countries (Table 5). It is usually in use starting at about 20 and ending at 60 years. In Morocco, an alternative to Pap test is visual inspection after application of acetic acid (VIA). Comparison of the cost-effectiveness of different cervical cancer (CC) screening tests in Lebanon using either cytology or HPV testing and at different coverage percentage concluded that 20% coverage by annual cytology testing reduced CC risk by 14% with an incremental cost-effectiveness ratio of I$80,670/year of life saved (YLS). Increased coverage to 50% with 3 and 6 years intervals yielded CC reduction of 26.1% and 21.4%, respectively at lower costs compared to 20% coverage with annual screening. Screening every 5 years with HPV DNA testing at 50% coverage resulted in 23.4% reduction in CC (than cytology at the same frequency) with an incremental cost-effectiveness ratio of I$12,210/YLS. The main implication of this research that repeated cytology covering a small percentage of women may be inefficient, while increased coverage up to

Table 5 Country

Cancer screening early detection strategies and palliative care policies in Middle East countries Cervical cancer early detection

Cervical cytology (PAP)

Acetic acid visualization (VIA)

Breast cancer Early detection Breast palpation/ clinical breast exam (CBE) Mammogram

Colorectal cancer Early detection Faecal occult blood test or faecal immunological test

Cancer treatment and palliative care

Screening by exam or colonoscopy Radiotherapy

Total high energy teletherapy units/ Number of million radiotherapy centers inhabitants

Number of radiation oncologists

Availability at the public primary health care (PPHC) level (yes/no) No Yes Yes No No Yes No Yes Yes Yes No No Yes Yes No Yes No No No Yes No Yes No No

No No Yes Yes Yes Yes No No No Yes Yes No No No No No NR No No No No

No Yes Yes Yes

No Yes Yes

No Yes Yes

No

No Yes Yes No Yes

Yes Yes Yes Yes Yes No Yes Yes No Yes Yes Yes No Yes Yes Yes Yes Yes

No Yes Yes No Yes No Yes Yes No No No No No Yes No Yes Yes No

Yes Yes No No No No Yes No Yes No No No Yes No Yes Yes No

Yes No No No No Yes Yes No No No No No Yes No Yes No No

PPHC: public primary health care level. DK: Country responded “don’t know”. NR: Country replied to survey but did not give a response to specific question. – Data not available. Source: WHO Cancer country profiles (http://www.who.int/cancer/country-profiles/en/).

No

Community/ Oral morphine home care for (formulation people with not specified) advanced stage

Availability at the public health system(yes/no) No Yes Yes No No Yes Yes Yes Yes Yes Yes Yes Yes Yes No Yes Yes No Yes Yes Yes Yes Yes Yes

0 0 2.6 – 0.8 0.9 0.2 3.4 0.8 1.2 1.9 1 0.4 0.6 0.1 0.9 0.1 – 0.2 0.3 1.6 2 0.6 0.1

– 1 1 – 34 40 8 9 5 1 9 4 17 1 26 1 12 – 2 2 10 96 3 1

– 8 7 – 237 147 27 35 28 – 15 7 106 5 31 3 55 – 19 9 26 501 6 –

No Yes Yes No No Yes Yes Yes Yes Yes Yes Yes Yes Yes No Yes Yes No Yes Yes Yes Yes Yes Yes

Yes Yes Yes No Yes No

No Yes Yes No No No No No No

No No

No No

Yes DK No No Yes No No Yes Yes

No DK No No DK No Yes No No

Cancer in the Middle East

Afghanistan Bahrain Cyprus Dibouti Egypt Iran Iraq Israel Jordan Kwait Lebanon Libya Morocco Oman Pakistan Qatar Saudi Arabia Somalia Sudan Syria Tunisia Turkey UAE Yemen

Chemotherapy (medicines not specified)

239

240

Cancer in the Middle East

50% and extending screening interval are cost-effective. More attention should be paid toward novel HPV screening strategies providing greater CC reduction and more cost-effective than cytology. Colorectal cancer (CRC) screening program adoption is variable. While some countries as Lebanon did not provide any information either optimistic or organized about it, Turkey has a wellorganized screening program. The immunochemical faecal occult blood test (i-FOBT) is the most commonly adopted test. In Oman, although CRC screening test could be implemented immediately upon physician’s order, an operational cancer policy or action plan does not exist in addition to the lack of official statistics measuring regular uptake of screening services. A survey among Omani physicians and nurses to identify their practice toward CRC screening concluded that a few percentage did not exceed 40% who practice the following for screening:

• • • • •

Order CRC screening for patients Refer patients to get CRC screening Health educates patients about CRC screening Order genetic testing for patients suspected of inherited susceptibility to CRC Health educate patients with suspected inherited susceptibility to CRC and refer them whenever possible for genetic testing

The following recommendations were perceived by Omani physicians and nurses to positively influence the process of CRC screening:

• • • • •

Availability of system to identify patients eligible for CRC screening services Continuing professional education on cancer prevention Availability of a specialist in cancer Evidence published in scientific journals (somewhat influential) Policy on cancer screening

Barriers for implementation of CRC screening test were: (a) Patients-related barriers

• • • • • • • • • • •

Fear of finding out that they have cancer. Believe screening is not effective. Embarrassment or anxiety about screening tests. Unawareness of CRC screening tests and when they should be done. Patients do not perceive CRC as a serious health problem. Culture is not favorable to procedures used for CRC screening. (b) System-related barriers: Screening costs are expensive. Long waiting time for screening appointments. Lack of hospital policy or protocol on cancer screening. Health care providers do not actively recommend CRC screening to patients. Shortage of trained health care workers to conduct CRC screening and follow-up with invasive procedures.

One of effective primary prevention tools for HCC is regular follow-up of cirrhotic patients with ultrasound. Early cancer detection measures for screening of HCC are serum alpha-fetoprotein (AFP), but still expensive with poor sensitivity and urinary markers. The urinary metabolite panel, comprising inosine, indole-3-acetate, galactose, and an N-acetylated amino acid (NAA), showed an area under ROC curve of 0.88 to differentiate between cases with HCC and only Cirrhosis. The measurement of Alfa-fetoprotein should not be used in isolation for screening or surveillance of the development of HCC in patients with hepatitis C, but, combined with ultrasound every six months and should be confined to patients with cirrhosis. Prostate cancer incidence is relatively low in ME with high incidence in Cyprus, Israel, and Somalia. Screening by total prostate-specific antigen (tPSA) may enhance early detection. A multicentered study from 22 Arab countries of the ME and North Africa investigates the proportion of patients with tPSA > 10 ng/mL who had prostate cancer. Although evidence from Europe and USA denotes around 50% chance of prostate cancer with a high level of tPSA > 10 ng/mL, there was prominent overlap in proportion of high PSA among ME patients with Prostatitis (20.5), benign prostatic hyperplasia (63.8), and prostate cancer (11.2). Absence of indicative tumor marker in light of low sensitivity and specificity of tPSA necessitates close follow-up of patients with high tPSA for a progressive decline in or return to normal PSA values with or without treatment of prostatitis, but certainly after transurethral resection of prostate (TURP) for large symptomatic BPH, to confirm the absence of prostate cancer. Persistent rising of tPSA is preliminary suggestive of prostate cancer and needs further investigation by trans rectal ultrasonography (TRUS) of the prostate and biopsy of suspicious lesions for histological diagnosis.

Palliative Care Policies Cancer remains one of the most important challenges for the population in the ME. Around 50% of cancer patients consult physicians for the first time at stage 3 or 4. Nothing is left except palliation. Palliative care (PC) is defined according to WHO as “an approach that improves the quality of life (QOL) of patients and their families facing the problem associated with

Cancer in the Middle East

241

life-threatening illness, through the prevention and relief of suffering by means of early identification, assessment, treatment of pain and other problems, physical, psychosocial and spiritual.” PC is still in its infancy in many ME countries. Application of cancer care model from western developed to developing countries proves to be unrealistic due to many reasons:

• • •

The dysfunctioning primary health care system: Lack of well-trained family physicians in many countries compels cancer patients to seek treatments at tertiary care centers for minimal symptoms that could be managed by trained local treating physicians. Current beliefs that hinder the best cancer care: These include cultural, religious, and family opinions. Fear and stigma associated with cancer let many patients hesitate to seek treatment. Cultural concepts affect the female patients’ adherence to treatment and follow-up due to insufficient professional female caregivers. Also, many terminal cases refuse cancer care due to “God’s will” concept; therefore, patients are less demanding all possible treatments. Limited available cancer treatment in terms of radiotherapy and chemotherapy: While there is debate in developed countries about the use of newly advanced technology radiotherapy (e.g., proton beam, IGRT, IMRT, etc.), there are debates in some countries of the ME about who gets treated and who does not. Radiotherapy is cost-effective and 40% of cancer cure could be achieved by radiotherapy that should be available in every ME country. Analgesic opioids consumption is also the lowest in ME. Many physicians face patients’ treatment rejection for fear of addiction.

One of the strategic actions for cancer prevention and control in the Eastern Mediterranean region was to improve cancer management and support PC and pain relief through the following action plans:

• • • • • •

Strengthen cancer diagnosis and treatment programs through all levels of care. Promote and implement interventions in childhood cancers at different levels of the health system. Strengthen the development of human resources in cancer management. Develop or strengthen PC services, including promotion of community nursing and home care. Ensure accessibility and affordability of PC medicines. Support integration of cancer management and PC in primary health care.

It is crucial to put resources difference apart from in ME countries and to develop a third part collaboration to integrate PC like Middle East Cancer Consortium (MECC). It is a nongovernmental unique valuable organization that collaborates with regional ministers of health and international health care organizations. Members of it include Egypt, Jordan, Palestine, Israel, Cyprus, Turkey, and the USA. Countries such as Pakistan, United Arab Emirates, Lebanon, Morocco, Iraq, Sudan, Qatar, and Oman are actively participating in MECC’s training programs. MECC seeks to over bridge the barriers of palliative care by offering training for oncologists, developing palliative care programs, and improving communication between health care givers and patients. Many countries are in the capacity building stage to improve PC. Turkey initiated a model to other countries through a novel program whereby population-based and home-based palliative care teams were developed throughout the country, including peripheral areas where appropriate care was not available. Bahrain started PC unit in 2009 that serves patients through followup, pain clinics, and a hotline for home care patients. Jordanian Ministry of health is now integrating PC in inpatient and outpatient services with around 400 new patients per year and care for about 60–80 patients a month (30–40 patients through home care). However, attention should be paid to postgraduate education and training programs for health professionals. Oman is still in the earliest stage of PC services. The increasing incidence of cancer in Oman necessitates training of oncologists especially for opioid administration, and awareness programs for patients with cancer to help with the opioidphobia. Egypt with > 90 million residents in 2015 and with cancer prevalence of 215,000 cases in 2012 has no oral morphine which is problematic in patients with end-stage cancer. Slow-release morphine tablets (30 mg) were previously prescribed and manufactured under licence by a single supplier in the United Kingdom of Great Britain and Northern Ireland. Nowadays, it is not available in Egypt. Immediate action should be taken to avoid such shortage and support cancer pain relief. PC is not also well established in other countries as in Iraq and Syria relative to advances in Saudi Arabia where specialized PC centers are available and embrace intensive management unit, consultation service, outpatient clinics, and home health care programs. The Saudi PC team consists of physicians, nurses, social workers, and other supportive care providers (dieticians, physical therapist, a unit translator, home health care nurses, health educator, and pharmacists). Dedicated PC centers in UAE and Qatar were well established and efforts are made to best integrate home health care services.

See also: Cancer Disparities. Cancer in Sub-Saharan Africa. Cancer in Populations in Transition.

Further Reading El-Akel, W., El-Sayed, M.H., Kassas, M.E., et al., 2017. National treatment programme of hepatitis C in Egypt: Hepatitis C virus model of care. Journal of Viral Hepatitis. https:// doi.org/10.1111/jvh.12668. Fadhil, I., Lyons, G., Payne, S., 2017. Barriers to, and opportunities for, palliative care development in the eastern Mediterranean region. Lancet Oncology 18 (3), e176–e184. https://doi.org/10.1016/s1470-2045(17)30101-8.

242

Cancer in the Middle East

Forman D, Bray F, Brewster DH, et al. 2013. Cancer Incidence in Five Continents (CI5). vol. 10. Retrieved from http://publications.iarc.fr/Databases/Iarc-Cancerbases/CancerIncidence-In-Five-Continents-Ci5-Volumes-I-To-X-2014. Giordano, L., Bisanti, L., Salamina, G., et al., 2016. The EUROMED CANCER network: State-of-art of cancer screening programmes in non-EU Mediterranean countries. European Journal of Public Health 26 (1), 83–89. https://doi.org/10.1093/eurpub/ckv107. Hacikamiloglu, E., Utku, E.S., Cukurova, Z., et al., 2016. Community palliative care in Turkey: A collaborative promoter to a new concept in the Middle East. Journal of Public Health Management and Practice 22 (1), 81–88. https://doi.org/10.1097/phh.0000000000000252. Kim, J.J., Sharma, M., O’Shea, M., et al., 2013. Model-based impact and cost-effectiveness of cervical cancer prevention in the extended Middle East and North Africa (EMENA) (provisional abstract). Vaccine 31 (Suppl. 6), G65–G77. http://onlinelibrary.wiley.com/o/cochrane/cleed/articles/NHSEED-22013054067/frame.html. Lehman, E.M., Wilson, M.L., 2009. Epidemiology of hepatitis viruses among hepatocellular carcinoma cases and healthy people in Egypt: A systematic review and meta-analysis. International Journal of Cancer 124 (3), 690–697. https://doi.org/10.1002/ijc.23937. Lemoine, M., Eholié, S., Lacombe, K., 2015. Reducing the neglected burden of viral hepatitis in Africa: Strategies for a global approach. Journal of Hepatology 62 (2), 469–476. Lemoine, M., Thursz, M.R., 2017. Battlefield against hepatitis B infection and HCC in Africa. Journal of Hepatology 66 (3), 645–654. https://doi.org/10.1016/j.jhep.2016.10.013. Omar, S., Alieldin, N.H., Khatib, O.M., 2007. Cancer magnitude, challenges and control in the eastern Mediterranean region. Eastern Mediterranean Health Journal 13 (6), 1486–1496. Sancho-Garnier, H., Khazraji, Y.C., Cherif, M.H., et al., 2013. Overview of cervical cancer screening practices in the extended Middle East and North Africa countries. Vaccine 31 (Suppl. 6), G51–57. https://doi.org/10.1016/j.vaccine.2012.06.046. Sharma, M., Seoud, M., Kim, J.J., 2017. Cost-effectiveness of increasing cervical cancer screening coverage in the Middle East: An example from Lebanon. Vaccine 35 (4), 564– 569. https://doi.org/10.1016/j.vaccine.2016.12.015. Silbermann, M., Daher, M., Fahmi-Abdalla, R., et al., 2015. The Middle East cancer consortium promotes palliative care. Lancet 385 (9978), 1620–1621. https://doi.org/10.1016/ s0140-6736(15)60791-7. Silbermann, M., Epner, D.E., Charalambous, H., et al., 2013. Promoting new approaches for cancer care in the Middle East. Annals of Oncology 24 (Suppl. 7), vii5–vii10. https:// doi.org/10.1093/annonc/mdt267. Strategy for cancer prevention and control in the Eastern Mediterranean Region 2009–2013. Retrieved from. http://www.emro.who.int/health-topics/cancer/index.html (30 May 2017)

Relevant Websites http://globocan.iarc.fr/dWHO. International Agency for Research on Cancer. GLOBOCAN 2012: Estimated Cancer Incidence, Mortality and Prevalence Worldwide in 2012. http://www.emro.who.intdWorld Health Organization. Regional Office for the Eastern Mediterranean.

Cancer Risk Reduction Through Lifestyle Changesq Christina Bamia, National and Kapodistrian University of Athens, Athens, Greece © 2019 Elsevier Inc. All rights reserved.

Glossary Epidemiology The science concerned with the study of factors determining and influencing the frequency and distribution of disease and other health-related events and their causes in a defined human population. Micronutrient A vitamin or mineral that the body must obtain from outside sources. Micronutrients are essential to the body in small amounts because they are either components of enzymes (the minerals) or act as coenzymes in managing chemical reactions. Nutrigenetics The science to identify how genetic variation affects response to nutrients. Nutrigenomics The science of the effects of foods and food constituents on gene expression. Oncogene Mutated and/or overexpressed version of a normal gene that in a dominant fashion can release the cell from normal restraints on growth and thus, alone or in concert with other changes, convert a cell into a tumor cell. Pharmacotherapy The treatment of diseases or conditions by medicines. Polymorphism The occurrence of different forms (alleles) of a gene (typically > 1%) in individual organisms or in organisms of the same species, independent of sexual variations. Primary prevention The identification, control, and avoidance of environmental factors related to cancer development. Randomized, controlled trials A clinical trial that uses a control group of people given an inactive substance (placebo) and an intervention group given the substance or action under study. Sunscreen A substance applied to the skin to protect it from the effects of the sun’s rays; sunscreens act by either absorbing ultraviolet (UV) radiation or reflecting incident light. Their effectiveness is rated by their sun protection factor (SPF); for example, a sunscreen with an SPF of 15 allows only 1/15 of the incident UV radiation or light to reach the skin.

Introduction Cancer remains a leading cause of mortality. It is estimated that by 2020 there will be 16 million new cancer cases, and 10 million related deaths, whereas there may be > 20 million new cases of cancer in 2030. Lifestyle behaviors are modifiable key factors for cancer prevention and cancer survivorship. It is now acknowledged that lifestyle habits including tobacco use, poor diet, physical inactivity, alcohol abuse and overexposure to sunlight are associated with 50% to 70% of all cancers, meaning that 281,000 to 395,000 cancer deaths each year could be prevented through changes in health behaviors in population level. Effective interventions have been suggested in clinical trials for modifying lifestyle habits and specifically so for reducing tobacco use, increasing physical activity, reducing sun exposure, and reducing alcohol excess intake. Nonetheless millions of adults are engaged in risky behaviors: approximately 23% smoke, < 25% follow dietary recommendations (e.g., about fruits and vegetables) and about 40% are physically inactive. The previous figures may indicate a failure in transferring evidence-based findings to the general public. Therefore, disseminating the scientific knowledge about the importance of behavioral changes in cancer prevention and survivorship, as well as, delivering, through health care systems, effective interventions towards healthier lifestyle choices, are major, urgently needed tasks for public health policy makers worldwide.

Diet and Cancer The association of diet with health outcomes, including cancer is of special interest due to the essential role of diet in human being. On the other hand, diet is a modifiable risk factor which can be directed towards healthier choices, in individual and population level through clinical advice and public health policies. Many studies focused on single foods, food groups or nutrients in association to cancer incidence/mortality worldwide. In addition, studies have also focused on the association of dietary patterns with cancer incidence and survival. Notably, the importance of the growing body of research regarding the benefits of healthy eating in cancer prevention/mortality is reflected in the recent reports of the World Cancer Research Fund (WCRF)/American Institute

q

Change History: November 2017. C. Bamia updated the text and updated Table 1 with two additional smoking cessation methods. This article is an update of Peter Greenwald, Darrell E. Anderson, and Sharon S. McDonald, Cancer Risk Reduction (Diet/Smoking Cessation/Lifestyle Changes), in Encyclopedia of Cancer (Second Edition), edited by Joseph R. Bertino, Academic Press, 2002, Pages 311–317.

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.96060-4

243

244

Cancer Risk Reduction Through Lifestyle Changes

for Cancer Reduction (AICR) and in certain dietary guidelines (e.g., US Department of Health and Human Services and US Department of Agriculture, 2015).

Individual Foods, Food Groups and Nutrients The strength of the overall evidence for specific dietary constituents in relation to cancer incidence depends on the number and quality of published studies, as well as, on the consistency of the reported findings. The level of evidence can be ranked from “low/unlike” to “convincing”. In the 2017 major review of WCRF/AICR on the role of diet, somatometry and physical activity in cancer risk and mortality, the global research based on studies published up to 2017, was evaluated. The evidence for increased cancer risk was rated as “convincing” for the following associations: aflatoxins and liver cancer; processed meat and colorectal cancer; beta-carotene, in the form of high dose supplements, and lung cancer; alcohol intake and cancers of: (a) mouth, pharynx and larynx, (b) esophagus (squamous cell carcinoma), (c) liver, (d) colorectum and (e) postmenopausal breast. In addition, the evidence for increased cancer risk was evaluated as “probable” for the following associations: red meat and colorectum cancer; processed meat, foods preserved by salt and alcohol and stomach cancer; alcohol and premenopausal breast cancer. On the other hand, the evidence concerning foods and nutrients associated with decreased risk of cancer has been evaluated at the most as “probable’ (but not convincing). Such associations have been highlighted for: colorectum cancer and consumption of wholegrains, foods rich in dietary fiber, dairy products and calcium intake, either from diet or from through supplements (200– 1000 mg/day); lung cancer and fruit intake; mouth, pharynx and larynx and fruit, as well as, starchy vegetables intake; liver and endometrium cancers with increased coffee consumption, and; kidney cancer with alcohol intake. Associations that have been investigated, but were judged as “unlikely” were those for beta carotene intake either through diet or as supplement intakes in relation to prostate and skin cancer risk.

Dietary Patterns Dietary patterns (DPs) are an alternative approach to assess diet in a holistic way, that is, as a combination of, rather than as isolated, dietary components. A DP can be hypothesis-driven (a priori DP), expressing adherence to a distinct diet for example, the Mediterranean Diet (MD), or, level of compliance with formal dietary guidelines such as the WCRF/AICR guidelines for cancer prevention. DPs can be also empirically derived (a posteriori DPs) through the application of mathematical/statistical methods which identify suitable combinations of a set of dietary variables.

A-priori DPs The Mediterranean Diet Score (MDS), which quantifies the traditional diet of people living in olive-growing areas around the Mediterranean basin is characterized by high intake of vegetables, fruits, fish (high in n-3 fatty acids), legumes, cereals and olive oil (high in monounsaturated fatty acids), low intake of dairy and meat, and moderate alcohol intake, and has been repeatedly investigated with respect to cancer incidence/mortality. Other frequently-used indices in studies of cancer outcomes include: (a) the Healthy Eating Index (HEI) based on the 1992 USDA Food Guide Pyramid and the 1995 Dietary Guidelines for Americans, (b) the Dietary Approaches to Stop Hypertension (DASH) diet index, (c) the Diet Diversity scores, and (d) WCRF/AICR index, as well as variations of the above.

A-posteriori DPs A-posteriori DPs may include both, “healthy” and “unhealthy” dietary choices and are labeled in various ways, the most frequent ones being: “prudent”, “vegetable/fruit”, “traditional”, “western-type”, “animal”, “fat and salty”, “refined”, “alcohol-based”, “drinking”, based on the foods mainly characterizing these patterns and their suspected relation with specific cancer sites. Aposteriori DPs that share common labels (e.g., “healthy”) may, however, include different foods/food groups.

Dietary patterns and cancer incidence/mortality Many systematic reviews and meta-analyses for adult populations have been undertaken on the indicated a priori and a posteriori DPs in relation to cancer risk/mortality overall, as well as, by cancer site. A 2016 comprehensive review examined 64 studies of prospective design, including many large multiregion cohorts reporting on a priori dietary patterns. Meta-analyses have been also undertaken for quantifying the overall evidence regarding the association of MDS, as well as, HEI, AHEI, and DASH diet with cancer incidence or mortality. A posteriori patterns have been also studied in systematic reviews in relation to colorectal, breast, gastric, esophageal and lung cancer. Accumulated evidence from the aforementioned research indicates that there may exist a priori and a posteriori DPs that are associated with reduced cancer risk, and specifically so for colon cancer, breast cancer (especially for the postmenopausal breast cancer) and gastric cancer. From the a priori DPs, collective evidence seems to consistently indicate inverse associations of overall cancer risk with MDS, HEI, AHEI and DASH dietary indices and variations of those. The same holds for a posteriori DPs that have been considered/labeled as “healthy”. Evidence is less conclusive for DPs that have been found to be associated with increased cancer risk, mainly the a posteriori “unhealthy” DPs. Interactions likely occur among many dietary constituents. However, neither these interactions nor their influence on cancer risk are well understood. Thus, at present it is difficult to tease out the specific effects of individual dietary components from diet and cancer research data.

Cancer Risk Reduction Through Lifestyle Changes

245

Randomized, Controlled Trials Most of the evidence relating diet with cancer risk/mortality relies on observational studies. On the other hand randomized controlled trials (RCTs) offer one of the best means for testing diet and cancer hypotheses developed from the insights provided by epidemiologic and experimental studies. Specific dietary constituents have been investigated in RCTs such as vitamins A, C, D and E, folic acid, calcium, and selenium, beta-carotene etc. Interventions in these trials consisted of high-dose daily supplements of the aforementioned vitamins compared to placebo or to no special dietary advice in the group of controls. From these, vitamin D, calcium and selenium have shown promise in reducing overall/site-specific cancer risk. A few RCTs have been also undertaken with respect to dietary patterns such as low-fat diets for the prevention of breast cancer and MD for the prevention of cancer in general, but due to limited number of studies results are still inconclusive.

Gene–Nutrient Interactions Recently, research on the relationship between diet and cancer has also focused in gene–nutrition interactions, since genetic factors play a key role in the development of cancer and are affected by nutrition. Exposure to the same quantitative level of dietary factor(s)/carcinogens can increase cancer risk in one individual but not in another, depending on specific susceptibilities to gene–nutrient interactions. Therefore, studying the gene–nutrition interactions may help to explain inter-individual differences in response to the same nutritional intakes and, therefore, may shed light to the underlying mechanisms linking diet to the risk of developing cancer. Nutrigenetics and nutrigenomics are the related research areas and are defined according to Wikipedia as: “the science to identify how genetic variation affects response to nutrients (nutrigenetics)”, and “the science of the effects of foods and food constituents on gene expression (nutrigenomics)”, respectively. Genes are involved in carcinogenesis through metabolic activation/detoxification, DNA repair, chromosome stability, activity of oncogenes or tumor suppressors, cell cycle control, signal transduction, hormonal pathways, vitamin metabolism pathways, immune function, and receptor or neurotransmitter action. To date many animal and observational studies have evaluated a number of hypotheses regarding specific diet–gene interactions with respect to cancer risk and prognosis, focusing on cancer sites for which certain diet-related genes, as well as, nutrients/foods have been identified as possible risk factors. Perhaps, some results on colorectal cancer risk may hold promise, for example, the interplay between high consumption of well-done meat (as a source of exposure to heterocyclic amines (HAAs)), and NAT2 and CYP1A2 phenotypes; the association of folic acid and vitamin 12 with DNA methylation and DNA synthesis processes; the association of fruit and vegetable intake with K-ras mutations. Results, however, from such studies are still in very early stage. Understanding gene–nutrient interactions and individual differences in genetic susceptibilities will enhance, in the long run, the development of personalized nutrition strategies for cancer prevention. Moreover it will help researchers to design future dietary intervention strategies to reduce cancer risk. Focusing on polymorphisms in intervention studies will allow investigators to develop study designs, stratify participants, and analyze results based on genetic differences within the study population. For example, polymorphisms in genes affecting the use of vitamin D seem to confer different levels of risk for prostate cancer. This kind of information is important for researchers in designing trials to investigate the role of vitamin D, or its analogs, in prostate cancer risk. Notwithstanding the promising results of nutrigenetics and nutrigenomics, it should be acknowledged that, at this point in time, any evidence coming from this area is still at premature stage. The challenge for the future is to use any knowledge gained about how genetic polymorphisms influence cancer risk through specific dietary constituents or patterns in order to allow for tailor-made dietary recommendations based on genetic testing or gene expression, as well as, metabolic/nutritional status biomarkers. Under this concept of individualized dietary recommendations, diagnosis, prevention and treatment of cancer could be optimized.

Dietary Guidelines As research into the role of diet in cancer risk continues, it is important that clinicians and the public be advised about the importance of modifying diets to reduce cancer risk. Since the mid-1970s, various scientific organizations around the world, including the WCRF/AICR, the American Cancer Society (ACS), and the US National Cancer Institute (NCI), have developed dietary guidelines to promote cancer risk reduction as a population strategy. As an example, the WCRF dietary recommendations for the prevention of cancer advise to: (1) avoid high-calorie foods (e.g., processed foods high in fat or sugar intake) and sugary drinks, (2) eat more of wholegrains, vegetables, fruit and pulses (e.g., legumes and beans), (3) limit red meat intake to no more than 500 g/week (cooked weight) and avoid totally, if possible, processed meat (e.g., ham and bacon), (4) avoid alcohol intake or limit alcoholic drinks to the level suggested by national guidelines, (5) avoid eating foods preserved in salt and limit salt intake to less than 6 g/day, (6) avoid mouldy grains and cereals because they may be contaminated with aflatoxins, and, (7) do not rely on supplements but follow a healthy diet in order to protect against cancer. Dietary guidelines are not followed, however, by the general public. For example the average American consumes too much fat (too much saturated fat, too little monounsaturated fat), too little fiber, and too few vegetables, fruits, and whole grains. This pattern is seen in many developed countries worldwide. It is therefore urgently needed that physicians and nutritionists try to better educate people about components of healthful and cancer-preventing diets and small changes in eating patterns that can lower cancer risk and improve cancer outcomesdthis is particularly important nowadays that information on “good” or “bad” diets has become readily available but with questionable scientific gravity and validity.

246

Cancer Risk Reduction Through Lifestyle Changes

Smoking Cessation Smoking and Cancer Risk In the United States, health conditions related to smoking are responsible for 480,000 deaths each year, 12 years in advance of what would be normally expected (given sex and age distributions). These figures translate to an aggregated annual loss of > 5 million life-years. Moreover, about 32% of cancer deaths and 20% of all premature deaths in the United States are attributable to smoking. Tobacco use is a major avoidable contributor to the cancer burden, accounting for one third of all cancer deaths and 87% of lung cancer deaths each year. Cigarette smoking is a risk factor contributing highly to the amount of cancers not only of the lung, but also of trachea, bronchus, larynx, pharynx, oral cavity, nasal cavity, and esophagus, and is also related to cancers of the pancreas, kidney, bladder, stomach, and cervix. At least 20 carcinogens (e.g., the polycyclic aromatic hydrocarbons (PAHs), including the carcinogen benzo[a]pyrene (BaP)) and the nicotine-derived tobacco-specific nitrosamine, 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK)) that are components of tobacco smoke cause lung tumors in either animals or humans. Smoking also affects nonsmokers. Secondhand tobacco smoke as a result of household or occupational exposure, is responsible for inhalation and metabolization of components of the smoke and, thus, has been associated with increased incidence of lung cancer. In 2014, estimated rates of current smokers in the US adults were 18.8% and 14.8% among men and women, respectively. The prevalence of smoking is inversely associated with education level and with years of education. In the last 30 years, the prevalence of current smoking decreased overall and especially so among men. This decline in smoking prevalence since 1980s was followed by a decline in mortality rates from lung cancer in men. Smoking cessation has a dramatic impact in reducing the risk of cancer among former smokers even if these quit smoking at older ages. Smokers who stop smoking before the age of 50, succeed in halving mortality risk within 15 years after quitting as compared with people who continue to smoke. In specific, lung cancer risk after 10 years of abstinence is 30% to 50% lower than that of people who continue smoking, whereas the risk of oral and esophageal cancer is 50% lower in former (as compared to current) smokers after 5 years of smoking cessation. From the above figures it is evident that educational strategies to prevent the start of smoking and promotion of effective approaches to permanently stop tobacco use by smokers are very important elements in reducing smoking-related cancer risk. Policy strategies encouraging reduction of tobacco use at the community level in various countries have been credited with notable reductions in smoking prevalence. These strategies include, pursuing smoke-free laws, media campaigns, increasing the minimum legal age of access to tobacco products, increasing the cost and/or apply excise taxes to tobacco, and offering quitting programs to people through primary health and other health care organizations. About 70% of smokers claim to be interested in quitting smoking. When smokers, however, try to quit on their own, only about 7% succeed. Long-term success rates can be increased to 15%–30% if smoking cessation interventions are used. These include pharmacotherapies, as well as, behavioral therapies such as individual or group counseling, intensive physician advice, or combinations of the above.

Pharmacotherapies for Smoking Cessation Pharmacotherapies are effective interventions for smoking cessation. These can be divided to nicotine replacement therapies (NRTs) and non-nicotine medications (Table 1). Nicotine, present in all tobacco products, is an addictive substance. For most users, tobacco use results in drug dependence, although in different levels, depending on patterns and quantity of daily smoking. Evidence demonstrating that pharmacotherapies can help people addicted to nicotine to quit smoking is strong and consistent, and NRTs such as nicotine gum, patch, spray, lozenge, and inhaler have been approved for smoking cessation by the US Food and Drug Administration. A 2004 meta-analysis of 110 randomized trials concluded that NRT treatments, alone or in combination, improved significantly cessation rates over placebos after 6 months of use. There are also non-nicotine pharmacotherapies that have been efficacious for smoking cessation, by lessening nicotine craving and withdrawal symptoms, namely bupropion, an antidepressant, and varenicline, a selective alpha-4-beta-2 nicotinic receptor. Based on 31 randomized trials comparing bupropion to placebo, bupropion was associated with a statistically significant increase in the likelihood of quitting smoking after 6 months of follow-up. Varenicline also improves cessation rates over placebo as shown Table 1

Pharmacotherapies for smoking cessation

Pharmacotherapy

Availability

Side effects

Nicotine gum Nicotine patch Nicotine Lozenges Nicotine nasal spray Nicotine inhaler Bupropion Varenicline

Over-the-counter only Over-the-counter and prescription Prescription only Prescription only Prescription only Prescription only Prescription only

Mouth soreness, indigestion, stomatitis, sore throat Skin reaction at patch site, insomnia Local irritation (warmth and tingling) Nasal/throat irritation, sneezing, coughing Mouth/throat irritation, coughing Insomnia, dry mouth, dizziness, rhinitis Nausea, insomnia

Cancer Risk Reduction Through Lifestyle Changes

247

in a meta-analysis of nine trials including 7267 participants (4744 under varenicline) where the pooled risk ratio (RR) for continuous abstinence at  6 months for varenicline versus placebo was 2.33. Moreover in this meta-analysis varenicline was shown to be superior to bupropion or NRTs after 1 year of follow-up. Both bupropion and varenicline have certain adverse effects. In 2009 and after post marketing surveillance the risk of serious neuropsychiatric symptoms associated with both products were added to the Boxed Warnings Symptomsdthese include: “changes in behavior, hostility, agitation, depressed mood, suicidal thoughts and behavior, and attempted suicide.”

Combinations of pharmacotherapy Using more than one NRTs seems to increase smoking cessation rates as compared to using a single type NRT. Some studies have shown that NRT in combination with bupropion may be more efficacious than bupropion alone.

Behavioral Interventions Many behavioral interventions are available to encourage smoking cessation in adults. These interventions can be delivered in the primary or in the community care settings and include in-person behavioral support and counseling, telephone counseling, and self-help materials. It has been shown that behavioral interventions may increase smoking abstinence from a baseline 5%–11% in control groups to 7%–13% in intervention groups. Effective individual or group counseling is essential to help a smoker, who is willing to quit, achieving long-term success in smoking cessation and can be delivered by various types of primary care providers, such as physicians, nurses, psychologists, social workers, and specialized counselors. Counseling aims to helping smokers to organize a specific plan on how to quit, acknowledge situations that increase their risk of smoking and, accordingly, develop resistance skills. Effective counseling, therefore, offers basic information about smoking/quitting, describes coping strategies and identifies sources of social support that can be useful (e.g., community resources). Physician-advice of either minimal (< 20 min in 1 visit) or intensive ( 20 min plus > 1 follow-up visit) type is also effective for successful smoking cessation. Relevant studies seem to suggest a dose–response relationship of the intensity of counseling and smoking cessation rates. Accordingly, the Public Health Service guidelines of the United States, state that “smokers should receive at least 4 in-person counseling sessions”. Telephone counseling provided by trained counselors or health care providers can be also effective. Self-help materials targeting to the needs of the individual smoker rather than simply presenting the adverse health consequences of smoking can be also helpful for smoking abstinence. Recently, computer- or mobile phone-based programs have been also proposed as behavioral interventions for smoking cessation.

Combinations of behavioral and pharmacotherapy interventions Interventions that combine behavioral and pharmacotherapy strategies seem to increase cessation rates from about 8% to 14% as compared to these interventions given separately. These combinations frequently consist of NRTs and a variety of behavioral approaches usually delivered in > 1 sessions from specialized staff.

Tobacco Cessation Guidelines In 2008 the US Public Health Service, published the updated 2008 guidelines, “Treating Tobacco Use and Dependence” which are available online. In brief, the broad elements of these guidelines include: the need from primary care providers of: recording the smoking status of every patient at every visit, enquiring about his/her willingness to quit (if smoker) and informing those who wish to quit about effective interventions, brief or intense. The guidelines also emphasize on the importance of: (a) encouragement and assistance by the clinicians to those who are trying to quit smoking, (b) close monitoring of the process of smoking cessation in patients’ every visit and, (c) training patients under smoking cessation programs in order to confront with situations that may provoke their smoking-abstinence program.

Lifestyle Changes Physical Activity Physical activity (PA) is any form of movement using skeletal muscles. Physical activity increases energy expenditure and is a factor of energy balance. Different activities can be classified according to when (time window) and how (in terms of intensity) these are realized. According to “when”, activities are usually grouped as occupational, household, transport, and recreational. Based on “how”, PAs can be vigorous, moderate, light, or sedentary. People can be moderately or vigorously physically active at work (manual/labor) or can be occupied in jobs that request light or sedentary PA. In transport (walking, riding, cycling, driving or using public transportation) people can be lightly or moderately physically active. Exercise and other forms of physical training but also activities done during household are types of recreational physical activitydthese may be of light, moderate, or vigorous intensity depending on the

248

Cancer Risk Reduction Through Lifestyle Changes

nature/intensity of activities, hobbies, and pursuits. Total PA levels for a given timeframe (e.g., day, month, year, etc.) are evaluated as the combination of frequency, intensity, and duration of all activities that take place at the specific time frame. In the 2017 WCRF/AICR report the only factor for which the evidence regarding its inverse association with cancer risk was judged as “convincing” was PA (moderate and vigorous) in relation to colon cancer. According to this report, PA also “probably” decreases the risk of post- and pre-menopausal breast, as well as, of endometrial cancers. The evidence suggesting that PA protects against cancers of the lung and pancreas was evaluated as “limited.” The report also notes that “. all types and degrees of physical activity are or may be protective, excluding extreme levels of activity: the evidence for any specific type or degree of physical activity is limited.” These results are consistent with the message that, the more physically active people are the better. Regular PA may be associated with reduced cancer risk through various mechanisms, including altering hormone levels and increasing immune system activity, energy expenditure, and antioxidant activity. Given that PA is a way to preserve energy balance and retain normal body weight it seems that PA protects against cancers the risk of which is increased by overweight, weight gain, and obesity, such as cancer of the colon, rectum, prostate, endometrium, breast, and kidney.

Physical activity guidelines Most people in urbanized and industrialized settings worldwide lead sedentary ways of life. The amount of PA needed to maintain a healthy weight, lose weight, and promote good health, including reducing cancer risk, as recommended by various organizations in the United States, including the NCI, ACS, and WCRF/AICR is, at least 30 min of moderate physical activity on most days of the week either continuously, or, in time intervals. This level of activity might include walking briskly (3–4 miles/hour) for about 2 miles, gardening and yard work, jogging, or swimming. In the recommendations of WCRF/AICR it is also emphasized that one: (a) should limit sedentary habits such as watching television and, (b) as fitness improves, he/she should aim for 60 min or more of moderate, or for 30 min or more of vigorous, physical activity every day.

Sun Exposure Cancer of the skin is the most frequent type of cancer. Basal cell and squamous cell carcinomas, both highly curable, account for the majority of skin cancers whereas malignant melanoma, the most dangerous form of skin cancer is less frequent. The incidence of skin cancer is increasing over the last 30 years, partly due to an increase in awareness about and detection of this cancer. It is estimated that non melanoma skin cancers (NMSCs) affected 3,000,000 subjects living in the United States in 2012. For melanoma skin cancer the number of people living in the United States who will be diagnosed with melanoma in 2017, is predicted to be about 87,000 whereas the associated deaths are expected to be about 10,000. People with red or blond hair and fair skin that freckles, tans poorly or burns easily are at especially high risk for skin cancer. Increased (in intensity and/or duration) exposure to sunlight, the main source of ultraviolet (UV) radiation, is implicated as a causative factor in the development of skin cancer in observational studies. Abundant evidence has established that skin cancer risk can be reduced by increasing awareness in people about skin type prone to skin cancer and of skin damage (burn and solar keratoses) due to sun exposure, as well as, by limiting exposure to sunlight and, thus, to UV radiation. For melanoma cancer there is some evidence that patterns of sun exposure may also be important. In 2009 a pooled analysis of 15 case–control studies (5700 melanoma cases and 7216 controls) investigated patterns of sun exposure, sunburn and solar keratoses (only three studies) with melanoma cancer risk. This study concluded that melanoma cancer risk at different body sites is associated with different amounts and patterns of sun exposure. Recreational sun exposure and sunburn were consistently associated with melanoma cancer at all latitudes, whereas occupational and total sun exposure were associated with melanoma risk, predominately at low latitudes. Suggested effective protective behaviors include avoiding the sun during midday (especially between 10 AM and 2 PM), wearing protective clothing and broad-brimmed hats, wearing sunglasses, avoiding tanning beds and sun lamps (these are also sources of UV radiation), and using sunscreen that has a sun protection factor (SPF) of 15 or higher, even on hazy or cloudy daysdhowever the respective evidence of these strategies acting as interventions for reducing NMSCs is not yet conclusive. The relationship between sunscreen use and melanoma cancer risk, as practiced in the general population, is, also, not absolutely clear. Many epidemiologic studies investigating this association have found either reduced risk or no clear association. There are also findings from some studies suggesting that sunscreen use is associated with an increased risk for melanoma. This may be attributed to the fact that people who use sunscreen primarily to avoid sunburn during intentional sun exposure, might increase their sun exposure time when using sunscreen, thus increasing their exposure to UV radiation and risk for melanoma. A meta-analysis of 18 studies that evaluated the association between sunscreen application and melanoma risk highlighted different degrees of quality among the included studies and estimated an overall weak inverse association. At present, accumulated evidence on sun exposure and skin cancer warrants using sunscreen as part of an overall sun protection strategy, during both intentional and unintentional exposures (e.g., in gardening or hiking).

See also: Cervical Cancer: Screening, Vaccination, and Preventive Strategies. Dietary Factors and Cancer. Diet and Cancer. Environmental and Occupational Exposures.

Cancer Risk Reduction Through Lifestyle Changes

249

Further Reading American Cancer Society, 2017. Cancer facts & figures 2017. American Cancer Society, Atlanta. https://www.cancer.org/content/dam/cancer-org/research/cancer-facts-andstatistics/annual-cancer-facts-and-figures/2017/cancer-facts-and-figures-2017.pdf. Braicu, C., Mehterov, N., Vladimirov, B., et al., 2017. Nutrigenomics in cancer: Revisiting the effects of natural compound. Seminars in Cancer Biology 46, 84–106. Byers, T., 1999. What can randomized controlled trials tell us about nutrition and cancer prevention? CA: a Cancer Journal for Clinicians 49, 353–361. Fenech, M., El-Sohemy, A., Cahill, L., et al., 2011. Nutrigenetics and nutrigenomics: Viewpoints on the current status and applications in nutrition research and practice. Journal of Nutrigenetics and Nutrigenomics 4, 69–89. Hu, F.B., 2002. Dietary pattern analysis: A new direction in nutritional epidemiology. Current Opinion in Lipidology 13, 3–9. Jamal, A., Agaku, I.T., O’Connor, E., King, B.A., Kenemer, J.B., Neff, L., 2014. Current cigarette smoking among adults–United States, 2005–2013. MMWR. Morbidity and Mortality Weekly Report 63, 1108–1112. Klein, W.M.P., 2014. Behavioral research in cancer prevention and control: A look to the future. American Journal of Preventive Medicine 46, 303–311. Lancaster, T., Stead, L., 2004. Physician advice for smoking cessation. The Cochrane Database of Systematic Reviews 4, CD000165. Lemmens, V., Oenema, A., Knut, I.K., Brug, J., 2008. Effectiveness of smoking cessation interventions among adults: A systematic review of reviews. European Journal of Cancer Prevention 17, 535–544. Potter, J., Brown, L., Williams, R., Byles, J., Collins, C., 2016. Diet quality and cancer outcomes in adults: A systematic review of epidemiological studies. International Journal of Molecular Sciences 17, 1052. Siu, A., US Preventive Services Task Force, 2015. Behavioral and pharmacotherapy interventions for tobacco smoking cessation in adults, including pregnant women: U.S. Preventive Services Task Force Recommendation Statement. Annals of Internal Medicine 163, 622–634. Stead, L.F., Perera, R., Bullen, C., Mant, D., Lancaster, T., 2004. Nicotine replacement therapy for smoking cessation. Cochrane Database of Systematic Reviews 3, CD000146. Saraiya, M., Glanz, K., Briss, P., Nichols, P., White, C., Das, D., 2003. Preventing skin cancer: Findings of the task force on community preventive services on reducing exposure to ultraviolet light. MMWRdRecommendations and Reports 52 (RR-15), 1–12. Task Force on Community Preventive Services, 2002. Recommendations to increase physical activity in communities. American Journal of Preventive Medicine 22, 67–72. World Cancer Research Fund and American Institute for Cancer Research, 2007. Food, nutrition, physical activity and the prevention of cancer: A global perspective. AICR, Washington, DC. World Cancer Research Fund and American Institute for Cancer Research, 2017. Cancer prevention and survival. Continuous Update Project. AICR, Washington, DC.

Cancer Survival and Survivorship Dana Hashim, International Agency for Research on Cancer, Lyon, France; and Icahn School of Medicine at Mount Sinai, New York, NY, United States Elisabete Weiderpass, Cancer Registry of Norway, Oslo, Norway; Karolinska Institute, Stockholm, Sweden; UiT The Arctic University of Norway, Tromsø, Norway; and Folkhälsan Research Center, Helsinki, Finland © 2019 Elsevier Inc. All rights reserved.

Glossary Actuarial life-table method The survival experience of a group is summarized by cumulative survival probabilities that are calculated only at the end of each equally spaced follow-up interval. Cancer fatality rate The proportion of individuals who experienced death within the population of individuals diagnosed with cancer, over the course of the disease. Cancer mortality rate The number of deaths due to a cancer within a population during a period of time (1 year or more). It is usually expressed per 100,000 person-years. Mortality rate ¼ (Cancer deaths/population)  100,000 person-years. Cancer survival The state of continued life in spite of a diagnosis of malignant cellular growth. Cancer survivorship Examines the state of the cancer survival experience of an individual or population, considering the complex reality of a present or long-term cancer diagnosis that can substantially influence survival. This includes physical symptoms and conditions, psychosocial concerns, health-related quality of life, health behaviors, tools and platforms for research, special populations, economic impact, and quality of care. Censoring A condition in which the survival experience of the study subject has not experienced the event of interest because (1) he/she is lost to follow up or drop out of the study, (2) if the study ends before the event can occur, or (3) have an event other than the outcome of interest. A censored subject is counted as alive or disease-free for the time he or she was enrolled in the study. Corrected survival rate When reliable data on cause of death are available, a correction is made for deaths due to causes other than the disease under study by censoring causes of death other than the cause of interest. Also sometimes called net survival rate, net survival, or cause-specific survival. Follow-up In a survival analysis and to calculate survival rates, enrolled subjects must be followed at intervals to assess vital status from the time elapsed between beginning and end of observation period. Hazard function l(t) The probability that the event of interest did not occur before the study observation period, t. Independent-censoring assumption The assumption that the event of interest in survival analyses is unrelated to whether or not a subject is censored. Kaplan–Meier method Estimates the proportion of people living after each increment of time during the study observation period. Cumulative survival probabilities are calculated when an event occurs or when a subject is censored. Lead-time bias A systematic error of apparently increased survival that occurs when disease is detected at an early stage. Leadtime bias is evident when disease detection does not alter the time from disease conception until the event of interest, but only appears so because the observation period after diagnosis ended before the event could occur. Also called information bias. Log-rank test A univariate statistical analysis comparing the survival distributions between two groups during the observation period, based on the null hypothesis that there is no difference in survival. Also called the Mantel–Cox test. Observed survival rates The proportion of individuals who did not experience death over the population diagnosed with cancer during an observed study period (e.g., 5 years). Also called the overall survival rate. Relative survival rate The ratio of the observed survival rate of cancer patients to the expected or background survival rate in a comparable set of cancer-free individuals. Often used for populations when cause-of-death data is not available. Survival function S(t) The probability that the event S did not occur before the study observation period, t. Time-to-event The time from entry into a study until the subject has a particular outcome. The “event” can be death, recurrence, symptom reduction, or hospital discharge.

Introduction Cancer survival is an all-inclusive term that refers to the state of continued life in spite of a diagnosis of malignant cellular growth. It is measured by survival statistics that display the proportion of patients alive over a specific observation period. In addition to life or death, survival statistics can examine outcomes that influence survival, such as cancer progression and recurrence within the observed time period. Cancer survivorship broadens the cancer survival definition, taking into consideration the complex reality of a present or long-term cancer diagnosis that can substantially influence survival. This includes physical signs and symptoms,

250

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65102-4

251

Cancer Survival and Survivorship Table 1

Positive influences that increase cancer survival rate for common cancers by cancer type

Lung Prostate Breast Colorectal Stomach Cervix Liver Acute lymphoblastic leukemia in children

Early diagnosis/screening

Treatment

()a ()b þ þ ()c þ  

 þ þ þ  þ  þ

a Screening is recommended in the United States only for current heavy smokers or those who have quit in the past 15 years and are between 55 and 80 years old, but has not been implemented in most other countries yet. Clinical trials evaluating the benefits and risks of lung cancer screening are ongoing. b Prostate-specific antigen level screening is recommended for some populations, but not widely recommended to all male population with average risk. c In countries with very low incidence, gastric cancer screening is not recommended. Gastric cancer screening programs exist nationwide for Japan and Korea and some regions of China.

psychosocial concerns, health-related quality of life, health behaviors, tools and platforms for research, special populations, economic impact, and quality of care. Cancer survival is influenced by two main determinants that are consistent across populations: early stage at diagnosis or screening and improved cancer management strategies (Table 1). These factors vary by cancer type and thus play a key role in observed differences in cancer survival rates. Acute lymphoblastic leukemia and testicular cancer, which have effective cancer treatment strategies, have 5-year survival rates greater than 85%, while pancreatic cancer, which has no effective treatment, has a 5-year survival rate of 1%. Access and utilization of oncological health care is another factor influencing each of the two survival determinants. These differ between populations based on a breadth of sociopolitical and economic factors and are reflected in survival rates across countries (Table 2). Survival and survivorship data are routinely collected and analyzed by cancer registries to monitor and assess cancer control programs in all patients registered in a population-based cancer registry. Survival data may also be collected in observational and randomized-control studies to compare groups of patient and effects of treatment and/or management strategies. The appropriate analysis and interpretation of survival data requires an understanding of the correct application of quantitative tools, metrics, assumptions, and limitations imposed by data source and availability. Cancer survival rates often use a 5-year survival rate, although 1-year and 10-year rates are also used. Survival rates of interest are cancer type, stage, and treatment groups, for a multitude of survival outcomes, including death, recurrence, and/or new symptoms. Although certain cancers can recur beyond the 5-year period, chance of a later recurrence for most cancers is very small. It is worth noting that survival rates and morality rates are not direct opposites. Survival rates differ from mortality rates in that they are highly dependent on when cancer is diagnosed or treated. While mortality rates of cancer within a certain time interval are absolute measures of cancer deaths regardless of when the cancer was diagnosed, cancer survival rates provide an outlook of the possibility of living past a fixed time period using the time of diagnosis or treatment as the starting point. The practical goal for analyzing survival data is to provide an estimated cancer prognosis for a patient based on the survival rates of a large number of patients diagnosed or treated for that cancer. However, as a prognostic benchmark, survival probability may be different for each individual patient and strongly depends on stage at medical diagnosis, his or her comorbidities, past medical history, and tumor characteristics.

Table 2

Five-year net survival, age-standardized estimates, %, adults (aged 15–99 years), 2005–2009

Lung Prostate Breast Colon Stomach Cervical Liver Childhood leukemia a

United Kingdom United States Norway Germany Italy

Slovenia Ecuador Brazil China India Japan Tunisia a Mauritius b

9.6 83.2 81.1 53.8 18.5 60.2 15.2 89.1

11.4 78.1 80.2 56 26.7 68.9 5.2 75.7

18.7 97.2 88.6 64.7 29.1 62.8 9.3 87.7

15 86.3 85.9 61.8 24.1 71.4 9.5 89.7

16.2 91.2 85.3 64.6 31.6 64.9 14.4 91.8

14.7 89.7 86.2 63.2 32.4 68.3 17.9 87.7

28.7 92.4 83.2 68.2 31.9 61.7 17.7 62.6

18 96.1 87.4 58.2 24.9 61.1 11.6 65.8

17.5 62.9 80.9 54.6 31.3 59.9 12.5 61.1

9.6 58.1 60.4 37.3 18.7 45.8 4.3 64.7

30.1 86.8 84.7 64.4 54 66.3 27 81.1

10.3 100 68.4 67.6 49 42.4 – 50.1

37.2 77.3 87.4 55.5 40.7 86.7 52.6 –

Survival estimates considered less reliable. Survival estimates are considered less reliable and are based on few cases. Data from Allemani, C., Weir, H. K., Carreira, H., et al. (2014). Global surveillance of cancer survival 1995–2009: analysis of individual data for 25,676,887 patients from 279 population-based registries in 67 countries (CONCORD-2). Lancet. https://doi.org/10.1016/S0140-6736(14)62038-9. b

252

Cancer Survival and Survivorship

Concepts of Survival Data A database suited for survival studies must have three key features: a defined event of interest, a specific start and end of each individual subject’s observation period, and a set observation time-scale, such as years or months. Randomized-control trials may use date of randomization, date of entry into the cohort, or date of diagnosis as start points. The dataset may also contain variables that could be evaluated for their relationships with survival. These variables can include sex, cancer stage and type, carcinogenic exposures, or treatment modalities. Survival statistics use time-to-event data. This is the time (t) from study entry until an individual is observed to have a particular outcome. Both the time enrolled in the study and the event of interest are recorded for each individual. A binary outcome of interest is typically used: no occurrence of event (0) or occurrence of event (1). Occurrence or nonoccurrence of the event is restricted to the observed study period only. If an individual leaves a study before the event of interest, the outcome is censored. Individuals are censored if they are lost to follow-up, drop out from study for any reason, or if the study is terminated before the outcome of interest occurs. The dataset would contain a record that the individual survived at least to time t. A study proposing to assess the probability that the event did not occur before t is interested in the survival function S(t). The proportion of persons not observed to experience the event by the time the study ends is reported as the survival experience for that group. Vital status of study subjects can be determined through two methods of follow-up: active and passive. Active follow-up collects vital status information by repeated residential visits, surveys, physician notes, death registry offices, or other social or community gatherings. Passive follow-up is more common in regions with established cancer registries and is accomplished by linking a subject’s personal identifiers to an institutionally recorded vital status. Although survival data often use life or death as the events of interest, other outcomes may be of clinical interest. Examples of other outcomes include developing a complication of disease, treatment of a side effect, resolution of symptoms, progression, and recurrence.

Methods for Survival Analyses: Descriptive Measures The first step in survival analysis is to estimate the overall survival experience of the entire population or cohort. Unlike incidence or mortality statistics where the total population is included in the ratio denominator, only cancer-diagnosed patients are included in survival fractions. The observed survival rate or overall survival rate is the proportion of individuals alive after receiving a cancer diagnosis or treatment. It includes all deaths for these patients, regardless of whether the cause was from the cancer itself, a cardiovascular disease, injury, or other cause. When reliable information on cause of death is available, a corrected, net, or cause-specific survival rate can be calculated. Deaths occurring only due to the disease of interest are counted as an observation of the event, while deaths due to other causes are treated as a nonevent. Treating other causes is controversial, particularly for studies examining the relationship between exposures and cancer. The cancer may still have occurred had an individual not experienced the competing cause of death. In these circumstances, competing risks can be calculated (see section “Competing Risks”). When cause of death data is unavailable or unreliable, a relative survival rate can account for survival from noncancer causes. Relative survival is a ratio of the observed survival rate of cancer patients to the expected or background survival rate in a comparable set of cancer-free individuals. An example of background survival rate can be cancer-free patients or overall survival of cancer patients when race, sex, and age, race, sex, or year information is missing. In practice, overall, corrected, and relative survival rates are very similar. Because they are calculated differently, relative and corrected survival estimates cannot be compared. Disease-free survival rate is another specific type of survival rate. This is the proportion of subjects with cancer who achieve complete remission. Progression-free survival rate is the proportion of subjects who still have cancer, but the cancer has not progressed over a specific time period. To be free of progression may have an impact on further therapy and is correlated to tumorspecific survival. When survival rates are compared between populations of varying age distributions, such as different countries, an age-adjusted survival rate is necessary to account the different age structures of different patient populations. Age-adjusted survival rates are calculated by multiplying specific survival rate for each age group with the corresponding proportion of persons in that age group in the population. Survival probabilities can be summarized in a tabular and graphical form in a survival curve. These cumulative survival probabilities can be calculated by the actuarial life-table method or the Kaplan–Meier method. In the actuarial life-table method, cumulative survival probabilities are calculated only at the end of each equally spaced follow-up interval. If the percent of the patients surviving until the end of the first, second, and third intervals are 90%, 83%, and 67%, the cumulative survival percentage (or the chance of surviving until the beginning of the fourth interval) is 50% (0.9  0.83  0.67 ¼ 0.50) (Table 3). Conversely, the Kaplan–Meier method calculates survival probabilities at the time that the death takes place, resulting in a stepped figure (Fig. 1). Both curves allow the estimation of the time beyond which 50% of study subjects are expected to survive (median survival time), comparison of survival of different groups in the study over time.

Cancer Survival and Survivorship

253

Calculating cumulative survival probabilities, actuarial life-table

Table 3 Time interval (months)

Number at risk during interval

Number of deaths during interval

Probability of death during interval

Probability of surviving during interval

Cumulative probability of surviving at the beginning of the time interval

0–4 5–9 10–14 15–19 20–24

100 90 75 50 25

10 15 25 25 3

0.10 0.17 0.33 0.50 0.12

0.90 0.83 0.67 0.50 0.88

1.00 0.90 0.75 0.50 0.25

Fig. 1

Five-year survival of cancer patients by treatment modalities A, B, and C using the Kaplan–Meier method.

Care must be taken when interpreting cumulative survival probabilities and two main underlying assumptions must be accounted for. Firstly, at any point during the study course, subjects who are censored at one time have the same survival probability as those who continue to be followed. Events such as undergoing new treatment or achieving complete remission within time t may otherwise alter the survival rate. Secondly, in studies with long enrollment periods, survival probabilities must be the same for subjects recruited early and late in the study. Some individuals recruited earlier in a study before effective treatment availability may be more susceptible to a cause of death that differs from subjects recruited later and the interpretation of the survival rate may be erroneous. Comparison of cumulative survival rates is calculated by a log-rank test, comparing the survival distributions, based on the null hypothesis that there is no difference in survival. This calculation has the form of a chi-square probability distribution, equally weighing the differences of observed survival rate compared to the expected survival rate for each group at each event time. The number of degrees of freedom is the number of groups to be compared,  1. If the difference between the observed and expected survival distribution has < 5% probability to be due to chance, then the null hypothesis is rejected and the alternative hypothesis of a significant difference in survival is expected. The log-rank test is only a test of significance and provides no estimate of the magnitude of difference between the groups studied. For comparative methods with ability to account for multiple variables influencing survival rates, multivariate models are used.

Methods for Survival Analyses: Basic Statistical Assessments for Comparisons Between Groups To evaluative the survival–predictor relationship or quantitatively compare groups, multivariate regression models such as the Cox model or proportional hazard regression analysis and Poisson regression are applied. These models both assume that the timeestimated hazard ratios (Cox) or incidence rate ratios (Poisson) are proportional to the length of time t and that the impact of each predictor in the model is constant in the relationship to the outcome for time t. Like any survival analysis, censoring must be independent or noninformative. That is, the censoring of subjects must not be related to the probability of an event occurring.

254 Table 4

Cancer Survival and Survivorship A comparison of the Cox model versus Poisson regression for survival analysis The Cox model or proportional hazards regression

Poisson regression

Distribution of response event

Semiparametric; does not estimate baseline hazard but Parametric; baseline hazard is estimated instead uses arbitrary baseline hazard Response variable in equation Hazard function gamma(t); reported as hazard ratio (HR) Incidence rate: natural log counts of event with a model offset for the amount of risk each individual had until the event (person-years) Dataset preconditions Nonaggregate or individual subject data are needed Can be used for both aggregate and nonaggregate data Survival distribution Does not make assumptions about survival distribution The incidence must fit a Poisson distribution (the mean and variance are equal) Use in longitudinal studies Allows to account for repeated observations for each Allows to account for repeated observations for each record record, but must be compressed prior to analysis Proportional hazard assumption Proportional to the predictor over time t and constant Proportional to the predictor over time t and also constant Limitations Cannot use if assumption of proportionality is violated, Difficult to set up dataset. Data on individual subjects must though stratification for some covariates can be done be organized into event-time tables that are stratified on time and other covariates Advantages No assumption is made about the shape of the baseline If Poisson distribution assumption is violated, alternative hazard function methods are available, including quasi-likelihood and binomial regression

The Cox model and Poisson regression differ in assumptions of survival distributions. For the event rate of interest, the Cox model assumes a semiparametric distribution whereas the Poisson model assumes a parametric distribution. Differences between the two model types are displayed in Table 4.

The Cox Model A Cox model is a statistical technique that can be used for survival-time (time-to-event) outcomes on one or more predictors. The response variable is the hazard function l(t), which assesses the probability that the event of interest (in this case, death) occurred before t. The equation models this hazard as an exponential function (exp) of an arbitrary baseline hazard (l0) when all covariates are null, and b is the regression coefficient of the covariate, x. lðt Þ ¼ l0 ðt Þexpðb1 x1 þ /bk xk Þ The Cox proportional hazards model makes two assumptions: (1) survival curves for different strata must have hazard functions that are proportional over the time t and (2) the relationship between the log hazard and each covariate is linear, which can be verified with residual plots. Examples of covariates can be categorical such as race or treatment group, or continuous such as biomarker concentrations.

Competing Risks So far, we have assumed that only a single event of interest occurs in a subject. In practice, one cause of death may alter the survival observed for another cause of death. An increase in cardiovascular deaths in a susceptible population may lead to an apparent increased survival for cancer deaths. The basic Cox model does not differentiate between subjects lost to the competing death and subjects who were alive at the end of his or her observation period. This leads to an erroneous estimation of the cancerspecific survival rate and, in comparisons between predictors or characteristics, an erroneous interpretation of the effect of these variables on survival. A second example occurs when the suspected predictor is linked to multiple causes of death. Smoking, for instance, is linked to multiple cancers of interest, but the event of interest is prostate cancer death, which is not as lethal as lung cancer. Lung cancer would be a competing cause of death. Another phenomenon can occur in clinical trials. A particular cancer management strategy may result in two outcomes: death or recurrence. Recurrence is related to death, but is a competing event that alters survival risk and would result in earlier death. Each of these scenarios violates the assumption that the risk of death is unrelated to censoring (independent-censoring assumption). There are a few strategies for handling competing risks that quantify more accurate hazard probabilities. Adjusting models for competing risks is useful in circumstances where it is highly suspected that the competing event alters the risk of the event of interest. For example, the patient could have died from the cause of interest given the predictor, but the competing cause of death obscured the true risk. Unlike censoring, which obstructs you from viewing the event probability, a competing event prevented the event of interest from occurring altogether. The most frequently used methodologies for adjusting analyses to take competing risks into account is the Fine and Gray method. The Fine and Gray method is a modification of the Cox model that postulates a model for the sub-hazard function of a failure event of primary interest. Instead of a binary outcome with one possible event of interest, the Fine and Gray method is able to account for additional competing events.

Cancer Survival and Survivorship

255

Poisson Regression Poisson regression uses the natural log of event counts and an exposure or offset parameter (person-years) to measure the relationship between predictors and the event of interest. Instead of comparing hazard rates, as the Cox model, the Poisson model compares incidence rates for predictors: In ðlÞ ¼ b0 þ b1 x1 .bk xk In this case, In is the natural logarithm of the incidence rate, l, b represents the coefficient of each predictor, x, and b0 is the baseline rate or intercept. The event rates over time t follow a Poisson distribution where Ln(l) has mean equal to its variance. Both Poisson regression and the Cox model can be used to analyze survival data. Both equations enable an analysis aimed at understanding the relationship between predictors to survivorship outcomes. However, Poisson regression is useful for aggregate data in which only cancer rates for the co-variates of interest, rather than each individual, are known. This may be the case for public use or registry-based datasets. Although the response variables differ (hazard ratios vs. incidence rate ratios), both methods provide an estimated relative risk (Table 4).

Cancer Survivorship at a Glance The number of cancer survivors has increased substantially over the past four decades. Many patients diagnosed with most cancers are now expected to survive. In countries with a high gross-domestic-product per capita, this survivor proportion is well over half of those diagnosed with cancer. In high-income countries, the increase has been by 40%–70% over the past four decades. Although cancer survival is increasing, aging populations in both high- and low-income countries suggest that the projected number of people living with or beyond a cancer diagnosis will grow. With more cancer survivors and advances in cancer diagnosis and treatment, greater emphasis is being placed on improving survival estimations, identifying key survival determinants, and utilizing these summary measures to improve cancer preventive planning and investment. Cancer survivorship studies merge clinical and public health efforts to improve patient cancer outcomes. Survival statistics’ accuracy and completeness requires systematic collaboration between clinicians and cancer registries. General practitioners, health institutions, hospitals, and death registries notify cancer registries using a standardized coding system. Cancer registries systematically collect data for the purposes of calculating cancer statistics and conducting cancer research. It must therefore be stated that the currently available survival statistics are less reliable for many low-income countries with fewer resources and infrastructure available than for high-income countries. Cancer survivorship is being recognized as a multifaceted field that is influenced not only by key differences in cancer survival, such as by cancer type and age at diagnosis, but also by several non-prediagnostic factors such as economics, culture, and society. These community-related factors include an individual’s socioeconomic status, healthcare-seeking behavior and consistency with follow-up, health care service efficiency, and personal treatment choices. In considering these many cancer survivorship variables, it is important to observe them within the context of stage at diagnosis and treatment access/availability. Cancer stage makes a significant impact on survivaldthose with higher clinical stage cancers are likely to have a lower survival rate than those with low-stage cancers. Cancers can be detected at a lower stage when the length of survival would be longer than the 5-year follow-up period. This cancer can be said to have a lead-time bias. Lead-time or information bias can lead to the erroneous interpretation that earlier cancer screening prolongs life, when in actuality, the cancer was only detected at an earlier starting point in time. In other words, the patients would have died at the same time whether they had screening or not, but would have known about the disease earlier. It is important to note that while 5-year survival is useful in monitoring progress for early detection and treatment of cancer, it does not represent the total proportion of people who are cured because cancer death can occur beyond 5 years after diagnosis. Lead-time bias is often observed for cancers in which a diagnostic method is discovered, but no known effective treatment exists regardless of stage at diagnosis. Factors that affect survival are unique to each type of cancer. For cancers without early detection or effective treatment, such as liver, lung, or pancreatic cancer, survival rates vary little across countries. In contrast, the largest survival differences between countries are observed for cancers with a better prognosis through screening and access to treatment: female breast, colorectal cancer, cervical cancer, and childhood leukemia (treatment only). Within countries, it has been widely recognized that survival differences are distributed differently across socioeconomic status groups and race. While survival has mostly improved over time, differences in survival by race that existed in 1975 are still evident in contemporary times. Indeed, survival rates for the most common cancers have improved among both US blacks and whites due to early detection and treatment, but US blacks still have an approximately 10% lower 10-year survival rate (Fig. 2).

Breast Cancer Among women worldwide, breast cancer survivors make up a third of those living with a cancer diagnosis in the past 5 years. When breast cancer is detected at an early stage, treatment is more effective and a cure is more likely. The 5-year survival rate for breast cancer in the United States in 2005–2009 was 88.6%, compared with 60.4% in India and 68.4% in Tunisia (Table 2). Within India, 5-year survival for early-stage (I or II) breast cancer ranged from 41% to 78% by region, indicating wide variation in health services

256

Cancer Survival and Survivorship

100

90 Survival

Race Black women White women 80

70 0.0

2.5

5.0

7.5

10.0

Year Fig. 2 Five-year relative survival by race for female breast cancer, United States, 1988–2013. Data from Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov) SEER*Stat Database: Incidence – SEER 9 Regs Research Data, Nov 2016 Sub (1973– 2014) dLinked to County AttributesdTotal U.S., 1969–2015 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, released April 2017, based on the November 2016 submission.

delivery. In high-income countries, over 75% of breast cancers are diagnosed at an early stage, which is reflected in the high 5-year overall survival rate (80%þ). By contrast, in several low-to-middle income countries, the majority of women are diagnosed with late-stage disease (clinical stage III or IV). Reasons for late-stage diagnoses include lack of awareness as well as limited access to adequate detection and diagnostic services. In some cultures, breast cancer or having any cancer at all, is associated with a social stigma and reduces a woman’s social status or the potential for her childrens’ marriage within her community. Cancer fatalism, a belief that cancer is preordained in one’s destiny and cannot be altered, can also hinder cancer screening programs and may delay or prevent cancer treatment. Additionally, women living in deprived or remote areas are less likely to receive breast-conserving surgery, and would have lower 5-year survival than women in more affluent areas.

Cervical Cancer When detected at an early stage, invasive cervical cancer is one of the most successfully treated cancers. The 5-year net survival rate ranges from 42.4% in Tunisia to 77% in South Korea and 71% in Norway, although it is between 60% and 70% in most countries. The Papanicolaou (Pap) test screens for precancerous cervical cells that can be treated to prevent cervical cancer, but this test requires a clinician to swab for a cell sample directly from the cervix and a trained cytologist, cytotechnician, or pathologist to examine these cells under a microscope. As the main cause of cervical cancer is a persistent infection with high-risk strains of human papilloma virus (HPV), HPV tests are being currently introduced as a mean of primary or secondary screening in several countries. This test has a higher sensitivity but lower specificity than the Pap test and using HPV in screening will be particularly useful in the future, when most women will hopefully become vaccinated for HPV. In many cultures, social barriers to screening exist, such as the reluctance to have a gynecological exam by a male doctor. Additionally, many remote areas or low-resource countries do not have the technology, expertise, or resources to support an effective cervical cancer screening program.

Colorectal Cancer Survival rates for colorectal cancer vary worldwide. In high-income countries, colon and rectum cancer 5-year net survival is about 50%–65%, while in Asia a wider range exists. Five-year colon and rectal cancer survival rates of more than 65% have been reported in Israel and South Korea, while they range from about 20% to 55% in other countries in Asia. As colorectal cancer is not one of the highest causes of cancer deaths in many Asian countries (with cervical cancer and cancer of the stomach being more common), there is less public awareness about screening in many areas. Survival is much higher when colorectal cancer is detected at an early stage.

Cancer Survival and Survivorship

257

Colorectal cancer is amenable to screening, and there are several screening modalities available such as fecal immunochemical test, fecal occult blood test, sigmoidoscopy, and colonoscopy. However, these screening modalities have been implemented almost exclusively in high-income countries. In some countries with routine screening practices for middle-age adults, colorectal cancers are diagnosed at an earlier and more treatable stage, and survival rates are consequently higher as observed in the United States and Germany, with 5-year survival rates of 64.7% and 64.6%, respectively.

Liver Cancer Liver cancer is one of the most fatal cancer types, with 5-year survival rates ranging between 10% and 20% across all countries. A cancer fatality rate is the proportion of individuals who experienced death within the population of individuals diagnosed with cancer, over the course of the disease. Liver cancer risk factors vary substantially by country. It main risk factors are alcohol abuse, liver cirrhosis, exposure to food contaminants such as aflatoxins, anddimportantlyddifferent types of hepatitis infections. Hepatitis infections, mainly hepatitis B and C, are the largest cause of liver cancer in many parts of the world. There is an effective vaccine for hepatitis B to be given to newborns/infants, and persons at risk, and hepatitis C can be effectively cured. Liver cancer incidence rate is increasing worldwide, making preventive studies and randomized clinical trials a high priority for patients’ prognostic improvement.

Lung Cancer Although there have been improvements in surgical techniques and combined therapies over the past several decades, lung cancer is one of the most lethal cancers and the most difficult to treat. For the general population, no screening is available and it is often not diagnosed until late stages. Most lung cancers are also difficult to remove surgically even for localized cases, although survival for this small proportion of patients is higher. For populations with less access to trained surgeons and tertiary health care centers, the 5year survival rate is only slightly lower (around 10%) than high-resource countries (up to 30% or slightly more). Clinical trials are currently underway to determine the risks and benefits of lung cancer screening modalities. Currently, the United States is the only country with lung cancer screening for certain populations. The US Preventive Task Force recommends annual screening for lung cancer with low-dose computed tomography in adults aged 55–80 years who have a 30 pack-year smoking history and currently smoke or have quit within the past 15 years. Once a person has not smoked for 15 years, he develops a health problem that limits life expectancy, or cannot have curative lung surgery, screening should be discontinued. Screening eligibility constitutes a fraction of all lung cancer patients and it should be noted that the US population 5-year survival rate for all lung cancers is 18.7%, which is not substantially higher than other countries.

Stomach Cancer In addition to differences in screening availability and treatment, international differences in stomach cancer survival rates are also affected by differences in screening policy and patient awareness, as well as access to treatment. For example, Japan has a high stomach cancer survival rate due to nationwide gastric screening; the 5-year net survival is of 54% and most cases are diagnosed at an early stage. In China, another country with high stomach cancer incidence rates but without nationwide screening, the average 5-year survival rate is 31.3%. In the United States, where only about 26% of cases are diagnosed at an early stage, the 5-year survival rate is 29%. Generally, the observed differences have been attributed to lack of awareness of signs and symptoms and greater delay in seeking care among particular groups.

Prostate Cancer Among men worldwide, over a quarter of all cancer survivors have had a diagnosis of prostate cancer. Since the widespread use of prostate-specific antigen testing in the late 1980s to early 1990s, the dramatic survival improvement in high-income countries has been attributed to lead-time bias. Many asymptomatic cancers would never have become clinically evident and for elderly men, the risks of invasive surgery outweigh the benefit. The 5-year net survival rate for patients diagnosed with prostate cancer is more than 90% in some countries (Israel, Canada, United States, Brazil, Ecuador, Germany, and many Western European countries). Survival rates are lower in Asia, where prostate cancer is less common or where public health programs or access to early diagnosis and treatment are unavailable.

Outlook in Cancer Survivorship Cancer has become a significant health burden internationally with survivorship not always being attainable for a variety of socioeconomic, cultural, and environmental reasons. There are currently more survivors in high-income countries than in low- and middle-income countries (Table 5). Given the economic and political transitions, diagnostic, treatment, and care may not be accessible and/or available and result in different survival rates. Efforts are underway by the SURVCAN project to estimate, compare, and identify survival in low-to-middle income countries. This new survival database established at the International Agency for Research

258

Cancer Survival and Survivorship Table 5

Prevalence of cancer survivors (number of survivors diagnosed with cancer within the past 5 years per 100,000 adults) (15 þ years), 2012

Country

Prevalence per 100,000 persons

United Kingdom United States Norway Germany Italy Slovenia Ecuador Brazil China India Japan Tunisiaa Mauritiusa

1594 1892.1 2020.1 1964.6 1933.8 1648.8 534.3 720.2 456 202.9 1830.7 310.1 530.2

a

Survival estimate considered less reliable. Data from Ferlay, J., Soerjomataram, I., Ervik, M., et al. GLOBOCAN 2012 v1.0, cancer incidence and mortality worldwide: IARC cancer base no. 11. Available from: http://globocan.iarc.fr (accessed July 16, 2014).

on Cancer allows a comparison for cancer survival across cancer types and by country, incorporating survival analyses methodology and recommendations for improving survival statistics in lower-resources areas. Patients face medical and nonmedical challenges in cancer survivorship. Results of cancer survivorship statistics reflect the impact of health-seeking behaviors, social inequities, natural histories, and health care service efficiencies in responding to diagnosis, treatment, and follow-up. Sharing expertise between clinical and population-based research will simultaneously enhance the inference potential of data collected in both types of resources and also provide a cost-effective solution to targeting key issues in oncological research. Increasing access to and improving screening program quality in high-risk groups, improving access to cancer treatment and management programs, and creating public health programs tailored toward health care barriers faced by each community are necessary to improve cancer survivorship.

Prospective Vision Accurate estimation and interpretation of survival statistics is necessary for understanding patient prognosis. Currently, survival statistics are mostly unreliable or unavailable for many low-resource countries. Methodologies and public health efforts aimed at building institutions and programs are currently underway. As with any clinical diagnosis, cancer patients are not all the same and have different survivorship struggles. Cancer survivorship is a new and burgeoning field that seeks to understand the barriers to cancer diagnosis and treatment faced by individuals and their communities. Several indicators are being explored by the collaborative efforts of epidemiologists and clinicians to make public health programs relevant to patients and their community, and to improve cancer survivorship.

See also: Aging and Cancer. Chemoprevention Trials. Symptom Control.

Further Reading Allemani, C., Weir, H.K., Carreira, H., et al., 2015journal. Global surveillance of cancer survival 1995–2009: analysis of individual data for 25,676,887 patients from 279 population-based registries in 67 countries (CONCORD-2). Lancet 385, 977–1010. American Cancer Society, 2016book. Cancer treatment & survivorship facts & figures 2016–2017. American Cancer Society, Atlanta. Available from: https://www.cancer.org/ content/dam/cancer-org/research/cancer-facts-and-statistics/cancer-treatment-and-survivorship-facts-and-figures/cancer-treatment-and-survivorship-facts-and-figures-20162017.pdf. Braaten, T., Weiderpass, E., Lund, E., 2009journal. Socioeconomic differences in cancer survival: the Norwegian Women and Cancer Study. BMC Public Health 9, 178. Brenner, H., Hakulinen, T., 2002other. Advanced detection of time trends in long-term cancer patient survival: experience from 50 years of cancer registration in Finland. American Journal of Epidemiology 156 (6), 566–577. Breslow, N.E., Day, N.E., 1987book. Fitting models to grouped data. In: Statistical methods in cancer research. Volume II – the design and analysis of cohort studies. World Health Organization, Lyon. IARC Scientific Publications No. 82. Available from: http://www.iarc.fr/en/publications/pdfs-online/stat/sp82/. Ferlay, J., Soerjomataram, I., Ervik, M., et al.. GLOBOCAN 2012 v1.0, cancer incidence and mortality worldwide: IARC cancer base no. 11. Available from: http://globocan.iarc.fr. (Accessed 16 July 2014).

Cancer Survival and Survivorship

259

Feuerstein, M. (Ed.), 2007. Handbook of cancer survivorship. Springer, New York. Hashim, D., Boffetta, P., La Vecchia, C., Rota, M., Bertuccio, P., Malvezzi, M., Negri, E., 2016journal. The global decrease in cancer mortality: trends and disparities. Annals of Oncology 27 (5), 926–933. https://doi.org/10.1093/annonc/mdw027. Hosmer, D., Lemeshow, S., May, S., 2008other. Applied survival analysis. Wiley series in probability and statistics, 2nd edn. Wiley, Hoboken. Marubini, E., Valsecchi, M.G., 2004book. Analysing survival data from clinical trials and observational studies. Wiley, New York. Sankaranarayanan, R., Swaminathan, R., Lucas, E., 2011other. Cancer survival in Africa, Asia, the Caribbean and Central America. IARC Scientific Publications No. 162. International Agency for Research on Cancer, Lyon. Available from: http://survcan.iarc.fr/indexsurvcan1.php. (Accessed 16 July 2017). Selvin S (2008). Survival analysis for epidemiologic and medical research. Cambridge: Cambridge University Press. Therneau, T.M., Grambsch, P.M., 2000book. Modeling survival data: extending the Cox model. Springer-Verlag, New York.

Relevant Websites http://eco.iarc.fr/euregdEuropean Cancer Observatory project at the International Agency for Research on Cancer. https://seer.cancer.gov/publications/survival-1975-2010/dCancer Survival From a Policy and Clinical Perspective: US Surveillance, Epidemiology, and End Results (SEER) Program, 1975–2010. http://survcan.iarc.fr/dCancer survival in Africa, Asia and Latin America (SURVCAN). https://surveillance.cancer.gov/survival/index.htmldMeasures of cancer survival including cancer survival downloadable statistical software.

Cancer Vaccines: Dendritic Cell-Based Vaccines and Related Approachesq Laurence Chaperot, Olivier Manches, and Caroline Aspord, Institute for Advanced Biosciences, Immunobiology and Immunotherapy in Chronic Diseases Team, Inserm U 1209, CNRS UMR 5309, University Grenoble Alpes, Grenoble, France © 2019 Elsevier Inc. All rights reserved.

Glossary C-type lectin receptors Receptors that bind to carbohydrates in a calcium-dependent manner. They are able to recognize microbial antigens and can be expressed on dendritic cells and monocytes/macrophages. Dendritic cells Professional antigen-presenting cells located in the skin, mucosa, lymphoid tissues, and circulating in blood. Dendritic cells form a cellular network of cells able to capture, process antigens and present them on the cell surface to activate T cells. They link the innate and the adaptive immune systems. Immune checkpoint blockers Immunotherapeutic drugs (usually antibodies) that help release the brakes cancer cells put on the immune system to prevent their destruction. They have revealed efficient, particularly in patients with metastatic melanoma. They target for example CTLA-4, PD-1, or PD-L1 coinhibitory molecules. Neoantigens Antigens specifically expressed by tumor cells and not by normal cells, generated from somatically mutated genes. They are ideal target for personalized cancer vaccine design. Therapeutic cancer vaccine Vaccines developed to treat an existing cancer by stimulating patients’ natural immune response against its own cancer. They are designed to activate cytotoxic T cells. Toll-like receptors Receptors expressed on sentinel cells such as macrophages and dendritic cells, that recognize structurally conserved molecules derived from pathogens. TLR play a key role in the initiating immune responses. Tumor-associated antigens Antigens differentially expressed by tumor cells compared to normal cells, corresponding to overexpressed proteins, tissue-specific molecules, cancer-testis or oncofetal antigens. Whole exome sequencing Exome sequencing allows to identify mutations found in the coding region of genes which affect protein sequences.

Vaccination was first developed to prevent infectious diseases, and in this specific context, its efficacy relies on the induction of neutralizing antibodies targeting viral or bacterial pathogens. In the past decades, vaccines have also been developed to prevent virus-induced cancers such as uterine cancers driven by papillomaviruses, or hepato-carcinomas associated with hepatitis B virus. In these cases, blocking cell infection prevents cancer development. Therapeutic vaccination is different, as it aims at curing patients with cancer or chronic viral infection by boosting immune responses targeting diseased cells, whether malignant or infected, for elimination. Cytotoxic T lymphocytes (CTL) are the main immune effectors involved in these responses, hence the need to develop new vaccinations strategies that cannot simply reproduce disease prevention schemes that aim at triggering B cell responses. The scope of therapeutic vaccination covers priming of naïve T cells to reactivation of existing but anergic or dormant antitumor memory T cells. Cellular cancer therapeutic vaccines can be based on killed cancer cells that carry a specific cancer-associated antigen(s) or on immune cells that are modified to present such an antigen(s) on their surface. Other types of cancer vaccines use viruses, yeast, or bacteria as vehicles to deliver both antigens and immunogenic signals. Alternatively, molecules of DNA or RNA coding for tumor-associated antigens (TAA) can be injected either alone (naked nucleic acid) or packaged into a vector. Among immune cells, dendritic cells (DC) are considered “professional” antigen presenting cells (APC), due to their unique capacity to prime naive T cell responses. They function as sentinels, searching for and integrating so-called “danger” signals, and upon infection, they drive the most suitable immune response (Steinman and Banchereau, 2007). For this reason, dendritic cells are central for the antitumor vaccines approaches we describe here (Saxena and Bhardwaj, 2018; Bol et al., 2016; Palucka and Banchereau, 2013a): the injection of DC loaded with tumor antigen (Ag), and the injection of medicinal products targeting DC in vivo. Priming of both CD4 þ helper and CD8þ cytotoxic antigen-specific T cells by DC is fundamental for therapeutic vaccine efficacy. These lymphocytes can recognize Ag expressed by tumor cells, either derived from mutation containing proteins (neoantigens), or overexpressed compared to healthy cells, or aberrantly expressed by tumor cells (spatially or temporally) (Coulie et al., 2014). Tumor-specific neoantigens are often considered as ideal tumor-rejection antigens, enabling the development of personalized vaccines. The Ag recognized by T lymphocytes are peptides, arising from processed endogenous proteins, and presented in association with major histocompatibility (MHC) complex molecules (human leukocyte antigen (HLA) system in human). Antitumor T

q

Change History: March 2018. Laurence Chaperot, Olivier Manches, and Caroline Aspord rewrote the entire chapter. This article is an update of James W. Young, Jean-Baptiste Latouche and Michel Sadelain, Cancer Vaccines: Gene Therapy and Dendritic Cell-Based Vaccines in Encyclopedia of Cancer (Second Edition) edited by Joseph R. Bertino, Academic Press, 2002, Pages 319–325.

260

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.99437-6

Cancer Vaccines: Dendritic Cell-Based Vaccines and Related Approaches

261

lymphocytes activated upon vaccination will be efficient if they succeed in (i) migrating toward the tumor and their metastases, (ii) escaping the immune-subversion mechanisms found inside tumors (e.g., regulatory T cells, immunosuppressive soluble or membrane-associated factors), and (iii) lysing malignant cells. In order to exert long-lasting effects and prevent relapses, these T lymphocytes must retain their capacity to proliferate and persist as memory cells (Palucka and Banchereau, 2013a).

Dendritic Cells: Professional Antigen Presenting Cells Dendritic cells represent critical regulators of the host immune system, involved in immune surveillance and bridging innate and adaptive immunity. They are native adjuvants, dictating the specificity and type of immune responses, depending on the microorganism or diseased cells they encounter. Multiple dendritic cell subsets have been described; they are produced in the bone marrow and derive from common myeloid progenitors (Gardner and Ruffell, 2016), and are found in almost all tissues, including blood, lymph nodes, spleen, skin or mucosae, where they act as sentinels and first activators for immune responses. Dendritic cells localized in tissues can take up Ag through specialized receptors, (by phagocytosis, endocytosis, or pinocytosis), detect associated danger signals, and migrate to secondary lymphoid organs. They integrate the environmental signals they receive, mature accordingly, process and present the Ag they have captured, in association with major histocompatibility complex class II and class I molecules, to finally activate both CD4 þ and CD8 þ T lymphocytes (Reis e Sousa, 2006). The diversity of dendritic cells in mice and humans is still under investigation, and the differential role that these subsets play in antitumor responses remains a matter of debate. DC subsets have been classified according to ontogeny, functionality, location or by transcriptome analyses. They are characterized by the expression of different surface molecules allowing the recognition of conserved pathogenic structures, their activation, migration, and interaction with various immune cells. Differential cytokine and chemokine secretion by different DC subsets accounts for the diverse mobilization, activation, and polarization of adaptive immune responses. A better understanding of DC heterogeneity is essential to improve the development DC-based antitumor immunotherapies.

Human Dendritic Cell Subsets In human blood, three major DC subsets have been identified: plasmacytoid DC (pDC), CD1c þ and CD141þ DC, that lack lineage markers (CD3, CD19, CD14, CD56, and GlycophorinA) and express high levels of MHC class II molecules (Dzionek et al., 2000). PDC express BDCA-2(CD303) and BDCA-4 (CD304), and are known to produce high levels of type I interferon (IFN) upon viral infection (Cella et al., 1999). Conventional or classical myeloid DC (cDC) produce interleukin (IL)-12 and are comprised of two subsets: BDCA-1 (CD1c)þ DC that may be particularly efficient for priming CD4 þ, and BDCA-3 (CD141) high DC, a minor population of DC in blood that are specialized in cross-presentation of exogenous antigen, allowing priming of CD8 þ T cells against pathogens that do not infect DC. The diversity of DC subsets is still under investigation, and recently, single-cell RNA-seq and CyTOF uncovered an underestimated phenotypic heterogeneity of cDC, identifying a new AXL þ SIGLEC6 þ circulating cDC subset named AS DC (Villani et al., 2017; Alcantara-Hernandez et al., 2017) (Table 1). In the skin, Langerhans cells (LC) are localized in the epidermis, while the dermis contains mostly DC expressing CD1a or CD14 (thought to be monocyte-derived cells), but also CD141hi DC (Gardner and Ruffell, 2016) and CD16 þ SLAN DC (Gulley et al., 2016). Upon inflammation, the skin can also present with increased numbers of inflammatory DC (CD11c þ, CD1c þ, HLA-DR þ) and pDC (Nestle et al., 2009). In the intestinal mucosae, three subsets of classical DC are found, characterized by their differential expression of Sirpa and CD103 (Watchmaker et al., 2014), whereas in the lung CD1c þ and CD141þ cDC have been described (Yu et al., 2013).

Table 1

Circulating DC subsets in human cDC CD1cþ

Phenotype

TLR expression C-type lectin expression Cytokine production

cDC CD141þ

DC CD16þ

CD45þ, lin (CD3, CD19, CD56, CD14, GlycophorinA)-, HLA-DRþ CD11cþ, CD123CD1cþ BDCA-3þ, CLEC9Aþ TLR1, 2, 3, 4, 5, 6, 8, 10 TLR1, 2, 3, 6, 8, 10 Dectin1, DEC-205, DCIR DC-SIGN, CD206, CLEC10a IL12p70, IL23

Dectin2, DC-SIGN, langerin, CLEC9A IL12p70, IFN-l

CD16þ TLR1, 2, 4, 5, 6, 8, 10

AS DC

PDC

CD11cþ/, CD123þ/ AXLþ, SIGLEC6þ ?

CD11c-, CD123þ

Dectin2

?

Dectin1, DEC-205, DCIR BDCA-2, ILT7

IL-1b, TNF-a

IL12p70

Type I IFN

Expression of surface markers, Toll-like receptors, C-type lectin, and cytokine production by human DC populations. , negative; þ, positive; bold numbers, main TLR expressed; ?, for controversial data, or not yet defined expression.

?

BDCA-4þ TLR1, 6, 7, 9, 10

262

Cancer Vaccines: Dendritic Cell-Based Vaccines and Related Approaches

Antigen Capture by Dendritic Cells In peripheral tissues, skin and mucosae where they are located, DC function as sentinels sampling self and foreign antigens for presentation. PDC and cDC express complementary sets of receptors allowing the recognition of different kinds of pathogens. Antigen (Ag) capture by cDC involves C-type lectin receptors (Caminschi et al., 2008; Sancho and Reis e Sousa, 2012): mannose receptor (MR, CD206), DC-SIGN (CD209), langerin (CD207), dectin-1 and 2, DEC-205 (CD205), LOX-1, DCIR, CLEC-9a, as well as receptors for immunoglobulin Fc regions (CD32 and CD64). PDC express CD32, allowing endocytosis of antibodyopsonized Ag; they may also capture Ag using BDCA-2, and DCIR. Following their uptake by DC, Ag are first directed to endosomes and endo-lysosomes, and processed via the exogenous pathway for presentation on MHC class II molecules to CD4 þ T lymphocytes. Ag-derived peptides can also be loaded on MHC class I molecules via two cross-presentation pathways: the cytosolic and the vacuolar pathways (Blum et al., 2013). Cross-presentation, that is, the presentation of peptides derived from exogenous Ag by MHC class I molecules (Joffre et al., 2012), is crucial for initiation of CD8 þ T cell responses, and is predominantly performed by DC, acting as bona fide professional APC.

Dendritic Cells Maturation and Migration Ag capture mediated by CLR must be associated with the recognition of a danger signal for DC to trigger a protective immune response. Pattern recognition receptors (PRR) are environmental sensors that include the surface or endosomal toll-like receptors (TLR), and intracellular NOD-like (NLR) and helicases (RIG-like) (Krishnaswamy et al., 2013; van Kooyk and Geijtenbeek, 2003; Geijtenbeek and Gringhuis, 2009; Goutagny et al., 2012). These receptors recognize highly conserved structures expressed by pathogens: the so-called pathogen-associated molecular patterns (PAMP). The expression of TLR is somehow difficult to evaluate, since specific antibodies are not available for all TLR, and most analyses were made by RT-qPCR or functional assays with TLR-ligands. While CD1c þ cDC mainly express TLR 2, 4, 5, allowing recognition of peptidoglycan, LPS, flagellin, or zymozan, CD141 þ cDC express high levels of TLR3, and pDC express TLR7 and 9, specific for viral single-stranded RNA (ssRNA) and hypomethylated CpG motifs of bacterial double-stranded DNA (dsDNA) (Hemont et al., 2013; Leal Rojas et al., 2017). Triggering of PRR modulates DC phenotype, DC maturation (upregulation of MHC class II molecules, expression of costimulatory molecules, for example, CD40, CD80, CD86, 4-1BBL, OX-40L) and cell shape modifications (dendritic morphology is acquired). DC migration to secondary lymphoid organs is driven by chemokine receptors such as CCR7 (receptor for the lymph node homing chemokines CCL19 and CCL21), where their cytokine secretion allows efficient T cell priming, activation, and polarization (Reis e Sousa, 2006). Mature DC phenotype can vary according to the nature of PAMP detected, and is further influenced by microenvironment clues, tissue integrity or local inflammation. Of note, DC are able to detect immunogenic signals produced by other cells, such as cell and tissue damage, and recognize “eat-me signals” expressed by necrotic and apoptotic cells (phosphatidyl-serine, calreticulin); they can also be activated by soluble factors released by dying-cells (ATP, HMGB-1) (Hemont et al., 2013). DC are also found infiltrating tumors, and correlations have been described between tumor prognosis and the presence of mature or immature cDC, or the presence of pDC (Vacchelli et al., 2013). However, the tumor-microenvironment may contain many immunosuppressive factors (e.g., IL-10, TSLP, TGFb, IDO), secreted by tumor cells, stromal cells, myeloid-derived suppressor cells (MDSC), or regulatory T cells that can locally suppress DC functions, blocking their recruitment, Ag presentation capacity, their activation or their maturation (Hargadon, 2013; Palucka and Banchereau, 2012; Faget et al., 2012; Le Mercier et al., 2013; Sisirak et al., 2012; Aspord et al., 2013). Understanding the complex interactions between dendritic cells and the tumor microenvironment is crucial to decipher how tumors subvert DC functions, in order to restore these functions (Palucka and Banchereau, 2012).

T Cell Responses Driven by Dendritic Cells In secondary lymphoid organs, mature DC loaded with Ag present antigenic peptides and initiate CD4 þ and CD8 þ responses, and their functional plasticity is responsible for triggering different types of response according to the context of Ag encounter. In addition, the functional specialization of different DC subsets may help drive various classes of T cell responses. PDC are specialized in initiating antiviral immune responses and through their capacity to secrete huge amounts of type I interferon (Liu, 2005), they are involved in innate and adaptive immune responses (Colonna et al., 2004; Lande and Gilliet, 2010; Reizis et al., 2011). In vitro, pDC have been shown to capture and cross-present tumor or viral antigens (Lui et al., 2009), and stimulate specific adaptive T cell responses. In vivo, they are crucial for the development of cytotoxic effector cells in the context of cancer or viral infections (Villadangos and Young, 2008). Because they secrete cytokines such as IFNa and express costimulatory molecules (ICOSL, GITRL, OX40L, or CD70.), they can interact with innate and adaptive immune cells, modulate natural killer (NK) and iNKT cell responses, and activate B lymphocytes (Swiecki and Colonna, 2015). PDC can also prime Th1 or Th2 responses following contact with virus or IL-3 (Ito et al., 2004). Functional specialization of cDC has also been described, and different subsets can produce IL12, IL-15, type I or III interferon, as well as regulatory cytokines, depending on the context of their activation. Among cDC, CD14 þ DC preferentially activate follicular helper T cells, facilitating/enhancing humoral responses; CD1c þ DC may polarize T cells toward Th1 or Th17 pathway through their IL-12p70 and IL-23 secretion, CD141þ cDC seem as efficient as CD1cþ cDC to prime Th1 polarization, however, they can also induce Th2 cells through their expression of OX-40L (Leal Rojas et al., 2017; Durand and

Cancer Vaccines: Dendritic Cell-Based Vaccines and Related Approaches

263

Segura, 2015; Yu et al., 2014). Ag-specific T cells recognizing their cognate Ag presented by DC differentiate and proliferate (Kapsenberg, 2003). DC-derived IFNa, IL-12, or IL-15 skew differentiation of CD4þ T lymphocytes toward the Th1 phenotype, that favors differentiation of CD8 þ CTL. Conversely, if DC express IDO (indoleamine 2,3-dioxygenase), secrete IL-10, TGFb, or express coinhibitory molecules such as CTLA-4 or PD-L1, they can drive the differentiation of CD4 þ regulatory T cells, with tolerogenic functions (Clark and Kupper, 2005). Of note, immature DC are intrinsically tolerogenic, and can induce specific T cell anergy. Altogether, the quality of the generated T cell response depends on the functional status of DC and the DC subtype, and DC plasticity is a key element for orchestrating this diversity of immune responses. Naïve T lymphocytes migrate to secondary lymphoid organs where they scan MHC-bound peptidic antigens on DC through their T cell receptor (TCR). Upon specific antigen encounter, T lymphocytes are activated, proliferate and differentiate, and then leave the lymph node to reach the tissues where the Ag is expressed. This specific migration is governed by expression of a set of chemokine receptors and adhesion molecules at the surface of T cells (Sheridan and Lefrancois, 2011). T lymphocytes expressing CLA, CCR4, and CCR10 migrate to the skin, if they express CD49a they will preferentially home to the lung, whereas CCR9 and a4b7 integrin expressing T lymphocytes migrate to the intestine. Thus, effector T cell function clearly depends on their capacity to migrate to the site where Ag are located, and to exert their function despite an immunosuppressive environment, to effectively counteract immune evasion (Hargadon, 2013; Gajewski et al., 2013).

Dendritic Cell-Based Therapeutic Vaccination The goal of cancer vaccines is to induce efficient CTL responses directed toward tumor-associated antigens. Several approaches have been developed, based on the injection of the defined antigens, in the form of whole proteins or antigenic peptides, or by the delivery of genes encoding these peptides or proteins, associated or not with adjuvants or costimulatory signals. Since dendritic cells are very efficient APC, many cancer vaccine approaches use these cells, either by injecting them loaded with the Ag, or by using vectors to address the Ag to specific DC subsets in vivo. Several parameters must be taken into account to optimize the design of therapeutic cancer vaccines based on DC or on other approaches of Ag delivery.

Diversity of Dendritic Cells As described above, DC belong to a complex multicellular system whose role in initiating potent immune responses is crucial for antitumor immunotherapy. Different strategies taking advantage of this potency are used at the forefront of cancer immunotherapy. However, due to the heterogeneity and plasticity of DC subtypes, several parameters have to be taken into account and tested individually to devise an optimal DC-based immunogen. Previous studies have shown how difficult it is to identify the best DC subset and DC differentiation protocol to be used for maximal clinical efficacy. Blood DC represent a tiny percentages of mononuclear cells (less than 1% for CD1c þ DC and even less for CD141 þ DC), so DC purified from blood have been used only sporadically in immunization protocols (see for example the clinical trial NCT01690377 in advanced melanoma that relies on injection of purified CD1c þ DC). To obtain sufficient numbers of DC, and for logistic and economic purposes, monocyte-derived DC have been the predominant source for vaccination in human clinical trials. Indeed, monocyte-derived DC (moDC) can be obtained in large numbers by isolation of monocytes, followed by incubation with GM-CSF and IL-4, yielding pure moDC populations in a relatively easy and scalable way. However, moDC may not be endowed with the same stimulation, migration, and functional capabilities as endogenous DC (transcriptomic analysis suggest a phenotype close to inflammatory DC) (Segura et al., 2013), although they are very potent at MHC-II presentation and CD4þ T cell activation. DC generated from monocytes by culture in GM-CSF and IFNa may be more efficient for cross-presentation to CD8 þ T cells (Lapenta et al., 2006). CD34 þ hematopoietic progenitor cells (HPCs) can be used as precursors for DC differentiation. HPC treated with c-kit ligand and tumor necrosis factor-a (TNF-a) yield subsets of myeloid DC in different stages of differentiation, including Langerhans cells (Caux et al., 1996). Alternatives to in vitro-derived DC include cell lines with similar functionalities as primary DC, such as the myeloid MUTZ-3 cell line derived from leukemic myeloid cells; however these cells have not yet been used in clinical trials. As described further below, a cell line was derived from the leukemic cells of a patient with plasmacytoid DC leukemia, which shares many phenotypic and functional features with primary pDC; this cell line (GENius-Vac) was recently used in a phase I clinical trial in melanoma. ProvengeÒ/Sipuleucel, developed by Dendreon Corporation, which was FDA approved in 2010, represents a unique case, as the injected cells are not strictly DC per se. Indeed, patients’ monocytes are incubated for a short period of time (4 days) with a fusion protein of GM-CSF and prostatic acid phosphatase (PAP). No differentiation factor, such as IL-4 or IL-15, is added, and no maturation is induced. The actual mechanisms of action of the vaccine are not known, but injected cells may complete their differentiation in vivo to stimulate CD4 þ and CD8 þ T cells. It is to be noted that injected DC may not directly prime specific T cells. Thus, it has been demonstrated in mouse models that antigen-loaded DC do not prime T cell responses without the help from endogenous DC (Yewdall et al., 2010). Antigen transfer between DC populations in the lymph node, through transfer of MHC-peptide (cross-dressing), exosome secretion, or capture of apoptotic cell, likely plays a role in CD4 þ T cells activation and cross-presentation to CD8 þ T cells. Studies of T cell activation in viral infection have identified choreographed interactions between antigen-carrying migratory DC and

264

Cancer Vaccines: Dendritic Cell-Based Vaccines and Related Approaches

lymph node resident XCR1 þ DC that are necessary to cross-present viral antigen to CD8 þ CTL (Hor et al., 2015; Eickhoff et al., 2015).

Nature and Formulation of Tumor Antigens A critical aspect of vaccination is the nature of the targeted antigens, as well as the form with which they are delivered to DC. Most clinical trials have used DC pulsed by MHC class I restricted peptides to specifically induce CD8 þ T cells. In order to optimize the uptake and presentation of antigens, in vitro-derived DC can be loaded with short (9-mer) or long peptides, whole proteins, tumor antigen coding RNA, DNA, or tumor lysate. In other approaches, the Ag (peptides, proteins, RNA, or DNA) can also be directly injected in diverse formulations, associating vectors and adjuvants for optimal targeting to tissue dendritic cells. Peptide or protein-based vaccines can be associated with heat shock proteins, or encapsulated into liposomes (Hu et al., 2018). DNAvaccines seem safe and promising. Autologous cells at the injection site, upon electroporation or other delivery processes, will up take the nucleic acids and synthesize the antigens, allowing development of immune responses directed toward these antigens. Different mechanisms of action are proposed for DNA vaccines, depending on the route of injection. If the vaccine is injected intranodally, or in the skin, Ag are directly produced by local dendritic cells, whereas if the DNA is injected intramuscularly, Ag are secreted by muscular cells, and must be taken-up by dendritic cells to trigger immune responses. The current strategies to improve the efficacy of DNA vaccine have been described recently (Tiptiri-Kourpeti et al., 2016): they include optimization of delivery systems, association with adjuvants, and introduction of sequences coding for costimulatory signals, cytokines, chemokines, or danger signals to improve vaccine immunogenicity. Whatever the form chosen for the vaccine candidate antigens, it is crucial to succeed in targeting them to dendritic cells, in order to ensure optimal T cell priming. Targeting multiple defined or undefined (lysate) antigens is likely to be advantageous, by limiting the risk of tumor escape through antigen loss. In addition, recent studies have shown that tumors can take advantage of MHC locus inactivation to prevent presentation of too diverse a repertoire of peptides, such that tumors with loss of heterozygosity in HLA class I, resulting in HLA-homozygosity, is a prevalent immune escape mechanism in lung cancer (McGranahan et al., 2017). Maximal heterozygosity at HLA class I loci (A, B, and C) and particular HLA class I supertypes are associated with increased overall survival after immune checkpoint blockade (Chowell et al., 2018). Long antigenic sequences (DNA, RNA encoded or long peptides) can potentially enable both CD4 þ and CD8 þ T cell activation. Concomitant activation of helper CD4 þ T cell responses and CD8 þ cytotoxic T cells has been associated with clinical benefit, as demonstrated with DC pulsed by MHC-I and MHC-II restricted peptides (Aarntzen et al., 2013). Antigen choice is crucial, as highlighted by the fact that not all tumor-associated antigenic sequences are immunogenic, and that not all immunogenic sequences can induce T cell-mediated tumor rejection, which defines a limited subset of tumor rejection antigens. The size and functional status of the specific T cell repertoire has to be considered. As an example, differentiation antigens, such as Melan-A or tyrosinase in melanoma, are immunogenic and a number of DC vaccine trials have shown potent CD8 þ T cell amplifications against these antigens. However, the functionality and avidity of specific T cell may be suboptimal to mediate tumor rejection, due to the potential peripheral tolerance to these self-antigens. Ideal tumor antigens are either selectively expressed by the tumor, but not by healthy somatic tissues, or are specifically and uniquely expressed by the tumor. The first category is exemplified by the so-called cancer-testis (CT) antigens (e.g., NY-ESO-1, MAGE antigens), whose expression is restricted to germline and placental tissue, and absent in somatic tissues. Tumors of diverse origins often express CT antigens due to epigenetic dysregulation. CT antigens are being used in a large number of clinical trials and have been shown to be highly immunogenic, despite relatively limited clinical responses up to now. In the latest years, progress in sequencing technologies has opened the door to the identification of the full spectrum of somatic mutations in one particular tumor. In conjunction with the improvements in MHC binding prediction algorithms, it is now possible to identify and select tumor neoantigens (derived from somatic mutations) expressed by the tumor in a highly specific fashion (Kroemer and Zitvogel, 2012). As an example, a large scale high-throughput analysis was performed by the Xpresident platform, combining sequencing of HLA-bound peptides, RNA sequencing and whole exome DNA sequencing of renal cell carcinoma and healthy autologous tissue, to identify a combination of 10 peptides specifically expressed by the tumor (TUMAP). These peptides are now being used for vaccination in conjunction with GM-CSF injection (Walter et al., 2012). An extensive catalog of somatic mutations in 7042 cancers was published, identifying 4,938,362 mutations and 20 distinct mutational signatures, of which some are shared among different cancers, and others are unique (Alexandrov et al., 2013). Such neoantigens are now being targeted in a large number of clinical trials (Hu et al., 2018). Preliminary results seem to indicate that both the quantity and the quality of neoantigens expressed by a tumor are important in determining the efficacy of vaccination. Thus, tumor with the highest somatic mutation and neoantigen burden (e.g., lung cancer, melanoma), patients with microsatellite instability, tend to respond better to immune checkpoint inhibitors (Rizvi et al., 2015; Van Allen et al., 2015). Most mutations are in nonessential genes (passenger mutations) and vastly outnumber mutations in driver genes. Recent studies have uncovered the role of presumably immunodominant neoantigenic sequences in determining clinical outcome in long-term cancer survivors (Luksza et al., 2017). Both the probability of binding to MHC molecules and homology with microbial peptide sequences signal a high probability of recognition by specific CD8 þ T cells, and tumor expression of these high quality neoantigens predicts a better clinical outcome. Characterization of the specificity of tumor infiltrating lymphocytes (TILs) will further validate these observations and enhance predictions of immunogenic tumor antigens (Linnemann et al., 2013). Finally, DC vaccination can potentially be enhanced by the use of epigenetic drugs that can augment the expression of tumor antigens and MHC molecules, in addition to their direct antioncogenic and pro-apoptotic effect (Heninger et al., 2015).

Cancer Vaccines: Dendritic Cell-Based Vaccines and Related Approaches

265

Dendritic Cell Maturation and Injection Route We mentioned the importance of DC maturation stage, which determines the quality and magnitude of induced T cell responses. The generation of local inflammation increasing the immunogenicity of a vaccine can be obtained by using adjuvants such as Montanide water-in-oil emulsions (Clancy-Thompson et al., 2013; Valmori et al., 2007). By using TLR or NLR agonists (mimicking microbial stimulation) as adjuvants, one can specifically activate one or the other DC subset, endowing them with the requisite immunostimulatory properties. Thus, imiquimod, a TLR7 agonist, can activate pDC to produce type I IFN, whereas poly-IC can activate cDC through TLR3 to produce IL-12. Poly-I:C stimulation seems to be optimal for CD8 T cell priming by moDC, whereas triggering other TLR, such as TLR4, may induce suppressive IL-10 secretion that could impede on IL-12 secretion and Th1 priming (Bogunovic et al., 2011). CD40 agonists can also be used for maturing and “licensing” DC, partially alleviating the need for CD4 þ T cell help. GM-CSF has been incorporated in several cancer vaccines to enhance their immunostimulatory capacities, for example in the GVAX strategy based on transduction of autologous or allogeneic whole tumor cells (Burkhardt et al., 2013; Lipson et al., 2015). Transduction of DC with constructs knocking down negative regulators of DC activation, such as SOCS1 (Evel-Kabler et al., 2006), can help potentiate cytokine secretion and prolong antigen presentation by injected DC to enhance antitumor immune responses (Shen et al., 2004). Many questions remain as to what is the best route for therapeutic vaccine injection. Regarding DC-based vaccines, subcutaneous injection is the most frequently used, while intranodal injection has also been tested to bypass DC migration to secondary organs. DNA vaccine can be administered by intradermal, intranodal, or intramuscular routes. Other approaches include device-mediated (gene gun) DNA delivery to keratinocytes or Langerhans cells. Electroporation by local application of electrical pulses causes transient cell membrane permeabilization, allowing DNA entry into cell cytoplasm. It is crucial to garner knowledge on DC circulation upon immunization, as this will likely impact on activated T cell trafficking, or on the differentiation of central and effector memory T cells. Thus, immunization against cutaneous and pulmonary tumors probably requires different routes of injection for optimal antitumor T cell activation and homing (Sandoval et al., 2013). Preconditioning of draining lymph nodes in animal models has shown that it can highly increase DC migration and vaccine efficacy (MartIn-Fontecha et al., 2003). Some strategies currently being tested in clinical trials involve direct intratumor injection of TLR agonists, with or without concomitant T cell costimulatory activation. They aim at modifying the tumor microenvironment, while recruiting DC and T cells, inducing DC migration to draining lymph nodes. Remarkably, in situ injection of CpG oligonucleotides (TLR9 agonists) in conjunction with OX40L agonistic antibodies in various murine tumors triggers systemic cytotoxic immune responses, eradicating metastasizing tumor cells and preventing tumor growth upon rechallenge (Sagiv-Barfi et al., 2018). Despite these advances, it is not clear yet which DC activation protocol and vaccine injection route will induce adequate T cell stimulation, in view of the frequent immune dysfunction observed in various malignancies.

Dendritic Cells in Cancer Immunotherapy Clinical Trials Increased knowledge of DC biology and optimized protocols for DC generation have translated into the design of immunotherapeutic strategies that are now being tested in multiple clinical trials. Safety and clinical evaluations are most of the time complemented with immunomonitoring protocols that quantify the induction of specific helper and CTL responses against tumor antigens.

Ongoing DC-Based Vaccines A pilot clinical trial in lymphoma utilizing DC pulsed by tumor-specific idiotype was performed more than two decades ago. However, as of today, only two commercial DC vaccines are approved as drug products. The most prominent is ProvengeÒ/Sipuleucel for the treatment of advanced prostate cancer. CreaVaxRCCÒ developed by the company CreaGene was approved in 2007 in Korea for the treatment of metastatic renal carcinoma. The National Institute of Health website (clinicaltrials.gov) represents an invaluable database cataloguing worldwide information about clinical trials. An analysis of the number of ongoing and completed DC vaccine trials highlights the positive dynamics of therapeutic DC vaccines for the treatment of cancer. Thus, a keyword search for “dendritic cell” and “vaccine” identifies 400 clinical studies. Sixty-seven of them are actively recruiting at the time of this writing. A large number of these studies are in phase 1 (257) and phase 2 (221), whereas only 10 protocols are currently tested in phase 3 trials, reflecting the challenges facing many early clinical trials and the difficulty in demonstrating clinical benefit. The ongoing trial NCT00045968 in glioblastoma multiforme is one of the few phase 3 DC-based studies. Developed by Northwest Biotherapeutics, DCVaxÒ-L consists in autologous DC pulsed with an autologous tumor lysate. Patients are injected with DCVaxÒ-L over a period of 120 weeks, while the placebo group receives mononuclear cells. Phase 1/2 evaluation of DCVaxÒ-L demonstrated an increase in median overall survival (> 4 years in 33% of patients and more than 6 years in 27%, compared to 14.6 months with conventional treatments) warranting further evaluation in a phase 3 trial. Many approaches are being tested to optimize immunization by DC. The vast majority (99%) of these studies focus on immunization by cDC, whether directly purified from blood, or differentiated from monocytes or CD34 þ HSC. However, two teams have also exploited the antigen presenting capabilities of pDC to immunize patients with metastatic melanoma. In the first trial (NCT0190377) (Tel et al., 2013), patients received intranodal injection of purified and tumor antigen-loaded autologous pDC.

266

Cancer Vaccines: Dendritic Cell-Based Vaccines and Related Approaches

In the second one (NCT01863108), the semi-allogeneic (HLA-A2 þ) pDC cell line GENius-Vac was used as vaccination platform upon loading pDC with melanoma-associated HLA-A2-restricted peptides. This approach, relying on the use of a wellcharacterized cell line, allows large cell numbers of a homogeneous cell preparation to be injected, at reduced costs. Similarly, in clinical trial NCT01373515 HLA-A2 þ patients with acute myeloid leukemia were injected with a semi-allogeneic (HLA-A2 þ) acute myeloid leukemia cell line that differentiates into functional mature dendritic cells upon anthracycline treatment. Rather than relying on injection of whole dendritic cells, an original approach exploits the immunogenic properties of DC-derived nanovesicles termed “exosomes” that can mediate MHC-dependent immune responses. A phase II trial of vaccination with tumor antigen-loaded dendritic cell-derived exosomes was performed in nonsmall cell lung cancer (NCT01159288).

Source of Antigen It is still unclear which antigenic targets will prove to be the most beneficial for long term clinical efficacy. Defined shared antigens, such as cancer-testis antigens or carcinoembryonic antigens are either loaded as peptides onto in vitro-derived DC, or targeted to DC through endocytic receptors. An example of this latter strategy is the intranodal injection of a DEC-205- NY-ESO-1 fusion protein with or without Sirolimus (rapamycin) administration in astrocytoma (NCT01522820). Targeting of antigens to DEC-205 on DC has been shown to potentiate cross-presentation to CD8 þ T cells (Bonifaz et al., 2002). As mentioned above, it is becoming widely appreciated that including multiple antigens in a vaccine could prevent tumor immune escape by outgrowth of antigen-loss variants. This may be one of the reason for the efficacy of the DCVaxÒ-L trial, where undefined but numerous antigens are loaded onto DC in the form of a cellular lysate. Several clinical trials are ongoing with DNA vaccines (Tiptiri-Kourpeti et al., 2016), mainly in melanoma, prostate or HPV-associated cancers, targeting various kinds of tumor-associated Ag (over-expressed, CT, or onco-viral). Recently, a clinical trial based on a DNA vaccine coding for two epitopes from gp100 melanoma-derived Ag inserted in a human IgG1 antibody sequence has been shown to be safe and stimulated T cell responses in melanoma patients (Patel et al., 2018). The new sequencing technologies (whole exome sequencing, RNAseq analysis) provide the opportunity to identify and target many tumor-specific neoantigens at the same time. So far, several phase I trials evaluating vaccination with personalized neoantigens are ongoing or have been completed in melanoma. The NCT00683670 study used DC loaded with short HLA-A2-restricted neoantigen peptides (Carreno et al., 2015). The NeoVax trial (NCT01970358) involved subcutaneous administration of up to 20 long neopeptides, admixed with poly-IC-LC (carboxymethylcellulose, polyinosinic-polycytidylic acid, and poly-L-lysine) (Ott et al., 2017). In the IVAC MUTANOME trial (NCT02035956) several doses of RNA encoding shared tumor antigen (tyrosinase and/or NY-ESO-1) and neoantigens were injected into inguinal lymph nodes (Sahin et al., 2017). The latter two studies rely on neoantigen uptake and presentation by tissue- or lymph-node resident DC to activate antitumor T cells. Although performed on a small number of patients, impressive immune responses against neoantigens were observed in all three studies. Some patients experienced vaccine-related objective clinical responses in the NeoVax and IVAC MUTANOME trials, warranting further evaluation in large scale studies. Neoantigens can also be used in DNA vaccine approaches, as in the ongoing clinical trial NCT03122106, in pancreatic cancer patients. This neoantigen DNA vaccine, administered with an electroporation device, incorporates neoantigens and personalized mesothelin polyepitopes, with the aim to activate both CD4þ and CD8 þ T lymphocytes.

Reversing Immunosuppression Due to the systemic and local immune dysregulation observed in the tumor microenvironment, and the sensitivity of DC to immunosuppressive cytokines and ligands, it is likely that the best immunization regimens will require coadministration of DC vaccines with other therapies. Current combination trials testing DC vaccines have been extensively described (Saxena and Bhardwaj, 2018), and we will briefly mention here a few important themes. One goal is to improve antigen delivery to DC in an immunogenic environment, which is being tested in different ways. DC vaccination combined with chemotherapy inducing “immunogenic cell death” (doxorubicine and cyclophosphamide) (e.g., NCT00082641), or with concomitant injection of antitumor T cells (unmodified autologous T cells (e.g., NCT00338377), antigen-specific TCR transduced T cells (e.g., NCT00910650, NCT01697527)), will test this concept in different malignancies. Another aim is to bypass immunosuppressive pathways, whether metabolic (e.g., IDO), induced by tumor-secreted mediators (e.g., prostaglandins) or through surface receptors (such as the PD-1/PD-L1 interaction). Thus, the breast cancer clinical trial NCT01042535 combines p53-targeted DC immunization with administration of the IDO inhibitor 1-methyl-D-tryptophan, whereas in trial NCT01560923 patients with advanced prostate cancer receive Sipuleucel with the IDO inhibitor indoximod. Immune checkpoint inhibitors (anti-PD-1 antibodies nivolumab, pembrolizumab, and anti-PD-L1 atezolizumab) are used to reawaken and prevent deactivation of T cell immune responses. Building on the results of the first DCVax-L trial, a phase 2 trial is being initiated, that combines vaccination with tumor lysate-loaded DC with nivolumab injection (NCT03014804). Finally, a few studies test sequential immunotherapy with multiple immunomodulators, in a multipronged strategy to enhance vaccine-induced immune responses. In trial NCT02677155, patients with lymphoma receive sequential intratumor injections of low-dose rituximab (anti-CD20) combined with local radiotherapy, and autologous DC with subcutaneous GM-CSF injection, with the aim of enhancing tumor antigen delivery and uptake by DC. Patients also receive intravenous antiPD-1 antibody (Pembrolizumab) to overcome tumor tolerance. This strategy reflects the growing trend toward combination of standalone promising approaches, such as DC vaccination, immunogenic conditioning by cytotoxic therapy, neoantigen targeting, and immune checkpoint inhibitors blockade, which when used together will hopefully translate into greater clinical benefits for cancer patients.

Cancer Vaccines: Dendritic Cell-Based Vaccines and Related Approaches

267

Defining Responders and Nonresponders Patients Understanding the effects of DC vaccines on the immune system is crucial to better characterize the pathways required to induce protective antitumor responses. The status of patient immune system affects the capacity to trigger antitumor T cell responses, modulates responses to immunotherapies, and also affects clinical responses. It is crucial not only to define the factors required for the activation of effector T cells, but also to define the interactions between DC and innate immune cells. The modulation of suppressive cells in peripheral blood but also at the tumor site have to be better characterized to potentiate vaccine efficacy and design. Identification of response biomarkers differentiating early responder and nonresponder patients will guide future developments. System-wide immunologic approaches, combining transcriptome profiling with multiparametric analysis by flow cytometry and proteomic analysis, may lead to the identification of new biomarkers of efficacy (Pulendran et al., 2010). Isolation of molecular and cellular signatures predicting the immunogenicity and efficacy of immunotherapies, and their correlation with induction of protective immune responses, will help optimize future vaccines.

Clinical Evaluation In addition, RECIST (response evaluation criteria in solid tumors) criteria of clinical evaluation (Eisenhauer et al., 2009) may not be appropriate to evaluate the clinical efficacy of immunomodulatory therapies. Indeed, in early time points after the beginning of immunotherapeutic treatment, a temporary tumor “growth” can be observed preceding a durable regression, the former reflecting inflammatory processes associated with active immune responses and tumor infiltration. Clinical evaluation based on irRC (immune-related response criteria) (Wolchok et al., 2009) will be more suitable to define the real clinical impact of immunotherapies.

Future Directions With Cancer Vaccines Therapeutic DC vaccines in clinic trials have often been disappointing, despite exciting results obtained in animals and preclinical trials. However with the development of new generation products such as combination with novel adjuvants or other immunotherapies, the use of new DC subsets (pDC) and clinical evaluation based on immunological criteria (Wolchok et al., 2009), these approaches should evolve toward potent anticancer therapies. Clinical trials performed in the last 20 years proved that DCbased vaccines are well tolerated, highly immunogenic, and trigger tumor-specific and functional T-cell responses in many patients. Despite limited objective clinical responses, these responses are often durable, correlate with the induction of antitumor immune responses, and have been associated with objective tumor regressions in some instances, demonstrating the capacity of the immune system to control tumor development. New clinical strategies are needed to improve the potential of DC vaccines to trigger tumor regression. One major problem of DC-based trials is that patients are treated in advanced stages of cancer, when the immune system is subverted by the unfavorable tumor microenvironment. Critical points to overcome for a better efficacy are (1) the absence of costimulatory signals within tumor microenvironment, (2) the presence of inadequate signals that lead to DC subversion and prevent triggering of protective antitumor immune responses, and (3) a better understanding of the mechanisms directing the response to immunotherapies. Advances in the knowledge of DC biology and in the understanding of interactions between tumor cells and the immune system have led to the emergence of new strategies that exploit the properties of various DC subsets and neutralize the immunosuppressive tumor microenvironment. Whereas past developments often required individual DC production, the trend is now to design universal products such as vaccines targeting DC in vivo, for broader use with simplified, standardized, and less expensive products. However, there is a parallel trend toward the development of personalized approaches using neoantigens. All in all, these strategies offer new therapeutic perspectives in the treatment of cancer and will hopefully translate into better clinical success (Fig. 1).

Exploitation of Various DC Subsets As mentioned earlier, DC used in immunotherapy are generated from monocytes, differentiated in vitro from CD34 þ HSC, or directly purified from circulating immune cells of patients. The discovery of new DC subsets with unique properties allows their use as vectors for cancer immunotherapy (Ueno et al., 2011; Palucka and Banchereau, 2013b). Distinct DC subsets trigger various T cell responses, and the quality of CTL induced by DC is crucial for antitumor efficacy. Various cytokine cocktails drive differentiation of DC with specific properties in vitro. Thus, moDC generated in the presence of GM-CSF and IL-15 induce CTL exhibiting higher antitumor activity compared to CTL induced by moDC generated in presence of GM-CSF and IL-4 (Dubsky et al., 2007), which may transfer into higher clinical efficacy. HPC can also be used to generate multiple DC subsets in sufficient numbers for immunotherapy. Differentiation of the cross-presenting CLEC9A þ DC or Langerhans cells from CD34 þ HPCs brings the opportunity to use specialized DC subsets for vaccination, and is currently being tested in clinical trials. Alternative strategies consist in directly purifying DC subsets from circulating immune cells of the patients. The use of primary circulating CD1c þ DC activated and loaded with tumor-derived peptides led to the induction of strong antitumor T cell responses associated with favorable progression free survival in stage IV melanoma patients (Schreibelt et al., 2016). pDC are also crucial actors of antitumor immune responses, and represent promising vectors for immunotherapy (Tel et al., 2012). In melanoma patients, injection of purified circulating pDC that

268

Cancer Vaccines: Dendritic Cell-Based Vaccines and Related Approaches

Fig. 1 Optimizing DC-based therapy. The current approaches are based on injection of autologous dendritic cells generated from monocytes or hematopoietic stem cells, or purified from patient’s blood, or of allogeneic DC lines. These DC are loaded in vitro with tumor Ag (neoantigens or shared Ag), activated with different adjuvants, and injected to the patients. DC in vivo targeting approaches are emerging; they aim at specifically targeting DC in vivo using pattern recognition receptors (PRR) or C-type lectin (CLR) ligands. CLR can be targeted through antibodies or glycans conjugated to an antigen, while TLR triggering activates a signaling cascade resulting in DC activation. Combination strategies may be the key for DC vaccine success, with the use of drugs neutralizing local immunosuppression, or debulking the tumor.

have been activated and pulsed with peptides derived from tumor antigens triggered significant CD4þ and CD8þ T cell responses, and was associated with promising clinical responses (Tel et al., 2013). However, large scale therapeutic use of autologous primary DC is difficult due to the low number of circulating DC, their potential functional alteration induced by the tumor, and the individual preparation required for each patient. These obstacles can be overcome by using semi-allogeneic DC cell lines. The MUTZ3 cell line derived from human leukemic myeloid cells harbor cDC phenotype and function (Santegoets et al., 2008). We developed an alternative strategy based on the exploitation of a human pDC line HLA-A*0201 þ derived from leukemic pDC. This pDC line shares phenotypic and functional properties with primary pDC. The GENiusVac strategy (Aspord et al., 2011a) is based on the use of the irradiated pDC line loaded with HLA-A*0201 þ-restricted peptides derived from tumor or viral antigens to trigger antigenspecific cytotoxic CD8 T cells in allogeneic but HLA-A*0201 þ matched settings. We demonstrated the strong potency of this strategy to induce multispecific and highly functional antitumor and antiviral T cell responses in vitro, its therapeutic efficacy in vivo in humanized mice, and its clinical relevance ex vivo with patients’ T cells (Aspord et al., 2010, 2011b, 2012). The GENiusVac strategy is currently tested in a phase I trial in metastatic melanoma. Such “one to many” strategy offers many advantages compared to existing immunotherapies, including its standardizable design, broad potential use for numerous tumoral and viral pathologies. Partial HLA matching between the vaccine and the patients further exploits the allogeneic effect to potentiate the immunogenicity of tumor or viral antigens (Fabre, 2001). Such semi-allogeneic settings have been explored in strategies using hybrid vaccines between DC and allogeneic tumor cells (Kugler et al., 2000), in another setting, allogeneic skin graft has been shown to trigger antitumor response in patients with cancer prostate (Muir et al., 2006).

Exploitation of Neoantigens As described above, neoantigens are generated by tumor-related (driver) or spontaneous (bystander) mutations that can be presented as highly specific tumor antigens. It has been shown that the mutational load and microsatellite instability in solid tumors

Cancer Vaccines: Dendritic Cell-Based Vaccines and Related Approaches

269

correlates with the response to immune checkpoint blockers (Rizvi et al., 2015; Van Allen et al., 2015). The use of DC loaded with patient-specific peptides derived from mutated neoantigens is a powerful alternative, currently under investigation, that may trigger epitope spreading, diversification of the TCR repertoire and elicit specific CD8 T cells in cancer patients (Carreno et al., 2015).

In Vivo DC Targeting DC scan the environment through PRR and take up antigens recognized by their CLR. Various DC subsets differentially express PRR, and CLR allowing their specific targeting to endow them with desired functionalities. Thus, emerging strategies aim at specifically targeting DC in vivo using PRR or CLR ligands. CLR can be targeted through an antibody or a glycan conjugated to an antigen that will be internalized, allowing the cross-presentation of peptides derived from the antigen. TLR triggering activates a signaling cascade resulting in DC activation: upregulation of costimulatory molecules expression and cytokine secretion, which favor innate and adaptive immune responses. Such in vivo DC targeting offers the advantage to develop universal off-the-shelf products, which is highly convenient for future transfer into the clinic.

Targeting DC through CLR A large number of CLR can be exploited to target different DC subsets (Ueno et al., 2011). Current approaches targeting CLR with antibodies show promising preclinical results: targeting of DEC-205, DC-SIGN or MR expressed by cDC, langerin expressed by LC, CLEC9A expressed by BDCA-3 þ DC, or dectin-1 and DCIR expressed by pDC (Palucka and Banchereau, 2013a; Ueno et al., 2011) allows the capture and cross-presentation of Ab-associated antigens, leading to induction of specific T cell responses. The use of DCtargeting antibodies combined to antigens together with adjuvants leads to efficient presentation of antigens by DC in vivo and to induction of specific CD4 and CD8 T cell responses (Tacken and Figdor, 2011). Many preclinical studies have demonstrated the efficacy of directly targeting DC in vivo to trigger specific immune responses (Caminschi et al., 2008; Tacken et al., 2005) and to control tumor growth (Sancho et al., 2008; He et al., 2007). CLR can also be targeted by carbohydrates ligands (Lepenies et al., 2013). Engineered glycans that have been optimized for increased binding affinity to CLR show promise as a means to enhance tumor antigen uptake and presentation; preclinical studies combining gp100 to glycans targeting DC-SIGN have demonstrated the feasibility and efficacy of these approaches. The location of targeted DC is important to determine the tropism of primed T cells, as DC from one tissue can induce T cells with migratory properties specific for that tissue. In addition, DC stimulated with different adjuvants can induce T cells with various migratory properties, so that the choice of adjuvant is critical to induce the desired type of immunity at the right location. Proof-of-concept clinical trials are on-going to determine the safety of such strategies. Targeting antigens to DC-specific receptors in vivo bypasses the need for ex vivo DC production, a long and critical step which can potentially alter DC functionality. It further allows antigen uptake and stimulation of specific DC subsets, even rare population with interesting properties. This strategy represents an alternative to strategies based on DC injections which are often costly, restricted to a limited number of patients, and limited to particular DC subsets (the ones that can be purified or produced in sufficient numbers).

Activating DC through TLR As described in detail above, TLR recognize conserved pathogenic motifs present on virus and bacteria, and TLR ligands (TLR-L) represent potent immunomodulatory molecules that can trigger DC maturation and modulate their function. Several TLR-L are under preclinical investigation due to their adjuvant properties for antitumor vaccines (Krieg, 2007) and tested in many clinical trials in cancer. By activating DC (upregulation of MHC and costimulatory molecules, secretion of inflammatory cytokines such as IL-12 and IFNs), they provide the missing signal in DC/T interactions within the tumor microenvironment, allowing induction of Th1-oriented adaptive immunity. TLR agonists already approved by the FDA (BCG, MPL, and imiquimod) together with others (TLR-4 agonist LPS, TLR-3 agonist poly-IC, TLR-7/8 agonist resiquimod) undergo promising clinical developments (Vacchelli et al., 2012). Synthetic TLR agonists efficiently trigger antitumor immunity in vivo in melanoma. In particular, agonists engaging TLR7 and TLR9, expressed by pDC, allow the induction of antitumor immune responses by stimulating pDC function. Imiquimod (TLR7 agonist) is approved for basal carcinoma, and systemic immune activation was observed in melanoma patients in response to TLR7-L (Dummer et al., 2008) or TLR9-L (Molenkamp et al., 2008) administration. The administration of CpG- oligodeoxynucleotides potentiates patients’ response to vaccination (Speiser et al., 2005) and stimulates pDC and NK cell functions (Pashenkov et al., 2006). Recently mRNA transfection-based delivery of a costimulatory molecule cocktail (CD40L, CD70, and constitutively active TLR4) called the TriMix-DC emerged as a potent DC maturation strategy leading to strong T cell responses (Van Lint et al., 2014). Vaccination with autologous TriMix-DC transfected with tumor associated antigens can trigger durable tumor responses in melanoma patients (Van Nuffel et al., 2012).

Combination of CLR and TLR A recent development combining DC targeting through CLR and DC activation via TLR is based on biodegradable nanoparticles constituted of PLGA (poly lactic–co-glycolic acid), covered with anti-DC-SIGN antibodies and containing tumor antigens and adjuvants (TLR3-L, TLR7/8-L) (Tacken et al., 2011). Another strategy is based on liposomes covered with anti-DEC-205 antibodies and containing tumor antigens and adjuvants (van Broekhoven et al., 2004). Such a combination of DC targeting via CLR, antigen delivery and direct in vivo activation of DC through TLR, seems promising (Zitvogel and Palucka, 2011) due to its simple and scalable design.

270

Cancer Vaccines: Dendritic Cell-Based Vaccines and Related Approaches

Toward Combinations of DC-based Immunotherapies With Other Antitumor Therapies With immune checkpoint blockers and inactivation of immunosuppressive cells The discrepancy between the potent immune responses and the disappointing clinical responses observed in clinical trials using DC vaccines is partly due to the presence of regulatory T cells and an immunosuppressive tumor microenvironment (Palucka et al., 2008). Inhibition of immunosuppressive mechanisms is required to improve the clinical efficacy of immunotherapies, especially the ones based on DC vaccines or DC targeting in vivo. Antitumor effector T cells are inhibited by engagement of inhibitory molecules (CTLA-4, PD1) and the often massive infiltration by immunosuppressive cells (regulatory T cells, MDSC) at the tumor site. Immunomodulatory therapies such as anti-CTLA4 and anti-PD1/anti-PD-L1 blockade have achieved prolongation of patient overall survival for 30%–40% of patients (Ott et al., 2013), demonstrating that inhibition of immunosuppressive mechanisms can lead to tumor control. The higher efficacy of combined DC vaccine with immune checkpoint blockers compared to monotherapy is supported by many preclinical studies (Vasaturo et al., 2013), and by a clinical trial in melanoma patients in which durable tumor regressions were observed following DC vaccine combined with anti-CTLA4 treatment (Ribas et al., 2009). Enhancement of tumor-specific immunity with DC vaccines combined to reversal of T cell exhaustion by immune checkpoint blockers is an attractive strategy. In addition, inhibition of MDSC can be achieved by COX-2 and arginase inhibitors (Wesolowski et al., 2013). Regulatory T cells and MDSC can also be targeted by inhibitors of the IDO pathway (Prendergast et al., 2014), a novel class of immunomodulators that may help reverse the dysfunction of antitumor effector cells. Combination of DC vaccines with IDO inhibitors are currently under clinical investigation. The combination of optimized DC vaccines with inhibition of immunosuppressive and immunoregulatory mechanisms is likely to trigger strong antitumor immune responses, priming potent effectors T cells while protecting them from exhaustion.

With chemotherapies Besides lowering tumor burden, many studies demonstrated that some classical antitumor chemotherapies can affect many aspects of antitumor immune responses and display immune-potentiating effects. DC may contribute to the efficacy of conventional antitumor chemotherapies that induce immunogenic cell death of tumor cells, defined as apoptotic processes favoring, rather than suppressing, immune responses. The positive immunologic consequences of certain chemotherapies are well characterized (Zitvogel et al., 2008, 2011). Immunogenic signals from tumor cells undergoing apoptosis can be detected by DC and potentiate DC maturation and cross-presentation of tumor antigens. In particular, chemotherapeutic drugs such as anthracyclin or oxaliplatin can induce an immunogenic tumor cell death of cells characterized by surface exposure of calreticulin which enhances phagocytosis of tumor cells by DC. Similarly, secretion of HMGB1 (a TLR4-L) by tumor cells inhibits phagosome–lysosome fusion in APC, preventing rapid tumor antigen degradation and enabling MHC-I restricted presentation and subsequent priming of T cells (Kroemer et al., 2013). The synergy of chemotherapies with DC vaccines has been demonstrated in colorectal cancer patients treated with DC vaccine combined to oxaliplatin (Lesterhuis et al., 2010).

With monoclonal antibodies and targeted therapies Many preclinical studies have demonstrated the beneficial immunomodulatory effects of antibodies and targeted drugs, and the potential synergy between immunotherapy and targeted therapy (Marabelle and Caux, 2012). The efficacy of therapeutic monoclonal antibodies may also involve DC (Galluzzi et al., 2012). Indeed, some antibodies such as cetuximab, anti-EGFR, not only attenuate oncogenic signals, but also induce antibody-driven effector mechanism (antibody-dependent cell-mediated cytotoxicity, phagocytosis, complement-dependent cytotoxicity) leading to tumor cell lysis or engulfment, and help elicit specific CD8 T cells by potentiating antigenic presentation by DC and their activation (Banerjee et al., 2008; Correale et al., 2012). Bevacizumab, an antiVEGF antibody, inhibits angiogenesis and also induces DC differentiation and tumor infiltration by CTL (Osada et al., 2008). Inhibition of angiogenic factors such as VEGF restores DC function within the tumor microenvironment and synergizes with DC-based vaccines (Tartour et al., 2011; Gabrilovich et al., 1999). Recent preclinical studies suggest that targeted therapies such as dasatinib or sunitinib (tyrosine kinase inhibitors) used in c-KIT mutated tumors, especially melanoma, gastrointestinal cancer (GIST) and mastocytomas (Yang et al., 2012), may decrease MDSC and immunosuppressive cytokines within the tumor microenvironment (Ko et al., 2009; Ozao-Choy et al., 2009) and potentiate DC vaccines in patients (Amin et al., 2015). JAK2 inhibitors not only interfere with tumor cell survival, but also potentiate DC functions by blocking STAT3 signaling, an inhibitory pathway leading to the induction of suppressive T cells (Lee et al., 2011). Indeed, in a preclinical model, the combination of anti-JAK2 targeted therapies with a DC vaccine enhanced DC maturation, induction of T cell responses, and suppressed tumor growth (Nefedova et al., 2005). Inhibition of mutated BRAF by vemurafenib blocks IL-10 and VEGF secretion by tumor cells, consequently improving DC functionality within the tumor microenvironment (Sumimoto et al., 2006). Thus, targeted therapies can directly antagonize the immunosuppressive microenvironment and restore DC functions. Overall, antitumor treatments or targeted therapies can be beneficial or deleterious for the immune system, and this needs to be taken into account for the design of optimal combinations to potentiate DC function and promote the emergence of a robust immunity. The synergistic effect of chemotherapies or targeted therapies with immunotherapies, may help prevent escapes due to mono-therapies while acting simultaneously at different points of control of tumor progression. Such combination strategies (Vanneman and Dranoff, 2012) may achieve a better clinical success. However, optimization of duration, sequence and delivery dose of the diverse therapies is crucial for the success of these combinatorial approaches and to avoid increased toxicity and side

Cancer Vaccines: Dendritic Cell-Based Vaccines and Related Approaches

271

effects. Numerous ongoing phase I/II clinical trials are testing combinations of targeted therapies or immunomodulatory therapies with DC vaccines.

Conclusion DC are crucial actors of the immune system, orchestrating a large panel of responses that can result in potent antitumor immune responses. However tumor cells hijack the immune system, often compromising DC function. Recent clinical trials have demonstrated that DC vaccines could trigger antitumor responses associated with a clinical benefit in subgroups of patients. DC-based cancer immunotherapy has evolved a lot, starting from initial ex vivo DC vaccines to a panel of new therapeutic options. Better knowledge of DC subsets, their biology and unique properties, together with a better understanding of the type of response triggered by specific DC subsets or engaged target receptors, open the door to the development of rationally optimized vaccines. In addition, characterization of tumor immune escape mechanisms and of DC subversion by tumors will help restoring DC functionality within the tumor microenvironment. Understanding the immunological consequences of conventional antitumor treatments will help potentiate immunotherapies. New approaches in preclinical development are promising, taking advantage of DC diversity and plasticity, of neoantigens identification, and exploitation of the synergies between immunotherapies and conventional therapies. Future developments will likely trend toward optimized personalized combined strategies depending on the properties of the immune system and characteristics of the patient’s tumor, to interfere with the cancer–immunity cycle (Chen and Mellman, 2013). Combination approaches are under development to optimize tumor control, involving both DC vaccines, small molecules inhibiting signal transduction pathways, and antibodies blocking immune checkpoints. The combined expertise of immunologists and oncologists is crucial to develop innovative and efficient antitumor treatments. The current challenge of antitumor immunotherapy is to reverse the balance from tolerance to immunity within a suppressive tumor microenvironment, but also to overcome many regulatory and financial issues associated with the transfer of new strategies in human clinic.

References Aarntzen, E.H., et al., 2013. Targeting CD4(þ) T-helper cells improves the induction of antitumor responses in dendritic cell-based vaccination. Cancer Research 73 (1), 19–29. Alcantara-Hernandez, M., et al., 2017. High-dimensional phenotypic mapping of human dendritic cells reveals interindividual variation and tissue specialization. Immunity 47 (6), 1037–1050 e6. Alexandrov, L.B., et al., 2013. Signatures of mutational processes in human cancer. Nature 500 (7463), 415–421. Amin, A., et al., 2015. Survival with AGS-003, an autologous dendritic cell-based immunotherapy, in combination with sunitinib in unfavorable risk patients with advanced renal cell carcinoma (RCC): Phase 2 study results. Journal for Immunotherapy of Cancer 3, 14. Aspord, C., et al., 2010. A novel cancer vaccine strategy based on HLA-A*0201 matched allogeneic plasmacytoid dendritic cells. PLoS One 5 (5), e10458. Aspord, C., et al., 2011a. GENiusVac, a novel antitumor vaccine strategy based on allogeneic plasmacytoid dendritic cells. La Revue de Médecine Interne 32 (5), 329–332. Aspord, C., et al., 2011b. Induction of antiviral cytotoxic T cells by plasmacytoid dendritic cells for adoptive immunotherapy of posttransplant diseases. American Journal of Transplantation 11 (12), 2613–2626. Aspord, C., et al., 2012. HLA-A(*)0201(þ) plasmacytoid dendritic cells provide a cell-based immunotherapy for melanoma patients. The Journal of Investigative Dermatology 132 (10), 2395–2406. Aspord, C., Leccia, M.T., Charles, J., Plumas, J., 2013. Plasmacytoid dendritic cells support melanoma progression by promoting Th2 and regulatory immunity through OX40L and ICOSL. Cancer Immunology Research 1 (6), 402–415. Banerjee, D., et al., 2008. Enhanced T-cell responses to glioma cells coated with the anti-EGF receptor antibody and targeted to activating FcgammaRs on human dendritic cells. Journal of Immunotherapy 31 (2), 113–120. Blum, J.S., Wearsch, P.A., Cresswell, P., 2013. Pathways of antigen processing. Annual Review of Immunology 31, 443–473. Bogunovic, D., et al., 2011. TLR4 engagement during TLR3-induced proinflammatory signaling in dendritic cells promotes IL-10-mediated suppression of antitumor immunity. Cancer Research 71 (16), 5467–5476. Bol, K.F., et al., 2016. Dendritic cell-based immunotherapy: State of the art and beyond. Clinical Cancer Research 22 (8), 1897–1906. Bonifaz, L., et al., 2002. Efficient targeting of protein antigen to the dendritic cell receptor DEC-205 in the steady state leads to antigen presentation on major histocompatibility complex class I products and peripheral CD8 þ T cell tolerance. The Journal of Experimental Medicine 196 (12), 1627–1638. Burkhardt, U.E., et al., 2013. Autologous CLL cell vaccination early after transplant induces leukemia-specific T cells. The Journal of Clinical Investigation 123 (9), 3756–3765. Caminschi, I., et al., 2008. The dendritic cell subtype-restricted C-type lectin Clec9A is a target for vaccine enhancement. Blood 112 (8), 3264–3273. Carreno, B.M., et al., 2015. Cancer immunotherapy. A dendritic cell vaccine increases the breadth and diversity of melanoma neoantigen-specific T cells. Science 348 (6236), 803–808. Caux, C., et al., 1996. CD34 þ hematopoietic progenitors from human cord blood differentiate along two independent dendritic cell pathways in response to GM-CSF þ TNF alpha. The Journal of Experimental Medicine 184 (2), 695–706. Cella, M., et al., 1999. Plasmacytoid monocytes migrate to inflamed lymph nodes and produce large amounts of type I interferon. Nature Medicine 5 (8), 919–923. Chen, D.S., Mellman, I., 2013. Oncology meets immunology: The cancer-immunity cycle. Immunity 39 (1), 1–10. Chowell, D., et al., 2018. Patient HLA class I genotype influences cancer response to checkpoint blockade immunotherapy. Science 359 (6375), 582–587. Clancy-Thompson, E., et al., 2013. Peptide vaccination in Montanide adjuvant induces and GM-CSF increases CXCR3 and cutaneous lymphocyte antigen expression by tumor antigen-specific CD8 T cells. Cancer Immunology Research 1 (5), 332–339. Clark, R., Kupper, T., 2005. Old meets new: The interaction between innate and adaptive immunity. The Journal of Investigative Dermatology 125 (4), 629–637. Colonna, M., Trinchieri, G., Liu, Y.J., 2004. Plasmacytoid dendritic cells in immunity. Nature Immunology 5 (12), 1219–1226. Correale, P., et al., 2012. Cetuximab þ/ chemotherapy enhances dendritic cell-mediated phagocytosis of colon cancer cells and ignites a highly efficient colon cancer antigenspecific cytotoxic T-cell response in vitro. International Journal of Cancer 130 (7), 1577–1589. Coulie, P.G., et al., 2014. Tumour antigens recognized by T lymphocytes: At the core of cancer immunotherapy. Nature Reviews. Cancer 14 (2), 135–146.

272

Cancer Vaccines: Dendritic Cell-Based Vaccines and Related Approaches

Dubsky, P., et al., 2007. IL-15-induced human DC efficiently prime melanoma-specific naive CD8 þ T cells to differentiate into CTL. European Journal of Immunology 37 (6), 1678–1690. Dummer, R., et al., 2008. An exploratory study of systemic administration of the toll-like receptor-7 agonist 852A in patients with refractory metastatic melanoma. Clinical Cancer Research 14 (3), 856–864. Durand, M., Segura, E., 2015. The known unknowns of the human dendritic cell network. Frontiers in Immunology 6, 129. Dzionek, A., et al., 2000. BDCA-2, BDCA-3, and BDCA-4: Three markers for distinct subsets of dendritic cells in human peripheral blood. Journal of Immunology 165 (11), 6037–6046. Eickhoff, S., et al., 2015. Robust anti-viral immunity requires multiple distinct T cell-dendritic cell interactions. Cell 162 (6), 1322–1337. Eisenhauer, E.A., et al., 2009. New response evaluation criteria in solid tumours: Revised RECIST guideline (version 1.1). European Journal of Cancer 45 (2), 228–247. Evel-Kabler, K., et al., 2006. SOCS1 restricts dendritic cells’ ability to break self tolerance and induce antitumor immunity by regulating IL-12 production and signaling. The Journal of Clinical Investigation 116 (1), 90–100. Fabre, J.W., 2001. The allogeneic response and tumor immunity. Nature Medicine 7 (6), 649–652. Faget, J., et al., 2012. ICOS-ligand expression on plasmacytoid dendritic cells supports breast cancer progression by promoting the accumulation of immunosuppressive CD4þ T cells. Cancer Research 72 (23), 6130–6141. Gabrilovich, D.I., et al., 1999. Antibodies to vascular endothelial growth factor enhance the efficacy of cancer immunotherapy by improving endogenous dendritic cell function. Clinical Cancer Research 5 (10), 2963–2970. Gajewski, T.F., Schreiber, H., Fu, Y.X., 2013. Innate and adaptive immune cells in the tumor microenvironment. Nature Immunology 14 (10), 1014–1022. Galluzzi, L., et al., 2012. The secret ally: Immunostimulation by anticancer drugs. Nature Reviews. Drug Discovery 11 (3), 215–233. Gardner, A., Ruffell, B., 2016. Dendritic cells and cancer immunity. Trends in Immunology 37 (12), 855–865. Geijtenbeek, T.B., Gringhuis, S.I., 2009. Signalling through C-type lectin receptors: Shaping immune responses. Nature Reviews. Immunology 9 (7), 465–479. Goutagny, N., et al., 2012. Targeting pattern recognition receptors in cancer immunotherapy. Targeted Oncology 7 (1), 29–54. Gulley, J.L., et al., 2016. Perspectives on sipuleucel-T: Its role in the prostate cancer treatment paradigm. OncoImmunology 5 (4), e1107698. Hargadon, K.M., 2013. Tumor-altered dendritic cell function: Implications for anti-tumor immunity. Frontiers in Immunology 4, 192. He, L.Z., et al., 2007. Antigenic targeting of the human mannose receptor induces tumor immunity. Journal of Immunology 178 (10), 6259–6267. Hemont, C., et al., 2013. Human blood mDC subsets exhibit distinct TLR repertoire and responsiveness. Journal of Leukocyte Biology 93 (4), 599–609. Heninger, E., Krueger, T.E., Lang, J.M., 2015. Augmenting antitumor immune responses with epigenetic modifying agents. Frontiers in Immunology 6, 29. Hor, J.L., et al., 2015. Spatiotemporally distinct interactions with dendritic cell subsets facilitates CD4 þ and CD8þ T cell activation to localized viral infection. Immunity 43 (3), 554–565. Hu, Z., Ott, P.A., Wu, C.J., 2018. Towards personalized, tumour-specific, therapeutic vaccines for cancer. Nature Reviews. Immunology 18 (3), 168–182. Ito, T., et al., 2004. Plasmacytoid dendritic cells regulate Th cell responses through OX40 ligand and type I IFNs. Journal of Immunology 172 (7), 4253–4259. Joffre, O.P., et al., 2012. Cross-presentation by dendritic cells. Nature Reviews. Immunology 12 (8), 557–569. Kapsenberg, M.L., 2003. Dendritic-cell control of pathogen-driven T-cell polarization. Nature Reviews. Immunology 3 (12), 984–993. Ko, J.S., et al., 2009. Sunitinib mediates reversal of myeloid-derived suppressor cell accumulation in renal cell carcinoma patients. Clinical Cancer Research 15 (6), 2148–2157. Krieg, A.M., 2007. Development of TLR9 agonists for cancer therapy. The Journal of Clinical Investigation 117 (5), 1184–1194. Krishnaswamy, J.K., Chu, T., Eisenbarth, S.C., 2013. Beyond pattern recognition: NOD-like receptors in dendritic cells. Trends in Immunology 34 (5), 224–233. Kroemer, G., Zitvogel, L., 2012. Can the exome and the immunome converge on the design of efficient cancer vaccines? OncoImmunology 1 (5), 579–580. Kroemer, G., et al., 2013. Immunogenic cell death in cancer therapy. Annual Review of Immunology 31, 51–72. Kugler, A., et al., 2000. Regression of human metastatic renal cell carcinoma after vaccination with tumor cell-dendritic cell hybrids. Nature Medicine 6 (3), 332–336. Lande, R., Gilliet, M., 2010. Plasmacytoid dendritic cells: Key players in the initiation and regulation of immune responses. Annals of the New York Academy of Sciences 1183, 89–103. Lapenta, C., et al., 2006. IFN-alpha-conditioned dendritic cells are highly efficient in inducing cross-priming CD8(þ) T cells against exogenous viral antigens. European Journal of Immunology 36 (8), 2046–2060. Le Mercier, I., et al., 2013. Tumor promotion by intratumoral plasmacytoid dendritic cells is reversed by TLR7 ligand treatment. Cancer Research 73 (15), 4629–4640. Leal Rojas, I.M., et al., 2017. Human blood CD1c(þ) dendritic cells promote Th1 and Th17 effector function in memory CD4(þ) T cells. Frontiers in Immunology 8, 971. Lee, H., et al., 2011. STAT3: A target to enhance antitumor immune response. Current Topics in Microbiology and Immunology 344, 41–59. Lepenies, B., Lee, J., Sonkaria, S., 2013. Targeting C-type lectin receptors with multivalent carbohydrate ligands. Advanced Drug Delivery Reviews 65 (9), 1271–1281. Lesterhuis, W.J., et al., 2010. A pilot study on the immunogenicity of dendritic cell vaccination during adjuvant oxaliplatin/capecitabine chemotherapy in colon cancer patients. British Journal of Cancer 103 (9), 1415–1421. Linnemann, C., et al., 2013. High-throughput identification of antigen-specific TCRs by TCR gene capture. Nature Medicine 19 (11), 1534–1541. Lipson, E.J., et al., 2015. Safety and immunologic correlates of melanoma GVAX, a GM-CSF secreting allogeneic melanoma cell vaccine administered in the adjuvant setting. Journal of Translational Medicine 13, 214. Liu, Y.J., 2005. IPC: Professional type 1 interferon-producing cells and plasmacytoid dendritic cell precursors. Annual Review of Immunology 23, 275–306. Lui, G., et al., 2009. Plasmacytoid dendritic cells capture and cross-present viral antigens from influenza-virus exposed cells. PLoS One 4 (9), e7111. Luksza, M., et al., 2017. A neoantigen fitness model predicts tumour response to checkpoint blockade immunotherapy. Nature 551 (7681), 517–520. Marabelle, A., Caux, C., 2012. Immunobiology. Combined targeted and immunotherapy: The future of personalized medicine. Blood 120 (23), 4454–4455. MartIn-Fontecha, A., et al., 2003. Regulation of dendritic cell migration to the draining lymph node: Impact on T lymphocyte traffic and priming. The Journal of Experimental Medicine 198 (4), 615–621. McGranahan, N., et al., 2017. Allele-specific HLA loss and immune escape in lung cancer evolution. Cell 171 (6), 1259–1271 e11. Molenkamp, B.G., et al., 2008. Local administration of PF-3512676 CpG-B instigates tumor-specific CD8 þ T-cell reactivity in melanoma patients. Clinical Cancer Research 14 (14), 4532–4542. Muir, G., et al., 2006. Preliminary evidence that the allogeneic response might trigger antitumour immunity in patients with advanced prostate cancer. BJU International 98 (5), 989–995. Nefedova, Y., et al., 2005. Regulation of dendritic cell differentiation and antitumor immune response in cancer by pharmacologic-selective inhibition of the janus-activated kinase 2/signal transducers and activators of transcription 3 pathway. Cancer Research 65 (20), 9525–9535. Nestle, F.O., et al., 2009. Skin immune sentinels in health and disease. Nature Reviews. Immunology 9 (10), 679–691. Osada, T., et al., 2008. The effect of anti-VEGF therapy on immature myeloid cell and dendritic cells in cancer patients. Cancer Immunology, Immunotherapy 57 (8), 1115–1124. Ott, P.A., Hodi, F.S., Robert, C., 2013. CTLA-4 and PD-1/PD-L1 blockade: New immunotherapeutic modalities with durable clinical benefit in melanoma patients. Clinical Cancer Research 19 (19), 5300–5309. Ott, P.A., et al., 2017. An immunogenic personal neoantigen vaccine for patients with melanoma. Nature 547 (7662), 217–221. Ozao-Choy, J., et al., 2009. The novel role of tyrosine kinase inhibitor in the reversal of immune suppression and modulation of tumor microenvironment for immune-based cancer therapies. Cancer Research 69 (6), 2514–2522. Palucka, K., Banchereau, J., 2012. Cancer immunotherapy via dendritic cells. Nature Reviews. Cancer 12 (4), 265–277. Palucka, K., Banchereau, J., 2013a. Dendritic-cell-based therapeutic cancer vaccines. Immunity 39 (1), 38–48.

Cancer Vaccines: Dendritic Cell-Based Vaccines and Related Approaches

273

Palucka, K., Banchereau, J., 2013b. Human dendritic cell subsets in vaccination. Current Opinion in Immunology 25 (3), 396–402. Palucka, A.K., et al., 2008. Dendritic cells: A critical player in cancer therapy? Journal of Immunotherapy 31 (9), 793–805. Pashenkov, M., et al., 2006. Phase II trial of a toll-like receptor 9-activating oligonucleotide in patients with metastatic melanoma. Journal of Clinical Oncology 24 (36), 5716–5724. Patel, P.M., et al., 2018. Targeting gp100 and TRP-2 with a DNA vaccine: Incorporating T cell epitopes with a human IgG1 antibody induces potent T cell responses that are associated with favourable clinical outcome in a phase I/II trial. OncoImmunology 7, e1433516. Prendergast, G.C., et al., 2014. Indoleamine 2,3-dioxygenase pathways of pathogenic inflammation and immune escape in cancer. Cancer Immunology, Immunotherapy 63 (7), 721–735. Pulendran, B., Li, S., Nakaya, H.I., 2010. Systems vaccinology. Immunity 33 (4), 516–529. Reis e Sousa, C., 2006. Dendritic cells in a mature age. Nature Reviews. Immunology 6 (6), 476–483. Reizis, B., et al., 2011. Plasmacytoid dendritic cells: Recent progress and open questions. Annual Review of Immunology 29, 163–183. Ribas, A., et al., 2009. Dendritic cell vaccination combined with CTLA4 blockade in patients with metastatic melanoma. Clinical Cancer Research 15 (19), 6267–6276. Rizvi, N.A., et al., 2015. Cancer immunology. Mutational landscape determines sensitivity to PD-1 blockade in non-small cell lung cancer. Science 348 (6230), 124–128. Sagiv-Barfi, I., et al., 2018. Eradication of spontaneous malignancy by local immunotherapy. Science Translational Medicine 10 (426), eaan4488. Sahin, U., et al., 2017. Personalized RNA mutanome vaccines mobilize poly-specific therapeutic immunity against cancer. Nature 547 (7662), 222–226. Sancho, D., Reis e Sousa, C., 2012. Signaling by myeloid C-type lectin receptors in immunity and homeostasis. Annual Review of Immunology 30, 491–529. Sancho, D., et al., 2008. Tumor therapy in mice via antigen targeting to a novel, DC-restricted C-type lectin. The Journal of Clinical Investigation 118 (6), 2098–2110. Sandoval, F., et al., 2013. Mucosal imprinting of vaccine-induced CD8(þ) T cells is crucial to inhibit the growth of mucosal tumors. Science Translational Medicine 5 (172), 172ra20. Santegoets, S.J., et al., 2008. Human dendritic cell line models for DC differentiation and clinical DC vaccination studies. Journal of Leukocyte Biology 84 (6), 1364–1373. Saxena, M., Bhardwaj, N., 2018. Re-emergence of dendritic cell vaccines for Cancer treatment. Trends Cancer 4 (2), 119–137. Schreibelt, G., et al., 2016. Effective clinical responses in metastatic melanoma patients after vaccination with primary myeloid dendritic cells. Clinical Cancer Research 22 (9), 2155–2166. Segura, E., et al., 2013. Human inflammatory dendritic cells induce Th17 cell differentiation. Immunity 38 (2), 336–348. Shen, L., et al., 2004. Silencing of SOCS1 enhances antigen presentation by dendritic cells and antigen-specific anti-tumor immunity. Nature Biotechnology 22 (12), 1546–1553. Sheridan, B.S., Lefrancois, L., 2011. Regional and mucosal memory T cells. Nature Immunology 12 (6), 485–491. Sisirak, V., et al., 2012. Impaired IFN-alpha production by plasmacytoid dendritic cells favors regulatory T-cell expansion that may contribute to breast cancer progression. Cancer Research 72 (20), 5188–5197. Speiser, D.E., et al., 2005. Rapid and strong human CD8 þ T cell responses to vaccination with peptide, IFA, and CpG oligodeoxynucleotide 7909. The Journal of Clinical Investigation 115 (3), 739–746. Steinman, R.M., Banchereau, J., 2007. Taking dendritic cells into medicine. Nature 449 (7161), 419–426. Sumimoto, H., et al., 2006. The BRAF-MAPK signaling pathway is essential for cancer-immune evasion in human melanoma cells. The Journal of Experimental Medicine 203 (7), 1651–1656. Swiecki, M., Colonna, M., 2015. The multifaceted biology of plasmacytoid dendritic cells. Nature Reviews. Immunology 15 (8), 471–485. Tacken, P.J., Figdor, C.G., 2011. Targeted antigen delivery and activation of dendritic cells in vivo: Steps towards cost effective vaccines. Seminars in Immunology 23 (1), 12–20. Tacken, P.J., et al., 2005. Effective induction of naive and recall T-cell responses by targeting antigen to human dendritic cells via a humanized anti-DC-SIGN antibody. Blood 106 (4), 1278–1285. Tacken, P.J., et al., 2011. Targeted delivery of TLR ligands to human and mouse dendritic cells strongly enhances adjuvanticity. Blood 118 (26), 6836–6844. Tartour, E., et al., 2011. Angiogenesis and immunity: A bidirectional link potentially relevant for the monitoring of antiangiogenic therapy and the development of novel therapeutic combination with immunotherapy. Cancer Metastasis Reviews 30 (1), 83–95. Tel, J., et al., 2012. Harnessing human plasmacytoid dendritic cells as professional APCs. Cancer Immunology, Immunotherapy 61 (8), 1279–1288. Tel, J., et al., 2013. Natural human plasmacytoid dendritic cells induce antigen-specific T-cell responses in melanoma patients. Cancer Research 73 (3), 1063–1075. Tiptiri-Kourpeti, A., et al., 2016. DNA vaccines to attack cancer: Strategies for improving immunogenicity and efficacy. Pharmacology & Therapeutics 165, 32–49. Ueno, H., et al., 2011. Targeting human dendritic cell subsets for improved vaccines. Seminars in Immunology 23 (1), 21–27. Vacchelli, E., et al., 2012. Trial watch: FDA-approved toll-like receptor agonists for cancer therapy. OncoImmunology 1 (6), 894–907. Vacchelli, E., et al., 2013. Trial watch: Dendritic cell-based interventions for cancer therapy. OncoImmunology 2 (10), e25771. Valmori, D., et al., 2007. Vaccination with NY-ESO-1 protein and CpG in Montanide induces integrated antibody/Th1 responses and CD8 T cells through cross-priming. Proceedings of the National Academy of Sciences of the United States of America 104 (21), 8947–8952. Van Allen, E.M., et al., 2015. Genomic correlates of response to CTLA-4 blockade in metastatic melanoma. Science 350 (6257), 207–211. van Broekhoven, C.L., et al., 2004. Targeting dendritic cells with antigen-containing liposomes: A highly effective procedure for induction of antitumor immunity and for tumor immunotherapy. Cancer Research 64 (12), 4357–4365. van Kooyk, Y., Geijtenbeek, T.B., 2003. DC-SIGN: Escape mechanism for pathogens. Nature Reviews. Immunology 3 (9), 697–709. Van Lint, S., et al., 2014. Optimized dendritic cell-based immunotherapy for melanoma: The TriMix-formula. Cancer Immunology, Immunotherapy 63 (9), 959–967. Van Nuffel, A.M., et al., 2012. Intravenous and intradermal TriMix-dendritic cell therapy results in a broad T-cell response and durable tumor response in a chemorefractory stage IVM1c melanoma patient. Cancer Immunology, Immunotherapy 61 (7), 1033–1043. Vanneman, M., Dranoff, G., 2012. Combining immunotherapy and targeted therapies in cancer treatment. Nature Reviews. Cancer 12 (4), 237–251. Vasaturo, A., et al., 2013. Clinical implications of co-inhibitory molecule expression in the tumor microenvironment for DC vaccination: A game of stop and go. Frontiers in Immunology 4, 417. Villadangos, J.A., Young, L., 2008. Antigen-presentation properties of plasmacytoid dendritic cells. Immunity 29 (3), 352–361. Villani, A.C., et al., 2017. Single-cell RNA-seq reveals new types of human blood dendritic cells, monocytes, and progenitors. Science 356 (6335), eaah4573. Walter, S., et al., 2012. Multipeptide immune response to cancer vaccine IMA901 after single-dose cyclophosphamide associates with longer patient survival. Nature Medicine 18 (8), 1254–1261. Watchmaker, P.B., et al., 2014. Comparative transcriptional and functional profiling defines conserved programs of intestinal DC differentiation in humans and mice. Nature Immunology 15 (1), 98–108. Wesolowski, R., Markowitz, J., Carson 3rd, W.E., 2013. Myeloid derived suppressor cellsdA new therapeutic target in the treatment of cancer. Journal for Immunotherapy of Cancer 1, 10. Wolchok, J.D., et al., 2009. Guidelines for the evaluation of immune therapy activity in solid tumors: Immune-related response criteria. Clinical Cancer Research 15 (23), 7412–7420. Yang, Y., et al., 2012. Antitumor T-cell responses contribute to the effects of dasatinib on c-KIT mutant murine mastocytoma and are potentiated by anti-OX40. Blood 120 (23), 4533–4543. Yewdall, A.W., et al., 2010. CD8þ T cell priming by dendritic cell vaccines requires antigen transfer to endogenous antigen presenting cells. PLoS One 5 (6), e11144. Yu, C.I., et al., 2013. Human CD1cþ dendritic cells drive the differentiation of CD103 þ CD8þ mucosal effector T cells via the cytokine TGF-beta. Immunity 38 (4), 818–830. Yu, C.I., et al., 2014. Human CD141 þ dendritic cells induce CD4þ T cells to produce type 2 cytokines. Journal of Immunology 193 (9), 4335–4343.

274

Cancer Vaccines: Dendritic Cell-Based Vaccines and Related Approaches

Zitvogel, L., Palucka, K., 2011. Antibody co-targeting of DCs. Blood 118 (26), 6726–6727. Zitvogel, L., et al., 2008. Immunological aspects of cancer chemotherapy. Nature Reviews. Immunology 8 (1), 59–73. Zitvogel, L., Kepp, O., Kroemer, G., 2011. Immune parameters affecting the efficacy of chemotherapeutic regimens. Nature Reviews. Clinical Oncology 8 (3), 151–160.

Further Reading Alcantara-Hernandez, M., et al., 2017. High-dimensional phenotypic mapping of human dendritic cells reveals interindividual variation and tissue specialization. Immunity 47 (6), 1037–1050 e6. Bol, K.F., et al., 2016. Dendritic cell-based immunotherapy: State of the art and beyond. Clinical Cancer Research 22 (8), 1897–1906. Chen, D.S., Mellman, I., 2013. Oncology meets immunology: The cancer–immunity cycle. Immunity 39 (1), 1–10. Hu, Z., Ott, P.A., Wu, C.J., 2018. Towards personalized, tumor-specific, therapeutic vaccines for cancer. Nature Reviews. Immunology 18 (3), 168–182. Saxena, M., Bhardwaj, N., 2018. Re-emergence of dendritic cell vaccines for cancer treatment. Trends Cancer 4 (2), 119–137. Tiptiri-Kourpeti, A., et al., 2016. DNA vaccines to attack cancer: Strategies for improving immunogenicity and efficacy. Pharmacology & Therapeutics 165, 32–49. Veglia, F., Gabrilovich, D.I., 2017. Dendritic cells in cancer: The role revisited. Current Opinion in Immunology 45, 43–51.

Relevant Websites https://cancerresearch.org/scientists/meetings-and-resources/peptide-databasedCancer Research Institute. Peptide Database. https://clinicaltrials.gov/dUS National Library of Medicine. Clinical Trials.

Cancer-Related Inflammation in Tumor Progression Alberto Mantovani, Humanitas Clinical and Research Center, Milan, Italy; Humanitas University, Milan, Italy; and Queen Mary University of London, London, United Kingdom © 2019 Elsevier Inc. All rights reserved.

Glossary Tumor microenvironment The ecological niche of cancer including inflammatory cells, immune cells and stroma.

Nomenclature TME Tumor microenvironment TAM Tumor-associated macrophages IL Interleukin

The perception that cancer and inflammation are linked processes goes back to the 19th century. This perception has not attracted the limelight for a long time. Different lines of work have led to a renaissance of the inflammation-cancer connection, leading to a generally accepted paradigm. Inflammatory cells and mediators are present in the microenvironment of most neoplastic tissues, including those not causally related to an obvious inflammatory process (e.g., breast cancer). Essential characteristics of cancerrelated inflammation include the presence of leukocytes, prominently tumor-associated macrophages (TAM); the presence of inflammatory cytokines (e.g., tumor necrosis factor (TNF), interleukin-1 (IL-1), IL-6, IL-23, IL-17, chemokines such as CCL2); prominent tissue remodeling and angiogenesis. In the last 20 years we have moved from a cancer cell centric view of the essence of cancer to one that includes the tumor microenvironment (TME) and immunity. A better understanding of innate and adaptive immunity in cancer has paved the way to development of immunotherapy approaches in its broad sense (Fig. 1), from checkpoint blockade inhibitors to inhibition of inflammatory cells and cytokines.

Fig. 1

A historical perspective of the development of tumor immunology and immunotherapy.

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65801-4

275

276

Cancer-Related Inflammation in Tumor Progression

The TME includes components of inflammation and adaptive immune responses. Pathways connecting inflammation, carcinogenesis and progression are complex (Fig. 2). Under selected conditions, smoldering local od systemic inflammation increases the cancer risk. We have referred to this pathway as “extrinsic.” The extrinsic pathway connection between local inflammation and cancer is illustrated by ulcerative colitis increasing the risk of developing colorectal cancer. At a systemic level, obesity causes a subclinical inflammation state and is associated to an increased risk to develop cancer, in this same extrinsic pathway perspective. Tumors epidemiologically unrelated to overt inflammatory conditions have inflammatory cells and mediators in their TME. For instance macrophages are a component of the TME in breast cancer and promote growth and metastasis. Moreover, in triple negative breast cancer tumor-associated macrophage (TAM) infiltration is associated with poor prognosis. Activation of the kinase MER TK which drives epithelial to mesenchymal transition is an important determinant of TAM recruitment under these conditions. Key orchestrators at the intersection of the intrinsic and extrinsic pathway include transcription factors (e.g., NFkB; STAT3), cytokines (e.g., IL-1; TNF) and chemokines. Thus, cancer-related inflammation (CRI) is an essential component of the tumor microenvironment and is now a recognized essential characteristic of cancer. TAM are now being considered as prognostic markers and therapeutic targets. Targeting TAM is a key component in the antitumor activity of Trabectedin, an EMA and FDA approved antitumor agent, thus providing proof of principle for TAM directed strategies. Based on preclinical work. Based on preckinical work, antibodies or simple molecules targeting the CSF1 pathway are undergoing clinical assessment per se or in concert with checkpoint blockade inhibitors. A variety of inflammatory cells and mediators are present in TME (Fig. 2). Macrophages have served as a paradigm of cancerrelated inflammation. Other inflammatory cells include neutrophils, eosinophils and basophils. Components of the humoral arm of innate immunity are present in the TME and have recently emerged as important drivers of tumor progression. Evidence now suggests that complement can be a component of tumor promoting inflammation. In the same perspective of complement mediated tumor promotion, the humoral pattern recognition gene PTX3 has been shown to be an extrinsic oncosuppressor gene. PTX3 interacts with factor H which regulates complement-driven inflammation in preclinical models and selected human tumors. Complement drives tumor promoting inflammation by recruiting macrophages which are the key cells in tumor promotion. Thus complement can have a yin yang role in cancer, on the one hand mediating tumor cell killing of recognition by effector cells after monoclonal antibody therapy. On the other hand, complement can be a driver of tumor promoting inflammation. The TME is remarkably different in tumors originating in different organs and among tumors which originate in the same organ or tissue. Within the same histological type, tumors are characterized by widely different TME. For instance, in colorectal cancer four TME phenotypes have been identified based on profiling. Assessment of components of the TME has prognostic significance. T cell infiltration (“Immunoscore”) is an independent prognostic indicator of favorable outcome. Conversely, TAM infiltration is generally associated with bad prognosis. A notable exception is colorectal cancer. Recent evidence indicate that TAM infiltration predicts response to 5FU chemotherapy. Dissection of the role and diversity of cancer-related inflammation may pave the way to improving on current immunotherapy strategies and may impact on diagnosis and prevention. The prototypic inflammatory cytokine IL-1 was shown early on to promote tumor progression. These early observations have now been translated. In the CANTOS study with over 10,000 patients treated with

Fig. 2 A schematic representation of the connection between inflammation and cancer. Reproduced with permission from Mantovani A. (2018). FEBS J. 285(4):638–640.

Cancer-Related Inflammation in Tumor Progression

277

anti-IL-1 (canakinumab) with high C reactive Protein. Over 50% protection against incidence and mortality from lung cancer were observed. Immunity is an essential component of the tumor microenvironment (TME) and a key determinant of metastasis. Inflammatory cells, tumor-associated macrophages (TAM) in particular, pave the way to tissue invasion and intravasation and provide a nurturing microenvironment for metastasis, serving as a component of the cancer cell niche at distant sites. NK cells are innate lymphoid cells (ILCs) which have long been considered to play a role in resistance against hematogenous dissemination of cancer cells, in particular to the lungs. Finally, tumor progression and escape are associated with activation of pathways of suppression of innate and adaptive anti-tumor responses which include, among others, immunosuppressive myeloid cells, activation of checkpoint blockade, induction and recruitment of T regulatory cells. Invasion and metastasis are key to malignancy. There is also evidence for a role for neutrophils and their major growth factor, granulocyte colony stimulating factor (G-CSF), in cancer progression to metastasis. The interplay between tumor-associated macrophages and neutrophils with the extracellular matrix shapes the TME and promotes tumor growth and dissemination. Quantification of the immune and inflammatory landscape of TME has provided novel prognostic indicators of cancer progression as shown by quantification of tumor infiltrating T cells and TAM. Genomic technologies have added a new dimension to the characterization of the TME and to classification of cancers. Finally, dissection of the mode of action of conventional cytoreductive strategies, the impact of checkpoint blockade inhibitors and adoptive cell therapy of hematological malignancies have proven the principle that the immune system can be harnessed to cope with advanced disseminated neoplastic diseases. Full exploitation of the diagnostic and therapeutic potential of innate and adaptive immunity will require an integrated in depth analysis of its components in primary versus spreading, metastatic tumors, dissection of the diversity of metastatic niches, identification and development of new molecular and cellular tools. Given the diversity of the involved innate and adaptive players and of the environmental clues involved, only an integrated effort with expertise from different fields can lead to the identification of new biomarkers and therapeutic strategies for metastatic disease.

See also: Helicobacter pylori-Mediated Carcinogenesis.

Further Reading Balkwill, F., Mantovani, A., 2001. Inflammation and cancer: Back to Virchow? Lancet 357, 539–545. Bonavita, E., Gentile, S., Rubino, M., Maina, V., Papait, R., Kunderfranco, P., Greco, C., Feruglio, F., Molgora, M., Laface, I., Tartari, S., Doni, A., Pasqualini, F., Barbati, E., Basso, G., Galdiero, M.R., Nebuloni, M., Roncalli, M., Colombo, P., Laghi, L., Lambris, J.D., Jaillon, S., Garlanda, C., Mantovani, A., 2015. PTX3 is an extrinsic oncosuppressor regulating complement-dependent inflammation in cancer. Cell 160, 700–714. Bonelli, S., Geeraerts, X., Bolli, E., Keirsse, J., Kiss, M., Pombo Antunes, A.R., Van Damme, H., De Vlaminck, K., Movahedi, K., Laoui, D., Raes, G., Van Ginderachter, J.A., 2018. Beyond the M-CSF receptordNovel therapeutic targets in tumor-associated macrophages. The FEBS Journal 285 (4), 777–787. Bottai, G., Raschioni, C., Székely, B., Di Tommaso, L., Szász, A.M., Losurdo, A., GyTrffy, B., Ács, B., Torrisi, R., Karachaliou, N., TTkés, T., Caruso, M., Kulka, J., Roncalli, M., Santoro, A., Mantovani, A., Rosell, R., Reis-Filho, J.S., Santarpia, L., 2016. AXL-associated tumor inflammation as a poor prognostic signature in chemotherapy-treated triplenegative breast cancer patients. NPJ Breast Cancer 2, 16033. Coussens, L.M., Zitvogel, L., Palucka, A.K., 2013. Neutralizing tumor-promoting chronic inflammation: A magic bullet? Science 339, 286–291. Fridman, W.H., Zitvogel, L., Sautes-Fridman, C., Kroemer, G., 2017. The immune contexture in cancer prognosis and treatment. Nature Reviews. Clinical Oncology 14, 717–734. Gonzalez, H., Robles, I., Werb, Z., 2018. Innate and acquired immune surveillance in the postdissemination phase of metastasis. The FEBS Journal 285 (4), 654–664. Hanahan, D., Weinberg, R.A., 2011. Hallmarks of cancer: The next generation. Cell 144, 646–674. Jeannin, P., Paolini, L., Adam, C., Delneste, Y., 2018. The roles of CSFs on the functional polarization of tumor-associated macrophages. The FEBS Journal 285 (4), 680–699. Ma, C., Zhang, Q., Greten, T.F., 2018. Nonalcoholic fatty liver disease promotes hepatocellular carcinoma through direct and indirect effects on hepatocytes. The FEBS Journal 285 (4), 752–762. Majety, M., Runza, V., Lehmann, C., Hoves, S., Ries, C.H., 2018. A drug development perspective on targeting tumor-associated myeloid cells. The FEBS Journal 285 (4), 763– 776. https://doi.org/10.1111/febs.14277. Mantovani, A., Allavena, P., Sica, A., Balkwill, F., 2008. Cancer-related inflammation. Nature 454, 436–444. Mantovani, A., Marchesi, F., Malesci, A., Laghi, L., Allavena, P., 2017. Tumor-associated macrophages as treatment targets in oncology. Nature Reviews. Clinical Oncology 14, 399–416. Mazzone, M., Menga, A., Castegna, A., 2018. Metabolism and TAM functionsdIt takes two to tango. The FEBS Journal 285 (4), 700–716. Mouchemore, K.A., Anderson, R.L., Hamilton, J.A., 2018. Neutrophils, G-CSF and their contribution to breast cancer metastasis. The FEBS Journal 285 (4), 665–679. Murray, P.J., 2018. Nonresolving macrophage-mediated inflammation in malignancy. The FEBS Journal 285 (4), 641–653. Pollard, J.W., 2008. Macrophages define the invasive microenvironment in breast cancer. Journal of Leukocyte Biology 84, 623–630. Porta, C., Sica, A., Riboldi, E., 2018. Tumor-associated myeloid cells: New understandings on their metabolic regulation and their influence in cancer immunotherapy. The FEBS Journal 285 (4), 717–733. Reis, E.S., Mastellos, D.C., Ricklin, D., Mantovani, A., Lambris, J.D., 2018. Complement in cancer: Untangling an intricate relationship. Nature Reviews. Immunology 18 (1), 5–18. Ridker, P. M., MacFadyen, J. G., Thuren, T., Everett, B. M., Libby, P. & Glynn, R. J. (2017) Effect of interleukin-1beta inhibition with canakinumab on incident lung cancer in patients with atherosclerosis: Exploratory results from a randomized, double-blind, placebo-controlled trial, Lancet, 390(10105):1833–1842. Varol, C., Sagi, I., 2018. PhagocytedExtracellular matrix crosstalk empowers tumor development and dissemination. The FEBS Journal 285 (4), 734–751.

Cancers as Ecosystems: From Cells to Population Graham A Colditz, Alvin J. Siteman Cancer Center and Institute for Public Health, Washington University School of Medicine, St. Louis, MO, United States © 2019 Elsevier Inc. All rights reserved.

Much of our research focuses on the causes and outcomes of cancer. Epidemiology studies the genetic and nongenetic risk factors (genes and environmental factors (MacMahon and Pugh, 1970)) and places these in the context of the assumed underlying disease process. Historically epidemiology has defined non-genetic factors as environmental, thought this embraces a broad range of lifestyle, occupational, and reproductive factors as well as medications. Using a reductionist approach, we work to separate the effects of each factor from the others, focusing on measurement and the role of exposures in etiology and modifying survivorship. However, humans live in communities and broader societies, and often the interplay between cells, human bodies, individuals, and society are ignored (Colditz and Sutcliffe, 2016). This is particularly important for considerations of the underlying causes of variation in cancer risk, and the role of lifestyle in survivorship (Colditz, 2001). Early studies of causes of cancer focused heavily on variation in incidence between countries and within countries according to occupation (Ziegler et al., 1993; MacMahon, 1965). Let me use the example of breast cancer. We have identified a range of reproductive and other lifestyle factors that modify risk (Hankinson et al., 2004). Each may be studied as an independent risk factor or combined into a model that summarizes risk making assumptions of the underlying disease process (Rosner et al., 1994, 2013; Rosner and Colditz, 1996), which Pike called breast tissue age (Pike et al., 1983). The individual risk factors such as first pregnancy may be studied in humans in relation to changes in circulating hormone levels (Hankinson et al., 1995), and intermediate markers such as breast mammographic density (McCormack et al., 2010). Pregnancy may also be studied in animal models to refine understanding of the cellular changes associated with a first full-term pregnancy (Medina, 2013). Other lifestyle causes of breast cancer include alcohol (IARC Working Group on the Evaluation of Carcinogenic Risks to Humans, 2010), obesity (Lauby-Secretan et al., 2016), physical activity (International Agency for Research on Cancer, 2002), and use of contraceptives and menopausal hormone therapy (IARC Working Group on the Evaluation of Carcinogenic Risks to Humans, 2007). These have been studied for action at the cellular level and then with more recent and larger epidemiologic studies they are related to risk of subtypes of breast cancer defined by receptor status (Colditz et al., 2004). For weight, the field has evolved to address timing of adiposity and weight gain (Rosner et al., 2015, 2017). In parallel to these studies of environmental causes of cancer, the details of genetic predisposition have been refined from a family history of breast cancer diagnosis (1800s) to genetic changes that carry varying levels of risk (Colditz et al., 1996; Miki et al., 1994; Couch et al., 2014). Molecular mechanism for genetic predisposition are described in other articles. After diagnosis of breast cancer, lifestyle is again a focus, as components of lifestyle may modify survival outcomes (adiposity increases mortality (Chan et al., 2014), physical activity may reduce mortality (Holmes et al., 2005)). Furthermore, socially isolated women have a greater reduction in their physical function after cancer diagnosis than those who are more socially integrated (Michael et al., 2002). Similar associations are studied for outcomes after colorectal cancer (Van Blarigan et al., 2018; Chan et al., 2008) and prostate cancer (Khan et al., 2018; Imm et al., 2017), to name a few. These risk factors do not occur in isolation. Rather they are the proximate measures that are easily accessible for epidemiologists to study. Many of these factors studied in isolation have been reviewed by the International Agency for Research on Cancer (IARC) and classified as carcinogens (IARC, 2011). However, many of these exposures and the attained level of risk is often driven by broader social forces. Take parity as an example. Before commercial contraception was readily available women had an average of 6–8 pregnancies. This level continued in rural China and other low-income countries until very recently (Lewington et al., 2014). Meanwhile, with industrialization and access to contraception parity has fallen. In addition, with increasing education and career options, women have deferred childbearing to later in life. Mean age at first pregnancy is now 30 or more in most OECD countries (OECD, 2012). Thus, the risk factor age at first birth is not isolated from the broader social changes of education, income, and opportunity (Bumpass et al., 1978). Changes in sun exposure over decades reflect changing social norms regarding sunbathing, swim suits, development of sunscreens, and now commercial tanning industries (Wehner et al., 2014). Likewise, consumption of sugar sweetened beverages has changed dramatically over decades with increasing serving size reflecting marketing and commercial targets over public health concerns. Age at menarche is another long-studied risk factor, marking the beginning of menstrual cycling and monthly breast proliferation and involution (Pike et al., 1993). The drivers of age at menarche have changed dramaticallydage at menarche has fallen from 18 to 12 or less in a century (Lewington et al., 2014), and with it more adiposity in childhood (Dietz and Gortmaker, 2001), greater attained height (Sung et al., 2009), and fewer infections have covaried. In Korea, for example, mean age at menarche decreased from 16.90  1.25 years for women born between 1920 and 1925 to 13.79  1.37 years for those born between 1980 and 1985, indicating a downward trend of 0.68 years per decade (95% CI, 0.64–0.71) in age at menarche (Cho et al., 2010). Some have worked to understand the relations between these societal changes and the change in age at menarche (Berkey et al., 2000). Nutrition is one clear driver of both adiposity (excess energy intake for energy expenditure) and sources (animal vs. vegetable) modifying age at menarche (Berkey et al., 2000). Higher animal protein intake is related to greater height and also earlier menarche. So, while the cellular changes in the breast may be the driver of breast cancer risk; societal changes in access to food, healthy living conditions, and so forth modify the expression of the underlying biologic process.

278

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65148-6

Cancers as Ecosystems: From Cells to Population

279

To refine understanding of pathways and modifiers of predisposition, studies have more recently included blood markers (hormone levels, nutrients, inflammatory markers) and related them to cancer risk (Colditz and Hankinson, 2005). Often this is in the context of understanding or refining understanding of exposures and cancers (McMichael, 1994). While these markers may serve as more proximate measures of exposure, to date they do not address the cellular environment in all levels of richness and are not specific to individual organ tissues.

Moving From Causes to Prevention The causal role of HPV in the etiology of cervical cancer demonstrates the refinement of understanding of causes of cancer over time, and the broader social context that can modify the effectiveness of prevention even when effective strategies are scientifically justified. The evolution of epidemiologic understanding from number of sexual partners as a risk factor for cervical cancer (hypothesized to reflect transmission of on infectious agent) (Pereyra, 1961; Schiffman et al., 1993; Beral, 1974) to defining the etiologic agent (Franco et al., 1999), and then developing an effective vaccine for cervical and anal cancer (Giuliano et al., 2011), has variable effectiveness as a prevention strategy at the population level (Walling et al., 2016). This is not due to varying biologic effectiveness at the cellular level, but rather due to differences in society approaches to implementing comprehensive vaccination programs (Walling et al., 2016). In the span of 20 years or less global evidence moved such that Australia has among the most effective programs (vaccine purchased by the Federal government and administered by school nurses) (Gertig et al., 2013) while the US on the other hand has hap hazard prescribing by primary care providers (USA) (Rahman et al., 2015). Australia has documented the dramatic decrease in premalignant cervical lesions on biopsy, and modified its surveillance program to every 5 year vaginal testing for HPV (Brotherton et al., 2016). The US still has a vaccination rate of only 43% of adolescents up to date on all recommended doses of HPV (https://www.cdc.gov/hpv/hcp/vacc-coverage.html).

Synthesis From these perspectives and others, the social ecologic model for prevention has been proposed as an approach to place the cellular level biologic processes of carcinogenesis in the broader context of society (Swinburn et al., 2011). For obesity, Story described how access to food and nutrients, as well as safe places to play and exercise modify energy balance; the built environment also supports or inhibits physical activity, active commuting, and so forth (Story et al., 2008; Shelton et al., 2011). A similar model applies globally (Swinburn et al., 2011). Food and agriculture policies modify access to foods/nutrients and the composition of diet. Marketing, home and workplace environments, also contribute to the enticement for food and drinks, such as sugar sweetened beverages, alcohol, etc.; and modify how we spend our timedwatching televisiondor in occupations that do not burn energy the way our forebears did in manual labor. In this spirit Warnecke and others have defined how the interplay of processes from the cellular level through individual and community level measure and broader social policies interact to drive cancer risk and outcomes (Warnecke et al., 2008). Moving forward to reduce disparities in cancer risk across populations and within populations greater focus from cell to society will be needed to maximize our return on investment in cancer research (Emmons and Colditz, 2017).

Conflict of Interest Authors declare no potential conflict of interest.

See also: Cancer Risk Reduction Through Lifestyle Changes. Prevention and Control: Nutrition, Obesity, and Metabolism.

References Beral, V., 1974. Cancer of the cervix: A sexually transmitted infection? Lancet 1 (7865), 1037–1040. Berkey, C.S., et al., 2000. Relation of childhood diet and body size to menarche and adolescent growth in girls. American Journal of Epidemiology 152 (5), 446–452. Brotherton, J.M., et al., 2016. HPV vaccine impact in Australian women: Ready for an HPV-based screening program. The Medical Journal of Australia 204 (5), 184. Bumpass, L., Rindfuss, R., Janosik, R., 1978. Age and marital status at first birth and the pace of subsequent fertility. Demography 15, 75–86. Chan, J.A., et al., 2008. Association of family history with cancer recurrence and survival among patients with stage III colon cancer. JAMA 299 (21), 2515–2523. Chan, D.S., et al., 2014. Body mass index and survival in women with breast cancer-systematic literature review and meta-analysis of 82 follow-up studies. Annals of Oncology 25 (10), 1901–1914. Cho, G.J., et al., 2010. Age at menarche in a korean population: Secular trends and influencing factors. European Journal of Pediatrics 169 (1), 89–94. Colditz, G.A., 2001. Cancer culture: Epidemics, human behavior, and the dubious search for new risk factors. American Journal of Public Health 91 (3), 357–359. Colditz, G.A., Hankinson, S.E., 2005. The Nurses’ health study: Lifestyle and health among women. Nature Reviews. Cancer 5 (5), 388–396. Colditz, G.A., Sutcliffe, S., 2016. The preventability of cancer: Stacking the deck. JAMA Oncology 2 (9), 1131–1133. Colditz, G.A., Rosner, B.A., Speizer, F.E., 1996. Risk factors for breast cancer according to family history of breast cancer. For the Nurses’ Health Study Research Group. Journal of the National Cancer Institute 88 (6), 365–371.

280

Cancers as Ecosystems: From Cells to Population

Colditz, G.A., et al., 2004. Risk factors for breast cancer according to estrogen and progesterone receptor status. Journal of the National Cancer Institute 96 (3), 218–228. Couch, F.J., Nathanson, K.L., Offit, K., 2014. Two decades after BRCA: Setting paradigms in personalized cancer care and prevention. Science 343 (6178), 1466–1470. Dietz, W.H., Gortmaker, S.L., 2001. Preventing obesity in children and adolescents. Annual Review of Public Health 22, 337–353. Emmons, K.M., Colditz, G.A., 2017. Realizing the potential of Cancer preventiondThe role of implementation science. The New England Journal of Medicine 376 (10), 986–990. Franco, E., Rohan, T., Villa, L., 1999. Epidemiologic evidence and human papillomavirus infection as a necessary cause of cervical cancer. Journal of the National Cancer Institute 91, 506–511. Gertig, D.M., et al., 2013. Impact of a population-based HPV vaccination program on cervical abnormalities: A data linkage study. BMC Medicine 11, 227. Giuliano, A.R., et al., 2011. Efficacy of quadrivalent HPV vaccine against HPV infection and disease in males. The New England Journal of Medicine 364 (5), 401–411. Hankinson, S.E., et al., 1995. Reproductive factors and family history of breast cancer in relation to plasma estrogen and prolactin levels in postmenopausal women in the Nurses’ health study (United States). Cancer Causes & Control 6 (3), 217–224. Hankinson, S.E., Colditz, G.A., Willett, W.C., 2004. Towards an integrated model for breast cancer etiology: The lifelong interplay of genes, lifestyle, and hormones. Breast Cancer Research 6 (5), 213–218. Holmes, M.D., et al., 2005. Physical activity and survival after breast cancer diagnosis. JAMA 293 (20), 2479–2486. IARC, 2011. Monographs on the evaluation of carcinogenic risks to humans. In: A review of carcinogendPart B: Biological agents, vol. 100. International Agency for Research on Cancer, Lyon. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans, 2007. I.A.f.R.o.C. World Health Organization Combined estrogen–progestogen contraceptives and combined estrogen–progestogen menopausal therapy IARC monographs on the evaluation of carcinogenic risks to humans. International Agency for Research on Cancer; Distributed by WHO Press, Lyon, France; Geneva, p. 528. ix. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans, 2010 (2007, Lyon France). Alcohol consumption and ethyl carbamate. IARC monographs on the evaluation of carcinogenic risks to humans. International Agency for Research on Cancer; Distributed by WHO Press, Lyon, France. Imm, K.R., et al., 2017. African American prostate cancer survivorship: Exploring the role of social support in quality of life after radical prostatectomy. Journal of Psychosocial Oncology 35 (4), 409–423. International Agency for Research on Cancer, 2002. Weight control and physical activity. In: IARC Handbook on cancer prevention, vol. 6. International agency for research on Cancer, Lyon, p. 315. Khan, S., et al., 2018. The association of weight change in young adulthood and smoking status with risk of prostate cancer recurrence. International Journal of Cancer 142 (10), 2011–2018. Lauby-Secretan, B., et al., Body fatness and CancerdViewpoint of the IARC working group. The New England Journal of Medicine, 2016. 375 (8): p. 794–8. Lewington, S., et al., 2014. Temporal trends of main reproductive characteristics in ten urban and rural regions of China: The China Kadoorie biobank study of 300 000 women. International Journal of Epidemiology 43 (4), 1252–1262. MacMahon, B., 1965. Epidemiologic methods in cancer research. The Yale Journal of Biology and Medicine 37 (6), 508–522. MacMahon, B., Pugh, T.F., 1970. Epidemiology: Principles and methods, 1st edn. Little, Brown and Company, Boston, MA, p. 376. McCormack, V.A., Perry, N.M., Vinnicombe, S.J., Dos Santos Silva, I., 2010. Changes and tracking of mammographic density in relation to Pike’s model of breast tissue aging: a UK longitudinal study. International Journal of Cancer 127 (2), 452–461. McMichael, A., 1994. Invited commentary: Molecular epidemiology. New pathway or new travelling companion. American Journal of Epidemiology 140, 1–11. Medina, D., 2013. Pregnancy protection of breast cancer: New insights reveal unanswered questions. Breast Cancer Research 15 (3), 103. Michael, Y.L., et al., 2002. Social networks and health-related quality of life in breast cancer survivors: A prospective study. Journal of Psychosomatic Research 52 (5), 285–293. Miki, Y., et al., 1994. A strong candidate for the breast and ovarian cancer susceptibility gene BRCA1. Science 266 (5182), 66–71. OECD, 2012. SF-2.3 Mean age of mothers at first childbirth. In: OECD Family database. OECD Social Policy Division, Paris, France. Pereyra, A., 1961. The relationship of sexual activity to cervical cancer. Obstetrics and Gynecology 17, 154–159. Pike, M.C., et al., 1983. “Hormonal” risk factors, “breast tissue age” and the age-incidence of breast cancer. Nature 303, 767–770. Pike, M.C., et al., 1993. Estrogens, progestogens, normal breast cell proliferation, and breast cancer risk. Epidemiologic Reviews 15, 48–65. Rahman, M., Islam, M., Berenson, A.B., 2015. Differences in HPV immunization levels among young adults in various regions of the United States. Journal of Community Health 40 (3), 404–408. Rosner, B., Colditz, G.A., 1996. Nurses’ health study: Log-incidence mathematical model of breast cancer incidence. Journal of the National Cancer Institute 88 (6), 359–364. Rosner, B., Colditz, G.A., Willett, W.C., 1994. Reproductive risk factors in a prospective study of breast cancer: The Nurses’ health study. American Journal of Epidemiology 139 (8), 819–835. Rosner, B.A., et al., 2013. Validation of Rosner-Colditz breast cancer incidence model using an independent data set, the California teachers study. Breast Cancer Research and Treatment 142 (1), 187–202. Rosner, B., et al., 2015. Short-term weight gain and breast cancer risk by hormone receptor classification among pre- and postmenopausal women. Breast Cancer Research and Treatment 150 (3), 643–653. Rosner, B., et al., 2017. Weight and weight changes in early adulthood and later breast cancer risk. International Journal of Cancer 140 (9), 2003–2014. Schiffman, M.H., et al., 1993. Epidemiologic evidence showing that human papillomavirus infection causes most cervical intraepithelial neoplasia. Journal of the National Cancer Institute 85, 958–964. Shelton, R.C., et al., 2011. The association between social factors and physical activity among low-income adults living in public housing. American Journal of Public Health 101 (11), 2102–2110. Story, M., et al., 2008. Creating healthy food and eating environments: Policy and environmental approaches. Annual Review of Public Health 29, 253–272. Sung, J., et al., 2009. Height and site-specific cancer risk: A cohort study of a korean adult population. American Journal of Epidemiology 170 (1), 53–64. Swinburn, B.A., et al., 2011. The global obesity pandemic: Shaped by global drivers and local environments. Lancet 378 (9793), 804–814. Van Blarigan, E.L., et al., 2018. Marine omega-3 polyunsaturated fatty acid and fish intake after colon cancer diagnosis and survival: CALGB 89803 (alliance). Cancer Epidemiology, Biomarkers & Prevention 27 (4), 438–445. Walling, E.B., et al., Interventions to improve HPV vaccine uptake: A systematic review. Pediatrics, 2016, 138(1). pii: e20153863. Warnecke, R.B., et al., 2008. Approaching health disparities from a population perspective: The National Institutes of Health centers for population health and health disparities. American Journal of Public Health 98 (9), 1608–1615. Wehner, M.R., et al., 2014. International prevalence of indoor tanning: A systematic review and meta-analysis. JAMA Dermatology 150 (4), 390–400. Ziegler, R.G., et al., 1993. Migration patterns and breast cancer risk in Asian-American women. Journal of the National Cancer Institute 85, 1819–1827.

Further Reading 1. Emmons, K., 2000. Behavioral and social science contributions to the health of adults in the United States. In: Smedley, B., Simes, S. (Eds.), Promoting health: Intervention strategies from social and behavioral research. National Academy Press, Washington DC, pp. 254–321.

Cancers as Ecosystems: From Cells to Population

281

2. Shelton, R.C., et al., 2011. The association between social factors and physical activity among low-income adults living in public housing. American Journal of Public Health 101 (11), 2102–2110. 3. Atwood, K., Colditz, G.A., Kawachi, I., 1997. From public health science to prevention policy: Placing science in its social and political contexts. American Journal of Public Health 87 (10), 1603–1606. 4. Colditz, G.A., Wei, E.K., 2012. Preventability of cancer: The relative contributions of biologic and social and physical environmental determinants of cancer mortality. Annual Review of Public Health 33, 137–156.

CarcinogendDNA Adductsq Annette M Krais, Lund University, Lund, Sweden Rajinder Singh, University of Leicester, Leicester, United Kingdom Volker M Arlt, MRC-PHE Centre for Environment and Health, King’s College London, London, United Kingdom; and NIHR Health Protection Research Unit in Health Impact of Environmental Hazards at King’s College London in partnership with Public Health England, London, United Kingdom © 2019 Elsevier Inc. All rights reserved.

Glossary Carcinogen-DNA adduct The addition product between two molecules, in this case a carcinogen and DNA, involving one or more covalent bonds. Cytochrome P450 A member of a superfamily of closely related heme-containing oxygenases that oxidize a wide variety of substrates. Exposure biomarker A xenobiotic, its metabolite, or the product of an interaction between a xenobiotic and some target molecule such as DNA or proteins that are reflective of an individual’s exposure. Metabolic activation The metabolism of a compound lacking the ability to react covalently with DNA (or proteins) to a product(s) that does. Xenobiotic Describing or relating to a foreign compound.

Introduction Some years ago Christopher Wild, director of the International Agency for Research on Cancer (IARC), introduced the “exposome” as a complementary concept to the “genome.” The exposome reflects all of the encountered exposures of an individual over the course of her/his lifetime. According to epidemiological studies, approximately 90% of cancer deaths can be attributed to the presence of certain environmental factors, and are not related to the genetic makeup of an individual. So far IARC has classified around 120 compounds as carcinogenic to humans (Group 1 carcinogens). The list of carcinogens contains organic chemicals, inorganic material, biological agents, mixtures, and radiation. The underlying pathways leading to cancer for most of those compounds are not yet fully understood. Chemical carcinogens can exert their effect either through genotoxic (involving DNA damage) or nongenotoxic mechanisms. Some highly reactive genotoxic carcinogens are capable of directly interacting with DNA (e.g., alkylating agents) but most chemical carcinogens (e.g., polycyclic aromatic hydrocarbons [PAHs], aromatic amines, heterocyclic aromatic amines, nitrosamines) are not chemically reactive as such and require metabolic activation to reactive intermediates capable of binding to DNA (i.e., nucleobases), forming covalent adducts. Xenobiotics that need metabolic activation to exert their genotoxic/carcinogenic properties are termed pro-carcinogens or indirect-acting carcinogens. A wide variety of enzymes are involved in xenobiotic metabolism and these xenobiotic-metabolizing enzymes can be divided into two groups. Phase I (functionalisation) reactions include oxidation, reduction and hydroxylation catalyzed by enzymes such as cytochrome P450s (CYPs), cyclooxygenases (COXs), and aldo-keto reductases (AKRs) with the aim to make hydrophobic xenobiotics more hydrophilic and thus excretable. Phase II (conjugation) reactions further add polar moieties, such as glucuronic, acetic or sulfuric acids or glutathione thereby increasing polarity, which further assist in excretion. Examples of phase II enzymes include uridine diphosphate glucuronosyltransferases, N,O-acetyltransferases, sulfotransferases and glutathione S-transferases. However, phase I and II enzymes can arguably be in either category. As an example Fig. 1 shows the bioactivation and DNA adduct formation of benzo[a]pyrene (BaP), a PAH that has been classified as human carcinogen (Group 1) by IARC. The major pathway in humans involves the oxidation of BaP by CYP1A1 or CYP1B1 to an epoxide (i.e., BaP-7,8-oxide) that is then converted to BaP-7,8-dihydrodiol by microsomal epoxide hydrolase (mEH). BaP-7,8dihydrodiol can be further activated by CYPs (i.e., CYP1A1 or CYP1B1) or COX to BaP-7,8-dihydrodiol-9,10-epoxide (BPDE) which is the ultimate reactive species capable of reacting with DNA to form covalent adducts preferentially at the exocyclic amino group at N2guanine (i.e., 10-(deoxyguanosin-N2-yl)-7,8,9-trihydroxy-7,8,9,10-tetrahydro-BaP [dG-N2-BPDE]) (see Table 1). Other activation pathways for BaP lead to radical-DNA interactions and DNA depurination but their role in BaP carcinogenesis is under discussion. Some representative carcinogens, their environmental sources, their reactive metabolites and their most abundant DNA adduct are shown in Table 1.

q

Change History: November 2017. Annette M. Krais, Rajinder Singh, and Volker M. Arlt wrote a new article. Only couple of sentences of the old chapter have been incorportated into the present version of chapter. This article is an update of Alan M. Jeffrey, M.I. Straub, Carcinogen–DNA Adducts, in Encyclopedia of Cancer (Second Edition), edited by Joseph R. Bertino, Academic Press, 2002, Pages 345–357.

282

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65219-4

CarcinogendDNA Adducts

283

1

12 11

2 3

10

CYP1A1 CYP1B1

AKR1A1 AKR1C1-1C4

mEH

9 8

4 7

6

HO

5

HO

O OH

Benzo[ a]pyrene (BaP)

BaP-7,8-oxide

OH

BaP-7,8-dihydrodiol

BaP-7,8-catechol

NADP + CYP1A1 CYP1B1 COX

One electron oxidation

O2 + DNA

AKRs NADPH

H2O2

Oxidative damage to DNA (e.g. 8-oxo-dG)

O

+

O

HO

Radical cation

O

OH

BaP-7,8-dihydrodiol-9,10-epoxide + DNA

+ DNA

BaP-7,8-dione

+ DNA

Depurinating DNA adducts

Stable DNA adducts

Stable and depurinated DNA adducts

Radical cation pathway

Diol-epoxide pathway

o-Quinone pathway

Fig. 1 Pathways of BaP bioactivation leading to DNA adduct formation. CYP, cytochrome P450; mEH, microsomal epoxide hydrolase; COX, cyclooxygenase; and ARK, aldo–keto reductase. See text for details.

DNA adducts can also originate from endogenous processes, including normal metabolism, oxidative stress, lipid peroxidation and chronic inflammation (Fig. 2). The most abundant lesion linked to oxidative damage to DNA is 7,8-dihydro-8-oxo-20 -deoxyguanosine (8-oxo-guanine [8-oxo-dG]) which is formed by free radical attack on DNA (i.e., addition of OH to the C8 position of guanine). As shown in Fig. 1 BaP quinone redox cycling can generate reactive oxygen species which leads to the formation of 8-oxodG in DNA (see Fig. 2). Thus some genotoxic carcinogens that do not appear to directly modify DNA can instead damage it through inducing reactive oxygen species leading to oxidative base modifications. The DNA adductome is defined as the sum of all DNA adduct types and levels present in cellular DNA and can be considered reflective of the internal exposome. These DNA adducts, if not repaired before DNA replication, can lead to DNA mutations or chromosomal damage, thus possibly resulting in carcinogenesis. Therefore as adducts can lead to mutations, and mutations in critical genes are a characteristic feature of tumors, DNA adduct formation is considered to be the first critical step (i.e., initiation) in chemical carcinogenesis (Fig. 3). With few exceptions, however, carcinogens form DNA adducts in many tissues (in both target and nontarget tissues), leading to the view that adducts are necessary, but not sufficient, for carcinogenesis. Thus, although there are linear relationships between dose, adduct levels in target tissues and tumor incidence, it may be difficult to identify which tissues are targets for carcinogenicity by an agent purely from a consideration of the levels and persistence of DNA damage. Nevertheless, the identification and structural characterization of DNA adducts in human tissues can provide important clues on the etiology of human cancer. DNA adducts have the potential to be used as biomarkers of exposure (to carcinogens) and markers of cancer risk (from exposure to carcinogenic agents).

Application of DNA Adduct Analysis A major goal of adduct analysis is to improve our ability to extrapolate data on carcinogen exposure to human risk. It is reassuring to find that good correlations have been obtained between carcinogenicity of genotoxic compounds and their ability to bind to DNA, especially when closely related series of compounds are investigated. Sensitive DNA adduct detection methods make it possible to monitor carcinogen-DNA adduct formation in humans. Thus, DNA adduct formation in an exposed cell or tissue can be used: (i) to determine the mutagenic or carcinogenic potential of an agent; (ii) for monitoring human exposure to environmental carcinogens to study the etiology of cancer; (iii) for mechanistic investigations into carcinogen activation, tumor initiation or DNA repair; (iv) monitoring the environment for the presence of genotoxins (e.g., by detecting DNA adducts in aquatic species); and (v) assessing a patient’s response to cytotoxic cancer drugs. Since humans are exposed not only to one but to a complex mixture of carcinogens, direct proofs of an association of chemical exposure to the development of a specific cancer type are scarce. The plant carcinogen aristolochic acid and the mycotoxin aflatoxin B1 are two rare examples where distinct environmental exposures are linked to tumor development in humans. For both agents DNA adduct formation could be linked to cancer incidence, by combining molecular epidemiological data on chemical exposures,

Examples of environmental carcinogens and drugs that form DNA adducts

Carcinogen

Major active metabolite

284

Table 1

Major DNA adduct/site of modification

Environmental source

N

NH

N

N

HO DNA

OH

NH

Tobacco smoking combustion processes diet

HO

Benzo[a]pyrene (BaP) HO

BaP-7,8-dihydrodiol-9,10-epoxide (BPDE) O

O

O

O

O

O

O

O

O

O

O

O

O

OH

N2-guanine

O

CH

O

OH N

HN

DNA

N7-guanine

AFB1-8,9-epoxide

COOH

N

N

H N

O

O O

NO

O

O

N

OH

O

O H CO

HN OCH

OCH

HN

N

N

N

N

DNA

Aristolochic acid I

Mycotoxin

CH

O

Aflatoxin B1(AFB1)

O

N-Hydroxy-aristolactam I

N6-adenine

O

Aristolochia species

CarcinogendDNA Adducts

O

O

CH

CH

NH N

NH

N

N

O

CH O

N

N

C

N

O

N

NH

H N

CH

Food processing N

N

N

N

NH

DNA

2-Amino-1-methyl-6-phenylimidazo-[4,5-b]pyridine (PhIP)

C8-guanine N-Acetoxy-PhIP O

O NH NH

C

N

CH

O

NH

H N N

N-Acetoxy-4-aminobiphenyl

N

NH

Tobacco smoking, Azo dyes

DNA

4-Aminobiphenyl

C8-guanine NO

H N

OH O O N

O

N

O

Diesel exhaust air pollution

NH

N

N H

DNA

O

O

CH

NH

N2-guanine

N-Hydroxy-3-aminobenzanthrone

3-Nitrobenzanthrone

O

CH N

N

CH

CH

N

HN

HN

N

N DNA

H C

CH

H C

Anticancer drug

OSO

N H C

Tamoxifen

a-Hydroxy-tamoxifen sulfate

N2-guanine

O

CarcinogendDNA Adducts

HC

285

286

CarcinogendDNA Adducts O N

HN

N

N

N

OH N

H 2N

O

O

N

N

N

8-Oxo-guanine

N

N

N

N H

N

dR N 2,3-Ethenoguanine Fig. 2

N

dR

dR 5-Hydroxymethyluracil

N N

N

N

N

N N

O

dR

N

3,N 4-Ethenocytosine

1,N 6-Ethenoadenine

Structures of endogenous DNA adducts arising from oxidative stress, lipid peroxidation, and chronic inflammation.

XME

Detoxification

Excretion

Metabolic activation

Detoxification

Carcinogen +DNA

DNA repair

Normal cell

DNA adduct formation

DNA mutation

Cell death

Cell proliferation: Altered differentiation

Neoplastic cell

CANCER

The role of DNA adducts in the initiation process of carcinogenesis. See text for details.

N dR

dR

N 2,1-Ethenoguanine

Procarcinogen

Fig. 3

O

Thymidine glycol

N

N

OH

H

O N

HN

OH

H C

HN

OH N

O

Malondialdehydeguanine (M1-dG)

O

CH3

HN

dR

dR

O

CarcinogendDNA Adducts

287

chemical-specific DNA adducts in target tissues and mutation spectra in tumor-related genes. Thus aristolochic acid and aflatoxin B1 have both been classified as Group 1 human carcinogen by IARC. Aflatoxin B1 is a potent liver carcinogen that can be found as a contaminant in various crops and dietary exposure to aflatoxin B1 has been associated with hepatocellular carcinoma in certain areas of Asia and Africa. Those observations have been strengthened by the detection of N7-guanine adducts of aflatoxin B1 (see Table 1) in the urine of exposed individuals. These adducts lead to characteristic G:C / T:A transversion mutations selectively in codon 249 of the TP53 tumor suppressor gene in aflatoxin B1-induced liver cancers. Aristolochic acid, the plant extract derived from Aristolochia species, causes upper urothelial tract cancer in patients suffering from aristolochic acid nephropathy (AAN) and Balkan endemic nephropathy (BEN). Exposure to aristolochic acid has been demonstrated in many parts of the world and linked to medicinal herbal remedies or dietary intake. The 7-(deoxyadenosin-N6-yl)aristolactam I adduct (see Table 1) is the most abundantly found in urothelial tissue of AAN/BEN patients. It has even been detected in individuals over 20 years after exposure to aristolochic acid had ceased. The abundance and exceptionally long-term persistence of 7-(deoxyadenosin-N6-yl)aristolactam I in DNA can therefore be exploited as specific biomarker of AA exposure in human biomonitoring. This adduct also leads to otherwise rare A:T / T:A transversion mutations, which are frequently detected in TP53 in urothelial tumors of AAN/BEN patients. This characteristic mutation pattern is also found by whole-genome sequencing of AA-exposed tumors. It is clear from numerous studies that DNA adducts are intrinsically present in DNA. These “background” levels of adducts or DNA damage have been detected in individuals for whom no specific exposure to the agent of interest was known. Background DNA damage can have multiple origins including diet, air pollution or smoking and thus need to be taken into account when selecting control groups of individuals for comparison with a putative “exposed” population in molecular epidemiology studies. This will ensure that any difference in adduct levels between the study groups are due to the exposure of interest and not just linked to differences in other lifestyle factors. Besides “normal” environmental exposures, endogenous sources also contribute to background levels of DNA damage. DNA adducts have been detected in the range of 1 adduct per 107–1011 normal nucleotides in human tissues. The remainder of the chapter considers methods for carcinogen-DNA adduct detection in humans and how identified adducts in human tissues can be used to study cancer etiology.

Methods for Adduct Detection A variety of sensitive methods have been developed for the detection and characterization of carcinogen-DNA adducts (Table 2). In order to be applicable for human biomonitoring the assays used must be sufficiently sensitive to detect low levels of DNA adducts, require only small amounts of DNA, provide results quantitatively related to the exposure, be applicable to unknown DNA adducts that may be formed from complex mixtures and be able to resolve, quantitate, and identify DNA adducts. The main methods that have been used to detect DNA adducts include 32P-postlabeling, immunoassay, gas chromatography mass spectrometry (GC-MS), liquid chromatography (LC) coupled with electrochemical, fluorescence or mass spectrometry (MS) and accelerator mass spectrometry (AMS). Each of these methods adopts very different approaches for the detection of carcinogen-DNA adducts, and differs in their detection limits and requirements for quantities of DNA for analysis (Table 2). They all are sufficiently sensitive to be employed in human biomonitoring but they all have their strengths and weaknesses, so the method of choice has to be selected on a case-by-case basis.

Radioactive Labels Most early work on adducts required the use of radiolabeled compounds (labeled either with 3H or 14C). The DNA binding is measured by the detection of radioactivity in DNA isolated from exposed animal tissue or cells in culture achieving sensitivities of detection of 1 adduct in 108 nucleotides; 14C-labeling is less sensitive than 3H-labeling. The advantage of 3H is that it is relatively easily introduced into the carcinogen and can be obtained at higher specific activities than 14C because of the former’s shorter halflife. However, 3H is also very susceptible to loss during metabolism or by exchange reactions. This can result in the formation of tritiated water which in turn may become incorporated into nonexchangeable positions in newly synthesized DNA. Although Table 2

DNA adduct detection methods applicable to human biomonitoring and their limits of detection

Method

Variations

Amount of DNA required

Approximate detection limits

32

Nuclease P1 digestion, butanol extraction, TLC, HPLC, PAGE

1–10 mg

1 adduct per 109–1010 nucleotides

Up to 100 mg 20 mg 100–1000 mg Up to 100 mg

1 adduct per 109 nucleotides 1.5 adduct per 109 nucleotides 1 adduct per 109 nucleotides 1 adduct per 108–1012 nucleotides

P-postlabeling

Mass spectrometry Immunoassay Fluorescence Accelerator mass spectrometry

ELISA, DELFIA, CIA, IHC HPLC fluorescence, SFS

TLC, thin-layer chromatography; HPLC, high-performance liquid chromatography; PAGE, polyacrylamide gel electrophoresis; ELISA, enzyme-linked Immunosorbent assay; DELFIA, dissociation-enhanced lanthanide fluoroimmunoassay; CIA, chemoluminescence immunoassay; IHC, immunohisto-chemistry; SFS, synchronous fluorescence spectroscopy.

288

CarcinogendDNA Adducts

now at very low specific activity, this 3H can contribute significantly to the overall level of apparent binding. The major limitation of this approach to prelabel the carcinogens is the requirement of highly radioactive test compounds which cannot be used in human biomonitoring studies. The 32P-postlabeling assay also uses radiation as a measure to detect and quantify carcinogen-DNA adducts, the underlying principle is however that adducts are labeled after their formation. 32P can be obtained at about 300 times higher specific activity than 3 H and can be counted with higher efficiency due to the higher energy of the emitted b particle allowing the method to be intrinsically much more sensitive. The detection limit is approximately 1 adduct per 109–1010 nucleotides (see Table 2) using only a few micrograms of DNA which makes the assay widely applicable in human biomonitoring. As shown in Fig. 4, the method consists of four principle steps: (i) modified DNA is digested enzymatically to deoxyribonucleoside-30 -monophosphates using micrococcal

ATCAB Adducted DNA

5’ 3’

1) Micrococcal nuclease (endonuclease) 2) Spleen phospodiesterase (exonuclease)

DNA digestion G

T

A HO

HO P

HO

HO P

P

B

C

HO

P

P

Deoxyribonucleoside-3’-monophosphates

Enrichment

Nuclease P1 digestion G

Butanol extraction

C HO

HO

OH

OH

T

B HO P

A

HO

B HO

P

HO OH

OH

T4-polypolynucleotidekinase + [γ-32P]-ATP

Labeling reaction

32

P

B P

Deoxyribonucleoside-3’-[5’-32P]-bisphosphates

Chromatography

HPLC

PAGE

Multidirectional thin-layer chromatography

Autoradiography Quantitation by relative adduct labeling

Fig. 4 DNA adduct analysis using 32P-postlabeling. DNA (B indicates the modified base) is digested with micrococcal nuclease and spleen phosphodiesterase and subsequently phosphorylated by radioactive 32P using T4-polynucleotide kinase. The enrichment procedures include nuclease P1 digestion, which hydrolyzes most of the unmodified nucleotides to nucleosides, which are not kinase substrates. Many bulky DNA adducts, not linked to C8 purine position, are resistant to this enzyme. Butanol enrichment mainly extracts hydrophobic adducts. Radioactive-labeled adducts are separated by chromatography, mainly by using multidirectional thin-layer chromatography (TLC), but also high-performance liquid chromatography (HPLC) or polyacrylamide gel electrophoresis (PAGE) are occasionally used. Adducts separated by TLC are subsequently visualized by autoradiography and quantitation is achieved by calculating relative adduct labeling.

CarcinogendDNA Adducts

289

nuclease (endonuclease) and spleen phosphodiesterase (exonuclease); (ii) the adduct fraction is enriched using nuclease P1 digestion or butanol extraction; (iii) the DNA adduct is 50 -labeled by transfer of 32P-orthophosphate from [g-32P]-ATP using T4polynucleotide kinase resulting in deoxyribonucleoside-30 ,[50 -32P]-bisphosphates; and (iv) the chromatographic or electrophoretic separation of 32P-labeled adducts and quantitation by the measurement of their radioactive decay. As shown in Fig. 4, two principle enrichment procedures that follow DNA digestion have been developed. Digestion of deoxyribonucleosides-30 -monophosphates (modified or unmodified) with nuclease P1 prior to labeling, resulting in normal nucleotides but not many adducts, being converted to nucleosides which are no substrate for T4-polynucleotide kinase. Enrichment using nuclease P1 digestions works well for PAH-like adducts but is less suited for many aromatic amine adducts. Often adducts formed at C8-guanine appear not to be resistant to nuclease P1 digestion. The other enrichment procedure uses extraction of the aromatic/ hydrophobic adducts into butanol to separate them from the unmodified normal nucleotides (which remain in the water phase) for selective 32P-postlabeling. Chromatographic resolution of 32P-labeled adducts is mainly achieved by thin-layer chromatography (TLC) or alternatively by high-performance liquid chromatography (HPLC) (Fig. 4). Electrophoretic separation using polyacrylamide gel electrophoresis (PAGE) are also be utilized but is less frequently used. The major difficulty is the separation of the overwhelming excess of radioactivity associated with residual ATP, the normal nucleotides which also become phosphorylated, and any radiochemical decomposition products that might have been formed. Clearly, depending on the level of modification of the DNA by the carcinogen, which may be in the range of 1 adduct per 106–1010 nucleotides, only about that fraction of the disintegrations will be that of interest. Subsequently an elaborate separation technique was developed to remove any excess of radioactivity associated with residual ATP. This is illustrated here for the TLC version of the assay (see Fig. 4). The enriched fraction of adducts which have been labeled with 32P is placed in the center of a polyethylenimine cellulose TLC plate and the plate is washed first in one direction (D1) using a phosphate buffer, cutting away the edge of the plate with an attached filter paper which contains the bulk of the unwanted radioactivity moved. Multidirectional resolution in two directions of chromatography (i.e., D3 and D4), using solvent systems optimized for the adduct(s) of interest, allow for the separation of the carcinogen-modified bases. Chromatography in D2 is now often omitted. Subsequently, DNA adducts are visualized by placing the TLC plates in cassettes using autoradiography films or plates are scanned using radiation imaging systems (e.g., Instant imager or phosphoimager). 32 P-postlabeling can be applied to detect a variety of bulky carcinogen-DNA adducts as illustrated in Table 1. Prior structural characterization of adducts is not required, although some assumption about the likely chromatographic properties may be necessary. Thus the method can also be applied to the detection of DNA adducts formed by complex mixtures such as those present for example, in tobacco smoke. Total smoking-related adducts can be measured as a diffuse radioactive zone on the TLC plate that can be considered representative of PAH-DNA and other aromatic/hydrophobic adducts thereby providing a summary measure of a complex mixture of adducts present in the postlabeling chromatograms. Although the method does not provide direct structural information of adducts, identification of adducts can be achieved by co-chromatography with characterized synthetic standards, usually using a combined approach employing TLC and HPLC for confirmation.

Mass Spectrometry Mass spectrometry analysis is currently recognized as the “gold standard” for DNA adduct detection. MS analysis enables accurate quantitation and provides structural information about the DNA adduct, with only micrograms amounts of DNA or small volumes (from mL to mL) for adducts excreted in urine required for analysis. Detection is possible for DNA adducts at levels of up to one DNA modification per 109 deoxynucleosides (see Table 2), and sensitivity is still improving due to technical advancements in the MS instrumentation with regard to the efficiency of ionization, ion transmission and detection. Samples can be analyzed by using chromatographic separation coupled with various MS detection methods, for example, GC-MS, LC-MS/MS and AMS (see Section “Accelerator Mass Spectrometry”). Two different types of approaches can be used when studying genotoxin exposure using MS-based methods, targeted DNA adduct determination and screening of multiple DNA adducts (either targeted or untargeted); the latter adductomics approach is described in the “Adductomics” section. The targeted approach implies the identification and quantification of a limited number of DNA modifications. Using targeted DNA adduct detection, all known exposure types or exposure to specific chemicals can be analyzed in order to characterize the exposure. This implies that the MS system specifically scans for the presence of certain DNA adducts of interest, while the analysis of all other adducts that may be present in the sample is omitted. GC-MS can be used for the analysis of several DNA adducts, but samples require chemical derivatization prior to analysis to increase the volatility of the DNA adducts under investigation. It provides excellent separation and resolution due to the employment of capillary column chromatography as well as structural information about the adduct. Furthermore, GC-MS can use different MS modes for carcinogen-DNA adduct detection such as electronic impact (EI) and negative ion chemical ionization (NICI) mode. GC-EI-MS has been used to analyze polar deoxynucleosides and aglycone adducts as well as BaP-tetrols (i.e., r-7,t-8,t-9,c-10tetrahydroxy-7,8,9,10-tetrahydro-BaP) released by acid hydrolysis of DNA, as a marker for BPDE-DNA adduct formation. The main application of GC-EI-MS has been the detection of oxidized DNA bases, such as those resulting from the reaction of reactive oxygen species with DNA (e.g., 8-oxo-dG). The chemical derivatization of the DNA hydrolysate at high temperature has to be carefully monitored as artifacts of oxidized bases can easily be generated. Compared to the positive-ion EI mode, NICI possesses higher selectivity and superior sensitivity toward compounds with high electron affinities. GC-NICI-MS has been applied for DNA adducts caused by lipid peroxidation by-products, and tobacco smoke. However, GC-MS is less commonly used for carcinogen-DNA adduct

290

CarcinogendDNA Adducts

detection compared to LC-MS, due to labor-intensive sample preparation such as the extensive derivatization steps required to increase the volatility of the adduct and the potential to generate artifact adducts. LC-MS/MS is the primary method for the detection and quantitation of DNA adducts in DNA from biological matrixes. LC is more suited than GC for the analysis of DNA adducts since it allows the analysis of polar and nonvolatile molecules. Most analyses are performed using electrospray ionization (ESI) which is a soft ionization technique. LC-ESI-MS allows for the detection of adducted nucleobases, adducted deoxynucleosides, deoxyribose-phosphate adducts as well as modified oligonucleotides. LCMS/MS typically employs two quadrupole mass analysers in series separated from each other by a collision cell, which yields fragment ions following collision-induced dissociation (CID), and normally analyses are performed in selected reaction monitoring (SRM) mode. Several studies have used LC-MS/MS for the detection of modified bases in human urine including 8-oxo-dG, etheno-DNA adducts (see Fig. 2), N7-ethylguanine, and N7-guanine adducts of aflatoxin B1 (see Table 1). Examples for the application of LC-MS/MS in human tissues include 1,N2-propanodeoxyguanosine adducts derived from acetaldehyde and crotonaldehyde and acetaldehyde-derived adducts (as a possible marker of ethanol genotoxicity), 4-aminobiphenyl, aristolochic acid and 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) adducts (see Table 1).

Accelerator Mass Spectrometry AMS is an analytical technique that can quantify long-lived isotopes with attomole (amol) (10 18) sensitivity in isotope-labeled drugs and toxicants. AMS counts atoms of a rare isotope of interest and reports the ratio of the counted isotope to that of the total number of atoms of the element. Radiocarbon 14C (t1/2 ¼ 5730 years) and tritium 3H (t1/2 ¼ 12.3 years) are commonly used to label the compound or drug of interest for DNA adduct analyses. The excellent sensitivity of AMS enables analysis of DNA adduct formation for human biomonitoring. Human studies using AMS include adduct formation of the food carcinogen 2-amino-3,8dimethylimidazo[4,5-f]quinoxaline (MelQx) and tamoxifen, a compound used in the treatment of breast cancer. AMS has also been used to determine DNA adduct formation in rodents following the exposure of 14C-labeled benzene, ochratoxin A, and 2,3,7,8-tetrachlorodibenzo-p-dioxin. Despite the excellent sensitivity of AMS, sample preparation is labor-intensive and the instrumentation requires high levels of maintenance. Additionally, AMS does not provide any structural information about the DNA adduct, but it can be coupled to HPLC and identification can be achieved by using adduct standards. For example, HPLC separation followed by AMS analysis was employed for determining the distribution of [14C]oxaliplatin-induced DNA adducts in drugsensitive testicular and drug-resistant breast cancer cells at a clinically relevant drug concentration.

Immunological Methods These approaches rely on the preparation of antisera which recognize specifically the carcinogen-DNA adduct. To date around 20 different carcinogen-DNA adducts can be detected by immunological methods, limited by the need to raise antibodies against synthetically prepared and characterized adducts in order to detect them in human tissues. These antisera include those raised against alkylation adducts such as O6-methyl- and O6-ethylguanine, N7-methylguanine, and N6-methyladenine, or adducts formed by oxidative stress such 8-oxo-dG (see Fig. 2). Other antibodies that detect bulky DNA adducts include those derived from BaP, 4-aminobiphenyl and aflatoxin B1. Many variations exist on the method by which the antisera may be used. Most of the immunological studies have been based on enzyme-linked immunosorbent assay (ELISA) but modified assays such as the dissociationenhanced lanthanide fluoroimmunoassay (DELFIA) and chemiluminescence immunoassay (CIA) have been employed in molecular epidemiology because they have sensitivities approaching those of 32P-postlabeling. The underlying principle for the ELISA is that DNA to be tested is coated onto the surface of a suitable plate. Antiserum, specific for the adduct of interest, is added and after incubation any noncomplexed antibody is removed. A secondary antibody linked to a suitable marker enzyme, for example, alkaline phosphatase, is then added. This secondary antibody will bind to the plate only if there was an interaction between the adducted DNA and the primary antibody. After washing the plate, p-nitrophenylphospate is added as the enzyme substrate. As hydrolysis proceeds, which will be proportional to the extent of modification originally present in the DNA, the color formed from the released nitrophenyl anion can be measured. A standard curve is generated by serial dilution of either modified denatured DNA or monoadduct and mixing with the adduct-specific primary antibody. Samples with unknown levels of DNA adducts are tested in a similar way. The sensitivity of the ELISA can be further increased by using competitive versions of the assay, using substrates which yield fluorescent products. In the DELFIA assay the alkaline phosphatase conjugate is replaced by a biotin-europium-labeled streptavidin signal amplification system which releases europium into the solution forming a highly fluorescent lanthanide chelate complex. The assay principle of CIA is similar to that of competitive ELISA but uses chemiluminescene instead of a colored end product. Immunohistochemistry can also be used to detect DNA adducts in cells and tissue sections. Although the method is less sensitive than the immunological assays described above, it can provide valuable information on the localization of carcinogen-DNA adduct formation within a tissue or cell population. Assessment uses subjective estimation of staining intensity and/or number of positively stained cells with an arbitrary scale and thus need to be regarded as semiquantitative in its assessment of adduct levels. As with all immunohistochemical methods appropriate control experiments need to be conducted to validate the specificity of cell staining. Slot- or dot-blot immunological methods also provide a means to detect DNA adducts using antisera. The principle of the slot/ dot-blot methodology is that DNA is immobilized on nitrocellulose and then incubated with primary adduct-specific antibodies. Again enzyme-labeled secondary antibodies are subsequently used to measure the degree of binding using colored or

CarcinogendDNA Adducts

291

chemiluminescent end products. Immuno-dot/slot-blot methods have been developed to detect DNA adducts derived from PAHs, alkylating agents or malondialdehyde (see Fig. 2).

Fluorescence Techniques DNA adducts derived from PAHs (e.g., BaP) or aflatoxin B1 that contain fluorescent chromophores can be detected by means of their fluorescent emissions. This also includes some methylated DNA adducts. Usually HPLC separation coupled with fluorescence detection is used for the determination of carcinogen-DNA adduct formation. Combining the fluorescence characteristics (specific excitation and emission wavelengths) with HPLC separation makes it even possible to detect stereoisomers. The advantage of fluorescence analysis is that it requires no prior knowledge or preparation of the DNA adduct, only the knowledge that it is adequately fluorescent. However this is also a major limitation as only few carcinogen-DNA adducts are intrinsically fluorescent. HPLC with fluorescence detection has been applied in molecular epidemiological studies to determine DNA adducts but necessitated using fairly large quantities of DNA to achieve the required sensitivity. Other applications include using the fluorescence generated of BaP-tetrols, the hydrolysis products of BPDE-DNA adducts, as measures of BaP-derived DNA adduct formation by HPLC. Synchronous fluorescence spectrometry can also be used to detect BPDE-DNA adducts (sensitivity of  1 adduct/107 nucleotides). Samples are scanned synchronously with a fixed difference between the excitation and emission wavelengths (e.g., 34 nm for BPDE-DNA adducts) where the peak formed correlates linearly with the level of adducts. Fluorescent labeling of nonfluorescent DNA adducts is another approach to exploit fluorescence spectroscopy. Capillary electrophoresis and laser-induced fluorescence has been used to measure DNA adducts formed by BaP and aristolochic acid (sensitivity of  1.5 adducts/107 nucleotides).

Electrochemical Detection Electrochemically active DNA adducts which readily undergo reduction or oxidation reactions can be determined by HPLC coupled with electrochemical detection including 8-oxo-dG and N7-alkylguanines.

Adductomics The DNA adductome is part of the exposome. It is defined as the sum of all DNA adduct types and levels present in a DNA sample reflecting both known and unknown exogenous and endogenous exposures. Untargeted analysis refers to the total amount of all DNA adducts present, and in case of sufficient sensitivity, full scan MS data can be searched for the presence of known or unknown DNA adducts, in parallel or retrospectively (see also Section “Mass Spectrometry”). This approach however requires specialized untargeted screening technologies and methodologies as well as extensive data processing using specialized software. Despite its expense and complexity, screening studies have been suggested to be the future of cancer epidemiology. Recent improvements in MS scanning acquisition capabilities and instrument sensitivity potentially allow the use of MS in high-throughput applications in DNA adductomic approaches. DNA adductomics analyses usually takes advantage of the common structural feature of adducted deoxyribonucleosides which involves the cleavage of the labile glycosidic bond between the deoxyribose moiety (dR) and the adducted nucleobase. Hence, different approaches for the detection of these DNA adducts can be used by screening of the neutral loss of the dR moiety (116 u) and detecting the adducted base (Fig. 5A), which is the most common approach utilized or by analyzing the formation of the nucleobase-ion from the aglycone nuceloblase adduct (Fig. 5B).

(A)

+

+ A H

A

Neutral Loss of dR

Base

H

Base

+

dR

+

A

dR +

+

(B)

A H

Base

Base Ion Formation

H

Base

Fig. 5 The most common fragmentation pathways for adducted deoxynucleosides are (A) the neutral loss of the 20 -deoxyribose moiety and (B) formation of the nucleobase ion. Base, nucleobase; A, adduct; and dR, 20 -deoxyribose.

292

CarcinogendDNA Adducts

(A)

Constant neutral loss (CNL) Scan offset (116u)

Ion source

(B)

Fragmentation q2

Scanning Q3

Selected reaction monitoring (SRM)

Ion source

(C)

Scanning Q1

Static Q1

Fragmentation q2

Static Q3

Data dependent acquisition (DDA) -dR

Ion source

Full scan

MS2

MS3

Dynamic exclusion list: ions not further considered Fig. 6 Schematic illustrating the variety of data acquisition modes in the context for DNA adduct screening. The acquisition of DNA adductomics data can be performed in (A) CNL mode (only ions that fragment with the neutral loss of 116 u will be recorded as a signal), (B) in SRM mode (both Q1 and Q2 are operated in static mode and only ions of the user-specified m/z that produce specific fragments in q2 are detected), or (C) in DDA mode (the ions not showing the neutral loss of 116 u [green and yellow] are put into an exclusion list so that they are no longer eligible for fragmentation).

Several different modes of detection can be applied for DNA adductomic analyses, SRM, constant neutral loss (CNL), or data dependent acquisition (DDA) (Fig. 6). SRM methods offer the greatest sensitivity and are used primarily for targeted detection and quantitative analysis of DNA adducts, while the other two modes provide structural information and are useful for identification and characterization of novel DNA modifications. Constant neutral loss (CNL) scanning using a triple quadrupole mass spectrometer has been a common mode of detection for DNA adductomic analysis. This scan mode monitors the neutral loss of 20 -deoxyribose (from positively ionized 20 -deoxynucleoside adducts). As shown in Fig. 6A, quadrupole 1 (Q1) and Q3 are run in scan mode with an offset of mass difference 116 u, while ions are fragmented in the collision cell (q2). Only ions that fragment with the specific neutral loss of 116 u will be recorded as a signal. Alternatively, instead of using both quadrupoles in scan mode, a number of contiguous SRM transitions can be monitored all involving the neutral loss of 116 u. As shown in Fig. 6B, Q1 precursor ions are selected that undergo fragmentation (q2), followed by product ion selection Q3, thus allowing for selective detection of adducted 20 -deoxynucleosides using the [Mþ H]þ to [M þ H116]þ transition. The potential of using MS for the screening of DNA adducts by monitoring a large range of such SRM transitions led to the concept of producing an adductome map of all DNA adducts in a sample. This approach was first applied for the mapping of the DNA adductome by Kanaly and colleagues that used a series of SRM transitions totalling 374 for monitoring the neutral loss of 20 -deoxyribose. This LC-MS/MS approach for monitoring multiple SRMs has been used to analyze tissue from human lung for multiple DNA adducts derived from lipid peroxidation, furfuryl alcohol, and methyleugenol. Similar methods have been used for the simultaneous characterization of exocyclic DNA adducts derived from exogenous industrial chemicals and endogenous byproducts of lipid peroxidation in human DNA. An alternative LC-MS/MS adductomic approach can be used for the detection of N7-alkylguanine guanine adducts which are released from the DNA by thermal depurination. The transition of the adducted

CarcinogendDNA Adducts

293

nucleobase [Mþ H]þ precursor ion to the protonated guanine (m/z 152) product ion can be used as a generic method for monitoring N7-alkylguanines. An advancement for the detection and identification of unknown DNA adducts is the development of high resolution mass spectrometry (HRMS) and MSn detection. HRMS can provide accurate mass information enabling the determination of minute differences in mass between two compounds that would appear to be identical on a standard low resolution triple quadrupole instrument. By providing accurate mass measurements, HRMS provides data for the determination of elemental composition allowing identification of the DNA adduct. New approaches for DNA adductomics research may be achieved by using quadrupole-TOF (time of flight) instruments which provide accurate mass data. TOF-MS instruments can be used for the DNA adduct screening and characterization using DDA. DDA uses repeated acquisition of spectra from full scans as well as multistage scans (MSn). This scanning mode is performed in real-time and it thus requires preprogrammed decision making by the MS software during consecutive scans. For example, masses of ions selected for fragmentation can be placed into a dynamic exclusion list and these masses are no longer eligible for fragmentation in subsequent full scan spectra (Fig. 6C). Alternatively, an inclusion list containing masses of interest can be used to trigger analysis of targeted analytes. Furthermore, a reject mass list can be programmed containing a list of interfering background ions that will be excluded from analysis. DDA can also be used with ion trap instrumentation whereby ion trap MS analyzers allow the acquisition of multistage scan events (MSn) that can provide additional structural information about the DNA adduct. A linear ion trap LC-DDA-CNL-MS3 approach was developed, where the detection of a DNA adduct ion (listed in a targeted mass-list) in a limited m/z scan range leads to MS2 acquisition. Then the detection of the [Mþ H  116]þ ion among the top 10 of the most abundant MS2 ions triggers MS3 fragmentation. This approach has been used to study the formation of DNA adducts of derived from aromatic amines, heterocyclic aromatic amines, PAHs, and aldehydes. Other types of mass analysers such as the Orbitrap also provide very accurate mass detection information due to a high resolving power and mass accuracy. The Orbitrap mass analyzer is often coupled to a quadrupole or an ion trap and is well suited for small molecule analysis and untargeted adductomics applications and a high resolution LC-DDA-CNL-MS3 method was developed, using a linear ion trap-Orbitrap instrument. The instrumentation can also be applied to data-independent acquisitions (DIAs). In DIA, the MS data is collected systematically and independent of precursor ion information. The aim of DIA is to comprehensively obtain fragmentation data on all analytes that are present thereby allowing the data to be searched for the presence of known or unknown DNA adducts. Matrix-assisted laser desorption ionization (MALDI) is like ESI a soft ionization technique, but it provides far fewer multiply charged ions in the gas phase compared to ESI. This potentially simplifies data interpretation as well as analysis of complex mixtures. It has been a useful tool for the analysis modified oligonucleotides as well as determining their sequences. MALDITOF MS has been applied to characterize oligonucleotides carrying a DNA photoproduct as well as 5-hydroxymethylcytosine, 5formylcytosine and 6-oxothymine. Nontargeted analysis of modified nucleotides in DNA (as well as RNA) has been performed, involving derivatization of the deoxynucleotides with isotopologue benzoylhistamines followed by MALDI-TOF or MALDI-TOF/ TOF analysis. Triple quadrupole LC-MS/MS instruments operated in SRM mode are still the most sensitive mass spectrometers for detecting low levels of DNA adducts. New technological improvements enable very fast scanning times and rapid switching between SRM transitions without loss of sensitivity. Modern instruments operated in SRM mode can increase throughput with the capability of simultaneously acquiring hundreds of SRM transition channels per second while maintaining accuracy and precision, including multiple transitions per analyte. Though HRMS systems are excellent tools for identifying unknown DNA adducts, sensitivity may not be sufficient to detect background levels of DNA adducts that are present in human and animal DNA. However the sensitivity of these instruments can be improved by coupling with nano-LC and a nanospray ionization source.

Summary For many carcinogens the mechanism of tumor initiation is causally associated with carcinogen-DNA adduct formation. DNA adducts, if not repaired prior to DNA replication, can lead to mutation in critical genes of carcinogenesis (e.g., proto-oncogenes, tumor suppressor genes) which can drive malignant transformation of the cell and tumor development. DNA adduct formation is considered a critical step in the initiation of carcinogenesis and in many molecular epidemiology studies it has been shown that DNA adducts are valid biomarkers of exposure and cancer risk. A variety of sensitive methods have been developed for the detection and characterization of carcinogen-DNA adducts which are suitable for their applicability in human biomonitoring in order to study cancer etiology. The measurement of carcinogen-DNA adducts can also employed to evaluate mechanism(s) of carcinogenesis including host factors mediating the carcinogenic response/potency. All methods available for DNA adduct detection have their strengths and weaknesses, so the method of choice has to be decided on a case-by-case basis considering the study design and research question.

Prospective Vision Looking into the future, the field of DNA adductomics is still under development with many challenges still remaining however the hybrid HRMS/MSn methodology represents an important advance in the investigation of DNA adduct structures in complex

294

CarcinogendDNA Adducts

mixtures. This approach could be an extremely powerful tool for DNA adductome mapping. If sensitivity is sufficient, this could enable the discovery of yet unknown DNA adduct biomarkers. MS-based DNA adductomics is particularly suited for research on the exposure of the human body to both known and unknown hazardous agents (either exo- or endogenous) and any subsequently formed DNA adducts. However, limited availability of DNA adduct standards (in particular stable isotope labeled standards) is currently limiting full characterization and correct identification of unknown DNA adducts with MS. To improve DNA adductomics, a database could be set up with information on chemical structure and characteristics of DNA adducts, stability, prevalence, origin, and route of exposure. A multidisciplinary approach is needed, involving epidemiologists, clinicians, pathologists, analysts and bioinformaticians, to further unravel the DNA adductome.

See also: Aflatoxins. Cancer Risk Reduction Through Lifestyle Changes. Cell Responses to DNA Damage. Genetic Instability. Hereditary Risk of Breast and Ovarian Cancer: BRCA1 and BRCA2. Molecular Epidemiology and Cancer Risk. Radiation Therapy-Induced Metastasis and Secondary Malignancy. Role of DNA Repair in Carcinogenesis and Cancer Therapeutics.

Further Reading Arlt, V.M., Meinl, W., Florian, S., Nagy, E., Barta, F., Thomann, M., Mrizova, I., Krais, A.M., Liu, M., Richards, M., Mirza, A., Kopka, K., Phillips, D.H., Glatt, H., Stiborova, M., Schmeiser, H.H., 2017. Impact of genetic modulation of SULT1A enzymes on DNA adduct formation by aristolochic acids and 3-nitrobenzanthrone. Archives of Toxicology 91, 1957–1975. Arlt, V.M., Poirier, M.C., Sykes, S.E., John, K., Moserova, M., Stiborova, M., Wolf, C.R., Henderson, C.J., Phillips, D.H., 2012. Exposure to benzo[a]pyrene of hepatic cytochrome P450 reductase null (HRN) and P450 reductase conditional null (RCN) mice: Detection of benzo[a]pyrene diol epoxide-DNA adducts by immunohistochemistry and 32Ppostlabelling. Toxicology Letters 213, 160–166. Arlt, V.M., Singh, R., Stiborova, M., Gamboa da Costa, G., Frei, E., Evans, J.D., Farmer, P.B., Wolf, C.R., Henderson, C.J., Phillips, D.H., 2011. Effect of hepatic cytochrome P450 oxidoreductase deficiency on 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP)-DNA adduct formation in reductase inducible Cre (RIC) mice. Drug Metabolism & Disposition 39, 2169–2173. Arlt, V.M., Stiborova, M., Henderson, C.J., Thiemann, M., Frei, E., Singh, R., da Costa, G.G., Schmitz, O.J., Farmer, P.B., Wolf, C.R., Phillips, D.H., 2008. Metabolic activation of benzo[a]pyrene in vitro by hepatic cytochrome P450 contrasts with detoxification in vivo: Experiments with hepatic cytochrome P450 reductase null mice. Carcinogenesis 29, 656–665. Arlt, V.M., Schmeiser, H.H., Osborne, M.R., Kawanishi, M., Kanno, T., Yagi, T., Phillips, D.H., Takamura-Enya, T., 2006. Identification of three major DNA adducts formed by the carcinogenic air pollutant 3-nitrobenzanthrone in rat lung at the C8 and N2 position of guanine and at the N6 position of adenine. International Journal of Cancer 118, 2139–2146. Arlt, V.M., Ferluga, D., Stiborova, M., Pfohl-Leszkowicz, A., Vukelic, M., Ceovic, S., Schmeiser, H.H., Cosyns, J.P., 2002. Is aristolochic acid a risk factor for Balkan endemic nephropathy-associated urothelial cancer? International Journal of Cancer 101, 500–502. Balbo, S., Turesky, R.J., Villalta, P.W., 2014. DNA adductomics. Chemical Research in Toxicology (3), 356–366. Balbo, S., Hecht, S.S., Upadhyaya, P., Villalta, P.W., 2014. Application of a high resolution mass spectrometry-based DNA adductomics approach for identification of DNA adducts in complex mixtures. Analytical Chemistry 86, 1744–1752. Bessette, E.E., Goodenough, A.K., Langouet, S., Yasa, I., Kozekov, I.D., Spivack, S.D., Turesky, R.J., 2009. Screening for DNA adducts by data-dependent constant neutral losstriple stage mass spectrometry with a linear quadrupole ion trap mass spectrometer. Analytical Chemistry 81, 809–819. Brown, K., Tompkins, E.M., White, I.N., 2006. Applications of accelerator mass spectrometry for pharmacological and toxicological research. Mass Spectromtry Reviews 25, 127–145. Divi, R.L., Beland, F.A., Fu, P.P., Von Tungeln, L.S., Schoket, B., Camara, J.E., Ghei, M., Rothman, N., Sinha, R., Poirier, M.C., 2002. Highly sensitive chemiluminescence immunoassay for benzo[a]pyrene-DNA adducts: Validation by comparison with other methods, and use in human biomonitoring. Carcinogenesis 23, 2043–2049. ESCODD (European Standards Committee on Oxidative DNA Damage), 2002ESCODD(EuropeanStandardsCommitteeonOxidativeDNADamag. Comparative analysis of baseline 8-oxo7,8-dihydroguanine in mammalian cell DNA, by different methods in different laboratories: An approach to consensus. Carcinogenesis 23, 2129–2133. Farmer, P.B., Singh, R., 2008. Use of DNA adducts to identify human health risk from exposure to hazardous environmental pollutants: The increasing role of mass spectrometry in assessing biologically effective doses of genotoxic carcinogens. Mutation Research 659, 68–76. Biomarkers of carcinogen exposure and early effects. In: Farmer, P.B., Emeny, J.M. (Eds.), 2006. Environmental cancer risk, nutrition and Indvidual susceptibility (ECNIS), a network of excellence within the European Union Sixth Framework Programme, priority 5: Food quality and safety. The Nofer Institute of Occupational Medicine, Lodz. Gamboa da Costa, G., Singh, R., Arlt, V.M., Mirza, A., Richards, M., Takamura-Enya, T., Schmeiser, H.H., Farmer, P.B., Phillips, D.H., 2010. Quantification of 3-nitrobenzanthroneDNA adducts using online column-switching HPLC-electrospray tandem mass spectrometry. Chemical Research in Toxicology 22, 1860–1868. Gavina, J.M., Yao, C., Feng, Y.L., 2014. Recent developments in DNA adduct analysis by mass spectrometry: A tool for exposure biomonitoring and identification of hazard for environmental pollutants. Talanta 130, 475–494. Hemeryck, L.Y., Moore, S.A., Vanhaecke, L., 2016. Mass spectrometric mapping of the DNA adductome as a means to study genotoxin exposure, metabolism, and effect. Analytical Chemistry 88, 7436–7446. Kanaly, R.A., Matsui, S., Hanaoka, T., Matsuda, T., 2007. Application of the adductome approach to assess intertissue DNA damage variations in human lung and esophagus. Mutation Research 625, 83–93. Kanaly, R.A., Hanaoka, T., Sugimura, H., Toda, H., Matsui, S., Matsuda, T., 2006. Development of the adductome approach to detect DNA damage in humans. Antioxidant Redox Signaling 8, 993–1001. Klaene, J.J., Sharma, V.K., Glick, J., Vouros, P., 2013. The analysis of DNA adducts: The transition from 32P-postlabeling to mass spectrometry. Cancer Letters 334, 10–19. Krais, A.M., Speksnijder, E.N., Melis, J.P., Singh, R., Caldwell, A., Gamboa da Costa, G., Luijten, M., Phillips, D.H., Arlt, V.M., 2016. Metabolic activation of 2-amino-1-methyl-6phenylimidazo [4,5-b]pyridine and DNA adduct formation depends on p53: Studies in Trp53(þ/þ),Trp53(þ/) and Trp53(/) mice. International Journal of Cancer 138, 976–982. Krais, A.M., Speksnijder, E.N., Melis, J.P., Indra, R., Moserova, M., Godschalk, R.W., van Schooten, F.J., Seidel, A., Kopka, K., Schmeiser, H.H., Stiborova, M., Phillips, D.H., Luijten, M., Arlt, V.M., 2016. The impact of p53 on DNA damage and metabolic activation of the environmental carcinogen benzo[a]pyrene: Effects in Trp53(þ/þ), Trp53(þ/) and Trp53(/) mice. Archives of Toxicology 90, 839–851. Liu, S., Wang, Y., 2015. Mass spectrometry for the assessment of the occurrence and biological consequences of DNA adducts. Chemical Society Reviews 44, 7829–7854. Monien, B.H., 2014. Mass spectrometric DNA adduct quantification by multiple reaction monitoring and its future use for the molecular epidemiology of cancer. Advances in Experimental Medicine and Biology 806, 383–397.

CarcinogendDNA Adducts

295

Nortier, J.L., Martinez, M.C., Schmeiser, H.H., Arlt, V.M., Bieler, C.A., Petein, M., Depierreux, M.F., Pauw, L., Abramowicz, D., Vereerstraeten, P., Vanherweghem, J.L., 2000. Urothelial carcinoma associated with the use of a Chinese herb (Aristolochia fangchi). New England Journal of Medicine 342, 1686–1692. Phillips, D.H., 2013. On the origins and development of the 32P-postlabelling assay for carcinogen DNA adduct. Cancer Letters 334, 5–9. Phillips, D.H., Arlt, V.M., 2009. Genotoxicity: Damage to DNA and its consequences. Experientia Supplementum 99, 87–110. Phillips, D.H., 2008. Biomarker of exposure: Adducts. In: Wild, C., Vineis, P., Garte, S. (Eds.), Molecular epidemiology of chronic disease. John Wiley & Sons, Chichester, pp. 111–125. Phillips, D.H., Arlt, V.M., 2007. The 32P-postlabeling assay for DNA adducts. Nature Protocols 2, 2772–2781. Phillips, D.H., 2007. The formation of DNA adducts. In: Alison, M.R. (Ed.), The cancer handbook, 2nd edn. John Wiley & Sons, Chichester, pp. 338–350. Poirier, M.C., 2016. Linking DNA adduct formation and human cancer risk in chemical carcinogenesis. Environmental and Molecular Mutagenesis 57, 499–507. Santella, R.M., 1999. Immunological methods for detection of carcinogen-DNA damage in humans. Cancer epidemiology biomarkers and. Prevention 8, 733–739. Singh, R., Teichert, F., Seidel, A., Roach, J., Cordell, R., Cheng, M.K., Frank, H., Steward, W.P., Manson, M.M., Farmer, P.B., 2010. Development of a targeted adductomic method for the determination of polycyclic aromatic hydrocarbon DNA adducts using online column switching liquid chromatography/tandem mass spectrometry. Rapid Communication Mass Spectrometry 24, 2329–2340. Singh, R., Farmer, P.B., 2006. Liquid chromatography-electrospray ionization-mass spectrometry: The future of DNA adduct detection. Carcinogenesis 27, 178–196. Schmeiser, H.H., Nortier, J.L., Singh, R., Gamboa da Costa, G., Sennesael, J., Cassuto-Viguier, E., Ambrosetti, D., Rorive, S., Pozdzik, A., Phillips, D.H., Stiborova, M., Arlt, V.M., 2014. Exceptionally long-term persistence of DNA adducts formed by carcinogenic aristolochic acid I in renal tissue from patients with aristolochic acid nephropathy. International Journal of Cancer 135, 502–507. Stiborova, M., Arlt, V.M., Schmeiser, H.H., 2017. DNA adducts formed by aristolochic acid are unique biomarkers of exposure and explain the initiation phase of upper urothelial cancer. International Journal of Molecular Sciences 18, 2144. Tretyakova, N., Villalta, P.W., Kotapati, S., 2013. Mass spectrometry of structurally modified DNA. Chemical Reviews 113, 2395–2436. Villalta, P.W., Balbo, S., 2017. The future of DNA adductomic analysis. International Journal of Molecular Sciences 18, 1870. Wild, C.P., 2012. The exposome: From concept to utility. International Journal of Epidemiology 41, 24–32. Wogan, G.N., Kensler, T.W., Groopman, J.D., 2012. Present and future directions of translational research on aflatoxin and hepatocellular carcinoma. A review. Food Additives & Contaminants. Part A: Chemistry, Analysis, Control, Exposure & Risk Assessment 29, 249–257.

Carcinogenesis: Role of Reactive Oxygen and Nitrogen Speciesq Jean Cadet and J Richard Wagner, University of Sherbrooke, Sherbrooke, QC, Canada © 2019 Elsevier Inc. All rights reserved.

Glossary Complete carcinogen A carcinogen that causes cell initiation and promotes cell transformation and growth to benign tumors, as well as mediates progression to malignancy. Growth factors Low molecular weight substances produced by cells that induce growth of the same (autocrine response) or other (paracrine response) cells. Inflammation A very complex process that is initiated by a release of cytokines and chemokines leading to the infiltration of phagocytic cells. Upon stimulation, phagocytes generate copious amounts of reactive oxygen and nitrogen species, proteases, as well as other enzymes and proteins. It can be manifested by fever, edema, hyperplasia, phagocytic infiltration, and oxidative stress. Oxidative stress Excessive production of and/or impaired removal of oxidants with a concomitant decrease in reducing capacity of cells. Oxidatively generated DNA damage Direct oxidation is characterized by a decrease in electron density. Such a modification can occur by oxidation of a double bond in a normal DNA base, of a methyl group or by ionization. Indirect oxidatively produced modifications occur when another molecule or macromolecule is oxidized and the product or by-product of that oxidation reacts with DNA. Reactive oxygen species (ROS) and reactive nitrogen species (RNS) These are important players in normal physiology and signal transduction. They also participate in a plethora of damaging reactions, which lead to the dysregulation of normal cell controls, thus contributing to various pathologies, including cancer. Transcription factors Factors needed for the transcription of genes into mRNA by binding to the specific recognition sequences in the promoter or enhancer regions of DNA. Tumor promoters Agents that support the development of initiated cells into transformed cells and tumors. Tumor suppressor genes Genes that prevent cell transformation and growth of tumors. Mutation of such genes usually abrogates their normal functioning and facilitates tumor development.

Nomenclature NO Nitric oxide NO2 Nitrogen dioxide
OH Hydroxyl radical 1,N2-3Gua 1,N2-ethenoguanine 1 O2 Singlet oxygen 4-HNE 4-Hydroxy-2-nonenal 5-CaCyt 5-Carboxycytosine 5-ClCyt 5-Chlorocytosine 5-ClGua 5-Chloroguanine 5-FoCyta 5-Formyluracil 5-FoUra 5-Formylcytosine 5-HmCyt 5-Hydroxymethylcytosine 5-HmUra 5-Hydroxymethyluracil 5-OHCyt 5-Hydroxycytosine 8-ClAde 8-Chloroadenine 8-oxoAde 8-Oxo-7,8-dihydroadenine 8-oxodG 8-Oxo-7,8-dihydro-20 -deoxyguanosine 8-oxoGua 8-Oxo-7,8-dihydroguanine AID Activation-induced deaminase



q

Change History: May 2018. Jean Cadet updated the text, figures and references. This article is an update of Krystyna Frenkel, Carcinogenesis: Role of Reactive Oxygen and Nitrogen Species, in Encyclopedia of Cancer (Second Edition), edited by Joseph R. Bertino, Academic Press, 2002, Pages 359–367.

296

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.96062-8

Carcinogenesis: Role of Reactive Oxygen and Nitrogen Species

297

APE1 Apurinic/apyrimidinic endonuclease 1 APOBEC Apolipoprotein B mRNA editing enzyme catalytic polypeptide 1 BER Base excision repair CO3
L Carbonate radical anion CuD Cuprous ion DNMTs DNA methyltransferases DSB Double strand break DSBR Double strand break repair Endo III Endonuclease III FapyAde 4,6-Diamino-5-formamidopyrimdine FapyGua 2,6-Diamino-4-hydroxy-5-formamidopyrimidine Fe2 D Ferrous ion Fpg Formamidopyrimidine DNA N-glycosylase GC–MS Gas chromatography coupled to mass spectrometry H2O2 Hydrogen peroxide HOCl/OClL Hypochlorous acid/hypochlorite hOGG1 Human 8-oxoguanine glycosylase 1 HPLC-ECD High performance liquid chromatography coupled to electrochemical detection HPLC-ESI-MS/MS High performance liquid chromatography coupled to electrospray ionization tandem mass spectrometry HPLC-MS High performance liquid chromatography coupled to single mass spectrometer HR Homologous recombination IDH Isocitrate dehydrogenase ImidCyt 1-Carbamoyl-4,5-dihydroxy-2-oxoimidazolidine iNOS Inducible nitric oxide synthase LM-PCR Ligation-mediated polymerase chain reaction M1Gua Pyrimidopurinone 3-pyrimidopurin-10 (3H) one MBD4 Methyl CpG binding protein 4 MDA Malondialdehyde MMR Mismatch repair MUT1 Human mutT protein MUTHY mutY DNA glycosylase N2,3-3Gua N2,3-ethenoguanine NEIL1–3 Endonuclease 8-like proteins NER Nucleotide excision repair NHEJ Nonhomologous end joining NOS Nitrogen reactive species NTH1 Endonuclease III-like protein 1 O2
L Superoxide anion radicals OONOL Peroxynitrite Oz 2,2,4-Triamino-5(2H)-oxazolone ROS Reactive oxygen species ROO
Peroxyl radical SMUG1 Single strand specific mono-functional uracil-DNA glycosylase 1 SSB Single strand break TDG Thymine DNA glycosylase TET1–3 Ten-eleven translocation family of enzymes ThyGly 5,6-Dihydroxy-5,6-dihydrothymine ThyHyd 5-Hydroxy-5-methylhydantoin UNG Uracil-DNA N-glycosylase UraHyd 5-Hydroxyhydantoin UV Ultraviolet 6 3A 1,N -ethenoadenine 4 3Cyt 3,N -ethenocytosine

298

Carcinogenesis: Role of Reactive Oxygen and Nitrogen Species

Introduction Carcinogenesis is a complex multistage process often taking decades until malignancy appears. Conventionally, the carcinogenic process has been divided into three main stages: initiation, promotion, and progression. Initiation requires an irreversible genetic damage causing mutations in transcribed genes. Promotion consists of a potentially reversible oxidant-mediated conversion step followed by a clonal expansion of the initiated cells into benign tumors, which can progress to malignancy when they acquire many additional genetic changes. Those genetic changes include modifications of DNA bases, insertions and deletions, genetic instability consisting of loss of heterozygosity, chromosomal translocations, and sister chromatid exchanges, activation of oncogenes, and suppression of tumor suppressor genes. Although cell initiation is a frequent occurrence, tumor promotion and progression usually require a long time because of all of the genetic changes that have to accumulate within the same few cells. It has been known for many years that antioxidants inhibit formation of tumors, even though they might not decrease DNA adducts, thought to be the initiating lesions. Hence, antioxidants are likely to interfere with the oxidant formation during promotion and/or progression stages of tumor development. Since then, numerous publications have shown that various types of reactive oxygen species (ROS) are generated during all stages of carcinogenesis, but especially during that long time required to take an initiated cell to a fully disseminated cancer. This long period between the initiation stage and cancer development provides a large window of opportunity to interfere with and suppress the carcinogenic process. This can be accomplished more readily when processes of tumor promotion/progression and factors responsible for them are known. The importance of oxidative stress to human cancer is underscored by the existence of cancer-prone syndromes characterized by the production of high levels of oxidants, loss of adequate defense, or deficiencies in DNA repair processes. Human congenital syndromes related to oxidative stress include Fanconi’s anemia, Bloom’s syndrome, ataxia telangiectasia, Wilson’s disease, and hemochromatosis, among others. The last two conditions point to the contribution of an excess of bioavailable transition metal ions, such as copper and iron, leading to an overload of oxygen radicals in the liver and to the progression and outcome of those diseases. Hussain and colleagues clearly showed that livers of patients with Wilson’s disease and hemochromatosis contain increased levels of inducible nitric oxide synthase (iNOS), as well as mutations in the p53 tumor suppressor gene, especially G:C to T:A transversions at codon 249. This finding is particularly important because hydrogen peroxide (H2O2)/iron treatment, as well as lipid peroxide-derived mutagenic reactive aldehydes, such as 4-hydroxy-2-nonenal (HNE), cause the same type of mutations at codon 249. Etheno-base adducts are present in liver DNA from patients having these two diseases. Interestingly, HNE causes the formation of etheno derivatives in DNA and induces mutations at the 249 codon of the p53 gene in HNE-treated lymphoblastic cells. Notably, not only congenital syndromes exhibit mutations in the p53 gene, but also noncancerous colon tissues from ulcerative colitis patients, a colorectal cancer-prone chronic inflammatory disease. Frequencies of p53 mutations at codons 247 and 248 are strongly correlated with the progression of the disease being appreciably higher in inflamed than in noninflamed tissues. These studies strengthen the idea that chronic inflammation contributes to cancer, and the results are consistent with the hypothesis that p53 mutations at 247–249 codons are due to chronic inflammation-associated oxidative stress. Oxidants are continuously formed and scavenged during various normal cellular processes including mitochondrial respiration. ROS are required for the normal functioning of an organism and for the protection from invading bacteria. They are produced or utilized during mitochondrial respiration, metabolism of fats and xenobiotics, melanogenesis, and other peroxide reactions. Many types of cellular defenses keep these oxidants under control, but when they fail, extensive repair systems remove the damage attempting to restore DNA integrity. ROS produced in small amounts can serve as second messengers in signal transduction. However, when ROS are formed at a time when they are not needed and/or in amounts exceeding antioxidant defenses and DNA repair capacity, then ROS can contribute to various diseases, with cancer being prominent among them. Tumor promoters and complete carcinogens induce chronic inflammation, ROS production, and oxidatively generated DNA damage. Some of the oxidized DNA base derivatives are mutagenic, cytotoxic, and cross-linking agents. They can also cause a change in DNA methylation that leads to changes in gene expression. This process might perhaps account for the ROS-mediated enhanced levels of various growth and transcription factors, and genes involved in antioxidant defenses.

Oxidants: Formation and Reactivity The nucleus is subjected to continuous exposure to ROS that are generated during mitochondrial respiration as the result of incomplete one-electron reduction. In addition, the generation of ROS and also of reactive nitrogen species (RNS), likely involved in one-electron oxidation of guanine is enhanced during inflammation processes as further discussed below. Both types of oxidation reactions are also associated with exposure of exogenous physical agents including solar UV and ionizing radiation.

Reactive Oxygen and Nitrogen Species The three main reactive species involved in oxidative reactions included ROS, RNS and halogenating agents. Superoxide anion radical (O2
) is the predominant initial ROS that is generated during respiration and numerous various biochemical and chemical reactions. These include solar radiation activation enzymes such as of NADPH oxidase, UV and ionizing radiation and photosensitized reactions. Extensive amounts of (O2
) and RNS are produced during inflammation, a normal bactericidal and tumoricidal process. However, chronic inflammation contributes to the long-lasting pathologic effects of the carcinogenic course of action,

Carcinogenesis: Role of Reactive Oxygen and Nitrogen Species

299

regardless of the type of cancer. ROS and RNS are generated by activated phagocytic cells (PMNs, neutrophils, granulocytes) and both circulating and resident macrophages. Target cells (i.e., epidermal keratinocytes or hepatocytes) can also form these species in response to treatment with appropriate stimuli, such as tumor promoters or allergens. Although absolute amounts of ROS and RNS are much lower than those produced by stimulated phagocytes, these oxidants can set off the synthesis and release of a cascade of various cytokines and chemotactic factors, which are instrumental to the initiation of inflammatory responses by phagocytic cells. ROS and RNS also mediate increased formation of growth and transcription factors needed for the accelerated proliferation of initiated cells. In addition to O2
 four main other ROS are generated in cells, which include hydrogen peroxide (H2O2), hydroxyl radical
( OH), peroxyl radicals (ROO
) and singlet oxygen (1O2). H2O2 is mainly produced by enzymatic dismutation of O2
 by superoxide dismutase.
OH arises from either transition metal-mediated reduction of H2O2 according to Fenton reactions or radiolysis of water molecules through the indirect effects of ionizing radiation. Peroxyl radicals are generated by the addition of molecular oxygen to carbon-centered pyrimidine radicals. In contrast, the formation of 1O2 can be explained by the involvement of endogenous type II photosensitizers that are able upon excitation by UVA radiation/visible light to transfer energy to molecular oxygen. Mammalian heme peroxidases represent other sources of oxidizing agents that are activated in inflammatory tissues. They consist of myeloperoxidases, lactoperoxidases and eosinophil peroxidases that are able to trigger the formation of hypochlorous acid/hypochlorite (HOCl/OCl). RNS are also produced during inflammation and contribute to its pathologic effects. Nitric oxide (
NO) as a major player is generated through activation of epidermal inducible nitric oxide synthase (iNOS). The distinction between ROS and RNS is further blurred because of an avid reaction between O2
 and
NO at near diffusion rate giving rise to peroxynitrite (ONOO), a potent oxidizing and nitrating agent (Table 1).

One-Electron Oxidants Other main oxidative degradation pathways of DNA are initiated by the ionization of DNA bases and sugar moieties. Nucleobases are susceptible to several one-electron oxidants in the following decreasing order of reactivity guanine > adenine > cytosine z thymine. Thus, guanine preferentially undergoes one-electron oxidation by type I photosensitizers (riboflavin, menadione), bi-photonic ionization mediated by high intensity nanosecond UVC laser. In addition, guanine is oxidized by the carbonate radical anion (CO3
), a biologically relevant oxidizing radical that derives from the thermal decomposition of nitrosoperoxycarbonate, arising from the reaction of peroxynitrite (OONO) with CO2 or carbonates. Another one-electron oxidant includes bromate. The highly energetic photons emitted by ionizing radiation are able to oxidize both nucleobases and the 2-deoxyribose moieties through the so-called direct effect. Lastly, purine and pyrimidine radical cations generate aminyl radicals that are strong oxidants of neighboring DNA bases in DNA.

Reactivity Reactive oxygen species Various cellular ROS and oxidants exhibit a wide diversity of reactivity toward nucleic acids ranging from the totally unreactive O2
 to the highly oxidizing
OH. H2O2 is another poorly reactive ROS that unlike O2
 is able to cross membrane and migrate within cells to reach the nucleus. In the presence of reduced transition metals such as ferrous and cuprous ions, H2O2 is able to generate
OH that reacts essentially at the site where it is produced with the nearest molecule. For example, H O will react with DNA bound 2 2 Fe2 þ and Cuþ to generate
OH that subsequently reacts with DNA in a site specific manner. The diffusion of
OH in biological

Table 1

Major cellular oxidants and their reactivity toward nucleic acids

ROS or RNS

Reactivity

O2
 H2O2
OH Peroxyl radicals

Undetectable toward DNA constituents Very low reactivity with adenine; involved in Fenton reaction High reactivity with bases and the 2-deoxyribose moiety Pyrimidine peroxyl radicals may either abstract hydrogen atom from vicinal thymine or add to vicinal guanine giving rise to tandem base lesions Guanine the only reaction target via cycloaddition Give rise to oxidized bases and intra- and interstrand crosslinks through initial formation of base radical cation Chlorination of amino-substituted bases No detectable reactivity toward nucleic acid components at the exception of deamination of nucleobases One-electron oxidant of guanine through the transient formation of nitrosoperoxycarbonate with subsequent release of carbonate radical anion (CO3
) Reactive aldehydes generated through the thermal decomposition of lipid hydroperoxides form covalent adducts with amino-substituted bases

1 O2 One-electron oxidants HOCl/OCl
NO ONOO

Lipid hydroperoxides

300

Carcinogenesis: Role of Reactive Oxygen and Nitrogen Species

medium is very short (1–10 nm), and thus, difficult to scavenge with antioxidants. The reaction of
OH with nucleobases and the sugar moiety occur at close to diffusion-controlled rates with formation of modified bases and strand breaks, respectively. Peroxyl radicals (ROO
) in general exhibit a much lower reactivity than
OH. However, when inserted into DNA, pyrimidine ROO
have been shown to efficiently react with vicinal bases, preferentially when they are located on the 50 -terminus, giving rise to tandem base modifications. Two reactions have been identified in model studies: pyrimidine ROO
is able to either abstract a hydrogen atom from the methyl group of an adjacent thymine or add to vicinal guanine at C8. 1O2 is, in addition to
OH, a potent oxidizing ROS that among DNA components reacts exclusively with guanine. It may be noted that 1O2 due to its highly selective reactivity shows an intracellular diffusion distance within the range of 150–220 nm that is much higher than that of
OH.

DNA base radical cations Base radical cations of nucleobases produced by one-electron oxidation reactions are highly unstable intermediates that may undergo two main reactions. One is deprotonation that leads to the formation of carbon-centered radicals at the methyl group of thymine that is identical to the radical produced by
OH-mediated H-atom abstraction. Similarly, the competitive reaction for base radical cations involves hydration which leads to the formation of the same radicals that are generated by
OH although hydration occurs specifically at C6 of the pyrimidine ring. More generally the guanine radical cation is the predominant species in irradiated DNA partly due to redistribution on guanine bases of initially generated radical cation present on other bases. Furthermore, guanine radical cations are prone to nucleophilic addition leading to DNA–protein crosslinks and intrastrand crosslinks with opposite cytosines.

Reactive nitrogen species


NO is a poorly reactive species with the exception however of its ability to trigger the deamination of cytosine, adenine and guanine. This explains why
NO is able to migrate within cells and efficiently react with O2
, another unreactive radical, giving rise to ONOO. Evidence has been provided that one of the main modes of action of ONOO, a moderately reactive compound, is to react with CO2 or bicarbonates yielding unstable nitrosoperoxycarbonate that decomposes into carbonate radical anion (CO3
) and nitrogen dioxide (
NO2).

Hypochlorous acid/hypochlorite Hypochlorous/hypochlorite (HOCl/OCl) that is generated by peroxidases in inflammatory tissues is partly scavenged by amino containing compounds within membranes forming much-longer acting chloramines. However, HOCl/OCl is able to diffuse in cells and reach the nucleus where it reacts with DNA amino-substituted nucleobases.

Reactive aldehydes as breakdown products of lipid hydroperoxides Decomposition of lipid hydroperoxides formed by either the oxidation by
OH or 1O2 gives malondialdehyde and 4-hydroxy-2nonenal among the main reactive aldehydes. The released electrophilic species are able to migrate intracellularly from oxidized membranes to the genome where they are involved in the formation of adducts with amino-substituted nucleobases.

Oxidatively Generated Damage to DNA Both ROS and RNS can induce a plethora of various types of modifications to DNA with the bases and the 2-deoxyribose moiety as the main targets. For this purpose numerous studies have been dedicated to the identification of the main DNA degradation products arising from exposure of isolated DNA and model compounds to
OH, one-electron oxidants, 1O2, hypochlorous acid and reactive aldehydes. Thus, more than 100 different types of modifications including isomers have been identified so far in model studies and the mechanisms of their formation have been elucidated in most cases. In parallel, numerous attempts have been made to develop methods aimed at measuring oxidized bases in cellular DNA, still an analytical challenge as further discussed in the next section. It may be emphasized that in most cases, even under conditions of oxidative stress, the levels of the main oxidized bases, such as 8-oxo-7,8-dihydroguanine (8-oxoGua), is at best only a few modifications per 106 normal bases in nuclear DNA.

Measurement of Oxidized Bases in Cellular DNA The accurate detection and quantitative measurements of oxidized bases in cellular DNA have been hampered by the use of inappropriate methods. This was the case of the gas-chromatography–mass spectrometry (GC–MS) method that was found to lead to artifactual overestimation of the yields of oxidized bases by at least two orders of magnitude. The main origin of unreliable GC–MS data lies on efficient spurious oxidation during the hydrolysis and derivatization step of DNA bases involving heating of the samples at 140 C prior to the chromatographic analysis. Another questionable method that unfortunately is still widely applied particularly by clinicians and biologists involves the use of poorly specific immunoassays. In this respect, it was shown that monoclonal and polyclonal antibodies directed against 8-oxoGua suffer from a lack of antigenicity, showing a cross-reactivity with guanine on the order of 10 4. Other methods lead to inconsistent results including the [32P]-postlabeling technique, the ligation-mediated polymerase chain reaction (LM-PCR) assay and high performance liquid chromatography analysis associated with single mass spectrometry detector (HPLC-MS). The first sensitive measurement of oxidatively generated base damage in cellular DNA concerned

Carcinogenesis: Role of Reactive Oxygen and Nitrogen Species

301

8-oxo-7,8-dihydro-20 -deoxyguanosine (8-oxodG), which was achieved by HPLC coupled with electrochemical detection (ECD) operating in the oxidative mode. Versatile HPLC-electrospray-tandem mass spectrometry (EIS-MS/MS), which is more sensitive and quantitative than previous methods, allows for the unambiguous assignment of detected lesion and has become the gold standard method for measuring a wide range of oxidized nucleosides and bases. However, the possibility of adventitious oxidation of nucleobases still remains a problem if precautions are not taken to prevent this reaction from occurring during DNA extraction and the subsequent work-up prior the HPLC analysis. Thus, HPLC based analytical methods are limited to the measurement of oxidatively base damage only under conditions of severe oxidative stress. These difficulties can be partly overcome using enzymatic based assays: recognition and hydrolytic cleavage of the N-glycosidic modified bases is performed by DNA repair glycosylases including bacterial endonuclease III (Endo III) for oxidized pyrimidine bases and either bacterial formamidopyrimidine DNA N-glycosylase (Fpg) or human 8-oxoguanine DNA glycosylase (hOGG1) for modified purine bases. The resulting abasic sites are converted into single strand breaks, which are suitably measured in a highly sensitive way by either the alkaline comet assay or the alkaline elution technique. Despite a lack of specificity compared to HPLC-ESI-MS/MS measurements, the enzymatic assays are suitable tools for the detection of low variations in the levels of oxidized bases. In the subsequent sections only the oxidatively generated base lesions that were accurately measured by either HPLC based method or enzymatic assays are discussed.

Hydroxyl Radical-Mediated DNA Damage


OH is able to react with both the bases and the sugar backbone leading to the formation of a large spectrum of modifications.

Oxidized pyrimidine bases So far, the main
OH-mediated oxidation products that were identified in model studies in the presence of oxygen have been detected in cellular DNA. Two main classes of oxidized bases were found to be generated according to the nature of the initial
OH reaction. Thus, addition of
OH across the 5,6-double bond of pyrimidine bases gives rise to hydroxyhydroperoxides that are reduced or rearrange into more stable oxidation products. Two main stable products of thymine oxidation include 5,6dihydroxy-5,6-dihydrothymine, also called “thymine glycol” (ThyGly) and 5-hydroxy-5-methylhydantoin (ThyHyd). In the case of cytosine, 5-hydroxycytosine (5-OHCyt), 5-hydroxyhydantoin (UraHyd) and 1-carbamoyl-4,5-dihydroxy-2-oxoimidazolidine (ImidCyt) are the main cytosine oxidation products (Fig. 1). The second main
OH decomposition pathway of pyrimidine bases involves initial H-atom abstraction from the methyl group of thymine and 5-methylcytosine. In both cases, the transient formation of a peroxyl radical leads to the generation of alcohol and aldehyde derivatives that were identified as 5-hydroxymetyhyluracil (5-HmUra), 5-formyluracil (5-FoUra), 5-hydroxymethylcytosine (5-HmCyt) and 5-formylcytosine (5-FoCyt), respectively.

Modified purine bases The main degradation products of the purine bases arise from the transient formation of 8-hydroxy-7,8-dihydropurinyl radical formed by direct
OH addition at C8 of the bases or more indirectly through vicinal
OH-induced pyrimidine peroxyl radical. In a subsequent step, one-electron oxidation of 8-hydroxy-7,8-dihydroguanyl radical by O2 gives rise to 8-oxoGua while competitive one-electron reduction of the latter radical, likely by thiol compounds, generates an open-ring 2,6-diamino-4-hydroxy-5formamidopyrimidine (FapyGua). Similar reactions take place for adenine with the formation of 8-oxo-7,8-dihydroadenine (8-oxoAde) and 4,6-diamino-5-formamidopyrimidine (FapyAde) albeit with an about 10-fold lower efficiency than the formation of guanine products (Fig. 2). An additional oxidation product that arises from initial
OH at C4/C5 followed by a complex degradation pathway has been identified as 2,2,4-triamino-5(2H)-oxazolone (Oz). A major difference between
OH-mediated reactions and one-electron oxidation of DNA bases involves the efficient reaction of
OH with the sugar leading to DNA strand cleavage.

Strand breaks and other sugar degradation products Evidence has been provided that
OH-mediated hydrogen atom abstraction at C3, C5 and to a lesser extent C4 of the 2-deoxyribose moieties leads to the formation of DNA single strand breaks (SSBs). As a competitive reaction to SSB generation, the C4 oxidized abasic site is also involved in the formation of interstrand crosslinks. This is rationalized by the formation of a covalent bond between the reactive aldehyde generated from the oxidized abasic site and the exocyclic amino group of cytosine opposite to the abasic site. It may be pointed out that double strand breaks (DSBs) are not generated by one
OH attack. The formation of DSBs requires two independent radical hits on the two opposite DNA sites that are separated by less than 25 base pairs. This is typically achieved upon exposure of cellular DNA to ionizing radiations that also gives rise to non DSB oxidatively generated clustered damage.

One-Electron Oxidation of DNA One-electron oxidation of nucleobases generates purine and pyrimidine radical cations. These radical cations are able to migrate along the DNA chains with preferential trapping by guanines bases as sinks for the positive hole that are converted into stable degradation products identified as 8-oxoGua and FapyGua. In addition, minor degradation products arise from the partial conversion of other base radical cations as detected by HPLC-ESI-MS/MS. Thus, 8-oxoAde and FapyAde are formed by oxidation and reduction, respectively, of transiently generated 8-hydroxy-7,8-dihydroadenyl radical upon hydration of the adenine radical cation. Relatively minor pyrimidine oxidation products including ThyGly, 5-HmUra, 5-FoUra and 5-OHCyt have also been detected in cellular DNA,

302

Carcinogenesis: Role of Reactive Oxygen and Nitrogen Species

O

O CH3

HN 3

5 2 6

O

O

N

H N

CH3 OH O OH H ThyGly

HN N

Thy

CH3 OH ThyHyd

N

O

O CH2OH

HN O

O

CHO

HN

N

O

N 5-FoUra

5-HmUra

NH2

NH2

N O

N Cyt

NH2

R OH OH N H CytGly

N O

O

NH2

O

HN O

N 5-MeCyt

O

R

H2N

N

O

R=H or CH3

NH2

N

OH H OH

5-OHCyt

NH2 CH2OH

N

R OH OH N H UraGly

CytImid

NH2 CH3

N

N

O CH3

N

OH

N

CHO

N +

O

N

O 5-MeCyt

N

O 5-HmCyt

N 5-FoCyt

Fig. 1 Oxidation of thymine (Thy), cytosine (Cyt), and 5-methylcytosine (5-mCyt) by one-electron oxidation and
OH. One-electron oxidation and
OH give the same products except that the distribution changes, for example, one-electron oxidation gives greater methyl oxidation products (5-HmUra and 5-FoUra for Thy, and 5-HmCyt and 5-FoCyt for 5-MeCyt).

resulting from thymine and cytosine radical cations. It may also be pointed out that a guanine–thymine intrastrand cross-link is formed by nucleophilic addition of N3 of thymine to that guanine radical cation albeit the reaction is about 100-fold less efficient than the formation of 8-oxoGua.

Singlet Oxygen Oxidation of DNA Exposure of cellular DNA to 1O2 selectively reacts with guanine and leads to the predominant formation of 8-oxoGua with the exclusion of SSBs. Evidence has been provided that the UVA-induced generation of 8-oxoGua in the DNA of cells and human skin is accounted for the predominant involvement of 1O2 by type II photosensitization.

Modifying Base Activity of Hypochlorous Acid/Hypochlorite In addition to its oxidizing feature, HOCl/OCl is able to trigger the chlorination of amino-substituted bases leading to the generation of 5-chlorocytosine (5-ClCyt), 8-chloroadenine (8-ClAde) and 8-chloroguanine (8-ClGua) as measured by HPLC-ESI-MS/MS in cellular DNA (Fig. 3).

Indirect Oxidatively Generated Damage to DNA Bases In addition to the direct effects of oxidants on nucleobases, DNA may also be indirectly modified through interactions with electrophilic species produced by a ROS attack on other molecules or during ROS formation by such molecules. These consist essentially of etheno-base adducts formed in DNA by breakdown products of lipid peroxides. ROS including
OH and 1O2 react with

Carcinogenesis: Role of Reactive Oxygen and Nitrogen Species

O HN 1

5 2 4

H2N

O

7

N N

N

H N

HN

8

O

H2N

Gua

303

N

N

8-oxoGua

O

H N

HN H2N

CHO NH

N

FapyGua

O

H N

H2N

N

NH Oz NH2

NH2 N

N

N

N

H N

N

O N

N

Ade

8-oxoAde

NH2

H N

N

CHO NH

N FapyAde

Fig. 2 Oxidation of guanine (Gua) and adenine (Ade) by
OH, one-electron oxidants and singlet oxygen (1O2). The formation of 8-oxoGua and Fapy depends on the presence of reducing agents and oxygen. The reaction of 1O2 with Gua specifically gives 8-oxoGua. The yield of Ade oxidation products either by one-electron oxidation and
OH are 10-fold lower than those of Gua.

NH2

NH2

HOCl

N3

5 2 6

N

O

Cl

N N

O

Cyt O HN 1

5 2 4

H2N

N

5-ClCyt O

7

N

HOCl

8

N

NH2

Fig. 3

5-ClGua NH2 7

N1

N

N Ade

N

5 2 4

Cl N

N

H2N

Gua

N

HN

8

HOCl

N

N Cl

N N 5-ClAde

Halogenation of DNA bases (Cyt, Gua and Ade).

unsaturated lipids leading to the formation of lipid peroxides, which upon decomposition release aldehydes, such as malondialdehyde (MDA) and 4-hydroxy-2-nonenal (4-HNE). Both MDA and 4-HNE are mutagenic because they react with the exocyclic amino group of guanine, adenine and cytosine to form etheno derivatives including pyrimidopurinone 3-pyrimidopurin-10 (3H) one (M1Gua), 1,N6-ethenoadenine (εA), 3,N4-ethenocytosine (εCyt), N2,3-ethenoguanine (N2,3-εGua), and 1,N2-ethenoguanine (1,N2-εGua). The aldehyde adducts transform further into products with an additional ring attached to purines and

304

Carcinogenesis: Role of Reactive Oxygen and Nitrogen Species

O

N

N N

N

N

N

O

εAde

N

N

εCyt

N

M1Gua Fig. 4

N

O

N

N

N

N N

N

N2,3-εGua

O N

N

HN

N

N H

N

N

1,N2−εGua

Adducts of lipid peroxidation and DNA bases.

pyrimidines (Fig. 4). Etheno compounds are present in DNA isolated from various types of tumors or from nontumorous specimens of tumor-bearing animals and humans attesting to the increase of ROS-induced lipid peroxidation in cancerous cells.

DNA Repair and Mutagenesis of Oxidatively Generated Damage There are four basic pathways of DNA repair: Base excision repair (BER); nucleotide excision repair (NER); mismatch repair (MMR); and double strand break repair (DSBR) including nonhomologous end joining (NHEJ) and homologous recombination (HR). Each of these pathways involve various proteins that recognize DNA damage, undergo enzymatic steps (DNA cleavage, polymerization, ligation), provide access, support and stability to repair complexes, engage in crosstalk between repair pathways, and elicit DNA damage response signals. In the case of ROS-induced base and sugar damage, the main repair pathway is BER although certain types of damage are repaired by NER and MMR pathways. On the other hand, the presence of excessive BER substrates and intermediates at the replication folk can also lead to double strand breaks, and thereby set DSBR repair into motion. The conventional method to study DNA repair and mutagenesis of ROS-induced damage is to insert a modification into synthetic oligonucleotides and examine the ability of purified enzymes, nuclear extracts and whole cell extracts to cleave the oligonucleotide at the site of damage by gel electrophoresis. There have been considerable methodological advances in the study of DNA repair and mutagenesis of specific lesions applying mass spectrometry strategies, using DNA repair arrays, incorporating oligonucleotides containing lesions into cellular DNA by transfection, constructing nucleosomes core particles containing lesions. The location and efficiency of repair proteins can be monitored in real time by the incorporation of fluorescent nucleotides during DNA repair synthesis or labeling repair proteins directly. A typical feature of DNA N-glycosylases is their broad substrate specificity toward oxidized bases in DNA and their ability to backup or be redundant in cellular and animal model systems. For example, the removal of uracil from DNA involves uracil-DNA N-glycosylase (UNG), single strand specific mono-functional uracil-DNA glycosylase 1 (SMUG1) as well as two mismatch specific DNA glycosylases consisting of thymine DNA glycosylase (TDG) and methyl CpG binding protein 4 (MBD4). Other DNA glycosylases in mammalian cells include endonuclease III-like protein 1 (NTH1), 8oxoguanine DNA N-glycosylase (OGG1), endonuclease 8-like proteins (NEIL1–3) and apurinic/apyrimidinic endonuclease 1 (APE1), which collectively remove a broad spectrum of oxidatively induced lesions in single and double stranded DNA. To avoid mutations from 8-oxoGua formation, cells contain three other enzymes, which include hOGG1 for the removal of 8-oxoGua:C, mutY DNA glycosylase (MUTHY), which excises premutagenic 8-oxoGua:A mispairs, and human mutT protein (MUT1), which cleanses the cellular pool of 8-oxodG triphosphate. APE1 efficiently converts abasic sites to single strand breaks and also incises a number of pyrimidine modifications though a subpathway known as nucleotide incision. In contrast, the excision of purine 50 ,8-cyclo-20 -deoxyribonucleosides in which the C50 of the sugar is covalently attached to the C8 of either adenine or guanine are repaired by NER. The predominate occurrence of C to T transition mutations has largely been attributed to cytosine and 5-methylcytosine deamination or oxidation. Although thermally induced deamination is widely assumed to occur, it should be noted that ROS or RNS-induced reactions with pyrimidines lead to an enormous acceleration of the rate of deamination at the exocyclic amino group. The lifetimes of intermediate oxidation products of C and 5-mCyt, such as the corresponding 5,6-glycols and hydantoin derivatives, may explain the propensity of C to T transition mutations and the overwhelming C to T mutations at CpG dinucleotides in cancer. Once deamination occurs, the resulting modifications resemble uracil or thymine in their base pairing properties, and thus, they will mispair during replication and generate a C to T transition mutation.

Ten-Eleven Translocation Enzymatic Oxidation of 5-Methylcytosine and Epigenetics in Cancer The methylation of DNA, mediated by DNA methyltransferases (DNMTs), converts cytosine to 5-methycytosine (5-mCyt), predominantly at CpG dinucleotides such that 60%–90% of these sites are methylated in mammals. Cytosine methylation is a well-known

Carcinogenesis: Role of Reactive Oxygen and Nitrogen Species

305

epigenetic mark that generally silences the expression of genes during development and plays a regulatory role in maintaining genome integrity and the ability to modulate gene expression. In the past 8 years, a novel epigenetic mark of DNA has emerged that arises from the oxidation of 5-mCyt to 5-hydroxymethylcytosine (5-HmCyt) by ten-eleven translocation family of enzymes (TET1–3). The level of 5-HmCyt accumulates in the genome to levels as high as 20% of 5-mCyt. They are the highest in embryonic stem cells and as well in differentiated neurons that maintain a high degree of plasticity. Based on new sequencing methods that discriminate between 5-mCyt and 5-HmCyt, the distribution of 5-HmCyt in the genome is unique with respect to 5-mCyt and 5-HmCyt populated sites possess different binding partners compared to 5-mCyt. In contrast to 5-mCyt, 5-HmCyt appears to elevate gene expression with the levels of 5-HmCyt positively correlated with greater expression of hydroxymethylated genes. TET enzymes contain Fe2 þ, require a-ketoglutarate as a cosubstrate, and have the ability to oxidize not only 5-mCyt to 5-HmCyt but also 5-HmCyt into further oxidized forms: 5-formylcytosine (5-FoCyt) and 5-carboxycytosine (5-CaCyt). The levels of 5-FoCyt and 5-CaCyt in the genome, however, are one to two orders of magnitude lower than those of 5-HmCyt. The active removal of 5-FoCyt and 5-CaCyt by BER involving TDG keeps the levels of the latter two modifications low and completes the multistep conversion of 5-mCyt to cytosine. Other enzymes from BER, which include activation-induced deaminase (AID) apolipoprotein B mRNA editing enzyme catalytic polypeptide 1 (APOBEC), SMUG, MBD2, have also been implicated in 5-mCyt demethylation. The cycle of cytosine methylation and demethylation of cytosine may explain the ability of pluripotent cells to turn genes off and on as necessary for their function and survival. The level of 5-HmCyt in the DNA of cancer cells is generally much lower than that observed in normal cells in brain, melanoma, breast and various other types of cancer. The lack of 5-HmCyt in the genome of cancer tissues has been associated with loss of TET activity. This may occur by the presence of mutations within TET genes, isocitrate dehydrogenase (IDH) genes, or other genes. Mutations in IDH switch activity from wild-type activity in which isocitrate is transformed into a-ketoglutarate to the production of 2-hydroxyglutarate, an inhibitor of TET. Interestingly, vitamin C increases the global level of 5-HmCyt in cells in culture, underlining the importance of a reducing agent to maintain the iron center of TET in a reduced state for optimal activity. Low concentrations of oxygen observed in hypoxic solid tumors may also decrease TET activity. Lastly, a reduction of 5-HmCyt in the genome may be due to passive dilution as a result of the high proliferative state of cancer cells. Despite extensive work, the role of TET and 5-HmCyt, and their interplay with DNMT enzymes and 5-mCyt, remain central questions in cancer development and growth.

Prospective Vision There is increasing evidence today that oxidative stress, oxidatively induced DNA damage, deficiencies in DNA repair are intimately entrenched in the process of carcinogenesis and tumorigenesis. In particular, inflammation and expression of inflammatory cytokines lead to escalation of oxidative stress and the release of ROS and RNS. It is clear that mutations accumulate in cancer development and it is necessary to achieve a required number of mutations and mutant genes to achieve the transformation from normal to cancerous cells. Further structural analyses of damage has led to the characterization of many novel lesions in the past 10 years, including tandem base lesions, clustered lesions, DNA–DNA crosslinks, DNA–protein crosslinks and putative secondary purine oxidation products. The study of formation of intermediate and stable products in double stranded DNA and higher order DNA structures such as telomeres and nucleosomes has given a wealth of new information. However, the analysis in cellular DNA and the biological processing of many lesions remains incomplete. Sequencing techniques have greatly advanced our knowledge concerning the site of DNA damage and particularly toward the distribution of epigenetic marks in the genome. A major challenge today remains to home in on the location and distribution of various oxidatively generated modifications in the genome. It will be necessary to determine the effect of sequence on the formation of damage and biological processing to establish the cause of mutation signatures and align these mutations within passenger and driver genes involved in cancer. It will also be necessary to explain how epigenetic marks become profoundly altered in cancer. The development of single and multiple gene ablation and editing methods in cells and mice continues to shed much light on the biological importance of DNA damage and repair pathways in cancer.

See also: CarcinogendDNA Adducts. Cancer-Related Inflammation in Tumor Progression. Helicobacter pylori-Mediated Carcinogenesis.

Further Reading Bartsch, H., 2000. Studies on biomarkers in cancer etiology and prevention. A summary and challenge of 20 years of interdisciplinary research. Mutation Research 462, 255–279. Bochtler, M., Kolano, A., Xu, G.L., 2017. DNA demethylation pathways: Additional players and regulators. BioEssays 39, 1–13. Boiteux, S., Coste, F., Castaing, B., 2017. Repair of 8-oxo-7,8-dihydroguanine in prokaryotic and eukaryotic cells: Properties and biological roles of the Fpg and OGG1 DNA N-glycosylases. Free Radical Biology and Medicine 107, 179–201. Cadet, J., Wagner, J.R., 2013. DNA base damage by reactive oxygen species, oxidizing agents and UV radiation. Cold Spring Harbor Perspectives in Biology 5, a01559. Cadet, J., Douki, T., Ravanat, J.-L., 2015. Oxidatively generated damage to cellular DNA by UVB and UVA radiation. Photochemistry and Photobiology 91, 140–155. Cadet, J., Davies, K.J.A., Medeiros, M.H.G., Di Mascio, P., Wagner, J.R., 2017. Formation and repair of oxidatively generated damage in cellular DNA. Free Radical Biology and Medicine 107, 13–34. Drohat, A.C., Coey, C.T., 2016. Role of base excision “repair” enzymes in erasing epigenetic marks from DNA. Chemical Reviews 116, 12711–12729.

306

Carcinogenesis: Role of Reactive Oxygen and Nitrogen Species

Knutson, C.G., Mangerich, A., Zeng, Y., Raczynski, A.R., Liberman, R.G., Kang, P., Ye, W., Prestwich, E.G., Lu, K., Wishnok, J.S., Korzenik, J.R., Wogan, G.N., Fox, J.G., Dedon, P.C., Tannenbaum, S.R., 2013. Chemical and cytokine features of innate immunity characterize serum and tissue profiles in inflammatory bowel disease. Proceedings of the National Academy of Sciences of the United States of America 110, E2332–E2341. Kuraoka, I., Bender, C., Romieu, A., Cadet, J., Wood, R.D., Lindahl, T., 2000. Removal of oxygen free-radical-induced 50 ,8-purine cyclodeoxynucleosides from DNA by the nucleotide excision-repair pathway in human cells. Proceedings of the National Academy of Sciences of the United States of America 97, 3832–3837. Lee, A.J., Wallace, S.S., 2017. Hide and seek: How do DNA glycosylases locate oxidatively damaged DNA bases amidst a sea of undamaged bases? Free Radical Biology and Medicine 107, 170–178. Lindahl, T., 2016. The intrinsic fragility of DNA (Nobel lecture). Angewandte Chemie International Edition 55, 8528–8534. Lu, X., Zhao, B.S., He, C., 2015. TET family proteins: Oxidation activity, interacting molecules, and functions in diseases. Chemical Reviews 115, 2225–2239. Marnett, L.J., 2000. Oxyradicals and DNA damage. Carcinogenesis 21, 361–370. Menoni, H., Di Mascio, P., Cadet, J., Dimitrov, S., Angelov, D., 2017. Chromatin associated mechanisms in base excision repairdNucleosome remodeling and DNA transcription, two key players. Free Radical Biology and Medicine 107, 159–169. Nickoloff, J.A., Jones, D., Lee, S.-H., Williamson, E.A., Hromas, R., 2017. Drugging the cancers addicted to DNA repair. Journal of the National Cancer Institute 109 djx059. Pfeifer, G.P., Kadam, S., Jin, S.G., 2013. 5-Hydroxymethylcytosine and its potential roles in development and cancer. Epigenetics & Chromatin 6, 10.

Cell Adhesion During Tumorigenesis and Metastasis Lubor Borsig, University of Zurich and Zurich Center for Integrative Human Physiology, Zurich, Switzerland Heinz La¨ubli, University Hospital Basel, Basel, Switzerland © 2019 Elsevier Inc. All rights reserved.

Glossary Angiogenesis Is a physiological process of blood vessel formation from the preexisting vessels. Neoangiogenesis describes the formation of blood vessels in the tumor tissue. Epithelial-to-mesenchymal transition (EMT) Describes a process during cancer progression wherein carcinoma cells can change their epithelial phenotype into a mesenchymal phenotype. Mesenchymal cells are characterized by enhanced migratory and invasive properties required for metastasis. Extracellular matrix (ECM) Is a collection of extracellular molecules secreted by cells that provides structural and biochemical support within a tissue. The animal ECM includes the interstitial matrix filling out the space between cells and basement membrane, which is a structured ECM found in the epithelial tissues. Extravasation Is a process describing the migration of tumor cells or leukocytes out of the blood vessel during hematogenous metastasis or immune response. Glycan Is carbohydrate structure composed of many monosaccharides linked via glycosidic bonds. This term is used to refer to a carbohydrate portion of a glycoconjugate, such as glycoprotein or glycolipid. Hematogenous metastasis Is spread of cancer cells from the primary tumors through blood circulation. Homophylic/heterophylic interactions Homophylic interactions describe cell-cell interactions through the same adhesion molecule from the adjacent cells (e.g., cadherins). Heterophylic interactions describe binding of different proteins or glycans (e.g., integrins, selectins). Lectins Are carbohydrate-binding proteins, macromolecules that are highly specific for sugar moieties. Myeloid cells Are leukocytes of the innate immune system. To myeloid-derived cells belongs: granulocytes, monocytes, macrophages and immature myeloid precursors that are often detected during cancer progression and associated with suppression of anti-tumor immunity. Thromboembolism Is the development of blood clots within the vascular system and dissemination of such clots (embolism). Patients with cancer are prone to develop thromboembolic complications. Tumor embolus Embolus of tumor cells is composed of blood constituents including platelets and leukocytes and is known to promote hematogenous metastasis. Tumor microenvironment Is the cellular environment of a tumor, which is composed of blood vessel, immune cells, fibroblast, soluble signaling molecules and ECM. Tumor cells modulate the microenvironment by various cues that promote angiogenesis, immune suppression and thereby promote tumor growth and metastasis.

Introduction Cell–cell adhesion determines the polarity and the physiological function of cells within tissues. On every cell, adhesion molecules facilitate interactions within the cell microenvironment that consist of other cells and the extracellular matrix. Since cell adhesion receptors are connected to signal-transduction pathways, these cell–cell and cell–matrix interactions modulate cell phenotype, survival, differentiation, and migration. As a consequence, alterations in cell adhesion directly contribute to tumorigenesis and metastasis. In a developing tumor, there is general loss of stabilizing cell adhesions between cells enabling them to gain the migratory phenotype required for invasiveness and metastasis. On contrary, the invasive capacity of metastatic tumor cells depends on efficient cell adhesion during transit in the circulation, colonization of distant organs and interactions with the extracellular matrix at secondary metastatic sites. The metastatic dissemination of tumor cells is the primary cause of death in cancer patients. Therefore, the understanding of the cell adhesion mechanisms in cancer will enable to design new targeted therapies. Finally, cell adhesion plays a major role for the immune response to cancer and could be modulated to improve cancer immunotherapy. During tumorigenesis and metastasis several families of adhesion molecules including: cadherins, integrins, junctional-adhesion molecules, and selectins have been studied. This chapter covers the current understanding of cell adhesion mechanisms in cancer and describes the known pathophysiological functions associated with tumor progression, angiogenesis and modulation of immune responses. In solid tumors, mostly originating from epithelial tissues, the loss of cell polarity is a profound change associated with a malignant transformation resulting in a ubiquitous presence of adhesion molecules (Fig. 1). Changes in expression of cell adhesion receptors and their binding partners (ligands) define both the malignant progression and interactions within the tumor microenvironment.

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.64991-7

307

308

Cell Adhesion During Tumorigenesis and Metastasis

(A)

normal epithelium

BM

Blood Vessel

(B)

malignant epithelium BM

Blood Vessel

mucin carcinoma mucin integrin E-cadherin mesenchymal cadherins

Fig. 1 Cell adhesion changes leading to metastatic tumor cells during transformation from a normal to a malignant epithelium. (A) Normal epithelium consists of polarized epithelial cells that are bound together through E-cadherin adhesions. The epithelium is organized on the basement membrane (BM) consisting of various extracellular matrix proteins (e.g., type IV collagen, glycosaminoglycans, laminin, and small adaptor proteins). On a basolateral site of the epithelium integrins facilitates the interaction with BM. On the apical site, epithelial cells produce mucus that is composed of mucins, proteins containing a large amount of O-glycans. This glycosylated mucus creates a mechanical barrier between epithelial cells and the lumen of a tissue. (B) Malignant epithelium. The profound change during malignancy is associated with: loss of cell polarity, loss of E-cadherin-mediated epithelial adhesions; increased production of mucus carrying altered glycan structures, degradation of BM; and increased migration and invasiveness of the malignant cells. Malignant tumor cells with altered cell adhesion preferences are able to invade into blood vessels and thereby gain access to other organs in an organism.

Cadherin-Mediated Cell Adhesion Cadherins are cell adhesion molecules that play a major role in cell–cell interactions, tissue patterning during development and in cancer, particularly during tumor cell dissemination. E-cadherin is the most prominent cadherin expressed in epithelial cells (Fig. 1). Cadherins comprise of a large family of proteins with over 110 known members in humans with multiple functions. In a context of cancer, E-cadherin acts as a tumor suppressor. E-cadherin expression inhibits tumor progression by controlling epithelial to mesenchymal transition (EMT) that is associated with specific signaling pathways. Cadherins are a family of transmembrane glycoproteins with at least five subfamilies. Type I cadherins, so-called classical cadherins, includes also E-cadherin. While type II cadherins are closely related to type I cadherins, there are additional subfamilies with specific structures. Type I and II cadherins, have five extracellular cadherin domains, a single transmembrane domain and a highly conserved intracellular domain that interacts with the intracellular adaptor proteins. E-cadherin mediates homophilic calcium-dependent interactions between epithelial cells that affect intracellular signaling cascades. E-cadherin-based cell adhesions regulate multiple signaling pathways that are linked to intracellular recruitment of catenins and other intracellular signaling molecules. For instance, E-cadherin alters the Wnt-signaling pathway that is important for development, tissue homeostasis and cancer progression. In mature epithelial cells, b-catenin is bound to E-cadherin in the adherent junctions. Free cellular b-catenin is sequestered in the cell through binding to adenomatous polyposis coli (APC)-Axin complex, and is immediately phosphorylated through glycogen synthase kinase 3b (GSK3b). Phosphorylated b-catenin is further ubiquitinylated and destined for degradation in the proteosomes. Upon external activation of the Wnt pathway, which blocks GSK3b activity, b-catenin is not degraded and accumulates in the cytoplasm eventually reaching the nucleus, where it induces transcription program. The nuclear signaling

Cell Adhesion During Tumorigenesis and Metastasis

309

pathway of b-catenin has a major role in maintaining and promoting mesenchymal and stem cell-like phenotypes that are essential in cancer progression and metastasis.

Cadherins Modulate Cancer Progression E-cadherin is a major component of the adherens junctions of epithelial layers. The epithelial phenotype and the polarization of an epithelial tissue is dependent on E-cadherin expression. E-cadherin is considered as a key regulator of epithelial phenotype, since it is anchored in actin-myosin cytoskeleton and thereby contributes to stabilization of the epithelial architecture. During tumorigenesis, the loss of epithelial cell polarization (Fig. 1) is associated with reduced expression of E-cadherin coded by the CDH1 gene. Mutation in the CDH1 gene can lead to hereditary gastric cancer. E-cadherin loss has been found in different other cancers including hepatocellular carcinoma, squamous cell carcinoma of the skin, head and neck and esophagus, as well as in melanoma. In these cancers, E-cadherin is usually lost by epigenetic regulation via promoter methylation. A major pathway involved in the tumor suppressor function of E-cadherin is mediated by the interaction of the intracellular domain with b-catenin. E-cadherin binding to b-catenin inhibits the pro-tumorigenic b-catenin signaling. Downregulation or depletion of E-cadherin in experimental models leads to undifferentiated, highly invasive tumor cells. E-cadherin expression is usually lost in cells that undergo EMT while the expression of other mesenchymal cadherins, including N-cadherin (CDH2), R-cadherin (CDH4) or P-cadherin (CDH3) is upregulated. These changes are known as “cadherin switch” and are regulated by several transcription factors linked to EMT such as SNAIL, SLUG, TWIST, ZEB1, and ZeB2. EMT is considered as major prerequisite for many cancers to gain an invasive phenotype and to become metastatic. The p120 catenin is an important mediator of downstream signaling linked to the cadherin switch. EMT also induces a program in epithelial cells that can generate cancer stem cells and promote thereby metastatic progression. Binding of p120 to E-cadherin suppresses tumor progression while its association with mesenchymal cadherins, such as P-cadherins, mediates interactions with Rac1 leading to promotion of tumor cell invasiveness. In addition, N-cadherin promotes anchorage-independent growth and influences growth receptor signaling and the overexpression of P-cadherin leads to an increase of MMP secretion and enhanced tumor cell invasion. CDH2 or neuronal cadherin (N-cadherin) also has autonomous functions apart from cell adhesion. N-cadherin stimulates fibroblast growth factor receptor1 (FGFR1) and thereby promotes malignancy. Moreover, N-cadherin is also expressed on stromal cells and overexpression on tumor cells can mediate invasion and metastasis. Based on this knowledge, N-cadherin is an interesting target for cancer therapy. CDH3 or placental cadherin (P-cadherin) exercise a context-dependent role in cancer. P-cadherin upregulation during a cadherin switch has been observed in some cancers. In epithelial ovarian cancer, gonadotropin-releasing hormone receptors through interactions with p120-catenin leads to increased P-cadherin expression and tumor growth and cancer progression. While the type I cadherins E-cadherin, N-cadherin, P-cadherin and R-cadherin (CDH4) are associated with cancer progression, type II cadherins can also influence cancer progression. In addition, vascular endothelial cadherin (VE-cadherin, CDH5) is a major cadherin in endothelial cells. VE-cadherin is important for homophilic interactions between endothelial cells and tumor cells and this interaction potentiates tumor cell intravasation and metastasis. Desmosomal cadherins, such as desmoglein and desmocollins, have tumor suppressive function through modulation of oncogenic Wnt-b-catenin signaling. Protocadherins differ from cadherins by other domains that mediate homophilic interactions and different intracelullar domains. In many cases, intracellular interaction partners are not well described, but the expression of protocadherins is often tumor suppressive.

Integrin-Mediated Adhesions Integrins represents a family of cell adhesion molecules, which control tumor cell proliferation, migration and survival. Integrins directly binds components of extracellular matrix (ECM), and thereby provide traction necessary for cell motility and invasiveness. During cancer progression, integrins contribute to ECM remodeling by controlling the localization and activity of proteases. In addition, other integrin-expressing cells present either at the primary tumor site or at metastatic sites can contribute to the malignant process. Integrins are heterodimeric cell surface receptors that mediate cell adhesion to ECM proteins or to other cells through binding to immunoglobulin superfamily cell adhesion molecules (IgCAMs). Two major groups of receptors belong to IgCAMs: vascular cell adhesion molecule 1 (VCAM1), and intracellular cell adhesion molecules (ICAMs). More than 24 different integrin heterodimers exist, which are composed of a- and b-subunit and bind to different counter-receptors. There are at least 18 a- and 8 b-integrins identified, that combine many unique a/b integrin receptors, with a specific preference for distinct ligands. The presence of a particular integrin on a cell surface defines the adhesion and migratory properties of the cell in dependence on the adjacent ECM (e.g., fibronectin, vitronectin, laminin, or collagen). The first binding site for integrins that has been identified is defined by a specific tripeptide RGD (arginine-glycine-aspartic acid). This motif is recognized by a5b1-integrin and av-containing integrins. Different probes carrying the RGD epitope has been tested for integrin targeting in several pharmacological trials. Upon adhesion to ECM, integrins cluster in the cell membrane and form structures called focal adhesions, which further recruit signaling and adaptor proteins. Integrins facilitate through the adaptor proteins the coupling of ECM to the cytoskeleton and allow physical attachment, which affect the cell shape and migratory properties of the cells. The integrin-mediated recruitment and

310

Cell Adhesion During Tumorigenesis and Metastasis

activation of kinases, such as focal adhesion kinases (FAKs), Src family kinases or GTPases of the Ras/Rho family induces signal transduction, known as “outside-in” signaling. Conversely, intracellular signals (e.g., growth factors or cytokines) can induce alterations in the integrin conformation and thereby modulate the ligand-binding affinity of integrins to ECM that is known as an “inside–out” signaling. In addition to the well-established role of integrins during migration and invasion, integrins also regulate cell proliferation, cell survival and angiogenesis, all of which contribute to cancer progression. Integrins on the cell surface constantly interrogate their microenvironment through their capacity to bind to ECM. Ligated integrins convey cell survival, while unligated integrins may trigger pro-apoptotic pathways. The enhanced cell survival triggered by ligated integrins is mediated by several mechanisms, including expression of Bcl-2 or activation of PI3K-Akt or NF-kappaB signaling. This process is further dependent on a communication between integrin-growth factor pathways. On contrary, unligated integrins may trigger activation of Caspase-8 and initiate apoptosis, which is no longer blocked by the Bcl-2. Thus, integrins can paradoxically initiate either prosurvival or pro-apoptotic pathways that is dependent on integrin engagement.

Integrins Adhesion in Cancer and During Metastasis A wide variety of integrins on tumor cells are linked to cancer progression. Most solid tumors of the epithelial origin express integrins on a cell surface (Fig. 1). However, integrin expression may strongly vary between normal epithelial cells and malignant carcinoma cells. Particularly, avb3 and avb6 integrins are highly expressed in some tumors, while their presence in epithelial cells is negligible. The correlation between integrin expression and cancer patient’s survival was evaluated in many cancer types as discussed elsewhere (see further reading). For example, avb3 integrin expression correlates with a poor prognosis in many cancers including breast, colon, prostate, pancreatic, and ovarian carcinomas. Tumor cell seeding to a “metastatic niche” in a specific tissue is at least partially dependent on interactions between tumor cells and the available ECM. The unique binding specificities of integrins may therefore dictate the ability of tumor cells to colonize distinct tissues, and thus serve as a critical homing factor promoting metastasis. The metastatic susceptibility of certain tissue has been linked to a presence of specific ECM proteins at these sites. For example, tenascin C is a ligand for b1 and b3 integrins, and is present in the metastatic niche of the lungs, driving breast metastasis to this organ. Laminin is another ECM protein present in the lungs that is recognized by a6b4 or a6b1 integrins expressed both on tumor cells and platelets. avb3 of a4b1 integrins expressed on tumor cells play key roles in bone metastasis, where they enable tumor cell adhesion to ECM proteins such as osteopontin or type I collagen. The enhanced expression of a4b1 integrins on melanoma cells enables binding to VCAM1 on activated lymphatic endothelial cells thereby promoting lymph node metastasis. While the experimental evidence links integrin expression to tissue-specific metastatic seeding of tumor cells, the question how tumor cells reach ECM, which is normally not accessible from the lumen of blood vessels, remains open.

Integrins From the Tumor Microenvironment Promote Cancer Progression Tumor microenvironment consists of a variety of host cells, where the bone marrow-derived myeloid cells represent the major population of cells virtually in any cancer type. The expression of a4b1 integrin (also known as VLA-4) on bone marrow-derived cells enables their recruitment to the tumor microenvironment. The endothelial remodeling in growing tumor induces the expression of VCAM and cellular fibronectin that facilitates integrin-based adhesion of bone marrow-derived myeloid cells. In addition, VLA-4driven recruitment of monocytic cells to the activated endothelium, commonly observed in growing tumor and at metastatic sites, promotes angiogenesis. A positive role of platelets in tumor metastasis has been already defined. Apart from the initial P-selectindependent interactions, aIIbb3 integrin on platelets was shown to facilitate tumor cell-platelet interactions. Mechanistically, ECM proteins, such as fibrinogen serve as a bridge for platelet aIIbb3 integrin dependent adhesion and tumor avb3 integrin mediated interaction. This interaction contributes to arrest of tumor cell emboli leading to metastasis. Tumor-activated activated endothelial cells express avb3 integrin, which together with basic fibroblast growth factor (bFGF) or tumor necrosis factor a (TNF-a) stimulates angiogenesis. However, the expression of avb3 on activated endothelial cells is not unique to tumors, but has been detected in different pathophysiological situations. Finally, tumor-activated endothelium expresses VCAM1, ICAMs, which served as ligands for the integrin-mediated leukocyte recruitment (Fig. 2). The dominant members of leukocyte integrin families, facilitating leukocyte recruitment leading to their extravasation are the b2 integrins, e.g., aLb2 ¼ CD11a/CD18 or aMb2 ¼ CD11b/CD18, and b1 integrins, e.g., a4b1 ¼ VLA-4. In addition, to integrin-mediated firm adhesion of leukocytes to VCAM1 and ICAMs, a direct interaction between aLb2 or aMb2 on endothelial cells directly promotes leukocyte extravasation. Analogous to growth factor receptors, clustering of integrins in the plasma membrane directly regulate the amplification of signal transduction. In fact, there is accumulating evidence for a cooperative signaling between growth factor/cytokine receptors and integrins. This process is further regulated by other binding partners such as galectins. Galectins is a family of b-galactoside-binding lectins, which are expressed by tumor cells in soluble forms. Galectins influence tumor cell behavior through binding to carbohydrates on the cell surface receptors, such as integrins, and thereby promote their clustering. Galectin-1 expression in low metastatic lung carcinoma cells induced their migration, invasiveness and metastasis, through integrin activation. In addition, Galectin-3 induces b3 integrin-mediated tumor growth and drug resistance. Mechanistically, avb3 integrins on tumor cells, without binding to a ligand, recruit the oncogene KRAS to the tumor cell plasma and activate downstream signaling such as NF-kB pathway and thereby promote tumor growth. Thus, integrin avb3 was shown to be both necessary and sufficient for tumor initiation and resistance to receptor tyrosine kinase inhibitors in pancreatic carcinomas.

Cell Adhesion During Tumorigenesis and Metastasis

311

endothelial cell

platelet granulocyte tumor cell

monocyte

tumor cell

endothelial cell barrier opening

Fig. 2 Cell adhesion facilitates tumor cell intravascular survival and tumor cell extravasation. In the circulation, tumor cells interact with blood constituents (e.g., platelets, granulocytes, monocytes, and blood vessel wall ¼ endothelial cells), which is primarily mediated by adhesion of selectins to altered glycan structures (carcinoma mucins) presented on the tumor cell surface. Platelet-tumor cell interactions are directly facilitated by P-selectin, and indirectly through integrin binding both from platelet and tumor cells to extracellular components (e.g., fibronectin). Platelets “coating” of tumor cells protect them from immune responses, and contribute to higher tumor cell mobility of tumor cells. Tumor cell interactions with leukocytes are both L-selectin- or integrin-mediated. Tumor cell-endothelial cells adhesion may trigger endothelial activation through E-selectin or immunoglobulin superfamily cell adhesion molecules (IgCAMs) such as ICAM or VCAM that are recognized by integrins from tumor cells. The resulting increase in vascular permeability is linked to specific cytoskeletal rearrangements in the endothelial cells and loosening of endothelial cell junctions. Although tumor cells may individually migrate through the endothelium, trans-endothelial migration of tumor cells is significantly potentiated by monocytes. In addition, interactions between tumor and endothelial cells through junction adhesion molecules (JAMs) further modulate tumor cell extravasation.

In conclusion, it is important to emphasize that individual integrins may affect very different aspects of cancer progression. While unligated avb3 integrin promotes tumor cell stem-like properties, its ligation to ECM drives cell invasiveness and proliferations. Thus, the role of integrins in cancer biology is dependent on particular cues present in a tumor microenvironment.

Selectin-Mediated Cell Adhesion Selectins are vascular cell adhesion receptors that bind to glycans presented on cell surface proteins. The physiological function of selectins is to facilitate the initial tethering of leukocytes at inflammatory sites or secondary lymphoid organs before leukocyte adhesion and extravasation. Three different selectins (P-, E-, and L-selectin) can be found in mammals, which are conserved between species. Selectins are C-type lectins that bind preferentially to glycans with a terminal tetrasaccharide structure of sialyl-Lewis x (sLex) and sialyl-Lewis a (sLea) (Fig. 3), which are usually further modified with a sulfate group. The biosynthesis of sLex and sLea ligands is orchestrated by a set of glycosyltranferases, where the expression of sialyltransferases and fucosyltransferases is essential. In addition, selectins also bind to certain glycolipids (sulfatides). Heterophilic binding of selectins to glycans is usually calcium dependent. Structurally, all selectins possess an N-terminal C-type carbohydrate recognition domain (C-type lectin), an EGF domain, short consensus repeats, a transmembrane and an intracellular domain. The engagement of selectins by their ligands activates outside-in signaling through e.g., Ca2 þ and p38 MAP kinase pathway. P-selectin is stored in granules of platelets and endothelial cells, which bring P-selectin on the surface upon activation. E-selectin is present only on endothelial cells and its expression is regulated on a transcription level. L-selectin is constitutively expressed on

Fig. 3 Structures of sialyl Lewis x (sLex) and sialyl Lewis a (sLea). Selectins bind to glycans with a terminal tetrasaccharide structure known as sialyl Lewis x/a. The core of the terminal tetrasaccharide is a disaccharide of galactose linked to N-acetylglucosamine (GlcNAc), either in a1,4 (sLex) or a1,3 (sLea) linkage. This disaccharide is further modified by a2,3sialyltransferases linking the sialic acid to the galactose. The fucosyltransferases completes the tetrasaccharide by the addition of fucose to GlcNAc either in a1,3 linkage forming sLex or a1,4 linkage forming sLea. Depending on a selectin type, this structure is further sulfated either on 6-position of the Gal or 6-position of the GlcNAc. Alternatively, the protein backbone of the selectin ligand carrier is sulfated in a close proximity to the sLex structure.

312

Cell Adhesion During Tumorigenesis and Metastasis

most subsets of leukocytes, but lost upon antigen-mediated T cell memory formation. Selectins bind to properly posttranslationally modified glycan ligands, carrying mostly sLex or sLea structures. The most studied selectin ligand is P-selectin Glycoprotein Ligand 1 (PSGL-1), where the sLex is further sulfated. All three selectins bind to PSGL-1 that is mostly expressed in leukocytes. In addition, Eselectin binds to a variety of glycosylated protein ligands such as E-selectin Ligand-1 (ESL-1) or CD44. L-selectin binds to a group of glycans called addressins that includes: MADCAM, PNADs like CD34, and podocalyxin. These glycans are present in high endothelial venules of lymph nodes or Peyer’s patches, where they enable lymphocyte trafficking to secondary lymphoid organs. Moreover, P- and L-selectins also bind to glycosaminoglycans, like heparan sulfate and heparin, but the nature of this cell adhesion remains under investigation. On a functional level, selectins mediate the first steps of extravasation. Endothelial cells line up the blood vessels and circulating leukocytes do not interact with the vessel wall. Inflammatory stimuli activate endothelial cells that upregulate expression of adhesion molecules including P-selectin and E-selectin. Interactions of vascular selectins with their ligands lead to the leukocyte rolling on activated endothelium, followed by integrin-mediated firm adhesion and finally leukocyte extravasation into the parenchyma, where the effector functions of leukocytes can be executed. Selectin-mediated cell adhesion modulates two different aspects during cancer progression and metastasis: (A) tumor cell interactions with endothelial cells, platelets and leukocytes; (B) leukocyte recruitment and trafficking to developing tumors (Fig. 2).

Selectin-mediated tumor cell interactions Malignant transformation is associated with profound changes in cell surface glycosylation of tumor cells and this leads to upregulation of ligands for selectins. These changes in glycosylation are the result of epigenetic and genetic changes of glycosyltransferases and glycosidases. While changes can lead to increased branching of N-glycans, the prominent glycan alterations occur on O-linked glycans which are present on mucin producing tumor cells. Importantly, the upregulation of sialic-acid containing glycans on tumor cells together with an increased expression of truncated glycan structures has been detected across all cancer types. The upregulation of sialic acid-containing glycans is a result of alterations in expression of sialyltransferases or sialidases. Such sialylated glycans e.g., on mucins serve as ligands for selectins and enable interactions between tumor cells and selectin-carrying cells. Thus, reciprocal interactions between platelets, leukocytes, endothelial cells and tumor cells have been associated with cancer progression. While many of these cell–cell-interactions have been demonstrated in various assays in vitro, their relevance in cancer progression has been confirmed in mouse models using selectin-deficient mice. Importantly, enhanced expression of sLex and sLea structures is frequently associated with cancer progression and cancer patient’s poor prognosis in various tumors including colon, gastric, lung; renal and breast cancers; melanomas and others. Metastasis is a multistep process which begins with dissemination of tumor cells through the blood circulation and leads to colonization of a distant organ. Selectin-mediated interactions contribute to hematogenous dissemination and organ colonization. During hematogenous metastasis platelets bind to tumor cells and this platelet-“shield” protects tumor cells from elimination by NK cells. A strong reduction of metastasis was observed in mice lacking P-selectin, indicating that P-selectin could be a target for anti-metastatic therapy. Endothelial E-selectin has been shown to promote tumor cell adhesion and thus has been implicated in metastatic dissemination. A special glycoform of CD44 in human cells, named as HCELL, has been identified as an E-selectin ligand on tumor cells. In breast cancer cells, selectin-mediated interactions with HCELL on microvascular endothelial cells promotes homing of cancer stem cells to the bone marrow. Selectin-selectin ligand interactions can influence signaling both in tumor cells and in selectin-bearing cells; thereby modulating cancer progression. Cancer patients have a significantly increased risk to develop thromboembolic complications. This hypercoagulability first described by Trousseau is also to some part attributed to selectin-mediated interactions between cancer mucins, platelet Pselectin and L-selectin on leukocytes. Carcinoma-derived highly sialylated and glycosylated mucins enhance platelet interactions with myeloid leukocytes that promote thrombus formation. This finding might explain why heparin and heparin derivatives are potent inhibitors of thromboembolic events in cancer patients, when compared to other anticoagulants (e.g., Warfarin).

Selectin-mediated recruitment and trafficking of leukocytes in cancer Selectin-mediated interactions promote the leukocyte recruitment to tumors and enables their interactions within a tumor microenvironment. Increased leukocyte recruitment supports cancer progression through promotion of tumor cell invasion, intravasation, dissemination, and extravasation. In particular, selectins were shown to influence the recruitment of innate immune cells during metastatic colonization. Analysis of metastatic colonization in various mouse models have shown that selectin-mediated interactions is essential for the recruitment of myeloid cells and monocytes that enhances tumor cell extravasation and facilitate organ colonization. Increased production of chemokines (e.g., CCL2, CCL5, or CXCL12) within the metastatic microenvironment further promotes recruitment of monocytes and myeloid cells, which is further supported by selectin interactions. Interestingly, both endothelial E-selectin and the leukocyte L-selectin were shown to be implicated in the process of monocyte-assisted transendothelial migration of tumor cells in lungs. These studies show that metastatic tumor cells “highjack” the physiological function of selectins and leukocyte recruitment to promote tumor growth and metastasis. In the case of hematogenous malignancies, the tumor cells express not only selectin ligands, but also the L-selectin. The cell surface expression of L-selectin on leukemia cells can significantly influence their trafficking and distribution. For instance, in patients with chronic lymphocytic leukemia (CLL) tumor cells re-circulates to lymph nodes through L-selectin mediated rolling, which promotes cancer progression.

Cell Adhesion During Tumorigenesis and Metastasis

313

In conclusion, selectin-mediated interactions between tumor cells, platelets, endothelial cells and leukocytes are involved in various steps of metastasis and can facilitate the formation of a metastatic niche. Interestingly, glycosaminoglycans (GAGs) such as heparin and heparan sulfates bind to selectins and thereby inhibit interactions with tumor-associated ligands. Besides, glycomimetics of PSGL-1 can inhibit P-selectin mediated interactions and is currently tested in cancer therapy. The abundance of selectin ligands on virtually all cancer types, together with the overall access to vascular selectin during hematogenous metastasis strongly suggest that any interference in selectin-mediated cell adhesion in a temporal manner might have an antimetastatic potential also in clinical settings.

Other Cellular Adhesion During Cancer Progression There are other cell adhesion molecules associated with cancer progression, but their function is only partially elucidated. CD44 is a cell surface adhesion molecules, which was initially shown to be the receptor for an ECM glycosaminoglycan hyaluronan. Many cancer types express high levels of CD44 that upon cleavage from the cell surface serves as a marker for cancer progression. Besides the special glycoform HCELL found in cancer that can serve as ligands for selectins (see above), tumor-specific alternatively spliced CD44 has been described. The current evidence linked altered CD44 expression to cancer stem cell phenotype, while the intracellular domain is critically involved in this process. Junctional adhesion molecules (JAMs) are expressed on endothelial cells, where their homophilic interactions strengthen the endothelial barrier functions. During inflammation JAMs, together with endothelial ICAMs interact with aLb2 or aMb2 on leukocytes thereby promoting leukocyte extravasation through the retracted endothelium. Recent evidence suggests that JAM-c on melanomas actively contribute to lung metastasis though promotion of tumor cell extravasation through enhanced transendothelial migration of the tumor cells without affecting tumor cell adhesion (Fig. 2). Whether other tumor cells express JAMs or whether JAMs binding is relevant for metastasis to other organs remains to be defined.

Adhesion Molecules in Antitumor Immunity and Cancer-Associated Inflammation Recent advances in our understanding how the immune system recognizes and eliminates tumor cells and how tumor cells evade this immune control has led to the introduction of successful immunotherapies into the clinical routine. In particular, targeting of inhibitory receptors on T cells including PD-1 and CTLA-4 have revolutionized the treatment of patients with advanced cancers including melanoma, nonsmall cell lung cancer, kidney cancer or bladder cancer. While antitumor immunity has gained much interest, the virtually omnipresent cancer-associated inflammation associated with recruitment and polarization of myeloid-derived cells usually supports cancer progression by the generation of an immune-suppressive microenvironment. Dysfunction in the leukocyte recruitment with antitumor activity to tumors and increased recruitment of immune suppressive immune cells such as myeloid-derived suppressors, regulatory T cells or tumor-associated macrophages has been observed in many cancers. Cell adhesions facilitate leukocyte recruitment through selectin- and integrin-mediated interactions that influence also the adaptive immunity. Infiltration of CD8 positive T cells preferentially happens in the periphery of tumors, which is well studied in melanoma lesions. Moreover, in a murine melanoma model, endothelial cells downregulated ligands for the recruitment of cytotoxic T cells. PSGL-1 on T cells modulates T cell trafficking and the development of regulatory T cells therefore PSGL-1 represents an interesting drug target for cancer immunotherapy. T cells including regulatory T cells (Tregs) recirculate and control immune homeostasis including peripheral tolerance. Homing of naïve T cells to secondary lymphoid organs requires L-selectin binding to addressins on high endothelial venules. Selectin receptors as well as integrins have an important function in the recruitment and homing of Tregs to sites of inflammation and to tumors. Tregs play an important role in suppressing antitumor immune responses thus targeting of cell adhesion mechanisms leading to Tregs recruitment could be an interesting way to improve antitumor immune responses.

Prospective Vision Current evidence describes the role of cell adhesion in different processes associated with cancer progression, ranging from stemness of cancer cells; tumor growth survival, invasiveness, migration, and metastasis; to modulation of immune responses. While the biology of cell adhesion during cancer is rather complex, the function of certain integrins, cadherins and selectins has been extensively evaluated in several preclinical models. Nevertheless, no agent that specifically targets interactions of these molecules is currently used in clinical practice. Clearly, the complete identification of underlying mechanisms in the complexity of the tumor microenvironment is still ongoing and will hopefully help to develop therapeutic strategies. There are several studies testing targeted delivery of therapeutics to tumor sites through cell adhesion molecules. For example, tumor targeting by IL-2 cytokine-fusion proteins with RGD domains that bind to integrins have demonstrated promising results in mouse models together with checkpoint blockade. While timely-defined targeting of selectins resulted in reduced metastasis in many experimental models, selectins can also serve as targets for delivery of nanoparticles coated with selectin ligands. For example, P-selectin was targeted for cancer-directed drug-delivery with nanoparticles coated with a fucosylated polysaccharide. Interestingly, P-selectin expression could be initiated on endothelial cells by radiotherapy. This finding represents a very interesting concept; where drug targeting could be initiated by specific irradiation of a tumor lesion. In this context, new insights into function of cell-cell interactions, especially during

314

Cell Adhesion During Tumorigenesis and Metastasis

metastasis, will certainly open up new strategies to interfere in cancer progression. Interference in cell adhesion based processes during cancer lies in a combinational approach with other approaches e.g., direct tumor cell targeting or modulation of anticancer immunity.

See also: Metastatic SignaturesdThe Tell-Tale Signs of Metastasis.

Further Reading Desgrosellier, J.S., Cheresh, D.A., 2010. Integrins in cancer: Biological implications and therapeutic opportunities. Nature Reviews Cancer 10, 9–22. Hauselmann, I., Borsig, L., 2014. Altered tumor-cell glycosylation promotes metastasis. Frontiers in Oncology 4, 28. Jeanes, A., Gottardi, C.J., Yap, A.S., 2008. Cadherins and cancer: How does cadherin dysfunction promote tumor progression? Oncogene 27, 6920–6929. Kwan, B.H., Zhu, E.F., Tzeng, A., Sugito, H.R., Eltahir, A.A., Ma, B., Delaney, M.K., Murphy, P.A., Kauke, M.J., Angelini, A., Momin, N., Mehta, N.K., Maragh, A.M., Hynes, R.O., Dranoff, G., Cochran, J.R., Wittrup, K.D., 2017. Integrin-targeted cancer immunotherapy elicits protective adaptive immune responses. Journal of Experimental Medicine 214, 1679–1690. Läubli, H., Borsig, L., 2010. Selectins promote tumor metastasis. Seminars in Cancer Biology 20, 169–177. Obenauf, A.C., Massague, J., 2015. Surviving at a distance: Organ specific metastasis. Trends in Cancer 1, 76–91. Pearce, O.M., Laubli, H., 2016. Sialic acids in cancer biology and immunity. Glycobiology 26, 111–128. Pinho SS anf Reis CA, 2015. Glycosylation in cancer: Mechanisms and clinical implications. Nature Reviews Cancer 15, 540–555. Schneider, J.G., Amend, S.R., Weilbaecher, K.N., 2011. Integrins and bone metastasis: Integrating tumor cell and stromal cell interactions. Bone 48, 54–65. Shamay, Y., Elkabets, M., Li, H., Shah, J., Brook, S., Wang, F., Adler, K., Baut, E., Scaltriti, M., Jena, P.V., Gardner, E.E., Poirier, J.T., Rudin, C.M., Baselga, J., HaimovitzFriedman, A., Heller, D.A., 2016. P-selectin is a nanotherapeutic delivery target in the tumor microenvironment. Science Translational Medicine 8, 345ra387. van Roy, F., 2014. Beyond E-cadherin: Roles of other cadherin superfamily members in cancer. Nature Reviews Cancer 14, 121–134. Zarbock, A., Ley, K., McEver, R.P., Hidalgo, A., 2011. Leukocyte ligands for endothelial selectins: Specialized glycoconjugates that mediate rolling and signaling under flow. Blood 118, 6743–6751.

Cell Responses to DNA Damageq Jean YJ Wang, University of California San Diego, La Jolla, CA, United States © 2019 Elsevier Inc. All rights reserved.

Glossary AT Ataxia telangiectasia, which is an autosomal recessive genetic disease characterized by ataxia, immune deficiency, sterility, and mortality early in life, caused by loss-of-function mutation in the ATM gene BER Base excision repair, which is mechanism that repairs damaged bases in DNA, involving excision of damaged base by Nglycosylase excising the damaged base followed by endonuclease cleavage of DNA strand at the abasic site, followed by DNA polymerase filling and ligase sealing of the strand. BSE Bystander effect, which describes DNA damage-induced biological effects in cells that are not directly targeted by genotoxins, but are in the vicinity of directly damaged cells. DDR DNA damage response, which describes the compilation of cellular processes activated by lesions in genomic DNA. FA Fanconi anemia, which is an autosomal recessive genetic disease characterized by genome instability, defects in DNA repair, anemia and cancer susceptibility, caused by loss-of-function mutations in some 20 genes, collectively known as the FA genes. FHA Fork head associated, which is an evolutionarily conserved protein domain that specifically binds to a phospho-threonine epitope. MMR Mismatch repair, which is a mechanism that repairs mismatched base pairs and small insertions or deletions during DNA replication, involving protein complexes that recognize those lesions in newly synthesized daughter strand, followed by recruitment of nucleases to cleave the daughter stand, polymerases to fill the gap, and ligases to seal the strand. NER Nucleotide excision repair, which is a mechanism that repairs photo-adducts of bases in DNA, involving protein complexes that recognize those lesions in actively transcribed DNA (transcription coupled repair) or genomic DNA in general, followed by endonuclease cleavage of DNA strand surrounding the photo-adducts, DNA polymerase filling and ligase sealing. PIKK Phosphotidyl-inositol-3 (PI3K) kinase-related kinase, which is a protein kinase domain that is highly conserved in eukaryotic cells and is related to the lipid kinase PI3K. Upon DNA damage, three members of the PIKK family, namely ATM, ATR, and DNAPK, are activated by lesion recognition and sensing factors and these PIKKs function as master kinases to switch on DDR. XP Xeroderma pigmentosa, which is an autosomal recessive genetic disease characterized by UV sensitivity, defects in NER, developmental defects, and cancer susceptibility, caused by loss-of-function mutations in a dozen or so genes, collectively known as the XP genes.

Overview The study of DNA damage response (DDR) spans multiple disciplines and experimental approaches that defy a comprehensive review in one chapter. The goal of this chapter is therefore to provide a DDR framework that is amenable to expansion in biological directions by readers according to their interests. This DDR framework organizes factors and processes into three functional categories: (1) lesion recognition and damage sensing, (2) damage-signal transduction, and (3) downstream biological effectors (Fig. 1). DNA lesions can be generally divided into four major groups: base modifications (alkylation, oxidation), mismatched base pairs (non-Watson Crick base pairing, small insertions, or deletions), interstrand crosslink (ICL), and strand breaks (double-stranded or single-stranded breaks). These lesions are repaired by a number of DNA repair mechanisms, for example, base excision repair (BER), nucleotide excision repair (NER), mismatch repair (MMR), crosslink repair, nonhomologous end joining (NHEJ), and homologous recombination (HR). Each of these repair pathways is initiated by a lesion recognition factor, for example, the MSH2/MSH6 complex recognizes mismatched base pairs, Ku and MRN complexes recognize double-stranded breaks (DSB). The lesion recognition factors not only recruit DNA repair enzymes but also interact with signal transducers to activate additional DDR pathways (Fig. 1). Besides those factors that recognize specific DNA lesions, eukaryotic cells also possess an evolutionarily conserved “damage sensor,” that is the 9-1-1 hetero-trimeric DNA clamp. The 9-1-1 DNA clamp is loaded unto stalled replication forks and other damage sites where structures that resemble stalled replication forks are formed. The dual roles for lesion-recognition proteins to stimulate DNA repair and to activate damage-signal transduction ensures that DDR is terminated when lesions are repaired. With this design, the rate of DNA repair becomes an important determinant of the

q

Change History: June 2017. Jean YJ Wang modified text and Figure legends. JYJ Wang added the Glossary section, the recommended websites, and updated the Further Reading list. This is an update of Jean Y.J. Wang, Cellular Responses to DNA Damage, In Encyclopedia of Cancer (Second Edition) edited by Joseph R. Bertino, Academic Press, 2002, Pages 425–431.

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.98754-3

315

316

Cell Responses to DNA Damage

DNA Lesions Base modifications, mismatches, crosslinking, strand breaks

Assembly of Lesion-Specific Protein Complexes at Damaged Sites Lesion recognition proteins, 9-1-1 DNA clamp, PIKK, mediators, histones, and others

DNA Repair

Cell Cycle Checkpoints

Lesion-specific mechanisms Repair of lesions terminates signaling

G1/S, intra-S, G2/M Checkpoints are reversible

Gene Expression Alterations Transcription, alternative splicing, microRNA processing, chromatin remodeling

Inflammation

Necrosis

Apoptosis

Differentiation

Premature Senescence

Growth Arrest

Fig. 1 A framework of DNA damage response (DDR). A variety of DNA lesions occur under physiological conditions and are further induced by exposure to genotoxins. Damage recognition by lesion-specific detectors initiates the assembly of protein complexes through regulated protein– protein interactions involving covalent modifications of preexisting proteins (see Fig. 3 for an example of complex assembly at a double strained break site). These complexes serve two functionsdto stimulate DNA repair and to propagate the DNA damage signal. The immediate early action (within minutes of DNA damage) is to stimulate DNA repair. In proliferating cells, another immediate early action is to inhibit cell cycle progression. In both proliferating and nonproliferating cells, the third immediate action is to alter gene expression. Because transcription of new RNA and translation of new protein require time, gene expression-dependent biological responses are delayed (illustrated by the longer arrow). In DDR, gene expressiondependent responses include cell cycle arrest, premature senescence, inhibition or stimulation of cellular differentiation, apoptosis, necrosis, cell–cell communication, and inflammation. Because the cell type and developmental stages control the epigenetic landscape, the delayed responses driven by gene expression alterations exhibit cell-context dependency and cannot be easily generalized across all cell types.

biological outcomes in DDR. Because RNA transcription and protein translation require more time (longer arrow in Fig. 1) than covalent modifications of preexisting factors (short arrows in Fig. 1), a rapid extinction of DNA lesions (efficient repair) can prevent or reduce gene expression alterations to blunt the delayed responses. Therefore, it is the persistence of DNA lesions that triggers the downstream effects, such as premature senescence or cell death (Fig. 1). It must be emphasized that DDR exhibits not only temporal dependency but also cell type dependency (Fig. 1). In proliferating (cycling) cells, the immediate early effects (within minutes of lesion detection) are to stimulate DNA repair and to inhibit cell cycle progression. These rapid DDR maximizes genome protection, cell survival, and resumption of proliferation. By contrast, delayed effects such as premature senescence or death eliminate rather than protect the damaged cells. Therefore, the upstream factors in lesion detection and signal transduction can activate pro-survival or pro-death downstream effects. Under conditions when lesion removal (DNA repair) is highly efficient, the DDR signaling proteins contribute to cell survival. However, under conditions when DNA lesions overwhelm the repair capacity, DDR signaling proteins contribute to cell death.

Lesion Recognition and Damage Sensor Modified Bases Base modifications, such as guanine base alkylation (6-meG) or oxidation (8-oxoG), are recognized and repair by several different mechanisms. For example, MGMT (O6-methylguanine DNA methyltransferase) recognizes and demethylates O6-methylguanine. On the other hand, N-glycosylase enzymes recognize and cleave the N-glycosidic bonds to remove the modified bases in BER (base excision repair). In nucleotide excision repair (NER), which is the major repair mechanism of UV-induced photo-adducts, several XP protein complexes and the elongating RNA polymerase II serve as lesion detectors. The XP genes are those mutated in an autosomal recessive genetic disease Xeroderma pigmentosum (XP). XP patients display extreme sensitivity to UV; they also suffer from developmental defects and susceptibility to cancer.

Mismatched Base Pairs Mistakes during DNA replication occur at an appreciable rate despite the proofreading function of DNA polymerases and must be recognized and repaired to ensure the faithful transmission of genetic information. These replication-induced mismatched bases and small deletions or insertions are repaired by the mismatch repair (MMR) mechanism. In eukaryotic cells, heterodimeric

Cell Responses to DNA Damage

317

complexes of MSH2/3 or MSH2/6 proteins detect mismatched bases and then recruit other MMR proteins (MLH1, MLH3, PMS2) to initiate repair. In human, genetic defects in MSH2 or MLH1 cause cancer susceptibility in the Lynch syndrome.

Cross-Linked DNA Inter-strand cross-linking (ICL), where two strands of DNA are covalently linked, for example, by cisplatin, impedes transcription and DNA replication. The repair of ICL requires proteins encoded by more than 20 FA genes that are mutated in Fanconi anemia (FA), an autosomal recessive genetic disease characterized by anemia and cancer susceptibility. A complex of FA gene products recognize replication forks at ICLs to activate a complex repair mechanism that requires the coordination of nucleotide excision (NER), trans’lesion DNA synthesis, and homologous recombination (HR).

Double-Stranded Breaks (DSB) Double-stranded breaks (DSB) occur at stalled replication forks and during the process of repairing interstrand crosslinks. DSB are also induced by genotoxins, for example, ionizing radiation or topoisomerase inhibitors. Two evolutionarily conserved protein complexes are detectors of DSB: the hetero-dimeric Ku antigen (Ku-70, Ku-80) and the hetero-trimeric MRN complex (MRE11, RAD50, NBS1). In mammalian cells, the Ku antigen recruits DNA-PK to stimulate nonhomologous end joining (NHEJ) for the nonfaithful repair of DSB, whereas the MRN complex recruits ATM and many other factors to stimulate homologous recombination (HR) for the faithful repair of DSB by using an intact homolog of the broken DNA (Fig. 2).

The 9-1-1 Complex The 9-1-1 complex is a hetero-trimeric DNA clamp consisting of three evolutionarily conserved proteins: Rad9, Rad1, and Hus1. The 9-1-1 DNA clamp is structurally similar to the homo-trimeric PCNA DNA clamp required for DNA replication and repair. The 9-1-1 complex is loaded onto double- and single-stranded hybrid DNA at stalled replication forks through the action of a damageinduced Replication Factor C (RFC) clamp loader containing the RAD17 protein. The 9-1-1 clamp is also loaded onto similar DNA structures that are generated during DNA repair. The 9-1-1 complex differs from other lesion-specific binding proteins because it is a sensor of an abnormal DNA structure rather than a modified base or a DNA end. Similar to other lesion sensors, the 9-1-1 complex serves as a scaffold to promote DNA repair and damage signal transduction (Fig. 2).

DSB

DSB

Ku

DNA-PK

MRN

ATM

Stalled Fork TOPBP1

ATR ATRIP

RPA

RPA

911 Fig. 2 Activation of PIKK by DSB or stalled replication fork. Double-stranded breaks (DSB) are recognized by the Ku-antigen (a heterodimer of Ku70 and Ku-80) or by the MRN complex (a heterotrimer of MRE11, RAD50, and NBS1). Ku antigen interacts with DNA-PK (DNA-dependent protein kinase), a member of the PIKK family, which activates the nonhomologous end-joining mechanism for DSB repair. MRN interacts with ATM, another member of the PIKK family, which stimulates the phosphorylation of hundreds of substrates to regulate the homologous recombination mechanism for DSB repair, the cell cycle checkpoints and gene expression at several levels, including transcription initiation, alternative splicing and microRNA biogenesis. Stalled replication fork with double- and single-stranded hybrid DNA is sensed by the RAD17-containing RFC, which loads the 9-1-1 DNA clamp. The 9-1-1 complex and the single-stranded DNA binding protein, RPA, jointly recruit the ATR kinase associated with TOPBP1 (topoisomerase II binding protein 1) and ATRIP (ATR interacting protein). ATR, also a PIKK family of kinase, is essential to the repair and the restart of stalled replication forks, and therefore, ATR is essential to cell viability.

318

Cell Responses to DNA Damage

DNA Damage Signal Transduction PIKK Family of Protein Kinases In eukaryotic cells, DNA damage activates a family of evolutionarily conserved protein kinases belonging to the phosphoinositide 3kinase-like kinase (PIKK) family. The mammalian PIKKs central to the transduction of DNA damage signal are ATM (ataxia telangiectasia mutated), ATR (ATM and Rad3 related), and DNA-PK (DNA-dependent protein kinase) (Fig. 2). Loss of the ATM function causes Ataxia telangiectasia, an autosomal recessive genetic disease characterized by cerebellum degeneration, radiosensitivity, immune-deficiency, sterility, and cancer predisposition. Loss of DNA-PK is associated with severe combined immune deficiency (SCID) in mice. The mammalian ATR function is required throughout S-phase and plays an essential role in DNA replication. ATR is recruited to stalled replication forks or resected DSBs by a complex array of protein–protein interactions (Fig. 2). At the stalled fork, a long stretch of single-stranded (SS) DNA is formed and this ss-DNA is recognized and bound by RPA (an evolutionarily conserved ssDNA binding protein), which guides the loading of 9-1-1. Together, the 9-1-1 DNA clamp and RPA then facilitate the recruitment and activation of the ATR kinase in collaboration with TOPBP1 (topoisomerase II binding protein 1) and ATRIP (ATR interacting protein) (Fig. 2). Upon activation, the PIKKs phosphorylate hundreds of cellular proteins of diverse functions to mobilize an array of biological responses to DNA damage. Among the PIKK substrates, the most highly conserved are the mediators and the checkpoint kinases.

Mediators The mediators are scaffold proteins that play essential role in DNA damage signal transduction. The PIKKs phosphorylate the mediator proteins to regulate the assembly of protein complexes for DNA repair and signal propagation (Fig. 1). In DDR, the forkheadassociated (FHA) and the BRCA1 C-terminal (BRCT) domains serve as the phospho-epitope binding sites for the assembly of signaling and repair complexes. The FHA domain binds to phospho-threonine (pT) containing peptide motifs. The BRCT binding to phospho-peptide motifs requires two BRCT in tandem. The structural basis for phospho-epitope binding by the FHA and the BRCT domains has been elucidated. A prototypical mediator in DNA damage signal transduction is the mammalian MDC1 (mediator of DNA damage checkpoint 1), which is a large scaffolding protein with an FHA domain at the N-terminus, two tandem BRCT domains at the C-terminus, and numerous phosphorylation sites throughout the length of the protein (Fig. 3). The phosphorylated MDC1 can recruit proteins with the FHA, the tandem-BRCT, and other phospho-binding domains. At the same time, the FHA and BRCT domains of MDC1 can interact with other PIKK-phosphorylated substrates (Fig. 3).

Checkpoint Kinases Among the highly conserved PIKK substrates are the CHK1 and CHK2 protein kinases, which are members of the Protein Kinase superfamily, characterized by a highly conserved kinase domain that is distinct from the PIKK. CHK1 and CHK2 are phosphorylated by the PIKK, with CHK1 being primarily the substrate of ATR, and CHK2 of ATM. However, the PIKKs and the checkpoint kinases

s

i rax Ab 0 RAP8

BRCA1

Ub

Ub

ha

ha

NS

ATM

in

in

NS

P

RNF168

-c

-c

P

P NS

NS

NS

RFN8

NS

P P

P

P NS

NS P

MDC1

DSB

NS P

P MRN

P

ATM

Fig. 3 Assembly of protein complexes at DSB site. This diagram illustrates an example of protein complex assembly at a double-stranded break (DSB) site. The MRN complex (MRE11, RAD50, NBS1) binds to the DSB, recruits, and activates ATM. Activated ATM phosphorylates itself (phosphorylation sites are indicated as pink balls with P), the MRN, the H2AX histone in the nucleosomes (NS), and the mediator MDC1. The tandem BRCT domains of MDC1 interact with the phospho-epitope in the phosphorylated H2AX (aka gH2AX). Some phospho-epitopes in MDC1 bind to the BRCT and FHA domains of NBS1 in the MRN complex. Other phospho-epitopes in MDC1 bind to RNF8, which then ubiquitinates H2A and H2AX in the nucleosomes. The poly-ubiquitin chains, further elaborated by RNF168, recruit RAP80 and Abraxis, which then recruit BRCA1 that is also phosphorylated by ATM. Note that activated ATM is distributed beyond the DSB site to phosphorylate many other substrates. The BRCA1 complex at the DSB site facilitates homologous recombination repair.

Cell Responses to DNA Damage

319

are components of a highly dynamic signaling network with nonexclusive interactions. ATM phosphorylation of CHK2 causes oligomerization of this kinase via intermolecular phospho-epitope binding to its own FHA domain and this ATM-induced CHK2 oligomerization results in CHK2 kinase activation. In addition, checkpoint kinases interact with mediators, which control the compartmentalization and stability of CHK1 and CHK2. The checkpoint kinases play essential roles in the activation of cell cycle checkpoints. They also regulate DNA repair and gene expression by phosphorylating various downstream effector proteins.

Other Kinases Besides CHK1 and CHK2, other kinases of the Protein Kinase superfamily also participate in the transduction of DNA damage signal. These include the stress-activated protein kinases, for example, p38 and JNK, the ABL tyrosine kinase, and many others. The stress-activated protein kinases can activate the cell cycle checkpoints independent of CHK1 and CHK2. They also regulate gene expression. The ABL tyrosine kinase is localized to the cytoplasm and the nucleus. In response to DNA damage, the ABL kinase becomes accumulated in the nucleus through the action of ATM. Nuclear ABL tyrosine kinase activates pro-apoptotic gene expression by activating the p53 family of transcription factors (p53, p63, p73) and histone acetyltransferase, for example, Tip60 (KAT5).

Other Protein-Modifying Enzymes Besides phosphorylation, DNA damage activates other protein modifying enzymes to induce a variety of protein covalent modifications, including ubiquitination, poly-ubiquitination (Fig. 3), sumoylation, neddylation, ADP-ribosylation, acetylation, methylation, and possible other less studied modifications such as nitrosylation and succinylation. For example, the mediator protein BRCA1 (together with BARD1) functions as an E3 ubiquitin ligase to ubiquitinate proteins involved in DSB repair and gene expression. Some of these protein covalent modifications occur on preexisting proteins that are assembled into complexes at the damage sites (Fig. 3). However, covalent modifications in DDR can also occur on proteins that are induced in the delayed responses to regulate biological processes other than checkpoints and repair (Fig. 1). In other words, protein covalent modifications participate in the immediate and the delayed responses to DNA damage.

Assembly of DNA Damage Signaling Complexes at DSB Live cell imaging technology combined with laser micro-irradiation to generate DSB of defined patterns has observed a series of temporally distinct steps in the immediate response to DSB lesions (Fig. 3). Recruitment of the MRN complex and ATM at the DSB sites is followed by ATM acetylation, auto-phosphorylation, and the phosphorylation of H2AX (the ATM-phosphorylated H2AX is commonly known as gH2AX). Formation of gH2AX occurs initially at or near the DSB site and is then extended outward from the break site for several megabases. The mediator MDC1 is recruited to the DSB sites by binding to the phospho-epitope motif in gH2AX with its tandem BRCT domains (Fig. 3). MDC1 is then phosphorylated to generate binding sites for the FHA and BRCT domains of NBS1 in the MRN complex (Fig. 3). Other phospho-epitopes in MDC1 bind to the FHA domain of an ubiquitin ligase RNF8, which ubiquitinates H2A and H2AX at the DSB sites. The ubiquitinated histones then recruit another ubiquitin ligase RNF168 (Fig. 3). The polyubiquitin chains then recruit additional factors, such as the RAP80 protein, which contains a pair of ubiquitin interacting motifs (UIMs) that bind to polyubiquitin chains. RAP80, together with a coiled-coil scaffold protein Abraxis, in turn, create binding sites for BRCA1, which promotes homologous recombination to repair the DSB. Together, these live cellimaging results illustrate the complexity of protein assembly at DNA lesions (Fig. 3).

Downstream Biological Effects Cell Cycle Checkpoints DNA damage activates three major checkpoints to halt the progression of the cell cycle (Fig. 4). The G1/S checkpoint is activated in G1 cells to inhibit S-phase entry. The intra-S checkpoint is activated in S-phase cells to prevent the not-yet-activated origins from initiating DNA replication. The G2/M checkpoint is activated in G2 cells to inhibit the entry into mitosis. The cell cycle checkpoints are triggered by protein phosphorylation that is readily reversible upon protein dephosphorylation so that cell cycle progression can resume after DNA lesions are repaired (Fig. 4).

G1/S checkpoint A critical downstream effector in G1/S checkpoint is Cdc25A. DNA damage induces the phosphorylation, nuclear exclusion, and degradation of Cdc25A through PIKK-Mediator-CHK1/2. Inactivation of Cdc25A prevents the activation of Cdk2/Cyclin to block the initiation of DNA replication (Fig. 4). The G1/S checkpoint is an immediate early response to DNA damage as it is activated by protein phosphorylation within minutes of lesion detection. In mammalian cells, persistence of DNA lesions can prolong the G1/S checkpoint via the upregulation of p21/Cip1, which directly binds to and inhibits the Cdk2/Cyclin complexes. This p21Cip1mediated G1 arrest is a delayed response as it takes hours for the transcription and the translation machineries to produce this protein above the levels of Cdk2/Cyclin complexes.

320

Cell Responses to DNA Damage

(A)

(B)

Cdk1 Cyclin B M G2 G1

Cdk4/6 CyclinD

P

P Cdk

Cdk CDC25

Cyclin

Cyclin S Cdk2 CyclinA

Checkpoints

Cdk2 CyclinE G1/S Intra-S G2/M

Checkpoint Activated

Checkpoint Relieved

DNA damage signal

Fig. 4 DNA damage activates cell cycle checkpoints. (A) Cell cycle progression is coordinated by Cdk/Cyclin. D-type cyclins are expressed in G1 to activate Cdk4 and Cdk6. E-type cyclins are expressed in late G1 to stimulate G1/S transition by activating Cdk2. A-type cyclins are expressed throughout S and G2 to activate Cdk2. B-type cyclins are expressed in late G2 and early M to stimulate G2/M transition by activating Cdk1. DNA damage activates three checkpoints: the G1/S checkpoint (red) inhibits the entry of G1 cells into S-phase by blocking the initiation of DNA replication, the intra-S checkpoint (black) inhibits firing of replication origins that have not been activated in S-phase cells, the G2/M checkpoint (orange) inhibits the entry of G2 cells into mitosis. (B) Cell cycle checkpoint is activated when Cdk/cyclin kinase activity is inhibited. Cdc25 phosphatase, which dephosphorylates Cdk, is a critical activator of Cdk/cyclin kinase. In response to DNA damage, PIKK phosphorylates and activates the checkpoint kinases (CHK1 and CHK2) to phosphorylate, sequester, and degrade Cdc25, thus inhibiting cell cycle progression.

Intra-S checkpoint The intra-S checkpoint is a transient inhibition of DNA replication triggered by DNA damage in S-phase cells. The eukaryotic genome is divided into tens of thousands of replication units (replicons), each with an origin where replication initiates. These replication origins are activated (fired) nonsynchronously throughout S-phase, with the timing of origin firing being dependent on the chromatin conformation and other factors. When activated, the intra-S-phase checkpoint prevents the firing of uninitiated origins (Fig. 4). The ATM kinase activates an intra-S checkpoint that requires the phosphorylation of SMCs, which are components of the sister chromatin–cohesion complex. The ATM-induced intra-S checkpoint is an immediate early and short-lived response, possibly because the resumption of DNA replication can facilitate repair.

G2/M checkpoint The G2/M checkpoint mechanism is activated to prevent damaged cells from entering into mitosis. The Cdk1/Cyclin B kinase complex, also known as the Mitosis Promoting Factor (MPF), is the activator of mitotic entry. During late S and G2 phases of the cell cycle, Cdk1/Cyclin B is inhibited through threonine (Thr) and tyrosine (Tyr) phosphorylation of Cdk1 by the Wee1 kinase. Dephosphorylation of Cdk1 by the Cdc25 phosphatases (Cdc25A, B, and C) activates MPF to initiate mitosis (Fig. 4). DNA damage causes the inactivation of all three Cdc25 phosphatases through the actions of PIKK and checkpoint kinases to prevent MPF from being activated and thus block cells in G2. The p38 family of stress-activated protein kinases and the Polo-like kinases (Plk1 and Plk3) can also inhibit MPF to activate G2/M checkpoint.

Gene Expression The upregulation of p21Cip1 followed by the cell cycle arrest in G1 is one of the best established examples for how gene expression controls the cellular responses to DNA damage. DNA damage alters the expression of protein-coding mRNA as well as long noncoding RNA (ncRNA) and microRNA (miR) at the levels of transcription, splicing, and miR biogenesis.

Transcription The most well-studied transcription effector in DDR is the tumor suppressor p53, which is a classical transcription factor with a trans-activating domain (that interacts with transcription coactivators), a sequence-specific DNA binding domain, and an oligomerization domain. In response to DNA lesions, PIKK and CHK2 phosphorylate p53, MDM2 (an E3 ubiquitin ligase that targets p53 for degradation), and several other proteins to cause the stabilization and activation of p53. The activated p53 binds to enhancers containing p53-responsive elements (p53-RE) that are found in many promotes and enhancers. While p53 activation is a universal response to DNA damage across mammalian cell types, the tissue-specific responses to DNA damage are mediated by the combinatorial effects of p53 and many other tissue-specific transcription factors.

Cell Responses to DNA Damage

321

Alternative splicing Alternative splicing of nascent RNA is a key mechanism for diversifying the coding capacity. Downstream of DNA damage signaling are effectors that regulate alternative splicing. For example, DNA damage induces covalent modifications of splicing factors to skip or to include exons, thus altering the relative levels of protein isoforms in damaged cells. Alternative splicing of DDR signal transducers themselves, such as p53 and its related p63 and p73, can have deterministic influence on the biological outcomes. However, current knowledge on DNA damage-induced alternative splicing remains insufficient to identify such deterministic effects, possibly because of the highly context-dependent nature of splicing regulation and interactions among protein isoforms.

MicroRNA biogenesis MicroRNAs (miRs) are small ( 22 nt) single-stranded noncoding RNAs that are loaded into the RNA-induced silencing complexes (RISCs) and function as guides to target specific mRNAs for translation inhibition or degradation. The DDR-activated p53 can stimulate transcription initiation from specific miR genes, with the miR-34 family being the most notable example. The biogenesis of mature miRs requires (a) transcription of primary microRNA, pri-mir, that is generally a long RNA, (b) cleavage of pri-mir by the microprocessor complex (consisting of Drosha and DGCR8) to generate a  70 nt precursor microRNA hairpin, pre-mir, which is exported to the cytoplasm, and (c) processing of pre-mir by Dicer to form the mature miR that is loaded into the RISC complex. In response to DNA damage, the processing of pri-mir to pre-mir and from pre-mir to mir can be stimulated by ATM-dependent phosphorylation of specific RNA binding proteins. The ABL tyrosine kinase phosphorylates DGCR8 to stimulate pri-mir-34c processing to pre-mir-34c. The biological functions of miRs are highly dependent on the cellular context, because each miR is designed to target a large battery of different mRNAs. With its interesting features of stability and versatility in target selection, DNA damage-induced microRNAs can play important roles in the execution, the maintenance, or the termination of downstream biological effects in DDR.

Chromatin Modifications Chromatin modifications control the epigenetic landscape of a genome; thus, alterations in chromatin can have a profound and overarching influence on gene expression. Generally speaking, chromatin modifications involve regulations of: (a) the composition of histone isoforms in the nucleosomes; (b) the covalent modifications, for example, phosphorylation, acetylation, methylation, ubiquitination, sumoylation, etc., of histones; and (b) the positioning of nucleosomes. Each of these chromatin regulatory mechanisms is engaged in DDR to promote lesion recognition, repair, and signal transduction. As discussed above, PIKK-mediated phosphorylation of histone H2AX generates gH2AX (Fig. 3) that functions as a chromatin marker for the assembly of repair and signaling complexes. Phosphorylated gH2AX also recruits the INO80 complex that regulates the spacing of nucleosomes.

Cellular Differentiation In DDR, the differentiation checkpoint describes a biological response in which DNA damage inhibits the differentiation of tissue stem cells (Fig. 5). This response has been observed in adult muscle satellite cells and embryonic muscle stem cells. This differentiation checkpoint delays terminal differentiation to allow time for DNA repair in lineage-committed muscle stem cells. With

Myoblasts

Multinucleated myotubes

Myogenic differentiation

DNA damaged myoblast Neural stem cell (NSC)

Neuron

DNA damaged NSC

Oligodendrocyte

Astrocyte

Fig. 5 Effects of DNA damage signaling on cellular differentiation. In committed muscle stem cells (myoblasts), DNA damage signal inhibits the myogenic transcription factors, in particular, MyoD, to block muscle differentiation. In multipotential neural stem cells, DNA damage signal promotes growth arrest and differentiation toward astrocytes but not other neural cell types.

322

Cell Responses to DNA Damage

multipotential neural stem cells, on the other hand, DNA damage stimulates growth arrest and accelerates differentiation toward astrocytes but not other neural cell types (Fig. 5). These findings show that DNA damage signals can regulate differentiation in stem cells and the effects are dependent on the stem cell types.

Cell Death DNA damage activates distinct cell death mechanisms, including but not limited to apoptosis, necrosis, and mitotic death (premature senescence) (Fig. 6). Depending on the cell context, death induction can occur at widely different lesion concentrations. On one end of the spectrum is the induction of apoptosis by a single DSB. This hypersensitivity to DSB-induced apoptosis occurs in the developing central nervous system where a dose of ionizing radiation (IR) as low as 0.5 Gy is sufficient to activate ATM and p53-dependent apoptosis in neuroblasts within 6 h of irradiation. This hypersensitivity to IR-induced apoptosis is more of an exception than a rule as most mammalian cell types do not undergo apoptosis after 0.5 Gy of IR. On the other end of the spectrum is the resistance of tissue stem cells and terminally differentiated neurons or muscle cells to DNA damage-induced apoptosis. These resistances underlie the recovery of cancer patients after treatment with lesion-inducing agents, such as ionizing radiation (generates DSB) temozolomide (generates alkylated bases), cisplatin (generates ICL), and doxorubicin (generates DSB). The finding of such widely divergent sensitivity suggests that the death response in DDR is controlled not only by the concentrations and/or the persistence of DNA lesions, but it is also controlled by the proliferative status and the developmental lineages of the damaged cell (Fig. 6).

Apoptosis The word “apoptosis” was coined to describe a specialized form of suicidal cell death that is characterized by the morphological features of chromatin condensation and degradation, as well as cell shrinkage and fragmentation. The condensed cell fragments (apoptotic bodies) express surface-exposed “eat-me” signals that stimulate their uptake by macrophages or other healthy cells. This death mechanism is employed in embryonic development to eliminate excess cells. Apoptosis also eliminates many (but not all) types of terminally differentiated cells throughout adult life. In DDR, apoptosis induction requires alterations in gene expression (transcription initiation, alternative splicing, microRNA biogenesis). The most well established apoptotic pathway in DDR is that of p53-dependent transcription of pro-apoptotic genes, for example, PUMA (BBC3) and NOXA (PMAIP1). PUMA and NOXA are BH3-only members of the BCL2 family of apoptosis regulators. Accumulation of pro-apoptotic PUMA and NOXA proteins together with reduction in antiapoptotic BCL2, BCLxL proteins cause BAX/BAK-dependent release of cytochrome c from the mitochondria to activate the apoptosome and the effector caspases that bring about the morphological features of apoptosis. It should be noted that p53 activation is an immediate early response in DDR, mediated by PIKK phosphorylation of MDM2 and p53. However, apoptosis execution is a delayed response and only occurs in some but not other cell types. Because activated p53 not only stimulates pro-apoptotic genes but also stimulates p21Cip1 to cause G1 arrest and premature senescence (Fig. 6), p53 activation is necessary but not sufficient to determine the choice of apoptosis in DDR. Generally speaking, DNA damage is more likely to induce apoptosis in highly regenerative tissue where there is continuous turnover of differentiated cells.

Inflammation es kin to les Cy esic v

Necrosis

Apoptosis

DNA damaged cell

Bystander cell

Premature senescence Fig. 6 Cell death choices, cell–cell communication, and the induction of inflammation in DDR. A cell suffering from DNA damage (red) can choose three types of death: namely necrosis, apoptosis, or premature senescence. The senescent cell is not physically dead but it is dead to mitosis as senescence is an irreversible cell fate. A damaged cell can communicate with neighboring bystander cell by secreting cytokines and vesicles. The DNA damage can also trigger tissue inflammatory responses that are stimulated by those factors secreted by the damaged cell or dumped by a damaged cell that underwent necrosis.

Cell Responses to DNA Damage

323

Even within a single cell type, the timing of apoptosis execution is variable, depending on the status of cell cycle. The best example to illustrate this timing variation is with neuroblasts (hypersensitive to IR-induced apoptosis) in the developing central nervous system. Experimenting with three different genetically engineered mouse strains that are knockout for ATM, p53, or with a p53 mutant lacking the ATM phosphorylation site, it can be established that IR-induced neuroblast apoptosis absolutely requires the ATM-p53 pathway. Although ATM phosphorylation of p53 is activated immediately after IR, apoptosis occurs 6 h after IR in neuroblasts that are out of S-phase at the time of irradiation. However, apoptosis only occurs 24 h after IR in neuroblasts that are in S-phase at the time of irradiation. How S-phase cells suppress apoptosis execution despite the activation of ATM and p53 is not yet understood.

Necrosis Necrotic cell death is characterized by the loss of membrane integrity, cell bursting, and the release of cellular content. Necrosis can be accidental or regulated. The necrotic death that is actively induced by stress signals has been referred to as “programmed necrosis” or necroptosis. Upon infections or injuries, several inflammatory cytokines, for example, tumor necrosis factor-alpha (TNF), stimulate programmed necrosis of cells in the vicinity of the injuries to amplify the inflammatory signals. Either accidental or regulated, necrotic cell death contributes to the inflammatory responses triggered by infections or injuries (Fig. 6). In the most general and simplified terms, necrosis is activated when ATP (energy source) and/or NADH/NADPH (reducing power) become depleted and can no longer support the integrity of the plasma membrane. TNF-induced necrosis has been linked to the formation of a “mitochondrial attack complex,” which may also mediate necrosis activated by calcium and reactive oxygen species. In DDR, necrosis activation has been linked to the depletion of NAD by poly-ADP-ribose polymerase (PARP), which is a class of enzymes that convert NAD into poly-ADP-ribose chains on target proteins. PARP1 is an abundant chromatin protein that is activated by single-stranded ends in DNA and it can add poly-ADP-ribose onto a large number of proteins to stimulate DNA repair. Under conditions when excessive lesions or futile repair causes the accumulation and persistence of single-stranded ends, the continuous activation of PARP1 can lead to NAD depletion to trigger programmed necrosis. Other pathways linking DNA damage to regulated necrosis are under active investigation.

Mitotic death (premature senescence) Mitotic death is a phrase coined by radiobiologists to describe IR-induced “death to mitosis.” In the literature, this phrase is also used to describe mitotic catastrophe where cells die from catastrophic failure in chromosome segregation. In DDR, mitotic death and mitotic catastrophe are activated by different mechanisms. The death to mitosis is a premature senescence (Fig. 6), describing a prolonged or a permanent growth arrest in response to DNA damage. In DDR, premature senescence is induced in part through p53-dependent upregulation of p21Cip1, which inhibits Cdk2/Cyclin complexes to prevent phosphorylation of the RB family of pocket proteins, including RB, p107, and p130. The dephosphorylated RB-family proteins promote the assembly of repressive chromatin complexes at pro-proliferation genes to cause growth arrest. In DDR, mitotic catastrophe is induced when chromatin condensation and spindle assembly occur on incompletely replicated DNA to cause chromosome fragmentation. This event occurs as a result of failures to activate or to maintain the G2/M and/or the spindle checkpoints in cells suffering from DNA damage.

Cell–Cell Communications The initiation of DDR occurs within the cell that suffers directly from damage to its genomic DNA. Interestingly, DDR can also be propagated to neighboring cells that are not directly targeted by genotoxins. This propagation of DDR is demonstrated by media transfer experiments, in which media conditioned by irradiated cells can activate DDR in naïve, non-irradiated, bystander cells. The radiation-induced bystander effects (BSE) result from cell–cell communications where irradiated cells produce factors to induce DDR in non-irradiated bystander cells. The directly irradiated cell can affect neighboring non-irradiated cells through gap junction communication. The directly irradiated cell can also secrete factors to communicate with other cells in its vicinity (Fig. 6). In whole animals, DNA damage inflicted in one part of the body can generate DDR at a distance from the site of damage. These “off-field” or “abscopal” effects may involve production of cytokines, chemokines, and vesicles by the DNA-damaged cells. For example, IR can stimulate the directly targeted cells to produce CSF1 that attracts macrophages to the tissue microenvironment where the DNAdamaged cells reside. These activated macrophages can then elaborate an inflammation reaction with long-range effects on other tissues. Because radiation and genotoxins have remained as the mainstay cancer therapeutic modalities, mechanistic understanding of the bystander and abscopal responses to DNA damage in the body is likely to provide new insights on how to mitigate the side effects and enhance the efficacy of cancer therapy.

Prospective Vision Basic Research DNA is the blue print for all biological processes; thus, it is not surprising to find that DNA damage activates a wide array of responses. This chapter describes a framework for DDR, from molecules to cells to organisms. The present framework, however,

324

Cell Responses to DNA Damage

is still rather primitive and based mostly on studies in model systems such as yeast, mouse, and cultured cells. While model system research is continuingly required to address many pressing questions on the molecular mechanisms of DDR, future attention should be paid to the human tissue and cell-type specificity of these processes in order for us to understand the relevance of DDR in human diseases. The DDR in different human cell types, tissues, and the whole organism needs to be investigated directly in the human systems. In particular, how cells of different developmental lineages and proliferative potential decide to live or to die, and how apoptotic versus necrotic cell death affects neighboring or distant cells, are important but as yet unsolved problems. Because DNA damaging agents are widely used to treat many different cancer types, the ongoing Cancer Genomics efforts in finding genetic and genomic determinants for tumor response to genotoxic therapies can be leveraged to expand the knowledge of DDR in the human body.

Cancer Translational Research Knowledge gained from model organism research on DNA repair, which is the most highly conserved mechanism in DDR and is therefore readily applicable to the human system, has been translated to improve or develop cancer therapies. The best example for how knowledge on DNA repair can impact cancer therapy is that of PARP inhibitors, which were developed to target cancer with defects in homologous recombination (HR). PARP (poly-ADP-ribose polymerase) is activated by singlestranded DNA ends to facilitate DNA repair. PARP is not found in yeast cells, and its activity is not essential to cells with competent HR repair However, HR-defective cells become dependent on PARP and are thus sensitive to the PARP inhibitors. In the clinic, PARP inhibitors have improved progression-free survival, although not overall survival, of patients with BRCA1 or BRCA2 mutant cancers. The development of PARP inhibitors to target HR-defective cancers has provided the proof of concept that DDR defects in cancer cells can be exploited to develop targeted therapies. Cancer cells with HR defects are also hypersensitive to platinum-based drugs that generate ICLs in the genomic DNA. Because platinum resistance can be linked to recovery of HR function such as reversion of BRCA2 mutation back to wild type, deliberate disruption of HR function could in theory enhance the efficacy of platinum-based therapy. A flurry of research activities are ongoing in the public and the private sectors to find drugs that target DNA repair in cancer cells. DNA repair defects can also impact cancer immunotherapy as exemplified by the increased response of MMR-defective cancers to immune checkpoint blockade antibodies. As discussed above, MMR corrects replication errors and MMR-defective cancer cells suffer from hypermutations. The increased mutation rate causes a higher probability in the expression of neo-antigens to trigger antitumor immunity. In patients with MMR-defective cancer, therapeutic blockade of immune checkpoints, therefore, is more likely to activate tumor-specific cytotoxic T-lymphocytes to kill tumor cells expressing neo-antigens that are generated as a result of DNA repair defects. Given the already established therapeutic impacts from the current knowledge on DDR, it is not unreasonable to anticipate future developments of therapies to target not only repair but also the other DDR defects in cancer as we gain further understanding on DNA damage response in the human body.

Acknowledgment The author’s research on DNA damage response is supported by a grant (R01CA043054) from the National Cancer Institute of the National Institutes of Health, USA.

See also: Genetic Instability.

Further Reading Friedberg, E.C., Walker, G.C., Siede, W., Wood, R.D., Schultz, R.A., Ellenberger, T., 2006. DNA repair and mutagenesis, 2nd ed. ASM Press, Washington, DC. Article Series on “DNA Damage”, Nature Reviews Molecular Cell Biology. Blackford, A.N., Jackson, S.P., 2017. ATM, ATR and DNA-PK: The trinity at the heart of DNA damage response. Molecular Cell 66, 806–817. Czabotar, P.E., Lessene, G., Strasser, A., Adams, J.M., 2014. Control of apoptosis by the BCL-2 protein family: Implications for physiology and therapy. Nature Reviews Molecular Cell Biology 15, 49–63. D.T. Le, J.N. Durham, et al., Mismatch repair deficiency predicts response of solid tumors to PD-1 blockade, Science, https://doi.org/10.1126/science.aan6733. Lord, C.J., Ashworth, A., 2017. PARP inhibitors: Synthetic lethality in the clinic. Science 355, 1152–1158. Mukherjee, D., Coates, P.J., Lorimore, S.A., Wright, E.G., 2014. Responses to ionizing radiation mediated by inflammatory mechanisms. Journal of Pathology 232, 289–299. Reinhardt, H.C., Yaffe, M.B., 2013. Phospho-Ser/Thr-binding domains: Navigating the cell cycle and DNA damage response. Nature Reviews. Molecular Cell Biology 14, 563–580. Schneider, L., Pellegatta, S., Favaro, R., Pisati, F., Roncaglia, P., Testa, G., Nicolis, S.K., Finocchiaro, G., D’Adda Di Fagagna, F., 2013. DNA damage in mammalian neural stem cells leads to astrocytic differentiation mediated by BMP2 signaling through JAK-STAT. Stem Cell Reports 1, 123–138. Seeber, A., Hauer, M., Gasser, S.M., 2013. Nucleosome remodelers in double-strand break repair. Current Opinion in Genetics and Development 23, 174–184. Zhang, J., Walter, J.C., 2014. Mechanism and regulation of incisions during DNA interstrand cross-link repair. DNA Repair 19, 135–142.

Cell Responses to DNA Damage

Relevant Websites PubMed: https://www.ncbi.nlm.nih.gov/pubmed: Search with the listed Keywords to find review and research articles on DDR. GeneCards: http://www.genecards.org/: Search with Gene names to find basic information on genes mentioned in this article. Google Images: https://www.google.com/: Search Google Images with “DNA damage response” to view pathway diagrams and research data relating to this topic.

325

Cervical Cancer: Screening, Vaccination, and Preventive Strategies Paolo Giorgi Rossi, AUSL Reggio Emilia, IRCCS, Reggio Emilia, Italy Francesca Carozzi, ISPO, Cancer Prevention and Research Institute, Florence, Italy © 2019 Elsevier Inc. All rights reserved.

Glossary Ascertainment (second-level test) The test or set of tests used to ascertain the presence of a disease or clinically relevant lesion in women positive to first-level test and, in case of an algorithm including a triage test, to the triage test as well. In cervical cancer screening guidelines for high-income countries, ascertainment is always colposcopy with colposcopy-guided biopsy; see-and-treat strategies are not recommended. Treatment should follow a histological diagnosis on biopsy. Cotesting This term defines a screening algorithm in which two primary screening tests are performed in parallel; women are managed according to the results of both tests. This strategy has been proposed for HPV and Pap test by the US guidelines. The rationale is to maximize sensitivity, although the number of false positives and the number of tests needed to screen the population do increase. It is the opposite of the triage strategy, in which two tests are performed in sequence, the second only if the first one is positive. Detection rate The number of clinically relevant lesions found by screening divided by all the screened women. Organized screening program An organized preventive intervention aimed at screening the population in which the target population, the testing protocols, including age to start and stop screening, interval between tests, management of abnormal tests and treatment are all clearly defined. A monitoring and quality assurance system is also in place. Programs are populationbased if they actively invite the entire target population on a regular basis. The alternative model of care is opportunistic (or spontaneous) screening, where a physician takes the opportunity of any contact with each individual to recommend or prescribe the screening test. Participation rate The proportion of women participating in the screening program out of the total target population. In the absence of an organized screening program, this indicator can be called test coverage or uptake; it is computed as the proportion of the population having had a test within the recommended interval. Positive predictive value The probability that a woman who tests positive has a clinically relevant lesion, usually CIN2 or more severe. Primary screening test The screening test that is performed first in the screening algorithm and the only one that is administered to the entire population. HPV DNA, Pap test and VIA are recommended as primary screening tests. Referral rate The proportion of women that are referred to second-level test or ascertainment. Regressive lesion In the natural history of cervical neoplasia, HPV-induced lesions often regress. Cervical intraepithelial neoplasia grade 1 (CIN1) is typically a regressive lesion; its probability of progression to a more severe lesion is so low that it is no longer considered a precancerous lesion. CIN grades 2 and 3, which are considered high-grade lesions and precancer, are highly regressive, but the balance between the risk of cancer and the very low morbidity related to the treatment favors surgical excision of these lesions. All guidelines therefore recommend it for CIN3 and most guidelines recommend it also for CIN2. In recent years, some authors have suggested that strict follow up can be appropriate for CIN2 in young women, when the probability of regression is higher. Screening interval The period of time recommended between a negative screening test and the subsequent screening test. It defines the screening round, that is, the 3- or 5-year cycle in which a program should invite the whole target population to achieve complete invitation coverage. Screening The systematic application of a test to a healthy population to detect a disease or its precursors early in order to begin treatment immediately, thereby improving prognosis or, in the case of targeting precursors of the disease, to avoid the onset of the disease. Triage test In certain screening algorithms, women who test positive to the primary screening test are referred to a triage test in order to distinguish between those who need immediate ascertainment and those who can be followed up, thereby avoiding referring too many women to ascertainment test, which can be invasive and resource-consuming.

Nomenclature CIN Cervical intraepithelial neoplasia. See regressive lesion in the glossary for a distinction between different grades of CIN. High-risk HPV A restricted group of HPV types that infect human mucosae and that can induce cellular transformation. They are associated with some epithelial cancers in humans, and are virtually a necessary cause of cervical cancer. High-risk HPV types include the 12 types classified by the International Agency for Research on Cancer (IARC) as surely cancerogenic for humans (group 1): types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59. Some available commercial tests also include types 68 and 66, classified as probably (group 2a) and possibly (group 2b) cancerogenic, respectively, by IARC.

326

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65138-3

Cervical Cancer: Screening, Vaccination, and Preventive Strategies

327

HPV Human papilloma virus. DNA capsuled virus including > 200 types which infect human skin and mucosae. Some of them produce papillomas and other flat lesions, which are mostly asymptomatic. HPV DNA test Test for the presence of HPV DNA. In screening, the test should target only high-risk types and be validated on clinical endpoints, that is, having high sensitivity and specificity for clinically relevant lesions and not for any HPV infection (Meijer et al., 2009). Pap test Papanicolaou test is a cytological analysis of cells scraped from the cervix. Cells can be put directly on a slide, thus obtaining a smear, or suspended in a fixative medium and then transferred onto a slide in thin layer (liquid-based cytology).

Introduction Cervical Cancer Burden of Disease: The Role of HPV and Screening It is estimated that every year, > 527,000 new cervical cancer cases and 265,672 deaths occur worldwide. The vast majority of the burdend84% of cases and 87% of deathsdoccurs in less developed countries (Bruni et al., 2017). Cervical cancer affects relatively young women, with incidence peaking at between ages 40 and 55, then plateauing or decreasing slightly. Geographic distribution is extremely heterogeneous, ranging from an age-standardized rate of > 50/100,000 in some countries of East and Central Africa, to < 4/100,000 in many North African and Middle Eastern countries (Fig. 1A). There are large differences within Europe as well: Switzerland, Finland, and some Mediterranean countries have low incidence (about 4/100,000), while many Eastern European countries have an incidence of > 20/100,000 (Bruni et al., 2017; Li et al., 2011). Both molecular biology and epidemiological studies have demonstrated that cervical cancer is causally linked to human papillomavirus (HPV) infection; persistent infection of oncogenic HPV types (hereafter named hrHPV) is a necessary cause for the onset of high-grade precancerous cervical lesions (cervical intraepithelial neoplasia, CIN grades 2 or 3) and progression to invasive cancer (zur Hausen, 1999; Walboomers et al., 1999). Among the hrHPV types, HPV16 is responsible of > 65% of cancers worldwide, followed by HPV18 (Clifford et al., 2005; Bruni et al., 2017; de Sanjose et al., 2010; Plummer et al., 2016). The distribution of HPV types does not necessarily reflect the prevalence of infections in the healthy population; in fact, as cervical cancer is a relatively rare consequence of a very common infection, the proportion of cancers due to each type is mostly related to the ability of a certain type to induce cellular transformation (de Sanjosé et al., 2007; Guan et al., 2012). Based on epidemiological studies on HPV infection, CIN and cancer incidence distribution by age and through modeling studies, it is possible to estimate that the process from infection to cancer takes decades (Schiffman and Wentzensen, 2013; Baussano et al., 2013; Bruni et al., 2017), although this time varies according to the virus and to host factors that are not yet fully understood. The natural history includes long-lasting precancerous tissue modifications that can be detected through cytological or molecular tests and treated (Schiffman et al., 2016). The combination of all these characteristics of the cervical cancer pathogenetic process makes it possible to design an effective screening strategy. Observational studies based on time trend and geographical differences, as well more analytical case–control and cohort studies have shown that screening with Pap test can effectively prevent the incidence of cervical cancer through the identification and conservative treatment of preinvasive lesions (IARC, 2005). The impact of population-based programs with high test coverage of the population can be very strong, with a > 80% reduction of incidence (Anttila, 2007; Quinn et al., 1999). The prevalence of HPV infection (Fig. 1B), the most important risk factor, and the diffusion of effective screening programs, which is to date the most impactful preventive measure, explain most of the differences in cervical cancer incidence (Li et al., 2011; Giorgi Rossi and Ronco, 2013). In particular, countries with low screening coverage or ineffective screening programs show high variability in cervical cancer incidence, according to the prevalence of hrHPV infection in the population in previous decades, while countries with high screening test coverage and effective programs (Fig. 1C), which assure correct follow up and treatment, all have incidence below 10/100,000, with very small differences even when HPV prevalence is very different (Giorgi Rossi and Ronco, 2013). This phenomenon can be observed in women immigrating to high-income countries as well: they maintain the risk of their origin country for about 10 years or more after arrival (high if coming from countries with high prevalence of HPV and low if they come from low prevalence regions); afterward, thanks to having undergone more than one screening round, their excess risk is dramatically reduced. This is true even for women from high prevalence countries (Azerkan et al., 2008; Mousavi et al., 2012; di Felice et al., 2015). The sexual revolution at the end of 20th century caused an epidemic of HPV infection in most western countries, but the consequences of this epidemic in terms of cervical cancer incidence have been more or less completely controlled by the spread of wellorganized screening programs. Several epidemiologic studies provide evidence of this phenomenon (Bray et al., 2005a, b; Peto et al., 2004).

328

Cervical Cancer: Screening, Vaccination, and Preventive Strategies

Fig. 1 Worldwide distribution of cervical cancer and its determinants. (A) Age-standardized incidence rates of cervical cancer in the world (estimates for 2012). Rates per 100,000 women per year. For Sudan, South Sudan: Estimate for Sudan and South Sudan. Data sources: Ferlay, J., Soerjomataram, I., Ervik, M., et al. (2013). GLOBOCAN 2012 v1.2, Cancer Incidence and Mortality Worldwide: IARC CancerBase No. 11. Lyon, France: International Agency for Research on Cancer. (B). Prevalence of HPV among women with normal cervical cytology in the world. The samples for HPV testing come from cervical specimens (fresh/fixed biopsies or exfoliated cells). Systematic review updated 30 June 2015 (Bruni et al., 2017). (C) Worldwide status of cervical cancer screening programs. Availability of a cervical cancer screening program: Public national cervical cancer screening program in place (Cytology/VIA/HPV testing). Countries may have clinical guidelines or protocols, and cervical cancer screening services in the private sector but without a national public program. Publicly mandated programs have a law, official regulation, decision, directive or recommendation that provides the public mandate to implement the program with an authorized screening test, examination interval, target group and funding and copayment determined. Self-reported quality assurance: Organized programs provide for a national or regional team responsible for

Cervical Cancer: Screening, Vaccination, and Preventive Strategies

329

Pathogenesis The Natural History of the Disease: Preinvasive Phase The identification of HPV infection as the necessary cause of cervical cancer has led to an enormous improvement in the knowledge of its pathogenesis. HPV infection has been identified as a carcinogen for other sites as well, but cervical cancer accounts for most of the HPV-related burden of disease and is the only cancer with virtually 100% of population attributable fraction to HPV. More than 200 HPV genotypes have been identified and grouped into different genera. Papillomaviruses are classified into types depending on L1 sequencing; a new type is defined when its genome differs by at least 10% from that of all other classified types (Burk et al., 2013). Out of the five genera of HPVs, four (beta, gamma, mu and nu) contain only viruses that infect cutaneous epithelia. The fifth, the alpha genus, is unique in that it contains HPVs tropic to both cutaneous and mucosal epithelia. The oncogenic HPVs are a subset of the mucosotropic alpha-HPVs. Most of our knowledge about the natural history of HPV infection comes from studies conducted on the epithelium of the uterine cervix. Twelve types (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, and 59), also known as high-risk types, have been classified as certainly carcinogenic to humans according to the International Agency for Research on Cancer (WHO and IARC, 2012). Low-risk types, including HPV6 or HPV11, generally cause benign diseases such as genital warts (Egawa and Doorbar, 2017), while other types classified as probably or possibly carcinogenic are rarely found in large series of cancers, so their oncogenicity remains to be clarified (IARC, 2007). In contrast to low-risk HPV types, hrHPV can transform infected cells, although only a minority of hrHPV-infected individuals will develop precancerous lesions or invasive carcinomas during their lives. The papillomavirus family is a remarkably heterogeneous group of viruses that share the same genome structure and organization (Bravo and Felez-Sanchez, 2015). A circular double-stranded DNA genome of approximately 8 kb is structured into three main regions: (a) the early region (E) encodes genes that are necessary for the viral cycle and has an important role in cell transformation (E1, E2, E4, E5, E6, and E7); (b) the late region (L) encodes the L1 and L2 capsid proteins; (c) the upstream regulatory protein, referred to also as the long control region, is a noncoding region containing the replication origin and transcription factorbinding sites that contribute to regulate DNA replication by controlling the viral gene transcription. The expression of E6 and E7, together with that of E1, E2, E4, and E5, is essential for viral genome replication and virion synthesis and release, but they also play a key role in cell transformation (Doorbar et al., 2012). The HPV life cycle begins with the infection of the basal layer through microtraumas that compromise the epithelial barrier (Stubenrauch and Laimins, 1999). HPV genome is maintained at low copy number in the infected host basal cells. Upon differentiation of epithelial cells, the virus replicates to a high copy number and expresses the capsid genes (L1 and L2), resulting in the production of new progeny virions that are released from the epithelial surface. For persistence, HPV must infect basal cells showing stem cell-like features that are still able to proliferate (Egawa et al., 2015). This phenomenon is far less common in low-risk HPV types. The epithelial transition zones, and especially the endo-/ectocervix, are regions more susceptible to carcinogenesis by HPV type (Yang et al., 2015). High-risk types are more prone to activate cell proliferation in basal and differentiated layers, thereby promoting the transition from a productive infection to an infection that can activate several pathways essential for epithelial transformation (de Sanjose et al., 2017). One plausible explanation of the increased oncogenic capacity of the high-risk types resides in the activity of the E6 and E7 oncoproteins. Although E6 and E7 activity is present in high- and low-risk types, its role in low-risk types is limited to increasing viral fitness and viral production and is largely insufficient to trigger the development of preneoplastic lesions and cancer (Schiffman et al., 2016). Key functions of the HPV oncogenes are immune evasion, E7-mediated degradation of pRB family members, E6-mediated degradation of p53 and PDZ binding domain proteins and E6-mediated upregulation of telomerase. The main role of early E6 and E7 proteins in the carcinogenic process is through their inhibition of p53 and pRB tumor suppressors (Psyrri and DiMaio, 2008; McLaughlin-Drubin and Munger, 2008). E6 functions also include the activation of telomerase activity and deregulation of pathways involved in immune system response, of epithelial differentiation, of cell proliferation and of survival signaling. Besides cell cycle deregulation and proliferation, E7 enhances genomic instability and promotes the accumulation of chromosomal abnormalities. The deregulation of the cell cycle, the activation of the telomerase activity and genomic instability create a favorable environment for epithelial cell transformation. HPV integration can also drive the carcinogenic process through the inactivation of the E2 expression, the main inhibitor of E6 and E7, and the disruption of host genes because of the viral

=

implementation and require providers to follow guidelines, rules, or standard operating procedures. They also define a quality assurance structure and mandate supervision and monitoring of the screening process. To evaluate impact, organized programs also require ascertainment of the population disease burden. Quality assurance consists of the management and coordination of the program throughout all levels of the screening process (invitation, testing, diagnosis and follow up of screen-positives) to assure that the program performs adequately and provides services that are effective and in line with program standards. The quality assurance structure is self-reported as part of the national cancer programs or plans. For some countries, when 90%) clear within a few years after acquisition (Rodríguez et al., 2010; Winer et al., 2011). Innate and adaptive immune responses are interplayers in the resolution of such common infections (Schiffman et al., 2016). A Th1 proinflammatory response in the genital tract has been suggested as a potential mechanism for clearance (Scott et al., 1999), but cell-mediated immune response may be the most common mechanism. Several recent studies have observed a differential pattern of the vaginal microbiome, with more abundant Lactobacillus spp. among women clearing the infection compared with those with persistent infections (Shannon et al., 2017; Brotman et al., 2014; di Paola et al., 2017). Others have identified bacterial vaginosis associated with persistence of HPV infections among pregnant women (Kero et al., 2017). More data are needed to fully understand how HPV infections are modulated by the vaginal environment. HPV antibodies acquired through natural infection are observed in a limited number of infected people. A systematic review of the acquired natural immunity against subsequent genital HPV infection identified a range of seroprevalence estimates against HPV16 of 6.2%–45.5%, with a relative modest protection for HPV16 reinfection of 0.72 (0.62–0.82); it is thus unlikely to play a major role in clearance (Beachler et al., 2016). In some circumstances, after follow up of a positive HPV test, HPV may simulate a cleared infection and thus go undetected. It is now accepted that HPV infection may be latent and controlled by a cellular immune surveillance environment in which viruses are kept in very low copy numbers, escaping its detection. Reactivation may occur under moderate (menopausal state, aging) to relevant immunosuppressive situations (Maglennon et al., 2014), with or without hormonal influence. This reactivation could explain the second peak in HPV prevalence observed in postmenopausal women in some populations. Only a proportion of persistent infections progress via several infection stages toward neoplasia (Moscicki et al., 2006, 2014; Nyitray et al., 2011). Persistent carcinogenic HPV infection clearly predicts the risk of cervical cancer in women (Ronco et al., 2014; Khan et al., 2005). Although persistence does not have a standardized definition, is a very relevant concept, as many populations undergoing screening using HPV tests will rely on measures of persistent infection. Munoz et al. defined persistence as those infections that last more than the median duration, but this concept is relevant for natural history studies. Others define persistence as two consecutive positive HPV DNA tests with undetermined time interval (Marks et al., 2012; Munoz et al., 2009). Besides age, strong risk factors for the acquisition and persistence of an HPV infection and the subsequent development of neoplasia are immunosuppression and tobacco abuse (Schiffman and Kjaer, 2003; Rositch et al., 2013). Irrespective of the time window, major determinants of HPV persistence are HPV type and viral load at first detection. It remains unclear whether age is a key element in persistence. However, in the prospective study by Munoz et al. previously described, median duration of HPV incident infection was longer for high-risk HPV types than for low-risk types and for HPV16. Women younger than 30 years had a longer mean duration for HPV16 infections (16.6 months), while women > 30 years had a significantly shorter duration (9.5 months). Others suggest that persistence increases with age (Castle et al., 2011a). Moreover, the presence of concomitant HPV infections, a phenomenon very common in younger populations, is reported not to influence the duration of infection (Campos et al., 2011). During the period of persistent infection, there is no apparent immune detection of the virus (Stanley, 2012). This is partly due to the viral life cycle itself, which ensures that high levels of viral activity occur only in the upper, differentiated cell layers that are not exposed to immune defenses.

Conclusions The pathogenesis of cervical cancer historically has been interpreted as a continuum from infection to cancer in which progression to a more severe lesion was always less probable than regression until invasive cancer. The nomenclature reflects this interpretation: from normal epithelium to CIN1, CIN2 and CIN3, or from low-grade lesions to high-grade lesions. As has emerged in this section, the molecular interpretation of these morphological stages progressively suggests that in fact two different pathways can be distinguished in the course of an infection: the permissive and the transforming mode (Fig. 2). Some indirect molecular and clinical evidence suggests that this distinction between the modes of infections occurs very early: the viral methylation pattern of infection without any colposcopically and histologically detected CIN that will develop into a CIN3 in the subsequent 3 years is more similar to that of CIN3 than to that of CIN1 and of many CIN2 (Ronco et al., 2017); the probability of progression to CIN2 or 3, given an hrHPV infection, is the same whether in the presence of no detectable histological abnormality or whether a CIN1 is found (Mittal et al., 2017, Giorgi Rossi et al., 2013; Castle et al., 2011b; Sørbye et al., 2011), suggesting that CIN1 is not a necessary step for progression to abortive infections, but only the manifestation of an HPV infection, and therefore it represents the same risk as any initial detection of the virus.

Cervical Cancer Prevention The identification of HPV infection as a necessary cause of cervical cancer has opened the way to two new prevention tools: vaccines and a molecular test for screening. Although the focus of this article is the transformation of screening, it is impossible to address

Cervical Cancer: Screening, Vaccination, and Preventive Strategies

333

this issue without a brief description of the characteristics of the available vaccine and of the main strategies adopted for HPV immunization worldwide. While HPV testing is an improvement on an existing preventive measure, that is, screening with Pap test, vaccination is a completely new approach. Despite the fact that vaccination has the potential to decrease or to entirely eliminate the need for screening in the future, the two interventions will both be used until the near future. In fact, the vaccine is not effective in women who are infected, and mass vaccination has thus far been implemented for teenagers only (Obel et al., 2015; Dorleans et al., 2010; de Vuyst et al., 2015; GAVI Alliance Board, 2012; Bosch et al., 2016). Finally, to date, most women have been vaccinated only for HPV 16 and 18 and are thus still at risk of cervical cancer for other HPV types; this reduced but not negligible risk will require different, less intensive screening (Ronco and Giorgi Rossi, 2017; Franco et al., 2006, 2009; Cuzick et al., 2006). Screening in vaccinated women and how to implement these preventive measures more quickly is the focus of the “Prospective Vision” section of this article.

Vaccination The existing vaccines and the proof of efficacy and safety There are three registered vaccines against HPV: Cervarix (GlaxoSmithKline GSK, London, UK) covers HPV16 and 18; Gardasil (Merck, Kenilworth, New Jersey, USA) covers HPV 16, 18 and the low-risk types 6 and 11; Gardasil9 (Merck, Kenilworth, New Jersey, USA) covers the four types covered by Gardasil plus five other high-risk types: 31, 33, 45, 52 and 59. All the vaccines have shown almost complete protection against persistent vaccine-type HPV infections and precancerous lesions in studies recruiting 16–26year-old women (Lehtinen et al., 2012; Joura et al., 2015). Efficacy was limited only to those women that were not infected at the time of vaccination. Studies on clinical outcomes have been conducted only in women older than 16, but the most interesting target population in order to prevent infections are preadolescent girls. Since it is impossible to observe clinically relevant outcomes in a population that is still not sexually active, studies in 9–12-year-old girls could only measure the immunogenicity of the vaccine. These so-called “bridging studies” have shown that vaccines are more immunogenic in girls than in young women (Schiller and Müller, 2015). So far, vaccines have shown long-term protection of up to at least 10 years (Elfstrom et al., 2014; Kjaer et al., 2009). More recent studies have shown protection also in women up to age 40 (Castellsagué et al., 2011; Skinner et al., 2014) and in males for precancerous anogenital lesions (Giuliano et al., 2011b). Cross protection against nonvaccine types, in particular against 31, 33 and 45 for Cervarix (Wheeler et al., 2011) and 31 for Gardasil (Brown et al., 2009), has been observed. The vaccines are safe, and no serious adverse effect associated with the vaccine has been demonstrated (Phillips et al., 2017).

Implementation strategies and their effectiveness Vaccination of girls The first and most adopted immunization policy against HPV has been vaccinating only girls before the onset of sexual activity (in most countries, around the 11th year of life) (Obel et al., 2015; Dorleans et al., 2010; de Vuyst et al., 2015). The rationale for this choice is to target cohorts that are virtually all HPV-naïve, but delaying the vaccination as long as possible to assure the longest vaccine protection possible, thereby avoiding any need for a booster dose. The first vaccination campaigns started in 2007 and 2008 in Australia and in some European countries; now, almost all industrialized countries include HPV in their routine vaccine schedule. This strategy has achieved high coverage in countries with active invitation and particularly in those with school-based campaigns (de Vuyst et al., 2015). Obviously, this strategy will show benefits, that is, cancer incidence reduction, only several years after the start of the vaccination campaigns, when these cohorts will reach the age where at least precancerous lesions are worth treating and when screening starts; the effect on cervical incidence will be observed even later. Great emphasis has been given to equity in HPV vaccination for two reasons. The first is the intrinsic value given to equity of access to effective preventive measures, which all health systems recognize, particularly when vaccination is financed with public money. The second reason is the possible detrimental interaction of the inequity of access both to vaccination and to screening (Venturelli et al., 2017; Malagón et al., 2015). In the plausible hypothesis that those girls who are not vaccinated when they are 12 will also be those that do not participate in screening as adults, the impact of vaccination on the burden of disease in the population will be small. In fact, most invasive cervical cancers in many industrialized countries occur in the few women that are not screened or are under-screened (Sung et al., 2000; Leyden et al., 2005; Morrell et al., 2005; Bos et al., 2006; Andrae et al., 2008; Ingemann-Hansen et al., 2008; Lönnberg et al., 2010; Zucchetto et al., 2013). Furthermore, these women usually come from socioeconomically or culturally disadvantaged groups (Chiu, 2003; Giorgi Rossi et al., 2014); if the health system fails to vaccinate these women, it is missing the opportunity to really impact disease burden. Many authors have argued that the only way to be sure that vaccination is actually reaching a consistent part of these women is to have high population coverage, but results are inconsistent (Venturelli et al., 2017; Drolet et al., 2016). Catch-up strategies To accelerate the impact of vaccination campaigns, many countries have organized catch-up strategies, vaccinating girls aged 16 or even older. In most countries, these campaigns have achieved low coverage; in some cases, however, they have been quite successful (Pollock et al., 2014; Brotherton et al., 2015). These vaccinated cohorts of young women now make it possible to measure the effectiveness of the vaccine; the results are confirming the high protection of the vaccine against the targeted infection and related lesions, and in those areas with high coverage of the population, there is evidence of herd immunity (Drolet et al., 2015).

334

Cervical Cancer: Screening, Vaccination, and Preventive Strategies

Vaccination of boys Herd immunity is the main driver for the proposal of universal vaccination, that is, including boys and girls. With growing evidence of the involvement of HPV in a relevant proportion of head and neck cancers, in particular oropharynx (Plummer et al., 2016), it has become clear that the burden of vaccine-preventable disease in males is not negligible (de Martel et al., 2017). The vaccine has proven to be effective also on precancerous lesions of genital mucosae in males (Palefsky et al., 2011). Nevertheless, most models still predict that the largest benefit of vaccinating boys is the indirect effect on cervical cancer in women due to the reduction of circulating virus and to establishing of herd immunity faster (Brisson et al., 2011). The opportunity and cost effectiveness of vaccinating boys is still under debate, with the extreme heterogeneity of policies adopted in industrialized countries reflecting this uncertainty. Given the very high incidence of HPV-related cancers in males having sex with males, there is consensus that vaccinating this group is opportune (Markowitz et al., 2014; Kirby, 2015; Sauvageau and Dufour-Turbis, 2016), even though implementing effective strategies to target high-risk populations without indirectly fostering discrimination or stigmatization is challenging.

Screening The characteristics of cervical cancer’s natural history have made screening with Pap test one of the most successful preventive interventions ever implemented in industrialized countries. As already discussed in the Introduction of this article, screening has completely changed the epidemiology of this disease in some countries, reducing both incidence and mortality, and making cervical cancer a rare disease. Nowadays there are three possible tests to screen for cervical cancer, according to WHO guidelines: Pap test, HPV test (alone or in parallel with Pap test, so-called co-testing) and the visual inspection of the cervix after application of acetic acid (VIA). VIA is recommended only for low-resource countries, as this type of screening is strongly influenced by the attempt to overcome the lack of human and technological resources. The strategies proposed for Pap test and HPV-based strategies have been developed for application in industrialized countries, with a few exceptions. The aim is to find the best balance between possible harms and benefits for women; sustainability is rarely an issue, given the very high cost effectiveness of cervical screening and that screening with very intensive protocols is already in place in most high-income countries. In the following paragraph we will describe the recommendations given by the European Commission guidelines (Arbyn et al., 2008), with their recent update (Anttila et al., 2015), as an example of guidelines on how to implement organized screening programs, and of the most recent US guidelines as defined by the most influential scientific societies (ACS, ASCCCP, ASCP) (Saslow et al., 2012) and by the US Preventive Services Task Force (Moyer and US Preventive Services Task Force, 2012). For low-income countries, we report the 2013 recommendations of the World Health Organization (WHO, 2013).

Proof of efficacy and effectiveness Pap test The use of Pap test began to spread throughout most industrialized countries starting in the late 1960s, reaching very high population coverage by the end of the last century. Its introduction occurred in the absence of any experimental evidence of efficacy. Nevertheless, many observational studies demonstrated its effectiveness in reducing incidence and mortality just a few years after its introduction (IARC, 2005). The most convincing trend study on the effectiveness of Pap test screening reports the case of Finland, where organized screening was introduced in 1969, when opportunistic uptake of Pap test was still very low. Just 10 years after screening introduction, incidence decreased eightfold in the age class targets of screening (Anttila, 2007). The introduction of Pap test in many other countries was more gradual and the decrease in incidence was less dramatic. Nevertheless, most studies showed that an increase in Pap test coverage corresponded to a sharp decrease in cervical cancer incidence (Quinn et al., 1999; Serraino et al., 2015; Giorgi Rossi et al., 2015; Nygård et al., 2002). Case–control and cohort studies confirmed the trend studies (Sasieni et al., 2009; Ronco et al., 2005; Nieminen et al., 1999), and highlighted the additional protection of organized screening compared to opportunistic screening. (Ronco et al., 2005; Nieminen et al., 1999). Some studies have also reported examples of ineffective screening with Pap test. Observational studies from Latin America showed very low impact on cervical cancer incidence (Herrero et al., 1992), mostly because of low participation by the target population (Hernández-Avila et al., 1998) and due to the low quality of cytology and follow-up procedures (Lazcano-Ponce et al., 1996). One large cluster randomized trial conducted in India found no reduction in cervical cancer incidence and mortality after only one Pap test compared to no intervention. However, the screening intervention in this trial was minimal and the authors suggested that the low effectiveness of Pap test and VIA could have been due to low compliance to follow up and treatment rates (Sankaranarayanan et al., 2009). HPV test HPV DNA in cervical cancer screening was first introduced as a triage test in cases of borderline (atypical squamous cells of undetermined significance, ASC-US) Pap tests (Solomon et al., 2001). A large body of evidence on HPV DNA test accuracy shows that it is much more sensitive, but less specific, than Pap test in detecting CIN2 or more severe lesions (Arbyn et al., 2012). Four large trials nested in population-based screening conducted during the first decade of this century tested whether the HPV test’s higher sensitivity could also result in better prevention than Pap test screening (Naucler et al., 2007; Kitchener et al., 2009; Ronco et al., 2010; Rijkaart et al., 2012). In fact, the main endpoint of these four trials was the detection of high-grade lesions after the screening test.

Cervical Cancer: Screening, Vaccination, and Preventive Strategies

335

Although none of these trials was powered to detect a difference in the incidence of invasive cancers after a negative screening test, an extremely rare event in regularly screened women, the pooled analysis of the four trials clearly showed a significant reduction in invasive cancer incidence in the HPV arm compared with the Pap test arm (Ronco et al., 2014). Furthermore, these trials were designed to gather information necessary to understand which screening algorithms could optimize the process, minimize colposcopy referral and the risk of overtreatment and maximize the preventive efficacy. Thus, the vast majority of women recruited in the experimental arm received both HPV and Pap test and were managed according to the results of both. The very small number of lesions found in women HPV-negative and Pap test-positive made it possible to definitely attribute cervical cancer incidence reduction to the HPV test and not to the combination of the two tests. On the other hand, some trials referred all HPV-positive women to colposcopy, regardless of Pap test results, while others referred only HPV- and Pap test-positive women to immediate colposcopy, the so-called triage strategy, and referred to short interval follow up women HPV-positive and cytology-negative. The reduction in CIN2 or more severe lesion at the following round was similar in those trials with immediate referral to colposcopy for all women and in those adopting a triage strategy (Ronco and Giorgi Rossi, 2017). VIA Visual inspection methods, including those with naked eye or with minimal magnification after acetic acid or Lugol, have been proposed as first-level screening test in low-resource settings because the cost of materials and reagents is very low and because no infrastructural resources are necessary. If included in a see-and-treat strategy, they can be used in a one-step screening that does not require follow up, an important consideration in remote rural areas. Several studies have compared the accuracy of VIA versus HPV or cytology. Older studies, included in the systematic review conducted by the IARC (IARC, 2005), showed similar sensitivity and specificity of VIA compared with HPV test and similar sensitivity and lower specificity than cytology. However, the sensitivity of HPV test in these studies was implausibly low (< 80%) compared to what large population-based screening trials have now established. The recent systematic review by the WHO working group on cervical cancer prevention estimated a pooled sensitivity for HPV test of 95% versus 79% for VIA and HPV specificity of 84% versus 87% for VIA (WHO, 2013). Apart from accuracy studies, one large randomized trial compared the effectiveness of one VIA, one HPV, or one cytology in a lifetime to no intervention at all. HPV was the only screening strategy that could reduce cervical cancer mortality with only one screening episode (Sankaranarayanan et al., 2009). As already commented for the Pap test, the researchers who conducted the study suggest that low compliance to treatment could explain the lack of effectiveness of VIA in this context. This could also explain the heterogeneity between this trial comparing VIA, Pap and HPV and the trials conducted in southern India a few years before, where VIA showed a 25% incidence reduction and 35% mortality reduction (Sankaranarayanan et al., 2007).

The protocols for screening now: Comparison of existing guidelines High-income countries Until 2010, according to guidelines developed for high-income countries the only recommended primary screening test was cytology, that is, the Pap test. It is still the most used test, even if most guidelines now recommend HPV as preferable primary screening test. The protocols now used for Pap test-based screening vary for starting age from 21 to 30 and for the interval from 1 to 5 years, with most guidelines recommending 3-year interval after a negative Pap test. The second-level test is colposcopy in order to guide biopsies. Treatment is only recommended for CIN2 or more severe lesions, while CIN1 should be followed up. Many guidelines recommend HPV test to triage ASC-US or borderline lesions. The trend is to introduce a less intensive protocol, primarily for two reasons: (1) the target populations in most industrialized countries has a very low prevalence after years of screening; (2) observational and modeling studies from Northern European countries have shown that screening programs with long intervals and very low colposcopy referral are extremely effective in reducing cancer risk (IARC, 2005). With very few exceptions (Canadian Task Force on Preventive Health Care et al., 2013; Hamashima et al., 2010), the most recent guidelines in industrialized countries recommend HPV DNA test as primary screening. In some guidelines, Pap test-based and HPVbased strategies are considered comparable options (Moyer and US Preventive Services Task Force, 2012), while in other guidelines, HPV-based strategies are clearly identified as preferable (Anttila et al., 2015; Saslow et al., 2012). Guidelines for the use of HPV are quite consistent regarding the starting age (not before 30) and interval for HPV-negative women (at least 5 years; Fig. 3). The main difference in the recommendations is the use of cotesting or a triage strategy. The European guidelines have adopted a triage strategy in which HPV is a stand-alone test, followed by cytology in case of positivity; women testing positive to cytology are referred to immediate colposcopy, while a follow up in 6 or 12 months is recommended for those with negative Pap test (Anttila et al., 2015). The US guidelines recommend cotesting with HPV and cytology in parallel; women who are HPV-positive and Pap test-negative are referred to repeat cotesting after 1 year, as are those who have negative HPV and lowgrade cytology, while those with high-grade cytology, regardless of the HPV result, and those who are HPV-positive and have any kind of cytological abnormality are referred to immediate colposcopy (Saslow et al., 2012; Massad et al., 2013). Low-income countries In 2013, the WHO released new guidelines for screening, proposing a clear decision-making flowchart. According to these recommendations, whenever enough resources are available, HPV-based screening should be preferred to VIA, and cytology is recommended only if an ongoing program fulfilling the quality assurance criteria is in place (Fig. 4). VIA should be chosen only if there are no resources to implement HPV-screening or as second-level test after a positive HPV test. In this context, screen-and-treat approach is

336

Cervical Cancer: Screening, Vaccination, and Preventive Strategies

(A)

(B)

Co-testing HPV + cyto

HPV DNA

HPV-

ASC-US L-SIL

HPV+/ Cytology-

Cytology

HPV16/181

H-SIL

3yy

>=5yy

5yy

Negative

HPV+ or Cyto+ colposcopy

5yy

Co-testing

colposcopy

Cytology or HPV or HPV with cyto triage3

co-testing All negative

Positive2

Negative

colposcopy

1y

colposcopy 1y

HPV+

HPV+/ Cytology+

6/12m

HPV-/ Cytology+

>=5yy

HPV-/ Cytology-

Positive colposcopy

HPV

Co-testing Co-testing

HPV

Fig. 3 Cervical cancer screening algorithms as recommended by US and European guidelines. (A) US algorithm according to USPSTF (Moyer and US Preventive Services Task Force, 2012) and multisociety guidelines by American Cancer Society, American Society for Colposcopy and Cervical Pathology, American Society of Clinical Pathology (Saslow et al., 2012). Management of HPV-negative/cytology-positive women is reported in the American Society for Colposcopy and Cervical Pathology guidelines (Massad et al., 2013). (B) European algorithm according to the 2015 supplements of the Quality Assurance Guidelines for Cervical Cancer Screening (Anttila et al., 2015). (1) Women with HPV 16 or 18 can be referred directly to colposcopy even if cytology is negative. (2) Positivity threshold can be set at abnormal squamous cells of undetermined significance (ASC-US) or at high-grade lesions only. Thus, immediate colposcopy or follow up are both possible options for borderline and low-grade lesions. (3) The following optimal managements are recommended for HPV-positive/cytology-negative women: cytology after 6/12 months; HPV followed by colposcopy if test positive not earlier than 12 months; HPV with cytology triage not earlier than 12 months.

recommended, that is, when there are no resources and facilities to perform a colposcopy and to obtain histological confirmation on biopsy, women who test positive can be treated immediately.

Treatment of precancerous lesions As described previously, although the management of women with abnormal findings in cervical cancer screening is a complex process, when a clinically relevant lesion is found, it must be treated. In high-income countries, treatment should always follow a histological assessment on biopsy; see-and-treat approach during colposcopy is not recommended at all (Arbyn et al., 2008; Massad et al., 2013). As explained above, CIN1 is no longer considered a precancerous lesion; treatment, therefore, is not recommended except for when there are some specific conditions, that is, cytological findings strongly suggesting the presence of a high-grade lesion that the biopsy has not captured or a lesion persisting > 24 months (Arbyn et al., 2008; Massad et al., 2013). Treatment options should always be discussed with the woman. Very few studies have directly measured the progression rate of CIN2, and even fewer that of CIN3. Only one unfortunate cohort of untreated CIN3 had a long enough follow up to measure a 30% probability of progression to cancer in 30 years (McCredie et al., 2008). For CIN2, some studies reporting the rate of regression with a relatively short follow up have inconsistent results (Uchimura et al., 2012; Moscicki et al., 2010b); certainly, the probability of regression is higher in women below 35 years of age (Ronco et al., 2006, 2008). CIN3 are considered lesions with a relevant risk of progression for which the balance between the benefit of preventing invasive cancer is definitely greater than the harms of the treatment. With a few exceptions, treatment is recommended by almost all guidelines for CIN2 as well. In some cases, strict follow up may be an option for young women because they have a higher probability of rapid regression and progression to cancer usually takes a long time. The entire rationale for cervical cancer screening is based on the availability of very safe, noninvasive treatment options for precancerous lesions, making the balance between benefits and harms clearly in favor of the benefits, even when the probability of progression is quite low. Treatment options can be divided into ablative and excisional: in the former, as the tissue is destroyed, it is not available for postsurgery histological analysis; the latter maintains the integrity of the excised tissue. The most used

Cervical Cancer: Screening, Vaccination, and Preventive Strategies

337

Fig. 4 Decision-making flowchart for screening program managers as defined by the WHO guidelines for screening and treatment of precancerous lesions for cervical cancer prevention. Light pink bubbles identify the screening strategy, from the left: HPV test followed by VIA for HPV-positive to triage those to be treated immediately and those to be rescreened after 1 year; HPV test alone and VIA alone are screen-and-treat strategies; cytology or HPV followed by colposcopy are strategies in which an assessment through colposcopy-guided biopsy is possible although not mandatory. No triage test is considered by these recommendations. HPV-negative women should be rescreened after at least 5 years; VIA or cytology-negative women should be rescreened after 3–5 years. From WHO. (2013). WHO guidelines for screening and treatment of precancerous lesions for cervical cancer prevention. Geneva: World Health Organization.

commonly techniques today are cryotherapy and laser ablation for ablative treatments, and large loop excision of transformation zone (LLETZ or LEEP, loop electrosurgical excision procedure), laser excisional conization, needle excision of the transformation zone (NETZ) and cold knife conization for the excisional treatments. These techniques are all effective, and a recent systematic review did not find strong differences in recurrence rate (IARC, 2005; WHO, 2014). The treatment-related morbidity is minimal and similar for all these outpatient procedures, except for cold knife conization, which is a more invasive technique and is now recommended only when other options are not feasible. However, these interventions have all proven to have an impact on pregnancy, doubling the probability of preterm birth. It is difficult to quantify precisely the effect of treatment on this outcome because CIN2 and 3 lesions are also associated with preterm birth; it is thus difficult to distinguish between the excess risk due to treatment and that related to the preexisting lesion itself. This excess risk is higher for cold knife conization, while the risk is very similar for other types of treatment. Clearly, the risk increases with the depth of the tissue excised (Kyrgiou et al., 2016). Even though the harms of these treatments are very small compared to the consequence of an invasive cervical cancer, they are not negligible for women in their reproductive age. Therefore, the screening algorithm should minimize the probability of treating regressive lesions and false-positive lesions on biopsy, two different phenomena that nevertheless lead to the same kind of harms. False positives are an unavoidable consequence of any test; as for most tests, they are more frequent when the prevalence of the disease is low. Therefore, we can minimize false positives if we refer to colposcopy only those women with high prevalence of CIN2 or 3. In fact, it has been demonstrated that the probability of a false-positive biopsy, and thus of unnecessary treatment, increases when we refer to colposcopy women who have a very low probability of precancer, for example, HPV-negative women (Dalla Palma et al., 2008). The other cause of unnecessary treatment is over-diagnosis of regressive lesions. As mentioned previously, cervical cancer screening aims to treat precancer to avoid the incidence of cancer, and we accept that many of the lesions we treat will never become a cancer. Nevertheless, we should define protocols that try to minimize the diagnosis of regressive lesions to maintain the incidence of cancer as low as possible. Data from one of the European trials on HPV as primary screening, the ARTISTIC UK trial (Kitchener et al., 2009), clearly shows that lesions found in HPV-negative women are more regressive than HPV-positive lesions, at least when cytologists classify as positive a large proportion of the women, that is, they have low specificity. All these considerations led the working group that developed the European guidelines issued in 2015 (Anttila et al., 2015) to recommend against cotesting as primary screening strategy and to recommend an algorithm with HPV stand-alone followed by cytology triage. In this algorithm, only HPV-positive women are referred to colposcopy.

338

Cervical Cancer: Screening, Vaccination, and Preventive Strategies

Prospective Vision Screening in the Vaccine Era Screening should be adapted to the new disease epidemiology that will emerge in those countries that have high vaccination coverage against HPV16/18 when cohorts of women who were vaccinated in their 12th year of life arrive at the target age of screening. For those countries that started vaccination campaigns in 2007/2008, this will happen in the early 2020s. The main changes will be:

• • • •

> 60% decrease in the prevalence of clinically relevant cervical lesions (Bruni et al., 2017; Li et al., 2011). About 25% decrease in low-grade and borderline lesions, which are the vast majority of abnormal Pap tests (Bruni et al., 2017; Li et al., 2011). As a consequence, a strong decrease in the positive predictive value of Pap test (Kiviat et al., 2008). Pap test sensitivity will also probably decrease because finding lesions among an enormous number of negative slides is as difficult as finding a needle in a haystack (Franco et al., 2009; Evans et al., 2011), the so-called prevalence bias.

Given that Pap test is still the only recommended screening test below the age of 30, this is an issue. It is worth to note that also the HPV test will decrease its positive predictive value, but much less than Pap test, since the positive predictive value of HPV test is not influenced by the prevalence of the disease (Giorgi Rossi et al., 2012). On the other hand, we know that vaccinated women will have virtually no lesions from HPV16/18. These two HPV types are responsible for most cervical cancers overall, and in particular for early onset cancers, probably because their greater ability to transform the cell (Schiffman and Wentzensen, 2013) corresponds to a faster transition through the several progressive phases of the cancerogenic process and thus to a shorter lag time between infection and cancer. A lower absolute risk and longer lag time than in nonvaccinated women suggest that screening can safely be started later than today’s protocols. The only recommendation produced to date according to an evidence-based process is to delay the age to start screening. This is based on the estimate that starting screening at age 30 in vaccinated women would maintain the same or slightly smaller number of prescreening cancers now observed in nonvaccinated women starting at 25 (Giorgi Rossi et al., 2017). The same characteristicsdlow prevalence of disease and long lag time between infection and cancerdsuggest that longer intervals should be adopted to maintain the same level of protection and to avoid overtreatment-related harms (Fig. 5). The predicted low performance of Pap test and the probable increase of the screening starting age to 30 in those countries that now start screening at age 25 make the choice of HPV as primary test the most reasonable in vaccinated women. The scientific and public health community is discussing whether changes in screening protocols should be applied to the entire target population (one-size-fits-all strategy) or only to vaccinated women (tailored strategy). Obviously, vaccination status is a strong predictor of cervical cancer risk and can define the best screening algorithm for each woman; in the framework of an organized screening program, information on administered vaccine doses should be linked with screening history to correctly invite vaccinated and nonvaccinated women according to two different algorithms (Giorgi Rossi et al., 2017). On the other hand, herd immunity has been observed when coverage is high. Thus, beyond a certain level of vaccination coverage, the risk in nonvaccinated women will also be very low. Which coverage level and how long it should be sustained are subject matters for research. The answers are probably country-specific, as local sexual habits and social norms are important contributing factors (Baussano et al., 2016). The here-described changes in screening protocols were developed to apply to women vaccinated before their sexual debut. For those vaccinated when they were already possibly HPV-infected, we know the vaccine is only effective in preventing new infections and has no effect on prevalent infections. To apply less intensive screening, algorithms should be designed that take this into consideration. Finally, the introduction of 9-valent vaccine will further reduce cervical cancer risk in vaccinated women. How to screen, and whether it is opportune to do so, will be the next challenges of the scientific community.

Accelerating the Worldwide Control of Cervical Cancer The proposed vaccination strategies, together with the very low diffusion of effective HPV-based screening in low-income countries, imply that the benefits of this revolution in cervical cancer prevention will only be enjoyed in the quite distant future. Trials of HPV vaccination in women aged up to 55 years have shown very high efficacy against HPV16/18-related precancer in women not infected by HPV16/18-DNA when vaccinated. A group of authoritative researchers have advocated for extending routine vaccination programs to women of up to 30 years, paired with the use of HPV-screening with at least one test at age 30 years or older in low-resource countries and up to 5/6 screening episodes in high-income countries (Fig. 5) (Bosch et al., 2016). Furthermore, HPV test has good accuracy also on self-collected samples (Arbyn et al., 2014), meaning that it is possible to develop new strategies to reach under-screened women, thereby increasing test coverage (Verdoodt et al., 2015). All these strategies together have the potential to accelerate the decline in cervical cancer incidence and mortality in countries in Central and Eastern Europe, Latin America, Asia and Africa, where the burden of disease is higher because of the absence or low effectiveness of screening programs.

Cervical Cancer: Screening, Vaccination, and Preventive Strategies

339

40

45

PAP

PAP

PAP

PAP

60

HPV

45

55

HPV

40

PAP

PAP

PAP

PAP

PAP

PAP

35

50

HPV

30

HPV

25

45

HPV

30

40

HPV

PAP

25

35

50

55

60

55

60

11

HPV

HPV

50

HPV

HPV

HPV

HPV

30

HPV

Catch-up vaccine

Catch-up vaccine

25

HPV

HPV

35

35

40

45

50

55

60

35

40

45

50

55

60

Catch-up vaccine HPV

vaccine vaccine

High-income Low-income

11

HPV

vaccine

Vaccine era

11

HPV faster

HPV

11

30

PAP

HPV era

25

PAP

High-income countries only

11

PAP

PAP

Pap era

25

30

Fig. 5 The evolution of cervical cancer prevention. From programs based on Pap test requiring repeat test every 3 years, we now have HPV-based screening, recommended only after the age of 30, but with longer intervals (at least 5 years). In the near future, we will need to integrate vaccination and screening, with vaccinated women undergoing less intensive screening protocols which start later and which adopt longer intervals. Finally, the extension of more effective and faster strategies to reduce cervical cancer incidence in low-income countries as well, as proposed by Bosch et al. (2016) with the HPV faster program.

Most cost-effectiveness models suggest that the health benefits of extending the age of the vaccination target population (Jit et al., 2014; Brisson et al., 2016) have a very low cost, although this is still a subject matter of research. Nevertheless, this extension requires resources that are difficult to obtain in most of those settings that would reap the largest benefits; budget impact is thus much more an issue than is cost effectiveness.

Acknowledgment We would like to thank Jacqueline M. Costa for the careful editing of the final text.

See also: Cancer Risk Reduction Through Lifestyle Changes.

References Andrae, B., Kemetli, L., Sparén, P., et al., 2008. Screening-preventable cervical cancer risks: Evidence from a nationwide audit in Sweden. Journal of the National Cancer Institute 100 (9), 622. Anttila, A., 2007. Cervical cancer screening is effectivedThe Finnish experience. Entre Nous 64, 26–28. Anttila, A., Arbyn, A., de Vuyst, H., et al. (Eds.), 2015. European guidelines for quality assurance in cervical cancer screening, 2nd ed. Office for Official Publications of the European Union, Luxembourg. Supplements. Arbyn, M., Anttila, A., Jordan, J., et al. (Eds.), 2008. European guidelines for quality assurance in cervical cancer screening, 2nd ed. Office for Official Publications of the European Communities, Luxembourg. Arbyn, M., Ronco, G., Anttila, A., et al., 2012. Evidence regarding human papillomavirus testing in secondary prevention of cervical cancer. Vaccine 30 (Suppl 5), F88–99. Arbyn, M., Verdoodt, F., Snijders, P.J., et al., 2014. Accuracy of human papillomavirus testing on self-collected versus clinician-collected samples: A meta-analysis. The Lancet Oncology 15, 172–183. Azerkan, F., Zendehdel, K., Tillgren, P., Faxelid, E., Sparén, P., 2008. Risk of cervical cancer among immigrants by age at immigration and follow-up time in Sweden, from 1968 to 2004. International Journal of Cancer 123, 2664–2670.

340

Cervical Cancer: Screening, Vaccination, and Preventive Strategies

Baussano, I., Elfström, K.M., Lazzarato, F., et al., 2013. Type-specific human papillomavirus biological features: Validated model-based estimates. PLoS One 8, e81171. Baussano, I., Lazzarato, F., Brisson, M., Franceschi, S., 2016. Human papillomavirus vaccination at a time of changing sexual behaviour. Emerging Infectious Diseases 22, 18–23. Beachler, D.C., Jenkins, G., Safaeian, M., et al., 2016. Natural acquired immunity against subsequent genital human papillomavirus infection: A systematic review and metaanalysis. The Journal of Infectious Diseases 213, 1444–1454. Bos, A.B., Rebolj, M., Habbema, J.D., van Ballegooijen, M., 2006. Nonattendance is still the main limitation for the effectiveness of screening for cervical cancer in the Netherlands. International Journal of Cancer 119, 2372–2375. Bosch, F.X., Robles, C., Díaz, M., et al., 2016. HPV-FASTER: Broadening the scope for prevention of HPV-related cancer. Nature Reviews. Clinical Oncology 13, 119–132. Bravo, I.G., Felez-Sanchez, M., 2015. Papillomaviruses: Viral evolution, cancer and evolutionary medicine. Evolution, Medicine, and Public Health 2015, 32–35. Bray, F., Carstensen, B., Møller, H., et al., 2005a. Incidence trends of adenocarcinoma of the cervix in 13 European countries. Cancer Epidemiology, Biomarkers & Prevention 14 (9), 2191. Bray, F., Loos, A.H., McCarron, P., et al., 2005b. Trends in cervical squamous cell carcinoma incidence in 13 European countries: Changing risk and the effects of screening. Cancer Epidemiology, Biomarkers & Prevention 14, 677–686. Brisson, M., van de Velde, N., Franco, E.L., Drolet, M., Boily, M.C., 2011. Incremental impact of adding boys to current human papillomavirus vaccination programs: Role of herd immunity. The Journal of Infectious Diseases 204, 372–376. Brisson, M., Bénard, É., Drolet, M., et al., 2016. Population-level impact, herd immunity, and elimination after human papillomavirus vaccination: A systematic review and metaanalysis of predictions from transmission-dynamic models. Lancet Public Health 1, e8–e17. Brotherton, J.M.L., Saville, A.M., May, C.L., Chappell, G., Gertig, D.M., 2015. Human papillomavirus vaccination is changing the epidemiology of high-grade cervical lesions in Australia. Cancer Causes & Control 26, 953–954. Brotman, R.M., Shardell, M.D., Gajer, P., et al., 2014. Interplay between the temporal dynamics of the vaginal microbiota and human papillomavirus detection. The Journal of Infectious Diseases 210, 1723–1733. Brown, D.R., Kjaer, S.K., Sigurdsson, K., et al., 2009. The impact of quadrivalent human papillomavirus (HPV; types 6, 11, 16, and 18) L1 virus-like particle vaccine on infection and disease due to oncogenic nonvaccine HPV types in generally HPV-naive women aged 16–26 years. The Journal of Infectious Diseases 199, 926–935. Bruni, L., Barrionuevo-Rosas, L., Albero, G. et al. (2017). Human papillomavirus and related diseases in the world. Summary report 27 July 2017 Barcelona: ICO Information Centre on HPV and Cancer (HPV Information Centre). (Accessed 13/11/2017) Burk, R.D., Harari, A., Chen, Z., 2013. Human papillomavirus genome variants. Virology 445, 232–243. Campos, N.G., Rodriguez, A.C., Castle, P.E., et al., 2011. Persistence of concurrent infections with multiple human papillomavirus types: A population-based cohort study. The Journal of Infectious Diseases 203, 823–827. Canadian Task Force on Preventive Health Care, Dickinson, J., Tsakonas, E., et al., 2013. Recommendations on screening for cervical cancer. CMAJ 185, 35–45. Castellsagué, X., Bosch, F.X., Munoz, N., 2002. Environmental co-factors in HPV carcinogenesis. Virus Research 89, 191–199. Castellsagué, X., Muñoz, N., Pitisuttithum, P., et al., 2011. End-of-study safety, immunogenicity, and efficacy of quadrivalent HPV (types 6, 11, 16, 18) recombinant vaccine in adult women 24–45 years of age. British Journal of Cancer 105, 28–37. Castle, P.E., Gage, J.C., Wheeler, C.M., Schiffman, M., 2011a. The clinical meaning of a cervical intraepithelial neoplasia grade 1 biopsy. Obstetrics and Gynecology 118, 1222–1229. Castle, P.E., Rodríguez, A.C., Burk, R.D., et al., 2011b. Long-term persistence of prevalently detected human papilloma-virus infections in the absence of detectable cervical precancer and cancer. The Journal of Infectious Diseases 203, 814–822. Chiu, L.F., 2003. Inequalities of access to cancer screening: A literature review. Cancer Screening Series N 1. NHS Cancer Screening Programmes, Sheffield. Clarke, M.A., Wentzensen, N., Mirabello, L., et al., 2012. Human papillomavirus DNA methylationas a potential biomarker for cervical cancer. Cancer Epidemiology, Biomarkers & Prevention 21, 2125–2137. Clifford, G.M., Gallus, S., Herrero, R., et al., 2005. Worldwide distribution of human papillomavirus types in cytologically normal women in the International Agency for Research on Cancer HPV prevalence surveys: A pooled analysis. Lancet 366, 991–998. Cuzick, J., Mayrand, M.H., Ronco, G., Snijders, P., Wardle, J., 2006. Chapter 10: New dimensions in cervical cancer screening. Vaccine 24 (Suppl 3), S90–97. Dalla Palma, P., Giorgi Rossi, P., Collina, G., et al., 2008. The risk of false-positive histology according to the reason for colposcopy referral in cervical cancer screening: A blind revision of all histologic lesions found in the NTCC trial. American Journal of Clinical Pathology 129, 75–80. Doeberitz, M., Vinokurova, S., 2009. Host factors in HPV-related carcinogenesis: Cellular mechanisms controlling HPV infections. Archives of Medical Research 40, 435–444. Doorbar, J., 2006. Molecular biology of human papillomavirus infection and cervical cancer. Clinical Science (London, England) 110, 525–541. Doorbar, J., Quint, W., Banks, L., et al., 2012. The biology and life-cycle of human papillomaviruses. Vaccine 30 (Suppl 5), F55–70. Dorleans, F., Giambi, C., Dematte, L., et al., 2010. The current state of introduction of human papillomavirus vaccination into national immunisation schedules in Europe: First results of the VENICE2 2010 survey. Euro Surveillance 15 pii: 19730. Drolet, M., Bénard, É., Boily, M.C., et al., 2015. Population level impact and herd effects following human papillomavirus vaccination programmes: A systematic review and metaanalysis. The Lancet Infectious Diseases 15, 565–580. Drolet, M., Deeks, S.L., Kliewer, E., et al., 2016. Can high overall human papillomavirus vaccination coverage hide sociodemographic inequalities? An ecological analysis in Canada. Vaccine 34, 1874–1880. Egawa, N., Doorbar, J., 2017. The low-risk papillomaviruses. Virus Research 231, 119–127. Egawa, N., Egawa, K., Griffin, H., et al., 2015. Human papillomaviruses; epithelial tropisms, and the development of neoplasia. Virus 7, 386–390. Elfstrom, K.M., Smelov, V., Johansson, A.L., et al., 2014. Long term duration of protective effect for HPV negative women: Follow-up of primary HPV screening randomised controlled trial. BMJ 348, g130. Evans, K.K., Tambouret, R.H., Evered, A., Wibur, D.C., Wolfe, J.M., 2011. Prevalence of abnormalities influences cytologists’ error rates in screening for cervical cancer. Archives of Pathology & Laboratory Medicine 135, 1557–1560. di Felice, E., Caroli, S., Paterlini, L., et al., 2015. Cervical cancer epidemiology in foreign women in Northern Italy: Role of human papillomavirus prevalence in country of origin. European Journal of Cancer Prevention 24, 223–230. Franco, E.L., Cuzick, J., Hildesheim, A., de Sanjosé, S., 2006. Chapter 20: Issues in planning cervical cancer screening in the era of HPV vaccination. Vaccine 24 (Suppl 3), S171–77. Franco, E.L., Mahmud, S.M., Tota, J., Ferenczy, A., Coutlée, F., 2009. The expected impact of HPV vaccination on the accuracy of cervical cancer screening: The need for a paradigm change. Archives of Medical Research 40, 478–485. GAVI Alliance Board, 2012. The announcement by the GAVI Alliance Board to support human papillomavirus (HPV) vaccines in the world’s poorest countries. Vaccine 30 (Suppl 4), xii. Giorgi Rossi, P., Ronco, G., 2013. Future of cervical cancer screening in Europe. Current Pharmaceutical Design 19, 1490–1497. Giorgi Rossi, P., Franceschi, S., Ronco, G., 2012. HPV prevalence and accuracy of HPV testing to detect high-grade cervical intraepithelial neoplasia. International Journal of Cancer 130, 1387–1394. Giorgi Rossi, P., Benevolo, M., Vocaturo, A., et al., 2013. Prognostic value of HPV E6/E7 mRNA assay in women with negative colposcopy or CIN1 histology result: A follow-up study. PLoS One 8, e57600. Giorgi Rossi, P., Baldacchini, F., Ronco, G., 2014. The possible effects on socio-economic inequalities of introducing HPV-testing as primary test in cervical cancer screening programs. Frontiers in Oncology 4, 20.

Cervical Cancer: Screening, Vaccination, and Preventive Strategies

341

Giorgi Rossi, P., Caroli, S., Mancini, S., et al., 2015. Screening history of cervical cancers in Emilia-Romagna, Italy: Defining priorities to improve cervical cancer screening. European Journal of Cancer Prevention 24, 128–134. Giorgi Rossi, P., Carozzi, F., Federici, A., et al., 2017. Cervical cancer screening in women vaccinated against human papillomavirus infection: Recommendations from a consensus conference. Preventive Medicine 98, 21–30. Giuliano, A.R., Lee, J.H., Fulp, W., et al., 2011a. Incidence and clearance of genital human papillomavirus infection in men (HIM): A cohort study. Lancet 377, 932–940. Giuliano, A.R., Palefsky, J.M., Goldstone, S., et al., 2011b. Efficacy of quadrivalent HPV vaccine against HPV infection and disease in males. The New England Journal of Medicine 364, 401–411. Gravitt, P.E., 2011. The known unknowns of HPV natural history. The Journal of Clinical Investigation 121, 4593–4599. Guan, P., Howell-Jones, R., Li, N., et al., 2012. Human papillomavirus types in 115,789 HPV-positive women: A metaanalysis from cervical infection to cancer. International Journal of Cancer 131, 2349–2359. Hamashima, C., Aoki, D., Miyagi, E., et al., 2010. The Japanese guideline for cervical cancer screening. Japanese Journal of Clinical Oncology 40, 485–502. zur Hausen, H., 1999. Immortalization of human cells and their malignant conversion by high risk human papillomavirus genotypes. Seminars in Cancer Biology 9, 405–411. Herfs M., Y. Yamamoto, A. Laury, X., et al. (2012). A discrete population of squamocolumnar junction cells implicated in the pathogenesis of cervical cancer. Proceedings of the National Academy of Sciences of the United States of America 109, 10516–10521. Hernández-Avila, M., Lazcano-Ponce, E.C., de Ruíz, P.A., Romieu, I., 1998. Evaluation of the cervical cancer screening programme in Mexico: A population-based case-control study. International Journal of Epidemiology 27, 370–376. Herrero, R., Brinton, L.A., Reeves, W.C., et al., 1992. Screening for cervical cancer in Latin America: A case-control study. International Journal of Epidemiology 21 (6), 1050. IARC, 2005. Handbooks of cancer prevention Vol. 10 Cervix cancer screening. IARC, Lyon. IARC, 2007. Working group on the evaluation of carcinogenic risks to humans, vol. 90. IARC, Lyon. Ingemann-Hansen, O., Lidang, M., Niemann, I., et al., 2008. Screening history of women with cervical cancer: A 6-year study in Aarhus, Denmark. British Journal of Cancer 98, 1292–1294. International Collaboration of Epidemiological Studies of Cervical Cancer, 2006. Cervical carcinoma and reproductive factors: Collaborative reanalysis of individual data on 16,563 women with cervical carcinoma and 33,542 women without cervical carcinoma from 25 epidemiological studies. International Journal of Cancer 119, 1108–1124. International Collaboration of Epidemiological Studies of Cervical Cancer, Appleby, P., Beral, V., et al., 2007. Cervical cancer and hormonal contraceptives: Collaborative reanalysis of individual data for 16,573 women with cervical cancer and 35,509 women without cervical cancer from 24 epidemiological studies. Lancet 370, 1609–1621. Jit, M., Brisson, M., Portnoy, A., Hutubessy, R., 2014. Cost-effectiveness of female human papillomavirus vaccination in 179 countries: A PRIME modelling study. The Lancet Global Health 2, e406–14. Joura, E.A., Giuliano, A.R., Iversen, O.E., et al., 2015. A 9-valent HPV vaccine against infection and intraepithelial neoplasia in women. The New England Journal of Medicine 372, 711–723. Kalantari, M., Garcia-Carranca, A., Morales-Vazquez, C.D., et al., 2009. Laser capture microdissection of cervical human papillomavirus infections: Copy number of the virus in cancerous and normal tissue and heterogeneous DNA methylation. Virology 390, 261–267. Kanodia, S., Fahey, L.M., Kast, W.M., 2007. Mechanisms used by human papillomaviruses to escape the host immune response. Current Cancer Drug Targets 7, 79–89. Kelly, H., Mayaud, P., de Sanjose, S., 2015. Concomitant infection of HIV and HPV: What are the consequences? Current Obstetrics and Gynecology Reports 4, 213–219. Kero, K., Rautava, J., Syrjanen, K., et al., 2017. Association of asymptomatic bacterial vaginosis with persistence of female genital human papillomavirus infection. European Journal of Clinical Microbiology & Infectious Diseases 36, 2215–2219. Khan, M.J., Castle, P.E., Lorincz, A.T., et al., 2005. The elevated 10-year risk of cervical precancer and cancer in women with human papillomavirus (HPV) type 16 or 18 and the possible utility of type-specific HPV testing in clinical practice. Journal of the National Cancer Institute 97, 1072–1079. Kines, R.C., Thompson, C.D., Lowy, D.R., Schiller, J.T., Day, P.M., 2009. The initial steps leading to papillomavirus infection occur on the basement membrane prior to cell surface binding. Proceedings of the National Academy of Sciences of the United States of America 106, 20458–20463. Kirby, T., 2015. UK committee recommends HPV vaccination for MSM. The Lancet Oncology 16, e7. Kitchener, H.C., Almonte, M., Thomson, C., et al., 2009. HPV testing in combination with liquid-based cytology in primary cervical screening (ARTISTIC): A randomised controlled trial. The Lancet Oncology 10, 672–682. Kiviat, N.B., Hawes, S.E., Feng, Q., 2008. Screening for cervical cancer in the era of the HPV vaccinedThe urgent need for both new screening guidelines and new biomarkers. Journal of the National Cancer Institute 100, 290–291. Kjaer, S.K., Sigurdsson, K., Iversen, O.E., et al., 2009. A pooled analysis of continued prophylactic efficacy of quadrivalent human papillomavirus (types 6/11/16/18) vaccine against high-grade cervical and external genital lesions. Cancer Prevention Research (Philadelphia, Pa.) 2, 868–878. Kyrgiou, M., Athanasiou, A., Paraskevaidi, M., et al., 2016. Adverse obstetric outcomes after local treatment for cervical preinvasive and early invasive disease according to cone depth: Systematic review and meta-analysis. BMJ 354, i3633. Lazcano-Ponce, E.C., Rascón-Pacheco, R.A., Lozano-Ascencio, R., Velasco-Mondragón, H.E., 1996. Mortality from cervical carcinoma in Mexico: Impact of screening, 1980-1990. Acta Cytologica 40, 506–512. Lehtinen, M., Paavonen, J., Wheeler, C.M., et al., 2012. Overall efficacy of HPV-16/18 AS04-adjuvanted vaccine against grade 3 or greater cervical intraepithelial neoplasia: 4-year end-of-study analysis of the randomised, duble-blind PATRICIA trial. The Lancet Oncology 13, 89–99. Leyden, W.A., Manos, M.M., Geiger, A.M., et al., 2005. Cervical cancer in women with comprehensive health care access: Attributable factors in the screening process. Journal of the National Cancer Institute 97, 675–683. Li, N., Franceschi, S., Howell-Jones, R., Snijders, P.J., Clifford, G.M., 2011. Human papillomavirus type distribution in 30,848 invasive cervical cancers worldwide: Variation by geographical region, histological type and year of publication. International Journal of Cancer 128, 927–935. Lönnberg, S., Anttila, A., Kotaniemi-Talonen, L., et al., 2010. Low proportion of false-negative smears in the Finnish program for cervical cancer screening. Cancer Epidemiology, Biomarkers & Prevention 19, 381–387. Maglennon, G.A., McIntosh, P.B., Doorbar, J., 2014. Immuno-suppression facilitates the reactivation of latent papillomavirus infections. Journal of Virology 88, 710–716. Malagón, T., Drolet, M., Boily, M.C., Laprise, J.F., Brisson, M., 2015. Changing inequalities in cervical cancer: Modeling the impact of vaccine uptake, vaccine herd effects, and cervical cancer screening in the post-vaccination era. Cancer Epidemiology, Biomarkers & Prevention 24, 276–285. Maran, A., Amella, C.A., di Lorenzo, T.P., et al., 1995. Human papillomavirus type 11 transcripts are present at low abundance in latently infected respiratory tissues. Virology 212, 285–294. Markowitz, L.E., Dunne, E.F., Saraiya, M., et al., 2014. Human papillomavirus vaccination: Recommendations of the Advisory Committee on Immunization Practices (ACIP). MMWRdRecommendations and Reports 63 (RR-05), 1–30. Marks, M.A., Castle, P.E., Schiffman, M., et al., 2012. Evaluation of any or type-specific persistence of high-risk human papillomavirus for detecting cervical precancer. Journal of Clinical Microbiology 50, 300–306. de Martel, C., Plummer, M., Vignat, J., Franceschi, S., 2017. Worldwide burden of cancer attributable to HPV by site, country and HPV type. International Journal of Cancer 141, 664–670. Massad, L.S., Einstein, M.H., Huh, W.K., et al., 2013. 2012 ASCCP Consensus Guidelines Conference. 2012 updated consensus guidelines for the management of abnormal cervical cancer screening tests and cancer precursors. Journal of Lower Genital Tract Disease 17 (5 Suppl 1), S1–S27. McBride, A.A., Warburton, A., 2017. The role of integration in oncogenic progression of HPV-associated cancers. PLoS Pathogens 13, 1006211.

342

Cervical Cancer: Screening, Vaccination, and Preventive Strategies

McCredie, M.R., Sharples, K.J., Paul, C., et al., 2008. Natural history of cervical neoplasia and risk of invasive cancer in women with cervical intraepithelial neoplasia 3: A retrospective cohort study. The Lancet Oncology 9, 425–434. McLaughlin-Drubin, M.E., Munger, K., 2008. Viruses associated with human cancer. Biochimica et Biophysica Acta 1782, 127–150. Meijer, C.J., Berkhof, J., Castle, P.E., Hesselink, A.T., Franco, E.L., Ronco, G., Arbyn, M., Bosch, F.X., Cuzick, J., Dillner, J., Heideman, D.A., Snijders, P.J., 2009. Guidelines for human papillomavirus DNA test requirements for primary cervical cancer screening in women 30 years and older. International Journal of Cancer 124 (3), 516–520. https:// doi.org/10.1002/ijc.24010. Mirabello, L., Yeager, M., Cullen, M., et al., 2016. HPV16 sublineage associations with histology-specific cancer risk using HPV whole-genome sequences in 3200 women. Journal of the National Cancer Institute 29, 108. Mittal, S., Basu, P., Muwonge, R., et al., 2017. Risk of high-grade precancerous lesions and invasive cancers in high-risk HPV-positive women with normal cervix or CIN 1 at baseline-A population-based cohort study. International Journal of Cancer 140, 1850–1859. Morrell, S., Taylor, R., Wain, G., 2005. A study of Pap test history and histologically determined cervical cancer in NSW women, 1997–2003. Journal of Medical Screening 12, 190–196. Moscicki, A.B., Schiffman, M., Kjaer, S., Villa, L.L., 2006. Chapter 5: Updating the natural history of HPV and anogenital cancer. Vaccine 24 (Suppl 3), S3/42–51. Moscicki, A.B., Ma, Y., Jonte, J., et al., 2010a. The role of sexual behavior and human papillomavirus persistence in predicting repeated infections with new human papillomavirus types. Cancer Epidemiology, Biomarkers & Prevention 19, 2055–2065. Moscicki, A.B., Ma, Y., Wibbelsman, C., et al., 2010b. Rate of and risks for regression of cervical intraepithelial neoplasia 2 in adolescents and young women. Obstetrics and Gynecology 116, 1373–1380. Moscicki, A.B., Ma, Y., Farhat, S., et al., 2014. Natural history of anal human papillomavirus infection in heterosexual women and risks associated with persistence. Clinical Infectious Diseases 58, 804. Mousavi, S.M., Sundquist, K., Hemminki, K., 2012. Morbidity and mortality in gynecological cancers among first- and second-generation immigrants in Sweden. International Journal of Cancer 131, 497–504. Moyer, V.A., US Preventive Services Task Force, 2012. Screening for cervical cancer. U.S. Preventive Services Task Force recommendation statement. Annals of Internal Medicine 156, 880–891. Munoz, N., Hernandez-Suarez, G., Mendez, F., et al., 2009. Persistence of HPV infection and risk of high-grade cervical intraepithelial neoplasia in a cohort of Colombian women. British Journal of Cancer 100, 1184–1190. Naucler, P., Ryd, W., Tornberg, S., et al., 2007. Human papillomavirus and Papanicolaou tests to screen for cervical cancer. The New England Journal of Medicine 357, 1589–1597. Nieminen, P., Kallio, M., Anttila, A., Hakama, M., 1999. Organised vs. spontaneous Pap-smear screening for cervical cancer: A case-control study. International Journal of Cancer 83, 55–58. Nygård, J.F., Skare, G.B., Thoresen, S.Ø., 2002. The cervical cancer screening programme in Norway, 1992–2000: Changes in Pap smear coverage and incidence of cervical cancer. Journal of Medical Screening 9, 86–91. Nyitray, A.G., Carvalho da Silva, R.J., Baggio, M.L., et al., 2011. Six-month incidence, persistence, and factors associated with persistence of anal human papillomavirus in men: The HPV in men study. The Journal of Infectious Diseases 204, 1711–1722. Obel, J., McKenzie, J., Buenconsejo-Lum, L.E., et al., 2015. Mapping HPV vaccination and cervical cancer screening practice in the Pacific regiondstrengthening national and regional cervical cancer prevention. Asian Pacific Journal of Cancer Prevention 16, 3435–3442. Palefsky, J.M., Giuliano, A.R., Goldstone, S., et al., 2011. HPV vaccine against anal HPV infection and anal intraepithelial neoplasia. The New England Journal of Medicine 365, 1576–1585. di Paola, M., Sani, C., Clemente, A.M., et al., 2017. Characterization of cervico-vaginal microbiota in women developing persistent high-risk human papillomavirus infection. Scientific Reports 7, 10200. Peto, J., Gilham, C., Fletcher, O., Matthews, F.E., et al., 2004. The cervical cancer epidemic that screening has prevented in the UK. Lancet 364, 249–256. Phillips, A., Patel, C., Pillsbury, A., Brotherton, J., Macartney, K., 2017. Safety of human papillomavirus vaccines: An updated review. Drug Safety. https://doi.org/10.1007/s40264017-0625-z. Plummer, M., de Martel, C., Vignat, J., et al., 2016. Global burden of cancers attributable to infections in 2012: A synthetic analysis. The Lancet Global Health 4, e609–e616. Pollock, K.G., Kavanagh, K., Potts, A., et al., 2014. Reduction of low- and high-grade cervical abnormalities associated with high uptake of the HPV bivalent vaccine in Scotland. British Journal of Cancer 111, 1824–1830. Prigge, E.S., von Knebel Doeberitz, M., Reuschenbach, M., 2017. Clinical relevance and implications of HPV-induced neoplasia in different anatomical locations. Mutation Research, Reviews in Mutation Research 772, 51. Psyrri, A., DiMaio, D., 2008. Human papillomavirus in cervical and head-and-neck cancer. Nature Clinical Practice. Oncology 5, 24e31. Pyeon, D., Pearce, S.M., Lank, S.M., Ahlquist, P., Lambert, P.F., 2009. Establishment of human papillomavirus infection requires cell cycle progression. PLoS Pathogens 5, e1000318. Quinn, M., Babb, P., Jones, J., Allen, E., 1999. Effect of screening on incidence of and mortality from cancer of cervix in England: Evaluation based on routinely collected statistics. BMJ 318, 904–908. Rijkaart, D.C., Berkhof, J., Rozendaal, L., et al., 2012. Human papillomavirus testing for the detection of high-grade cervical intraepithelial neoplasia and cancer: Final results of the POBASCAM randomised controlled trial. The Lancet Oncology 13, 78–88. Rodríguez, A.C., Schiffman, M., Herrero, R., et al., 2010. Longitudinal study of human papillomavirus persistence and cervical intraepithelial neoplasia grade 2/3: Critical role of duration of infection. Journal of the National Cancer Institute 102, 315–324. Ronco, G., Giorgi Rossi, P., 2017. Role of HPV DNA testing in modern gynaecological practice. Best Practice & Research: Clinical Obstetrics & Gynaecology pii:S1521-6934, 30119-0. Ronco, G., Pilutti, S., Patriarca, S., et al., 2005. Impact of the introduction of organised screening for cervical cancer in Turin, Italy: Cancer incidence by screening history 1992-98. British Journal of Cancer 93, 376–378. Ronco, G., Giorgi-Rossi, P., Carozzi, F., et al., 2006. Human papillomavirus testing and liquid-based cytology in primary screening of women younger than 35 years: Results at recruitment for a randomised controlled trial. The Lancet Oncology 7, 547–555. Ronco, G., Giorgi-Rossi, P., Carozzi, F., et al., 2008. Results at recruitment from a randomized controlled trial comparing human papillomavirus testing alone with conventional cytology as the primary cervical cancer screening test. Journal of the National Cancer Institute 100, 492–501. Ronco, G., Giorgi-Rossi, P., Carozzi, F., et al., 2010. Efficacy of human papillomavirus testing for the detection of invasive cervical cancers and cervical intraepithelial neoplasia: A randomised controlled trial. The Lancet Oncology 11, 249–257. Ronco, G., Dillner, J., Elfstrom, K.M., et al., 2014. Efficacy of HPV-based screening for prevention of invasive cervical cancer: Follow-up of four European randomised controlled trials. Lancet 383, 524–532. Ronco, G., Fiano, V., Grasso, C., et al., 2017. In: Association of HPV methylation with present and future high grade cervical lesions in the NTCC cohort. 31st International Papillomavirus Conference & Clinical and Public Health Workshops. International Papillomavirus Conference, Cape town. Rositch, A.F., Koshiol, J., Hudgens, M.G., et al., 2013. Patterns of persistent genital human papillomavirus infection among women worldwide: A literature review and meta-analysis. International Journal of Cancer 133, 1271–1285.

Cervical Cancer: Screening, Vaccination, and Preventive Strategies

343

de Sanjosé, S., Diaz, M., Castellsagué, X., et al., 2007. Worldwide prevalence and genotype distribution of cervical human papillomavirus DNA in women with normal cytology: A meta-analysis. The Lancet Infectious Diseases 7, 453–459. de Sanjose, S., Quint, W.G.V., Alemany, L., et al., 2010. Human papillomavirus genotype attribution in invasive cervical cancer: A retrospective cross-sectional worldwide study. The Lancet Oncology 11, 1048–1056. de Sanjose, S., Brotons, M., Pavón, M.A., 2017. The natural history of human papillomavirus infection. Best Practice & Research. Clinical Obstetrics & gynaecology pii: S15216934, 30133-5. Sankaranarayanan, R., Esmy, P.O., Rajkumar, R., et al., 2007. Effect of visual screening on cervical cancer incidence and mortality in Tamil Nadu, India: A cluster-randomised trial. Lancet 370, 398–406. Sankaranarayanan, R., Nene, B.M., Shastri, S.S., et al., 2009. HPV screening for cervical cancer in rural India. The New England Journal of Medicine 360, 1385–1394. Sasieni, P., Castanon, A., Cuzick, J., 2009. Effectiveness of cervical screening with age: Population based case-control study of prospectively recorded data. BMJ 339, b2968. Saslow, D., Solomon, D., Lawson, H.W., et al., 2012. American Cancer Society, American Society for Colposcopy and Cervical Pathology, and American Society for Clinical Pathology screening guidelines for the prevention and early detection of cervical cancer. American Journal of Clinical Pathology 137, 516–542. Sauvageau, C., Dufour-Turbis, C., 2016. HPV vaccination for MSM: Synthesis of the evidence and recommendations from the Québec Immunization Committee. Human Vaccines & Immunotherapeutics 12, 1560–1565. Schiffman, M., Kjaer, S.K., 2003. Chapter 2: Natural history of anogenital human papillomavirus infection and neoplasia. Journal of the National Cancer Institute. Monographs (31), 14–19. Schiffman, M., Wentzensen, N., 2013. Human papillomavirus infection and the multistage carcinogenesis of cervical cancer. Cancer Epidemiology, Biomarkers & Prevention 22, 553–560. Schiffman, M., Doorbar, J., Wentzensen, N., et al., 2016. Carcinogenic human papillomavirus infection. Nature Reviews Disease Primers 2, 16086. Schiller, J.T., Müller, M., 2015. Next generation prophylactic human papillomavirus vaccines. The Lancet Oncology 16, e217–e225. Scott, M., Stites, D.P., Moscicki, A.B., 1999. Th1 cytokine patterns in cervical human papillomavirus infection. Clinical and Diagnostic Laboratory Immunology 6, 751–755. Serraino, D., Gini, A., Taborelli, M., et al., 2015. Changes in cervical cancer incidence following the introduction of organized screening in Italy. Preventive Medicine 75, 56–63. Shannon, B., Yi, T.J., Perusini, S., et al., 2017. Association of HPV infection and clearance with cervicovaginal immunology and the vaginal microbiota. Mucosal Immunology 10, 1310–1319. Skinner, S.R., Szarewski, A., Romanowski, B., et al., 2014. Efficacy, safety, and immunogenicity of the human papillomavirus 16/18 AS04-adjuvanted vaccine in women older than 25 years: 4-year interim follow-up of the phase 3, double-blind, randomised controlled VIVIANE study. Lancet 384, 2213–2227. Solomon, D., Schiffman, M., Tarone, R., ALTS Study Group, 2001. Comparison of three management strategies for patients with atypical squamous cells of undetermined significance: Baseline results from a randomized trial. Journal of the National Cancer Institute 93, 293–299. Sørbye, S.W., Arbyn, M., Fismen, S., Gutteberg, T.J., Mortensen, E.S., 2011. HPV E6/E7 mRNA testing is more specific than cytology in post-colposcopy follow-up of women with negative cervical biopsy. PLoS One 6, e26022. Stanley, M.A., 2012. Epithelial cell responses to infection with human papillomavirus. Clinical Microbiology Reviews 25, 215–222. Stanley, M., Pinto, L.A., Trimble, C., 2012. Human papillomavirus vaccine immune responses. Vaccine 30 (Suppl 5), F83–7. Stubenrauch, F., Laimins, L.A., 1999. Human papillomavirus life cycle: Active and latent phases. Seminars in Cancer Biology 9, 379–386. Sung, H.Y., Kearney, K.A., Miller, M., et al., 2000. Papanicolaou smear history and diagnosis of invasive cervical carcinoma among members of a large prepaid health plan. Cancer 88, 2283–2289. Uchimura, N.S., Uchimura, T.T., Martins, J.P., Assakawa, F., Uchimura, L.Y., 2012. Evaluation of conservative management of high-grade cervical squamous intraepithelial lesion. Revista de Saude Publica 46, 466–471. Venturelli, F., Baldacchini, F., Campari, C., et al., 2017. Association between mothers’ screening uptake and daughters’ HPV vaccination: A quasi-experimental study on the effect of an active invitation campaign. BMJ Open 7, e016189. Verdoodt, F., Jentschke, M., Hillemanns, P., et al., 2015. Reaching women who do not participate in the regular cervical cancer screening programme by offering self-sampling kits: A systematic review and meta-analysis of randomised trials. European Journal of Cancer 51, 2375–2385. de Vuyst, H., Howell-Jones, R., Levy-Bruhl, D., Giorgi Rossi, P., Franceschi, S., 2015. Implementation of vaccination against human papillomavirus in Europe. In: Anttila, A., Arbyn, M., de Vuyst, H., et al. (Eds.), European guidelines for quality assurance in cervical cancer screening, 2nd ed. Office for Official Publications of the European Union, Luxembourg, pp. 109–137. Supplements. Walboomers, J.M., Jacobs, M.V., Manos, M.M., et al., 1999. Human papillomavirus is a necessary cause of invasive cervical cancer worldwide. The Journal of Pathology 189, 12–19. Wheeler, C.M., Castellsague, X., Garland, S.M., et al., 2011. Cross-protective efficacy of HPV-16/18 AS04-adjuvanted vaccine against cervical infection and precancer caused by non-vaccine oncogenic HPV types: 4-year end-of-study analysis of the randomised, double-blind PATRICIA trial. The Lancet Oncology 13, 100–110. WHO, 2013. WHO guidelines for screening and treatment of precancerous lesions for cervical cancer prevention. World Health Organization, Geneva. WHO, 2014. WHO guidelines for treatment of cervical intraepithelial neoplasia 2–3 and glandular adenocarcinoma in situ: Cryotherapy, large loop excision of the transformation zone (LEEP/LLETZ), and cold knife conization. World Health Organization, Geneva. WHO & IARC, 2012. IARC monographs on the evaluation of carcinogenic risks to humans. Biological agents. In: Volume 100B: A review of human carcinogens. IARC, Lyon. Winer, R.L., Feng, Q., Hughes, J.P., et al., 2008. Risk of female human papillomavirus acquisition associated with first male sex partner. The Journal of Infectious Diseases 197, 279–822. Winer, R.L., Hughes, J.P., Feng, Q., et al., 2011. Early natural history of incident, type-specific human papillomavirus infections in newly sexually active young women. Cancer Epidemiology, Biomarkers & Prevention 20, 699–707. Yang, E.J., Quick, M.C., Hanamornroongruang, S., et al., 2015. Microanatomy of the cervical and anorectal squamocolumnar junctions: A proposed model for anatomical differences in HPV-related cancer risk. Modern Pathology 28, 994–1000. Zucchetto, A., Ronco, G., Giorgi Rossi, P., et al., 2013. Screening patterns within organized programs and survival of Italian women with invasive cervical cancer. Preventive Medicine 57, 220–226.

Further Reading Anon, 2014. Human papillomavirus vaccines: WHO position paper, October 2014. WHO Weekly Epidemiological Record 89, 465–492. Ferlay, J., Soerjomataram, I., Ervik, M., et al., 2013. GLOBOCAN 2012 v1.2, cancer incidence and mortality worldwide: IARC CancerBase No. 11. International Agency for Research on Cancer, Lyon, France. Lehtinen, M., Dillner, J., 2013. Clinical trials of human papillomavirus vaccines and beyond. Nature Reviews. Clinical Oncology 10, 400–410.

344

Cervical Cancer: Screening, Vaccination, and Preventive Strategies

Relevant Websites http://www.hpvcentre.net/index.phpdHPV information centre. http://www.iarc.fr/dInternational Agency for Research on Cancer. http://www.who.int/reproductivehealth/topics/cancers/en/dWorld Health Organization, Cervical Cancer. http://www.gavi.org/support/nvs/human-papillomavirus/dGAVI the vaccine alliance. http://www.cervicalcanceraction.org/home/home.phpdCervical Cancer Action. https://www.cancer.org/health-care-professionals/american-cancer-society-prevention-early-detection-guidelines/cervical-cancer-screening-guidelines.htmldAmerican Cancer Society. http://www.asccp.org/asccp-guidelinesdSociety of Lower Genital Tract Disorders ASCCP guidelines (formerly American Society for Colposcopy and Cervical Pathology).

Chemoprevention of Cancer: An Overview of Promising Agents and Current Research Elizabeth G Ratcliffe and Andrew D Higham, University Hospitals of Morecambe Bay NHS Foundation Trust, Lancaster, United Kingdom Janusz A Jankowski, University Hospitals of Morecambe Bay NHS Foundation Trust, Lancaster, United Kingdom; National Institute of Health and Care Excellence, Manchester, United Kingdom; and Royal College of Surgeons in Ireland, Dublin, Ireland © 2019 Elsevier Inc. All rights reserved.

Cancer diagnosis and treatment is distressing for patients and carries a high burden of morbidity and mortality. Preventable cases of cancer can be reduced through lifestyle changes and to augment this chemoprevention is becoming an increasingly vital area of research. The principle of chemoprevention is to find a medication or dietary supplement which when given to certain groups of the population can reduce the risk of carcinogenesis before invasion. Optimum agents need to be well understood from a pharmacokinetic and safety perspective, be cost effective for broad usage and be well tolerated by patients. A key draw back is chemoprevention agents must be taken for prolonged periods, creating a unique conundrum for clinicians in that within a population, for a proportion of people cancer may be delayed by a few years, yet this benefit may be outweighed by healthy patients who succumb to complications. Hence many agents that are studied are concurrently used for secondary prevention of other conditions and therefore may have multiple benefits to balance the risks (Fig. 1). Here we give an overview of some of the potential chemoprevention agents (Table 1), with an emphasis on aspirin which has shown the most promise.

Statins HMG-CoA reductase inhibitors act against the conversion of HMG CoA to mevalonate, thought to prevent disruption to Ras/Rho signaling proteins which in turn can affect normal cell differentiation, survival and growth. Statins also regulate the RAF–MAPK– ERK pathway which makes them proapoptotic, anti-inflammatory and immunomodulatory via both HMG-CoA independent and dependent pathways. Statins have possible associations with colorectal (CRC), hepatocellular (HCC), gastric, esophageal, and advanced prostate cancers. Tsan et al. in a population-based cohort study of chronic hepatitis found a 53% reduction in risk of HCC in patients on statins. In a meta-analysis of 10 studies by Singh et al. (2013) the statin group showed a 37% reduction in HCC (adjusted OR 0.63; 95% CI 0.52–0.76) though the number needed to treat (NNT) was 5209 in East–Asian men. Statins are well tolerated; common side effects include myalgia and deranged liver function tests. Concerns have been raised regarding an increased cancer risk in the elderly. Moreover a large epidemiological study into secondary effects of statins showed NNT (1266) to reduce esophageal cancer were markedly offset by number needed to harm (NNH) with severe liver derangement(136), moderate to severe myopathy (91 men 259 women) and renal impairment (434). Statins are already widely used and are an attractive chemoprevention agent given their high efficacy at lowering all-cause mortality, particularly cardiovascular, and cerebrovascular. Persuading the public of a firm association with a specific cancer will require further research to justify the risks.

Metformin The metabolic syndrome, sedentary lifestyle and high energy low complexity western diet have been linked with many cancers. Malignant cells are highly catabolic, and links between metabolic pathways, such as the AMPK/LKB1, and how these maybe affected by metformin chemoprevention are driving research. An energy stress response dependent on AMPK triggered by metformin is thought to reduce survival in cancer cell lines. A secondary effect of reducing proliferation of cells via inhibition of phosphoinositide 3-kinase/Akt/mammalian rapamycin signaling has been described. Meta-analysis of retrospective data for diabetic patients showed an overall reduction in all cancers (31%), this was most evident for pancreatic and HCC. A nonsignificant association was seen with breast, prostate, and CRC cancers. Meta-analysis of studies looking at HCC risk showed a 50% reduction of risk in patients who had ever used metformin versus those who had never, even accounting for adjustments for other medications. Higurashi et al. performed a multicenter double-blind, randomized controlled trial of metformin in nondiabetic patients and showed a significantly lower colonic adenoma recurrence rate (P ¼ .034, risk ratio 0.67 [95% CI 0.47–0.97]).

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65130-9

345

346

Chemoprevention of Cancer: An Overview of Promising Agents and Current Research

Cost Side effects Numbers needed to harm Tolerability Duration necessary for effect

Fig. 1

Safe Cheap Effective Multiple actions Works against common cancer

Considerations when choosing a chemoprevention agent.

Table 1

Overview of agents discussed

Agent

Cancer prevented

Risks

Statins

CRC, HCC, gastric, esophageal, prostate

Metformin Vitamin D Tamoxifen Aspirin

Pancreatic, HCC, all cancers Breast, esophageal, CRC ER positive breast cancer CRC, esophageal, hereditary CRC, breast, ovarian, pancreatic, prostate, lung

Liver injury, myopathy, renal derangement, increased cancer risk in the elderly Diarrhea, nausea, abdominal discomfort Increased risk of adenocarcinoma of esophagus VTE, Endometrial cancer, cataract formation GI Bleeding, Intracranial hemorrhage, all bleeding

Metformin can be troubling for patients with many experiencing diarrhea, nausea, and abdominal discomfort. More serious adverse effects such as lactic acidosis are dose dependent and studies into optimum dosing are still needed. The risk reduction of other conditions makes metformin a good agent which for the general public could seem appealing. The rising incidence of diabetes and risk of obesity related illnesses including cancers such as colorectal and postmenopausal breast cancer increase the case for its use in chemoprevention.

Vitamin D The mechanism explored in in vitro settings suggests vitamin D levels can affect apoptosis, cell proliferation, and differentiation. Dietary vitamin D is hydroxylated to its’ circulating form 25-hydroxyvitamin-D which can be measured in the circulation, the active form (1,25-HO-vitD) is created through a second hydroxylation allowing it to bind to VDR which results in a biological signaling cascade. Vitamin D levels have been linked to reductions in breast, esophageal and CRC among others. Dietary vitamin D intake showed a possible relationship that was protective for squamous cell carcinoma of the esophagus but increased risk of adenocarcinoma (nonsignificant). Gandhini et al. through meta-analysis found an inverse relationship between 25-HO-vit D and CRC but could not show significant associations with breast or prostate cancer. They felt some bias could come from many case-control studies taking vitamin D levels after cancer diagnosisdwhen patients’ diet, activity or sun exposure may have changed. Side effects from vitamin D are few, however in some studies an adverse link has been raised, for example an increased association with adenocarcinoma of the esophagus. This seems to particularly relate to the use of supplementation, whereas studies looking at lifetime UV exposure seemed to be more favorable in outcome. Vitamin D is certainly an attractive choice for patients, with few side effects as a supplement and when gained through sun exposure there is likely to be an association with physical activity, reducing other cancer risk factors such as sedentary behavior and obesity.

Tamoxifen Tamoxifen produces an antiestrogen effect in breast tissue and weakly estrogenic effects in the bone and uterus. Estrogen normally binds with the estrogen receptor (ER) which causes CoA to bind with the receptor creating a transcription complex. The complex is then able to commence DNA transcription from an estrogen-receptive gene. Tamoxifen binds with the ER-a part of the receptor

Chemoprevention of Cancer: An Overview of Promising Agents and Current Research

347

preventing it from recruiting CoA molecules therefore causing an antiestrogen effect in breast tissue. However, where CoA levels are greater a weakly estrogenic effect occursdin the uterus and bonedand through an association with fos/jun proteins stimulates the activating protein-1 (AP1) causing gene sequencing. Since these early mechanisms were described, studies into tamoxifen resistance have highlighted the importance of paired box-2 (PAX) in both signaling in breast tissue and the carcinogenesis in the endometrium. Patients who have been treated with tamoxifen were found to have fewer contralateral breast lesions than other patients raising the possibility of a role as a chemoprevention agent. Meta-analysis of 9 trials by Cuziak et al., found a significant risk reduction in estrogen receptor positive breast cancer in both average and higher risk women who had no prior disease. They calculated a NNT of 42. There was a nonsignificant increased risk of ER negative cases, which they hypothesized could relate to mutation from ER sensitive to hormone resistance. Significant adverse risks are associated with tamoxifen particularly venous thromboembolism (VTE) (reported up to twice as often as placebo) and endometrial cancer (2–5 times above the baseline rate). The risks are less for premenopausal women, with no increased risk of VTE or endometrial cancer, however in postmenopausal women there was an increase in VTE, Endometrial cancer and cataract formation. As a treatment of breast cancer tamoxifen is well tolerated, the routine side effects are few and as chemoprevention for high risk women it is recommended by NICE and approved for use by the FDA in the United States. For the wider, “healthy” population, the more serious complications, particularly endometrial cancer, will be harder to justify.

Aspirin The mechanism of chemoprevention via aspirin (Fig. 2) has been linked to prostaglandins, particularly PGE2, via COX signaling. Prostaglandin E2 has been associated with stimulating angiogenesis, increasing a cell’s proliferation, migration potential, and invasiveness whilst modulating immunity and preventing apoptosis. In animal colonic models deletion of receptors for PGE2 resulted in reduced abnormal crypt formation and adenomas. Nan et al. through gene versus environment case-control study in CRC identified two genes which showed a greater prevention outcome with aspirin and NSAIDS that are functionally linked with increased PGE2 production. Inhibition by aspirin of COX-2-mediated platelet production of thromboxane has been hypothesized to affect tissues’ response to damage caused by disruption by neoplasia resulting in reduced cell proliferation. COX-2 inhibition in epithelial cells occurs at higher doses than in platelets, however, chemopreventative effects have been shown at low doses too which supports platelet mediated mechanisms. Aspirin also reduces the cellular quantity of b-catenin through inactivation of protein phosphatase

Fig. 2 Mechanism of aspirin chemoprevention. A diagram showing the pathways hypothesized to be affected by aspirin, Green arrows represent stimulation pathways and red arrows inhibitory. Aspirin affects pathways in four ways: affecting platelet activation (top right) by inhibiting the release of thromboxane; inhibiting the stabilization of b-catenin through inactivation of protein phosphatase 2A (PP2A) promoting ubiquitylation (bottom right panel);through inhibiting COX-2 mediated production of prostanoids (bottom left) by preventing COX-2 converting arachidonic acid to prostaglandin E2 (PGE2); and through the antiinflammatory effect of reducing circulating cytokines (CXCL1, C-X-C motif chemokine 1; MIC1, macrophage inhibitory cytokine 1; IFNg, interferon-g; IL, interleukin; PGT, prostaglandin transporter; sTNFR2, soluble TNF receptor 2; TNF, tumor necrosis factor). Modified from a diagram by Drew, D. A., Cao, Y., Chan, A. T. (2016). Aspirin and colorectal cancer: The promise of precision chemoprevention. Nature Reviews Cancer 16(3), 173–186. doi: 10.1038/nrc.2016.4.

348

Current trials pending publication or completion

Status

Trial name

Cancer target

Agent

Phase Participants Masking

Completion Randomization date Locations

Completed

A randomized, double-blind, placebocontrolled, multicentre efficacy and safety study of toremifene citrate for the prevention of prostate cancer in men with high grade prostatic intraepithelial neoplasia (PIN) Physicians health study II: trial of vitamins in the chemoprevention of cancer, CVD and eye disease Phase III placebo-controlled chemoprevention study of oral cavity squamous cell carcinoma patients A study to assess the efficacy and safety of enteric-coated acetylsalicylic acid in patients at moderate risk of cardiovascular disease (ARRIVE) Phase I multiple-dose safety, pharmacokinetic and pharmacodynamics clinical study of nitric oxide releasing aspirin (NCX 4016) The seafood (systematic evaluation of aspirin and fish oil) polyp prevention trial A phase III, randomized, study of aspirin and esomeprazole chemoprevention in Barrett’s metaplasia (AspECT) Pioglitazone for lung cancer chemoprevention Aspirin in reducing events in the elderly (ASPREE) Metformin hydrochloride in preventing breast cancer in patients with atypical hyperplasia or in situ breast cancer

Prostate

Toremifene

3

1590

Triple

Yes

Feb-10

35 Study locations

Prostate, colorectal, all cancers

Vitamin E, C, multivitamin, beta carotene 13-Cis retinoic acid

N/a

14641

Factorial Yes assignment

Jun-11

National Cancer Institute US

3

342

Single

Yes

Sep-12

National Taiwan University

Colorectal cancer (secondary end point)

Aspirin

3

12546

Triple

Yes

Nov-16

Bayer, 673 International Centres

Colorectal cancer

Nitric-oxide releasing acetylsalicylic acid

1

240

Double

Yes

Dec-16

Stony Brook University

Colorectal cancer

Aspirin

N/a

755

Double

Yes

Oct-17

University of Leeds, UK Centres

Esophageal

Omeprazole, aspirin

3

2513

Open label

Yes

May-17

Oxford

Lung cancer

Pioglitazone

2

391

Quadruple

Yes

Aug-17

Colorado/Tennessee

Colorectal cancer

Aspirin

4

19000

Quadruple

Yes

Jan-18

Breast cancer

Metformin

3

Aim 128

Double

Yes

Feb-19

Berman Centre for Outcomes and Clinical Research Alliance for Clinical Trials in Oncology, United States

Completed Completed Completed

Completed

Completed Active

Active Active Active

Oral cancer

Chemoprevention of Cancer: An Overview of Promising Agents and Current Research

Table 2

Active

Active Recruiting Recruiting

Recruiting

Recruiting

Recruiting

Breast cancer

Tamoxifen

Breast cancer

Anastrozole

Colorectal cancer

Mesalamine

Gastric cancer

3

1884

Triple

Yes

Dec-19

European Institute of Oncology

3864

Quadruple

Yes

Jan-22

2

Aim 540

Double

Yes

Sep-20

Queen Mary London 75 study locations Medical University of Vienna

Eflornithine

2

Aim 300

Double

Yes

Dec-20

Colombia/Honduras/Puerto Rico

Colorectal cancer

Aspirin

3

Aim 2000 Double

Yes

2023

CRUK and International

Colorectal cancer

Aspirin

3

Aim 852

Quadruple

Yes

Dec-24

Hospital Avicenne France

Breast, gastric, colorectal, esophageal, prostate

Aspirin

3

11000

Triple

Yes

Oct-26

University College London

Colorectal cancer

Aspirin

N/a

Aim 180

Double

Yes

Jul-28

Massachusetts General Hospital

Chemoprevention of Cancer: An Overview of Promising Agents and Current Research

Recruiting

The HOT study: hormone replacement therapy opposed by low dose tamoxifen. A phase III trial of breast cancer prevention with low dose tamoxifen in HRT users International breast cancer intervention study Mesalamine for colorectal cancer prevention program in Lynch syndromedMesaCAPP Targeted chemoprevention of gastric carcinogenesis in high risk populations A trial looking at different doses of aspirin to prevent cancer in people who have Lynch syndrome (CaPP3) Assessment of the effect of a daily chemoprevention by low-dose aspirin of new or recurrent colorectal adenomas in patients with Lynch syndrome Add-aspirin: A trial assessing the effects of aspirin on disease recurrence and survival after primary therapy in common nonmetastatic solid tumors Aspirin intervention for the reduction of colorectal cancer risk (ASPiRED)

349

350

Chemoprevention of Cancer: An Overview of Promising Agents and Current Research

2Ada protein which breaks down amino acid residues (T41 and S45) that are responsible for flagging b-catenin for ubiquitylation. b-catenin when it moves to the nucleus of a cell causes a sequence of gene expression resulting in increased cell migration and proliferation. Most of the cancers which seem to be prevented by aspirin have an inflammatory prodrome, such as Barrett’s to esophageal, aspirin is thought to reduce circulating levels of inflammatory cytokines and proteins which have been linked with these cancers. Parkin et al. estimated that if aspirin was taken for 10 years between the ages of 50–65 years a reduction in all cancers of 7%–10% could be achieved. The strongest associations appear to be with CRC, esophageal, gastric (all by around 1/3 reduction), prostate, breast, and lung cancers (10%–15% reduction). Rothwell et al. in meta-analysis of eight trials showed a significant reduction in 20 year cancer mortality verses placebo for all solid cancers, GI cancers and the benefit increased with age and duration of aspirin use. CRC cohort studies on large populations showed an inverse relationship between aspirin use and CRC incidencedwith the Cancer Prevention Study II describing a reduction of 40% in regular aspirin users. The CAPP-2 randomized controlled trial in Lynch syndrome patients showed a significant reduction in CRC cancer risk in this population which was sustained after the end of the randomization to 4 years. Many studies show a reduction in esophageal cancer with aspirin, even reduction in Barrett’s length and dysplasia progression. Further studies are underway to look at use in specific high risk groups such as in the AspECT trialdthe largest RCT using PPI and aspirin in combination with end points of developing dysplasia and esophageal cancer. Aspirin for CRC chemoprevention has been endorsed by The US Preventative Services Task Force for use in 50–59 year olds with a 10% or greater cardiovascular 10 year risk profile and low bleeding risk, and also on an individual basis for 60–69 year olds, though emphasizing the need for further investigation into the mechanism of action. The consequences of long term aspirin use are well understood, chiefly increased risk of bleeding due to the unselective nature of aspirin’s COX inhibition. COX-1 inhibition reduces the secretion of prostaglandin E2 in the gastric mucosa, increasing ulceration and risk of upper gastrointestinal bleeding. Aspirin reduces thromboxane production in platelets hence reducing atherosclerotic risk but increasing bleeding risk. The risk of severe adverse effects increases dramatically after the age of 60. The international consensus paper suggested that given premalignant lesions often present in middle age- around age 50–60, commencing treatment between the ages of 40–50 years may help prevent most premalignant lesions before the risk of adverse events goes up. Cuzick et al. estimated that for 100 average risk 55 year olds taking aspirin for 10 years there would be 2.29 fewer myocardial infarctions, cancers, and strokes in men and 1.32 in women over 15 years, yet only 0.49 more GI bleeds in men and 0.25 in women. They also estimated a clear reduction in mortality from cancer which outweighed the mortality risk from GI bleeding (a ratio of 7:1 in the general population). Aspirin is familiar to most lay populations, and has well studied risks meaning counseling healthy populations about it is very feasible. Many people understand its’ use in cardiovascular prevention hence the step to use in cancer should be straight forward and well received. Patient preference has been studied in Barrett’s esophagus patients with 76% open to the idea of using aspirin. The more knowledge that is gained through risk stratifying populations, as is already done in cardiovascular disease, the greater a case can be put to healthy subjects.

Conclusion Cancer chemoprevention as a principle is an exciting prospect for researchers, clinicians and patients. There are significant draw backs at present, centered around the risk of causing adverse events and treatment associated mortality in an otherwise healthy population. Rather than a one pill fits all approach the future of chemoprevention is likely to mimic the principles of cardiovascular prevention by targeting higher risk populations. Further studies are ongoing to define these populations and gene profiling is being used to narrow down even more specifically (Table 2).

See also: Chemoprevention Trials.

Disclaimer Funding: Cancer Research UK, Royal College of Surgeons Ireland; Declaration of Interests: Prof Jankowski is the Chief Investigator for the AspECT trial.

Further Reading Burn, J., Gerdes, A.M., MacRae, F., Mecklin, J.P., Moeslein, G., Olschwang, S., Bishop, D.T., 2011. Long-term effect of aspirin on cancer risk in carriers of hereditary colorectal cancer: An analysis from the CAPP2 randomized controlled trial. The Lancet 378 (9809), 2081–2087. https://doi.org/10.1016/S0140-6736(11)61049-0. Cuzick, J., 2017. Preventive therapy for cancer. The Lancet Oncology. https://doi.org/10.1016/S1470-2045(17)30536-3. Cuzick, J., Otto, F., Baron, J.A., Brown, P.H., Burn, J., Greenwald, P., Thun, M., 2009. Aspirin and non-steroidal anti-inflammatory drugs for cancer prevention: An international consensus statement. The Lancet Oncology 10 (5), 501–507. https://doi.org/10.1016/S1470-2045(09)70035-X. Cuzick, J., DeCensi, A., Arun, B., Brown, P.H., Castiglione, M., Dunn, B., Zwierzina, H., 2011. Preventive therapy for breast cancer: A consensus statement. The Lancet Oncology 12 (5), 496–503. https://doi.org/10.1016/S1470-2045(11)70030-4.

Chemoprevention of Cancer: An Overview of Promising Agents and Current Research

351

Cuzick, J., Sestak, I., Bonanni, B., Costantino, J.P., Cummings, S., DeCensi, A., Wickerham, D.L., 2013. Selective oestrogen receptor modulators in prevention of breast cancer: An updated meta-analysis of individual participant data. The Lancet 381 (9880), 1827–1834. https://doi.org/10.1016/S0140-6736(13)60140-3. Cuzick, J., Thorat, M.A., Bosetti, C., Brown, P.H., Burn, J., Cook, N.R., Umar, A., 2015. Estimates of benefits and harms of prophylactic use of aspirin in the general population. Annals of Oncology 26 (1), 47–57. https://doi.org/10.1093/annonc/mdu225. Dulai, P.S., Singh, S., Marquez, E., Khera, R., Prokop, L.J., Limburg, P.J., Murad, M.H., 2016. Chemoprevention of colorectal cancer in individuals with previous colorectal neoplasia: Systematic review and network meta-analysis. BMJ. https://doi.org/10.1136/bmj.i6188. Higurashi, T., Hosono, K., Takahashi, H., Komiya, Y., Umezawa, S., Sakai, E., Nakajima, A., 2016. Metformin for chemoprevention of metachronous colorectal adenoma or polyps in post-polypectomy patients without diabetes: A multicentre double-blind, placebo-controlled, randomised phase 3 trial. The Lancet Oncology 17 (4), 475–483. https://doi.org/ 10.1016/S1470-2045(15)00565-3. Hurtado, A., Holmes, K.A., Geistlinger, T.R., Hutcheson, I.R., Nicholson, R.I., Brown, M., Carroll, J.S., 2008. Regulation of ERBB2 by oestrogen receptor-PAX2 determines response to tamoxifen. Nature 456 (7222), 663–666. https://doi.org/10.1038/nature07483. Jankowski, J.A., Hawk, E.T., 2006. A methodologic analysis of chemoprevention and cancer prevention strategies for gastrointestinal cancer. Nature Clinical Practice. Gastroenterology & Hepatology 3 (2), 101–111. https://doi.org/10.1038/ncpgasthep0412. Nan, H., Hutter, C.M., Lin, Y., Jacobs, E.J., Ulrich, C.M., White, E., Chan, A.T., 2015. Association of Aspirin and NSAID use with risk of colorectal Cancer according to genetic variants. JAMA 313 (11), 1133. https://doi.org/10.1001/jama.2015.1815. Ricciardiello, L., Ahnen, D.J., Lynch, P.M., 2016. Chemoprevention of hereditary colon cancers: Time for new strategies. Nature Reviews. Gastroenterology & Hepatology 13 (6), 352–361. https://doi.org/10.1038/nrgastro.2016.56. Singh, S., Singh, P.P., Roberts, L.R., Sanchez, W., 2014. Chemopreventive strategies in hepatocellular carcinoma. Nature Reviews. Gastroenterology & Hepatology 11 (1), 45–54. https://doi.org/10.1038/nrgastro.2013.143. U.S. Preventive Services Task Force.n.d. Recommendation Statement: Aspirin to prevent cardiovascular disease and cancer U.S. Preventive Services Task Force [online] https:// www.uspreventiveservicestaskforce.org/Page/Document/UpdateSummaryFinal/aspirin-to-prevent-cardiovascular-disease-and-cancer?ds¼ 1&s¼ aspirin.

Chemoprevention Trialsq Kunal C Kadakia, Ashley G Matusz-Fisher, and Edward S Kim, Levine Cancer Institute, Carolinas HealthCare System, Charlotte, NC, United States © 2019 Elsevier Inc. All rights reserved.

Introduction Epithelial cancers have historically been a cause for frustration by both patients and practitioners. Despite decades of cancer research, the mortality rate from epithelial malignancies has improved only 1%–2% per year for both men and women (Siegel et al., 2017), largely driven by decreased tobacco use (Jemal et al., 2008). Although some patients will present with early stage disease, the majority of patients will present with locally advanced or metastatic disease. Surgical intervention with adjuvant or neoadjuvant treatment, including radiotherapy and/or chemotherapy, has offered some improvements in long-term survival rates. However, local and distant relapse remain unacceptably high in many cancers. Thus, novel approaches to these epithelial cancers are needed, specifically the prevention of carcinogenesis prior to the onset of invasive cancer (Meyskens, 1992). Prevention strategies can occur at three broad levels: primary, secondary, and tertiary. A chemopreventive agent is a compound, either natural or synthetic, that prevents, reverses, or blocks the development of invasive cancer and can be considered at any level of prevention (Wattenberg, 1985). Primary prevention targets risk-reducing techniques in asymptomatic “normal” individuals and includes interventions such as diet and exercise. A high-risk individual such as one with a known inherited cancer syndrome who undergoes prophylactic surgery is also an example of primary prevention (e.g., BRCA carrier undergoing mastectomy). Treatments of individuals with premalignant findings to prevent progression to invasive cancer are considered secondary prevention. This can include surgical or ablative therapy for known precancerous conditions such as polypectomy for colonic adenomas or photodynamic therapy for Barrett’s esophagus as well as topical chemopreventive therapies for actinic keratoses, the precursor to squamous cell carcinoma of the skin. Tertiary prevention involves reducing the risk of recurrence after established malignancy and is exemplified by studies targeting secondary primary tumors of the aerodigestive tract following definitive treatment.

Biologic Rationale for Chemoprevention It is well established that advanced metastatic carcinomas are heterogonous invasive masses of genomically complex cells (Hanahan and Weinberg, 2011). Multiple dysregulated cellular-signaling pathways lead to the inevitable metastatic phenotype. The molecular heterogeneity observed occurs not only between patients with the same malignancy (interpatient heterogeneity) but also within metastatic sites of a given patient (intrametastatic heterogeneity). This complexity highlights the rational for targeting earlier neoplastic changes (i.e., intraepithelial neoplasia) prior to the onset of genomically diverse invasive carcinoma. As a result of a variety of environmental and genotoxic insults, normal epithelium transforms to neoplasia most often via a stepwise accumulation of molecular and genetic aberrations over the course of many years. This sequential accumulation of abnormalities was typified by the model of colorectal tumorigenesis by Fearon and Vogelstein in 1992 (Fearon and Vogelstein, 1990) and has since been described in multiple tumors. Chemopreventive agents have the potential to interrupt or perhaps reverse these changes throughout the carcinogenic process. Numerous potential molecular targets within neoplastic development are putative drivers. These include those that involve independence from growth signals (e.g., EGFR, PI3K), resistance to antigrowth signals (e.g., CDK, SMADs), evasion of apoptosis (e.g., mTOR, PTEN, Ras), immortal replicative potential (e.g., p53, pRb), angiogenesis (e.g., HIF-1a, VEGF, FGF), and tissue invasiveness and metastatic potential (e.g., E-cadherin, MAP-kinase) (Hanahan and Weinberg, 2011; Kelloff et al., 2006). Although inhibition of single transduction pathways has had clinical efficacy in the advanced cancer setting (e.g., anti-EGFR therapy in lung and colorectal cancers), the toxicity associated with complete inhibition of select pathways as well as the presence of bypass tracks (i.e., alternative proliferative pathways or downstream mutations) limits the applicability of single site blockade in complex tumors. This is highlighted by the failure of erlotinib, an anti-EGFR monoclonal antibody, to reduce oral cancer-free survival in high-risk individuals with oral premalignant lesions (William et al., 2016). In addition to aberrations in signal transduction, dependence of cancer cells on a single oncogene (i.e., oncogene addiction) remains an important potential target. This is suggested by the inherent sensitivity of the addicted cell to blockade of the oncogene. For example, preclinical data support the antiproliferative effect of even partial inhibition of cyclin D1, a purported addicted oncogene in esophageal cancer (Weinstein, 2002; Jain et al., 2002). Further clinical study of manipulation of oncogene addiction and tumor suppressor hypersensitivity is ongoing (Weinstein et al., 2008; Jacobi et al., 2017).

q

Change History: July 2017. Due to significant changes in the field, the vast majority of the content was updated and reworded in the context of the available data by Kunal C. Kadakia, Ashley G. Matusz-Fisher, and Edward S. Kim. This article is an update of Edward S. Kim, Fadlo R. Khuri, and Waun Ki Hong, Chemoprevention Trials, in Encyclopedia of Cancer (Second Edition), edited by Joseph R. Bertino, Academic Press, 2002, Pages 457–472.

352

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.96067-7

Chemoprevention Trials

353

Other significant chemopreventive avenues have evolved from the improved understanding of the role of inflammation in neoplasia. Multiple studies support the effects of chronic inflammation on promoting carcinogenesis (Caruso et al., 2004; Philip et al., 2004). Nuclear factor-kappa b, a proinflammatory transcription factor, along with regulatory proteins (e.g., tumor necrosis factor-a, cyclooxygenase-2, inducible nitric oxide synthase) play an integral role in the development and progression of neoplasia and have been proposed to serve as therapeutic targets as well as potential biomarkers (Aggarwal, 2004). Additionally, states of chronic inflammation caused by inflammatory bowel disease (IBD) and Barrett’s esophagus, or due to infections such as human papilloma virus (HPV) and hepatitis B (HBV) lead to prolonged inflammation that can progress to carcinogenesis (Shacter and Weitzman, 2002). Importantly, the development of vaccines for HPV and HBV has resulted in marked reduction in the incidence of cervical and hepatocellular carcinoma, respectively, representing the importance of targeting upstream molecular events prior to the development of frank carcinoma (Chang et al., 2009; Garland et al., 2016).

Drug Development of Chemopreventive Agents Similar to drug development for other indications, translational efforts for chemopreventive agents follow a similar process. The chemoprevention drug-screening program outlined by the United States National Cancer Institute’s Cancer Preclinical Drug Development Program (PREVENT) provides such a pragmatic approach involving a stage-gate process with go/no go decision points throughout the drug development process (Shoemaker et al., 2016). Ideally, preclinical testing includes biochemical prescreening assays, in vitro efficacy models, in vivo short-term screening, and animal efficacy prior to testing in human trials. Despite significant improvements in the preclinical sciences, the translation of findings from these models into clinical success in human chemoprevention trials has been limited. Although the reasons for this are multifactorial, murine/rodent models often consist of a genetically homogenous population that follows a strict dietary regimen in tightly controlled laboratory conditions. In contrast, even wellcontrolled human trials inherently include genetically diverse populations with varied dietary and environmental exposures that cannot be fully measured (Meyskens et al., 2016). Similar to other oncologic therapeutics, evaluation of the clinical safety and efficacy of chemopreventive agents undergo phased testing. Understanding the nuances specific to prevention, as compared to treatment, are integral to the successful design and implementation of chemoprevention trials. These include the wide therapeutic index required of the agents, the long latency to neoplastic transformation, the necessity for prolonged adherence, and the complexity underlying risk evaluation (Shureiqi et al., 2000). Therapeutic index is driven by a combination of the agent’s efficacy as well as its associated short- and long-term toxicities. Early phase studies are integral to understanding the optimal dosing, frequency, and acceptable toxicities prior to undertaking larger scale studies. The inherent principle of chemoprevention requires drug therapy in a largely asymptomatic patient and therefore greater understanding of detection, evaluation, and avoidance of toxicities is highly warranted. Such strategies include defining the minimally effective dose, evaluating the pharmacokinetic and pharmacodynamics of intermittent or combination therapies, and optimizing toxicity monitoring (Kelloff et al., 2006). In latter phases, the ideal endpoint of chemoprevention trials is a reduction in cancer incidence. Because cancer incidence is extremely low among the general population with no known cancer risk factors, demonstrating a treatment-induced reduction in cancer incidence among the general population requires large randomized studies, which are expensive and time-consuming. Targeting high-risk populations and utilizing validated intermediate biomarkers can significantly reduce the time and resources required for chemoprevention trials. Biomarkers can be defined as measurable and evaluable indicators of normal biologic processes, pathogenic processes, or pharmacologic responses to therapeutic interventions. Currently, the most common biomarker utilized in chemoprevention studies is histopathologic intraepithelial neoplasia (IEN). Examples of histopathologic IEN include adenomas in colorectal cancers, leukoplakia in oral cancers, bronchial dysplasia in lung cancers, and cervical intraepithelial neoplasia in cervical cancer. The goal of such a biomarker is to be a true surrogate of a future invasive process. The ideal biomarker should be variably expressed throughout the carcinogenic process, detectable early in the course or carcinogenesis, modulation with intervention should accurately be associated with clinical benefits, and ideally should be able to be quantified directly, reliably, noninvasively, and with technical and cost feasibility. As histopathologic changes are a result of a multiple molecular, proteomic, and genomic changes, establishing a biomarker that accounts for this interplay is most ideal. For example, gene set enrichment analyses where genes in a critical pathway are mapped to determine which are affected and not affected in cancer tissue might allow to better target tissue-specific molecular aberrations (Mootha et al., 2003). Although significant progress is being made with advanced technologies such as the gene set enrichment analysis described; newer proteomic, imaging, and microarray techniques might allow for a truly practical, prognostic, and predictive biomarker for each organ system to be fully realized (Fabian and Kimler, 2016).

Clinical Trials Involving Chemoprevention Cancer chemoprevention continues to evolve and its role in oncologic practice continues to expand. A multitude of chemopreventive agents have been studied in over 100 randomized trials over the past four decades. Some clinical activity has been demonstrated proving the potential utility of drug therapy for cancer prevention. Over 10 agents have been FDA approved for cancer risk reduction in select high-risk cohorts as well as the treatment of premalignant lesions (Tables 1 and 2). Although an exhaustive review of all chemoprevention trials is not possible in this article, a discussion regarding the major contemporary and historically significant chemoprevention trials by organ system is provided.

354

Chemoprevention Trials

Table 1

FDA approved therapies for cancer risk reduction

Agent

FDA indication and cohort a

Tamoxifen

1. Reduction in risk of invasive breast cancers in women with DCIS following breast surgery and radiation. 2. Reduction in incidence of breast cancer in women at high risk for breast cancer (“high-risk” defined as women at least 35 years of age with a 5-year predicted risk of breast cancer 1.67%, as calculated by the Gail Model.) 1. Reduction in the risk of invasive breast cancer in postmenopausal women at high risk for invasive breast cancer (high risk is defined as at least one breast biopsy showing lobular carcinoma in situ or atypical hyperplasia, one or more first-degree relatives with breast cancer, or a 5-year predicted risk of breast cancer > 1.66% [based on the modified Gail mode].) 1. Female 9 through 25 years of age for the prevention of the following diseases caused by oncogenic human papillomavirus (HPV) types 16 and 18: • cervical cancer. • cervical intraepithelial neoplasia (CIN) Grade 2 or worse and adenocarcinoma in situ. • cervical intraepithelial neoplasia (CIN) Grade 1. 1. Girls and women 9 through 26 years of age for the prevention of the following diseases: • Cervical, vulvar, vaginal, and anal cancer caused by human papillomavirus (HPV) types 16, 18, 31, 33, 45, 52, and 58. • Genital warts (condyloma acuminata) caused by HPV types 6 and 11. And the following precancerous or dysplastic lesions caused by HPV types 6, 11, 16, 18, 31, 33, 45, 52, and 58:

Raloxifene Cervarix (HPV Vaccine)

Gardasil9 (HPV Vaccine)

• Cervical intraepithelial neoplasia (CIN) grade 2/3 and cervical adenocarcinoma in situ (AIS). • Cervical intraepithelial neoplasia (CIN) grade 1 • Vulvar intraepithelial neoplasia (VIN) grade 2 and grade 3 • Vaginal intraepithelial neoplasia (VaIN) grade 2 and grade 3 • Anal intraepithelial neoplasia (AIN) grades 1, 2, and 3.

2. Boys and men 9 through 26 years of age for the prevention of the following diseases:

• Anal cancer caused by HPV types 16, 18, 31, 33, 45, 52, and 58. • Genital warts (condyloma acuminata) caused by HPV types 6 and 11.

And the following precancerous or dysplastic lesions caused by HPV types 6, 11, 16, 18, 31, 33, 45, 52, and 58:

• Anal intraepithelial neoplasia (AIN) grades 1, 2, and 3. a

As per the FDA product label. With permission, taken in part, from reference Maresso, K. C., Tsai, K. Y., Brown, P. H., et al. (2015). Molecular cancer prevention: Current status and future directions [Internet]. CA: A Cancer Journal for Clinicians 65, 345–383. Available from: http://doi.wiley.com/10.3322/caac.21287 [cited 7 August 2017].

Table 2

FDA approved therapies for the treatment of precancerous lesions

Agent

FDA indication a

Imiquimod

1. Immunocompetent adults for the topical treatment of clinically typical, nonhyperkeratotic, nonhypertrophic actinic keratoses on the face or scalp. 1. Males and females for the topical treatment of actinic keratosis on the face, scalp, trunk, and extremities. 1. Males and females for the topical treatment of actinic keratoses. 1. Males and females for the topical treatment of multiple actinic or solar keratosis. 1. Males and females for the topical treatment of minimally to moderately thick actinic keratoses of the face or scalp 1. Males and females with high-grade dysplasia in Barrett’s esophagus. Ablation of high-grade dysplasia in Barrett’s esophagus who do not undergo esophagectomy 1. Males and females with carcinoma in situ of the urinary bladder. Intravesical use in the treatment and prophylaxis of carcinoma in situ of the urinary bladder and for the prophylaxis of primary or recurrence stage Ta and/or T1 papillary tumors following transurethral resection (TUS). 1. Males and females with BCG-refractory carcinoma in situ. Intravesical therapy of BCG-refractory carcinoma in situ of the urinary bladder in patients for whom immediate cystectomy would be associated with unacceptable morbidity or mortality.

Ingenol mebutate Diclofenac sodium Fluorouracil Photodynamic therapy with 5-aminolevulinic acid Photodynamic therapy with photofrin Bacillus-Calmette-Guerin Valrubicin

a As per the FDA product label. With permission, taken in part, from reference Maresso, K. C., Tsai, K. Y., Brown, P. H., et al. (2015). Molecular cancer prevention: Current status and future directions [Internet]. CA: A Cancer Journal for Clinicians 65, 345–383. Available from: http://doi.wiley.com/10.3322/caac.21287 [cited 7 August 2017].

Chemoprevention Trials

355

Chemoprevention Trial Results by Cancer Type Head and neck cancers Currently, the primary risk factors for squamous cell carcinoma of the head and neck (SCCHN) include tobacco smoke, alcohol consumption, and HPV (Forastiere et al., 2001). Originally described in 1953 (Slaughter et al., 1953), field cancerization is the predilection of the surface epithelium of the upper aerodigestive tract, or field, to develop additional cancers in patients with known carcinogen-related SCCHN. The specific premalignant changes found in areas of carcinogen-exposed epithelium adjacent to tumors can progress concurrently to form second primary tumors (SPTs). SPTs are a leading cause of mortality in head and neck cancer and typifies the concept of field cancerization (Vokes et al., 1993). The reversal of oral premalignant lesions (oral IEN or dysplastic oral IEN as intermediate biomarkers of invasive cancers) and prevention of SPTs have been the focus of extensive research since the 1980s. Despite significant preclinical rational (Uray et al., 2016) and early phase trial success (Hong et al., 1990; Hong et al., 1986; Lippman et al., 1993) of retinoids, the most commonly tested chemopreventive agent, multiple large randomized clinical trials of these agents have failed to generate an FDA approval for oral cancer prevention (van Zandwijk et al., 2000; Bolla et al., 1994; Papadimitrakopoulou et al., 2009; Khuri et al., 2006). Although retinoids have not yet proven effective for SCCHN cancer prevention, numerous correlative studies within these trials have led to the discovery of markers of cancer risk including loss of heterozygosity (LOH) profiles (Mao et al., 1996), EGFR copy number gain (Taoudi Benchekroun et al., 2010), genetic variation (Lee et al., 2011), and gene expression signatures (Saintigny et al., 2011). For example, a large single nucleotide polymorphism (SNP) analysis of a cohort within the Retinoid Head and Neck Second Primary Trial of 13-cis-retinoic acid found that patients carrying the common genotype of rs3118570 in the retinoid X receptor were at a greater than threefold increased risk and that this locus along with two other genetic loci (CDC25C:rs6596428 and JAK2:rs1887427) identified those that had a 76% reduction in SPT/recurrence with chemoprevention (Lee et al., 2011). This data suggests that pharmacogenetic analysis might allow for improved targeting of “higherrisk” populations in future retinoid chemoprevention trials. The Erlotinib Prevention of Oral Cancer (EPOC) study represents the first chemoprevention trial that was based a priori on personalized molecular characterization (William et al., 2016). Erlotinib, an EGFR tyrosine kinase inhibitor, was chosen based on the preclinical data, suggesting that EGFR amplification is associated with oral premalignant lesions (OPLs) transformation (Taoudi Benchekroun et al., 2010) and that EGFR inhibition prevents oral dysplasia (Leeman-Neill et al., 2011). In this trial, patients with OPLs were first evaluated for LOH at select loci and were considered high risk if LOH was present at 3p14 and/or 9p21 in patients with a history of invasive cancer or LOH at 3p14 and/or 9p21 plus an additional site including 17p, 11p, 13q, 4q, or 8p and were low-risk if not meeting either of these criteria. High-risk LOH patients (N ¼ 150) were randomized to erlotinib or placebo and the primary endpoint was oral cancer free survival (CFS). The 3-year CFS rates in erlotinib versus placebo-treated patients were 70% and 74%, respectively, and were not statistically significant (P ¼ .45). The study, however, did validate LOH as a marker for oral cancer risk as the 3-year CFS rates for LOH-positive versus LOH-negative groups were 74% and 87%, respectively (P ¼ .01). HPV-positive oropharyngeal squamous cell carcinoma (OPSCC) has increased in incidence by 225% from 1998 to 2004 and continues to rise (Chaturvedi et al., 2011). Approximately 7% of the adults aged 14–69 population harbor active oral HPV infection, however only 1% of woman and 3% of men have the high-risk HPV16 subtype (Gillison et al., 2012). Although it is not certain that HPV vaccinations have directly reduced OPSCC, secondary evaluation of a large HPV16/18 vaccine trial suggested that oral HPV prevalence 4 years after vaccination was significantly reduced as compared to control (Herrero et al., 2013).

Lung cancers The rationale for prevention of lung cancer is similar to that in HNSCC. In both diseases, chronic exposure to tobacco is the major risk factor and dysplastic epithelial lesions are thought to represent a premalignant stage. Unfortunately, despite decades of work, a variety of agents including a-tocopherol (Alpha-Tocopherol, Beta Carotene Cancer Prevention Study Group, 1994), b –carotene (Hennekens et al., 1996; Omenn et al., 1996), retinoids (van Zandwijk et al., 2000; Lippman et al., 2001), selenium (Karp et al., 2013), and a combination of multivitamins and minerals (Kamangar et al., 2006) were all ineffective or deleterious (e.g., beta carotene) in large randomized phase III trials. These studies exemplify the limitations of utilizing epidemiological evidence and secondary endpoints in the absence of strong preclinical rational when planning chemoprevention trials. For example, three large randomized, double-blind, placebo-controlled trials of b-carotene were conducted primarily based on epidemiological data alone. The Alpha-Tocopherol, Beta Carotene (ATBC) Cancer Prevention study was a primary-prevention trial in which 29,133 Finnish male smokers received a-tocopherol 50 mg a day, b-carotene 20 mg a day, both, or placebo. These men were between 50 and 69 years of age and all smoked five or more cigarettes a day and were followed for 5–8 years. Lung cancer incidence, the primary endpoint, did not change with the addition of b-tocopherol alone, nor did overall mortality. However, both groups who received b-carotene supplementation (alone or with a-tocopherol) had an 18% increase in the incidence of lung cancer. There appeared to be a stronger adverse effect from b-carotene in those men who smoked more than 20 cigarettes a day. This trial raised the serious issue that pharmacologic doses of b-carotene could potentially be harmful in active smokers. The b-Carotene and Retinol Efficacy Trial (CARET) confirmed the results of the Finnish ATBC trial (Omenn et al., 1996). This trial tested the combination of b-carotene 30 mg and retinyl palmitate 25,000 IU against placebo in 18,314 men and women aged 50– 69 years at high risk for lung cancer; 14,254 had at least a 20 pack-year smoking history and were either current or recent former smokers. This trial was stopped after 21 months because no benefit and lung cancer incidence, the primary endpoint, increased 28%

356

Chemoprevention Trials

in the active intervention group. Overall mortality also increased 17% in this group. The Physicians Health Study included 22,071 healthy male physicians treated with b-carotene 50 mg on alternate days or placebo (Hennekens et al., 1996). In contrast to the ATBC and CARET, the use of supplemental b-carotene showed virtually no adverse effects, however no beneficial effects on cancer incidence or overall mortality during a 12-year follow-up were observed. Given these results, high-dose b-carotene is not recommended for chemoprevention of lung cancer. Although not based on epidemiologic data, the ECOG 5597 trial of selenium supplementation was largely based on a secondary analysis endpoint observed in a skin cancer prevention trial suggesting reduced lung cancer incidence (Karp et al., 2013; Clark et al., 1996). In this trial, 1561 patients with resected stage I NSCLC were randomly assisted to selenium 200 mg versus placebo daily for 48 months. The study was stopped early due to futility and showed no benefit of selenium over placebo. Given the inaccessibility of the lungs for repeated histologic evaluations, practical intermediate biomarkers remain an area of unmet need. An interesting avenue of research is the potential role of bronchial airway gene expression signatures as a measure of risk and as a treatment response biomarker. In a murine model, researchers performed transcriptome sequencing to analyze bronchial airway gene expression and found activation of PI3K and Myc signaling in normal bronchial epithelial cells with precancer lung squamous cell carcinoma lesions (Xiong et al., 2017). Most notably, the activation of PI3K and Myc was reversed by treatment with the PI3K Inhibitor XL-147 and pioglitazone, respectively. If validated, these and similar biomarker studies might provide improved tools to analyze the effect of chemopreventive agents in high-risk individuals. The negative randomized trials described earlier highlight the need for coordinated efforts integrating epidemiological data, validated intermediate biomarkers, and most importantly a strong biologic rational prior to undertaking large chemoprevention trials (Szabo et al., 2013).

Colorectal cancers Colorectal cancer (CRC) is associated with premalignant lesions, including adenomatous polyps and dysplastic epithelium. Large polyps with villous elements and dysplastic epithelia are more likely to progress to carcinoma than small, tubular polyps. The premalignant stages of colon carcinogenesis present an evaluable process for chemoprevention trials. Based on the wellestablished role of the cyclooxygenase (COX) enzymes and inflammation in colorectal carcinogenesis (Janakiram and Rao, 2014), aspirin and NSAIDs have been most extensively studied. In addition, fiber supplementation, hormone replacement therapy, and calcium salts have also underwent considerable evaluation. The effects of aspirin on primary prevention of colon cancer have been examined in two large population-based trials including the Physician’s Health Study (PHS) and the Women’s Health Study (WHS). In the PHS, there was no benefit observed for aspirin at 325 mg every other day in the primary prevention of CRC at 5 or 12 years of follow up (Stürmer et al., 2006). In contrast, the WHS of aspirin at 100 mg every other day found no reduction in CRC at 10 years but after a median of 18 years of follow up, there was a reduction in risk (HR 0.80), which was primarily for proximal cancers (Cook et al., 2013). Though the reason for the discrepant results is not fully established, the PHS study only tested 5 years of aspirin and longer duration of treatment might be needed given the relatively prolonged latency to CRC development (Chan et al., 2005). Further supporting the role of aspirin for primary CRC prevention comes from the pooled secondary analysis of four large European and British trials that evaluated the effect of aspirin on vascular events. This analysis included 14,033 patients and suggested that the 20-year incidence of colon, but not rectal cancer, was decreased by nearly 25% (HR 0.76) when given beyond four years and at a dose of at least 75 mg daily (Rothwell et al., 2010). This compelling evidence led the US Preventive Services Task Force (USPSTF) to recommend “low-dose aspirin use for the primary prevention of cardiovascular disease (CVD) and colorectal cancer in adults ages 50 to 59 years who have a 10% or greater 10year CVD risk, are not at increased risk for bleeding, have a life expectancy of at least 10 years, and are willing to take low-dose aspirin for at least 10 years (Bibbins-Domingo, 2016)”. For patients 60–69 years old, the USPSTF suggested consideration of low-dose aspirin after discussion of risks and benefits and for patients  49 or > 70 years, there was insufficient evidence to recommend aspirin as chemoprevention. The USPSTF suggested utilizing the freely available ASCVD Risk Estimator from the ACC/AHA for estimating 10-year cardiovascular risk (Goff et al., 2013). The role of aspirin as secondary prevention of recurrent adenomas after resection of adenomas is supported by a metaanalysis involving four clinical trials with 2967 randomly assigned patients (Cole et al., 2009). The risk of adenoma recurrence was lower for aspirin at any dose versus placebo (risk ratio [RR], 0.83), corresponding to an absolute risk reduction of nearly 7%. For advanced lesions, as defined as tubulovillous adenomas, villous adenomas, adenomas  1 cm in diameter, adenomas with high-grade dysplasia, or invasive cancer, the risk reduction was even greater (RR 0.72). These data suggest that aspirin is associated with a 20%–35% reduction in colonic adenomas and CRC incidence in average risk adults. The role of aspirin in hereditary CRC syndromes has also been evaluated in a pragmatic series of studies within the Colorectal Adenoma/Carcinoma Prevention Programme (CAPP). CAPP1 randomized 133 patients with familial adenomatous polyposis (FAP) to 600 mg aspirin daily or placebo and found a nonsignificant trend suggesting reduction in polyp count (RR 0.77) (Burn et al., 2011). In CAPP2, 937 patients with hereditary nonpolyposis colon cancer (Lynch Syndrome) were similarly randomized and no significant effect on incidence of adenoma formation or carcinoma was observed with aspirin therapy (Burn et al., 2008). Notably, in a perprotocol secondary analysis, patients in CAPP2 who received aspirin 600 mg daily for at least 2 years had a nearly 60% reduction in colorectal cancer incidence (Burn et al., 2011). An ongoing study evaluating different doses of aspirin in Lynch Syndrome is ongoing (CAPP3) and eagerly awaited. Similar to aspirin, COX-2 inhibitors and sulindac have also been extensively studied. Forms of the COX enzyme include COX-1, which is constitutively expressed, and COX-2, which is overexpressed in inflammatory cells and has been associated with colorectal adenomatous polyp and cancer formation. Most NSAIDs inhibit both enzymes, however inhibition of COX-2 specifically reduces

Chemoprevention Trials

357

untoward side effects such as ulcers and gastritis. The Prevention of Colorectal Sporadic Adenomatous Polyps (PreSAP, N ¼ 1561) and Adenomas Prevention with Celecoxib (APC, N ¼ 2035) were large randomized placebo-controlled trials in patients with resected adenomas and both found significant 30%–50% reduction in recurrent polyp formation (Bertagnolli et al., 2006; Arber et al., 2006). However, both trials also observed a significant increase in adverse cardiovascular events (two- to threefold) precluding the routine use of this agent. Similarly, a smaller study of celecoxib in FAP showed significant benefit and led, in part, to the accelerated approval of celecoxib for patients with phenotypic expression of FAP (Steinbach et al., 2000). However, this indication was subsequently voluntary removed in 2011 by the manufacturer due to delays in completing follow up studies required under the approval. Additionally, despite three small (N ¼ 10–24) studies of sulindac suggesting secondary preventive benefit of adenoma regression, number, and size in phenotypic FAP, a study of 41 genotypically confirmed, but not phenotypically affected, patients with FAP over four years failed to show sulindac prevented the development of adenomas compared with placebo (Giardiello et al., 2002). Given the established harm (cardiovascular and gastrointestinal) associated with long-term use of NSAIDs, these agents cannot currently be recommended for primary or secondary prevention of CRC. Fiber in the prevention of colon cancer has been examined in several studies. The Polyp Prevention Trial studied 2079 patients with a history of colorectal adenomas randomized to receive counseling, a low-fat, high-fiber diet rich in fruits and vegetables, or to continue their current diet without counseling (Schatzkin et al., 2000). Colonoscopy after one and four years found no difference in the incidence of recurrent adenomas. Another study by the Phoenix Colon Cancer Prevention Physician’s Network studied 1429 patients with a history of colorectal adenoma (Alberts et al., 2000). These patients were randomized to receive supplemental wheat bran, 2.0 or 13.5 g a day. Again, no difference in the incidence of recurrent adenomas was found between the two groups. Currently, there is no prospective evidence that fiber supplementation is effective for colorectal cancer prevention. Hormone replacement therapy in women in the form of combined estrogen plus progestin was observed to reduce the risk of CRC by nearly 45% in the Women’s Health Initiative (WHI). However, extended follow up of WHI suggested that with longer-term therapy, women receiving combination therapy had more advanced CRC diagnosis with a higher, albeit, nonstatistically significant increase in CRC morality rate (Simon et al., 2012). Given this data along with the negative data of unopposed estrogen (Lavasani et al., 2015), and the long-term cardiovascular concerns of HRT, HRT cannot be recommended for CRC prevention. High intake of calcium salts has been associated with lower risk of colorectal cancer along with synergistic chemopreventive effects when combined with vitamin D (Grau et al., 2003; Pence, 1993). Despite the biologic rational and epidemiological data, a large randomized trial of 2259 patients receiving vitamin D3 1000 IU, calcium carbonate 1200 mg, both, or neither daily showed no difference in colorectal adenoma formation over a 3–5-year period (Baron et al., 2015). Another similar randomized trial of 36,282 postmenopausal women showed no benefit of calcium and vitamin D over placebo in the development of CRC during a seven-year period.

Nonmelanoma and melanoma skin cancers Chemoprevention of skin cancers represents an ideal target organ for chemoprevention due to the sheer incidence in humans and practical accessibility of the skin. The vast majority of nonmelanoma skin cancers (NMSC) include cutaneous squamous cell carcinoma (cSCC) and basal cell carcinoma (BCC). cSCC accounts for 10%–20% of NMSC and occurs largely as a result of intraepithelial ultraviolet (UV)-induced damage which leads to a sequence of molecular changes, most often involving mutations of the p53 tumor suppressor gene (de Gruijl et al., 2001). cSCC has been the focus of the most chemoprevention trials due to the presence of the identifiable premalignant lesion, the actinic keratosis (AK) in 60%–70% of cSCC (Fernandez Figueras, 2017). Currently, the primary strategy for lesion-directed therapy for a single or few AKs involves ablative modalities with curettage, cryosurgery, or electrodessication. Field-directed topical therapies, including 5-fluorouracil cream, imiquimod cream, diclofenac gel, ingenol mebutate gel, and delta-aminolevulinic acid photodynamic therapy have all been FDA-approved for the treatment of AKs (see Table 2). These topical agents have been extensively studied in randomized controlled trials; however, no large trials comparing the efficacy between these agents have been undertaken. Despite inherent limitations of cross trial comparisons, their effect on AK clearance and adverse effects are comparable (Haque et al., 2015; Gupta et al., 2012). Retinoids, including retinol, acitretin, isotretinoin, and etretinate, have been the mainstay of oral chemoprevention trials directed at reduction of NMSCs. In a phase III trial, 2297 patients who were at moderate risk for skin cancer, including patients with a history of > 10 AKs and  2 prior NMSCs, received oral retinol 25,000 IU or placebo daily for 5 years (Moon et al., 1997). Retinol treatment was effective in reducing the incidence of cSCC (HR 0.74), but not BCC. However, another much larger (N ¼ 22,071) primary prevention trial of b-carotene compared with placebo showed no effect on cSCC or BCC (Frieling et al., 2000). Importantly, even in higher-risk patients with a history on NMSCs, three phase III trials that tested the effects of retinol (Levine et al., 1997), b-carotene (Greenberg et al., 1990), and low-dose isotretinoin (Tangrea et al., 1992) demonstrated that these retinoids had no effect on the incidence of SPTs. The immunosuppression required for organ transplant patients is an established strong risk factor (65–250-fold) for the development of aggressive cSCC (Tufaro et al., 2015). The second-generation retinoid acitretin has shown some clinical benefit particularly in high-risk organ transplant recipients. For example, a randomized placebo-controlled trial of 44 renal-transplant patients showed that acitretin at 30 mg daily over a 6-month period reduced new cSCC and AKs compared with placebo (11% vs. 47% and 13% vs. 28%, respectively)(Bavinck et al., 1995). However, in another small trial (N ¼ 26) of low (0.2 mg/kg daily) versus high-dose (0.4 mg/kg daily) acitretin, no difference in cSCC was observed (de Sévaux et al., 2003). Acitretin in nontransplant high-risk patients failed to

358

Chemoprevention Trials

show benefit over placebo (Kadakia et al., 2012). Although systemic retinoids are used in high-risk patients in practice (Otley et al., 2006), neither acitretin nor any other systemic retinoid is FDA approved. Importantly, use of these agents is limited to the duration they are taken and rapid AKs/NMSC devolvement upon discontinuation can occur (George et al., 2002). Despite preclinical data supporting the efficacy of COX-2 inhibition and the ornithine decarboxylase inhibitor a-difluoromethylornithine (DFMO), neither agent in randomized placebo-controlled trials met their primary endpoint of reduction in AKs (Elmets et al., 2010) and NMSCs (Bailey et al., 2010), respectively. Nicotinamide (Vitamin B3) at 500 mg twice daily, which is protective against UV damage, was tested in a randomized placebo-controlled trial in 386 patients with a history of  2 NMSCs (Chen et al., 2015). There was a significant 23% reduction in new NMSC, with a slightly higher reduction in cSCC compared with BCC (30% vs. 20%). Although promising, longer-term safety and efficacy are needed prior to recommending nicotinamide for secondary prevention. Lastly, a phase II randomized placebo-controlled trial of vismodegib, an oral hedgehog inhibitor approved for locally advanced or metastatic BCC, at 150 mg daily showed significant chemopreventive action in patients with the highly penetrant nevoid basal cell carcinoma syndrome, however 54% of patients discontinued treatment due to toxicity (Tang et al., 2012). Although significant advances in the field of immuno-oncology have changed the prognosis of advanced melanoma, no agent, including vitamin D, NSAIDs, sulindac, and lipid-lowering drugs, has yet been proven to successfully prevent neoplastic melanocytic transformation (Chhabra et al., 2017).

Breast cancer Chemoprevention approaches in breast cancer focus on preventing breast cancer in high-risk patients. This strategy stemmed from the success of tamoxifen, a selective estrogen-receptor modulator (SERM), in reducing recurrent breast cancer by 40%–50% following definitive therapy for estrogen-receptor (ER) positive breast cancer (Early Breast Cancer Trialists’ Collaborative Group (EBCTCG) et al., 2011). Multiple large trials of tamoxifen compared with placebo in both high-risk and average risk women have been conducted since the 1990s. A 2013 meta-analysis revealed that compared to placebo tamoxifen decreased the incidence of invasive breast cancer, primarily as a reduction in ER positive breast cancer risk by 30% (RR 0.70) and a reduction in nonvertebral factures risk by 34% (RR 0.66) but had no impact on breast cancer-specific or all-cause mortality (Nelson et al., 2013). The longterm follow up of the STAR (Study of Tamoxifen and Raloxifene) trial found that the second-generation SERM raloxifene confers 78% efficacy of tamoxifen in breast cancer prevention, however was not associated with an increase in uterine cancer (Nelson et al., 2013; Vogel et al., 2010). Despite the chemopreventive benefits observed, a patient-level meta-analysis of 83,399 patients involved in SERM chemoprevention trials showed that multiple rare but serious risks occur including a slight increase in uterine cancer (105 vs. 63 events, HR 1.56) and thromboembolic disease (375 vs. 215, HR 1.73)(Cuzick et al., 2013). Based on these data, tamoxifen and raloxifene were FDA-approved for the primary prevention of breast cancer in high-risk population (see Table 1). High-risk in these studies was defined as women at least 35 years of age with a 5-year predicted risk of breast cancer  1.67%, as calculated by the Gail Model (Rockhill et al., 2001). Aromatase inhibitors, which have also showed significant benefit over tamoxifen in the adjuvant breast cancer setting in postmenopausal women (Early Breast Cancer Trialists’ Collaborative Group (EBCTCG) et al., 2015), have also been evaluated for primary prevention in high-risk patients. Both exemestane (NCIC-MAP3 trial, N ¼ 4560) and anastrozole (IBIS-II trial, N ¼ 3864) as compared to placebo revealed significant reductions in invasive breast cancer by 50%–65% (Goss et al., 2011; Cuzick et al., 2014). As with SERMs, neither anastrozole or exemestane significantly reduced. Anastrozole and exemestane are not yet FDAapproved for risk reduction. Despite the minimal negative effect on quality of life reported in the major phase III chemoprevention trials of SERMs and AIs, uptake of these agents in the clinic has been very low. For women aged 35 years or older with increased risk of breast cancer and low risk of adverse effects, the USPSTF provides a grade B recommendation and supports shared and informed decision making when offering tamoxifen or raloxifene. Both agents are effective in pre- and postmenopausal women (Nelson et al., 2013). For high-risk patients who wish to pursue chemoprevention, raloxifene is a reasonable choice if concerns for uterine cancer and thromboembolic disease predominate and tamoxifen if risk of breast cancer prevention predominates. To date, there are no proven therapies to prevent HER2-positive breast cancers, although trials are ongoing targeting anti-HER2 therapy for HER2-positive ductal carcinoma in situ and might provide rational for future chemoprevention studies.

Esophageal and gastric cancer Esophageal adenocarcinoma (EAC) and esophageal squamous cell carcinoma (ESCC) are the two predominate histologic subtypes of esophageal cancer. Though historically less common, the incidence of EAC has dramatically risen in the US and westernized countries and is suspected to be related to multiple risk factors including obesity and gastroesophageal reflux disease (Zhai et al., 2010; Drahos et al., 2016). Given the precursor premalignant lesion (Barrett’s esophagus) in EAC, secondary prevention utilizing endoscopic screening in high-risk patients as well as local ablative and excisional therapies for established high-grade Barrett’s esophagus is the current clinical approach to reduce the risk of progression to EAC. The FDA approved porfimer sodium in conjunction with photodynamic therapy (PDT) in 2003 based on a randomized trial of this combination versus omeprazole alone in 208 patients with highgrade dysplasia (Overholt et al., 2005). Complete ablation of high-grade dysplasia and progression to invasive cancer was improved with porfimer sodium than with omeprazole alone (77% vs. 39% and 13% vs. 28%, respectively), however serious complications

Chemoprevention Trials

359

occurred in 12% and strictures in 36%. Radiofrequency ablation with mucosal resection has replaced PDT due to the improved efficacy and safety (Shaheen et al., 2009). A randomized trial of celecoxib in patients with Barrett’s esophagus showed no difference in regression of dysplasia as compared to placebo (Heath et al., 2007). A large study of 2500 patients (AspECT trial) comparing aspirin in combination with low or high-dose esomeprazole to esomeprazole alone and has completed accrual and results are eagerly awaited. In contrast to EAC, ESCC remains prevalent in developing countries and is associated with tobacco and alcohol exposure as well as nutrient deficiencies and exposure to N-nitroso compounds. Multiple large randomized chemoprevention studies involving residents of Linxian, China tested the effects of combinations of agents such as retinol, riboflavin, zinc, selenomethionine, and vitamin E without a clear signal of benefit (Blot et al., 1993; Wang et al., 2013; Limburg et al., 2005). A 10-year follow up of the largest trial involving 29,584 patients suggested a 17% reduction in ESCC mortality in those aged < 55 years, but an increase of 14% in mortality in those aged  55 years (Qiao et al., 2009). The interpretation of these studies is made difficult by the use of readily available vitamin supplements in the control arm, blurring potential differences from the treatment arm. This represents an inherent flaw in trials utilizing vitamin supplements and dietary interventions. Worldwide, Helicobacter pylori, a Gram-negative microaerophilic bacterium, are frequently associated with noncardia gastric cancers. Although treatment of H. pylori with a combination of antibiotics and a proton pump inhibition (“triple therapy”) is clearly effective, data to support a chemopreventive role of such therapy on decreasing gastric cancer occurrence are becoming increasingly realized. One large trial involving 3365 patients randomized patients to amoxicillin and omeprazole for two weeks and showed that at 15 years follow up the incidence of gastric cancer was reduced in those receiving antibiotic therapy (3% vs. 4.6%, odds ratio of 0.61) (Ma et al., 2012). COX-2 inhibition with celecoxib for 24 months was tested alone or following H. pylori treatment versus placebo in a randomized trial of 1024 patients with H. pylori and advanced gastric lesions (as defined as severe chronic atrophic gastric, intestinal metaplasia, indefinite dysplasia, or dysplasia) (Wong et al., 2012). Interestingly, regression of gastric lesions was increased by both celecoxib and anti-H. pylori treatment alone (53% vs. 41% and 59% vs. 41%, respectively), however celecoxib following anti-H. pylori treatment showed no benefit over placebo. A meta-analysis of aspirin from cardiovascular disease trials revealed that aspirin was associated with reduction in a variety of cancers including gastric cancer mortality if treated for > 10 years (HR 0.42) (Rothwell et al., 2011). Based on the available evidence, screening and local ablative therapies for high-risk Barrett’s esophagus to prevent EAC and treatment of H. pylori in high-risk areas is recommended. Although data regarding NSAIDs and aspirin appear encouraging, large phase III studies are needed before firm recommendations can be made.

Bladder cancer Urothelial carcinoma of the bladder is most commonly attributed to tobacco as well as environmental and occupational carcinogen exposures. Although a number of trials for primary and secondary chemoprevention have been conducted, none have proven effective with the exception of intravesical agents. Only Bacillus-Calmette-Guerin (BCG) and intravesical valrubicin in BCG-refractory patients are FDA-approved (Table 2). A comprehensive review of intravesical therapy to prevent recurrence of nonmuscle invasive bladder cancer will not be reviewed here. Instead, the results of major chemoprevention trials of oral agents will be discussed. Previous studies using retinoids to prevent recurrence in bladder cancer patients have been small and limited by toxicity. Two trials with prolonged low-dose etretinate appeared to have some efficacy. One double-blind, placebo-controlled trial studied the preventive effect of etretinate in 30 patients with superficial bladder cancers (Alfthan et al., 1983). A reduction in recurrence was seen, however a larger multicenter randomized study of 90 patients observed significant toxicity including deaths from myocardial infarction precluding further study of this agent (Studer et al., 1995). Fenretinide, a better tolerated retinoid, failed to show any significant difference in recurrence rates over placebo in two randomized trials (Sabichi et al., 2008; Decensi et al., 2000). Notably, no benefit of retinoids as primary prevention of bladder cancer was observed in the ATBC trial (described in Lung Cancers section) or a meta-analysis of six case-control studies (Alpha-Tocopherol, Beta Carotene Cancer Prevention Study Group, 1994; Steinmaus et al., 2000). A number of other agents including vitamins, selenium, DFMO, and NSAIDs have been studied with largely disappointing results. Pyridoxine (vitamin B6) was hypothesized to counteract the high level of tryptophan metabolites observed in patients with bladder carcinoma. Despite encouraging results from a smaller trial, a larger European Organization for Research and Treatment of Cancer double-blind randomized phase III trial of 291 men with superficial bladder cancer showed no difference between pyridoxine and placebo with regards to recurrence rate or time to first recurrence (Newling et al., 1995). Notwithstanding promising preclinical rational, the phase III Selenium and Bladder Cancer Trial (SELEBLAT) randomized141 patients with noninvasive bladder cancer to selenium-yeast supplementation at 200 mg/day for 3 years and observed no significant decrease in recurrence compared with placebo (Goossens et al., 2016). Similarly, DFMO, a polyamine synthesis inhibitor, at 1 gram daily for 12 months failed to prevent recurrent bladder cancer compared with placebo (Messing et al., 2006). NSAIDs have also failed to show benefit in preventing secondary bladder cancer. In one trial patients were treated with celecoxib 200 mg twice daily for an average of 1.25 years with no difference in recurrence rate compared with the placebo group (Sabichi et al., 2011). Another larger double-blind phase III trial of celecoxib 200 mg twice daily for two years found no difference in recurrent disease and observed an increase in cardiovascular events (Kelly et al., 2015). Currently, there is no role for any nonintravesical chemoprevention agent in the primary or secondary prevention of bladder cancer.

360

Chemoprevention Trials

Cervical cancer Following the regular implementation of the Papanicolaou (Pap) test for cervical cancer screening, a marked reduction in cervical cancer incidence and mortality in the US and worldwide was observed (Peirson et al., 2013). HPV is the chief etiologic agent for cervical cancer carcinogenesis. Although there are at least 118 HPV types, around 40 are specific to the anogenital epithelium and among these, several high-risk HPV types are considered carcinogenic including HPV types 16 and 18, which account for 70% of diagnoses (de Villiers et al., 2004). Both squamous cell carcinoma and adenocarcinoma of the cervix undergo wellrecognized premalignant dysplastic stages (i.e., cervical intraepithelial neoplasia [CIN] 1–3 and adenocarcinoma in situ [AIS], respectively) that are accessible for chemoprevention study. Primary prevention with HPV vaccination has markedly reduced the risk of cervical cancer and the development of CIN 2 and 3. The FUTURE II trial randomized 12,167 women between the ages of 15 and 26 to three doses of either HPV-6/11/16/18 vaccine or placebo, given at day 1, month 2, and month 6 with a primary composite endpoint of CIN 2 or 3, AIS, or cervical cancer related to HPV 16 or 18. At three years, the quadrivalent vaccine prevented the primary composite endpoint by 98% compared with 44% in the placebo group (FUTURE II Study Group, 2007). In a subsequent randomized trial of 14,215 women aged 16–26, the 9-valent HPV vaccine was equally effective at preventing cervical, vulvar, and vaginal cancers and warts as the quadrivalent vaccine (Joura et al., 2015). Other large trials of the bivalent HPV vaccine also showed similar efficacy in preventing CIN 2 or more severe disease due to HPV infection (Hildesheim et al., 2014). Quadrivalent HPV vaccination of 4065 boys and men aged 16–26 prevented infections with HPV-6/11/16/18 and led to 60% less genital lesions compared with placebo (Giuliano et al., 2011). These data led to the FDA approval of HPV vaccines: Gardasil9 for males and females as well as Cervarix for females (see Table 1). Prior to the development of HPV vaccination, chemoprevention utilizing a variety of agents including interferon, folic acid, b-carotene, DFMO, and retinoids were tested with largely negative results. The efficacies of HPV vaccination along with the improved techniques of cervical screening remain the cornerstone of cervical cancer prevention.

Prostate cancer The high incidence, long latency, dependence on androgen, and known intermediate biomarkers (high-grade prostatic intraepithelial neoplasia [HG-PIN] and prostate specific antigen [PSA]) of prostate cancer has led to numerous efforts focusing on chemoprevention. The 5-areductase (5-AR) inhibitors and nutrients including selenium and vitamin E have been best studied. The Prostate Cancer Prevention Trial (PCPT) was launched in 1993 with its principle endpoint being a reduction of biopsyproven prostate cancer incidence. This double-blind placebo-controlled trial included 18,882 males aged 55 years and older who were randomized to finasteride, a 5-AR inhibitor, at 5 mg or placebo daily for seven years. The trial was closed early due to meeting the primary study endpoint with a 25% decrease in biopsy-proven prostate cancer, however controversy ensued due to the observed absolute increase in number and proportion of patients with high-grade (Gleason score  7) prostate cancers (Thompson et al., 2003). At 18 years of follow up, the risk of high-grade prostate cancer remained elevated in patients randomized to finasteride (3.5% vs. 3%) and no between-group difference in the rates of overall or cancer-specific survival was observed (Thompson et al., 2013). In another trial (REDUCE trial) testing the 5-AR inhibitor dutasteride in men with an elevated PSA of 2.5–10 ng/mL and a negative prostate biopsy showed a nearly 25% reduction in biopsy-proven prostate cancer. However, similar to PCPT, higher rates of high-grade prostate cancer were observed (Andriole et al., 2010). Selenium and vitamin E also had been observed to decrease prostate cancer incidence in several early phase trials. Based on these early phase studies, the National Cancer Institute launched the SELECT trial which tested if selenium 200 mg/day, vitamin E 400 IU/ day, both, or placebo in 35,533 men with PSA  SA 33 menbolenium 200 mg/day, exam (Klein et al., 2011). Study treatment was stopped after a three-year interim analysis suggested futility and vitamin E was noted to increase the risk of prostate cancer (HR 1.17). Similarly, selenium alone or in combination with vitamin E showed no chemopreventive benefit even in those with low selenium status. Importantly, selenium increased the risk of high-grade prostate cancer among those with high selenium status (Kristal et al., 2014). Currently, the evidence to support any 5-AR inhibitor for prostate cancer chemoprevention is limited due to the potential for increasing high-grade prostate cancer and the lack of any demonstrable impact on mortality. Nutrients including selenium and vitamin E, above the recommended dietary intake, should not be endorsed for the chemoprevention of prostate cancer.

Hepatocellular carcinoma HBV and hepatitis C virus (HCV) account for the major risk factors for hepatocellular carcinoma (HCC) due to the development of chronic liver disease. HBV-related HCC has declined with the increasing use of the HBV vaccination. In a 20-year follow-up study from Taiwan, where universal vaccination was first implemented in 1984, there was a nearly 70% risk reduction of HCC in vaccinated compared with unvaccinated cohorts (Chang et al., 2009). Although an effective HCV vaccine has yet to be created, observational data of antiviral therapy in the form of interferon or direct-acting antivirals might reduce the risk of HCV-related HCC, particularly those resulting in sustained viral response (Chou et al., 2014; Smith et al., 2012). Studies of retinoids in HCC have been described as well. A prospective randomized trial of 89 patients who were definitively treated for HCC received polyprenoic acid (600 mg daily) or placebo for 12 months. There was a significant decrease in recurrent HCC in the retinoid-treated group (27% vs. 49%) which lasted up to 199 weeks after randomization (Takai et al., 2005; Muto et al., 1996; Chu et al., 2010). Further studies are needed to verify these encouraging results.

Chemoprevention Trials

361

Future Directions Cancer Risk Models Optimally, treatment of epithelial malignancies should move toward prevention. Since the early 1980s, many patients at high risk for epithelial cancers, mostly those with premalignant lesions or a personal history of epithelial malignancy, have been enrolled in chemoprevention trials (as described earlier). While this accounts for a substantial number of patients, the vast majority of epithelial malignancies arise in patients with no history of either of these risk factors. This particular population is, at present, generally unrecognizable and is therefore most often not entering chemoprevention trials. This highlights the need to develop risk models to define high-risk people from the general population. Historically, cancer risk models have relied largely on risk factors obtained from population-level data. The Gail Model of breast cancer risk represents a relatively simple example of this and is related to multiple variables accounting for estrogen exposure (Rockhill et al., 2001). Although the Gail Model is validated in breast cancer, similar models for other cancers as well as more refined breast cancer risk assessments ideally require the incorporation of measurable objective biologic data. A simple example of such an objective risk assessment is diagnosing heredity breast and ovarian cancer syndrome via germ-line molecular testing of the BRCA-1 and BRCA-2 genes. This thereby allows identification of a high-risk cohort available for further study. However, for more common non-hereditary cancers it is likely that a combination of histologic, genetic, epigenetic, and proteomic analyses will elucidate those at highest risk. Utilizing transcriptome meta-analysis, scientists found specific gene signatures that drive HCC risk in cirrhosis (Nakagawa et al., 2016). Other studies utilizing novel technologies such as hierarchical gene clustering and computer-assisted imaging analysis (e.g., morphometric signatures) are ongoing (Gann et al., 2013; Ciriello et al., 2015). If validated, such comprehensive tools can be used to select the “highest-risk” patients for chemoprevention trials, thereby avoiding diluting any possible signal of efficacy by inclusion of lower-risk patients.

Summary The ultimate goal of oncology is to prevent rather than treat morbid invasive cancer. Despite multiple chemoprevention trial failures described earlier, the successes of endocrine therapy in breast cancer primary prevention and HPV vaccinations in cervical cancer provide proof of principle that chemoprevention has the potential to significantly decrease the burden of cancer in society. Ongoing advances in our understanding of the molecular complexities driving carcinogenesis provide a plethora of new avenues to investigate. Current concerted multidisciplinary efforts such as the Precancer Atlas represent a major shift toward better defining the earliest features of neoplasia and exploiting all the technological advances in contemporary science to successfully undertake future precision chemoprevention trials (Spira et al., 2017).

See also: Cancer Survival and Survivorship. Chemoprevention of Cancer: An Overview of Promising Agents and Current Research. Hereditary Cancer Syndromes: Identification and Management.

References Aggarwal, B.B., 2004. Nuclear factor-kappaB: The enemy within [Internet]. Cancer Cell 6, 203–208. Available from: http://linkinghub.elsevier.com/retrieve/pii/S1535610804002442 [cited 2017 June 2019]. Alberts, D.S., Martinez, M.E., Roe, D.J., et al., 2000. Lack of effect of a high-fiber cereal supplement on the recurrence of colorectal adenomas [Internet]. The New England Journal of Medicine 342, 1156–1162. Available from: http://www.ncbi.nlm.nih.gov/pubmed/10770980 [cited 24 June 2017]. Alfthan, O., Tarkkanen, J., Gröhn, P., et al., 1983. Tigason (etretinate) in prevention of recurrence of superficial bladder tumors. A double-blind clinical trial [Internet]. European Urology 9, 6–9. Available from: http://www.ncbi.nlm.nih.gov/pubmed/6337054 [cited 25 June 2017]. Alpha-Tocopherol, Beta Carotene Cancer Prevention Study Group, 1994. The effect of vitamin E and beta carotene on the incidence of lung cancer and other cancers in male smokers [Internet]. The New England Journal of Medicine 330, 1029–1035. Available from: http://www.nejm.org/doi/abs/10.1056/NEJM199404143301501 [cited 24 June 2017]. Andriole, G.L., Bostwick, D.G., Brawley, O.W., et al., 2010. Effect of dutasteride on the risk of prostate cancer [Internet]. The New England Journal of Medicine 362, 1192–1202. Available from: http://www.nejm.org/doi/abs/10.1056/NEJMoa0908127 [cited 25 June 2017]. Arber, N., Eagle, C.J., Spicak, J., et al., 2006. Celecoxib for the prevention of colorectal adenomatous polyps [Internet]. The New England Journal of Medicine 355, 885–895. Available from: http://www.nejm.org/doi/abs/10.1056/NEJMoa061652 [cited 24 June 2017]. Bailey, H.H., Kim, K., Verma, A.K., et al., 2010. A randomized, double-blind, placebo-controlled phase 3 skin cancer prevention study of -difluoromethylornithine in subjects with previous history of skin cancer [Internet]. Cancer Prevention Research 3, 35–47. Available from: http://www.ncbi.nlm.nih.gov/pubmed/20051371 [cited 25 June 2017]. Baron, J.A., Barry, E.L., Mott, L.A., et al., 2015. A trial of calcium and vitamin D for the prevention of colorectal adenomas [Internet]. The New England Journal of Medicine 373, 1519–1530. Available from: http://www.ncbi.nlm.nih.gov/pubmed/26465985 [cited 24 June 2017]. Bavinck, J.N., Tieben, L.M., Van der Woude, F.J., et al., 1995. Prevention of skin cancer and reduction of keratotic skin lesions during acitretin therapy in renal transplant recipients: A double-blind, placebo-controlled study [Internet]. Journal of Clinical Oncology 13, 1933–1938. Available from: http://ascopubs.org/doi/10.1200/JCO.1995.13.8.1933 [cited 25 June 2017]. Bertagnolli, M.M., Eagle, C.J., Zauber, A.G., et al., 2006. Celecoxib for the prevention of sporadic colorectal adenomas [Internet]. The New England Journal of Medicine 355, 873– 884. Available from: http://www.nejm.org/doi/abs/10.1056/NEJMoa061355 [cited 24 June 2017]. Bibbins-Domingo, K., 2016. U.S. Preventive Services Task Force: Aspirin use for the primary prevention of cardiovascular disease and colorectal cancer: U.S. Preventive Services Task Force recommendation statement [Internet]. Annals of Internal Medicine 164, 836–845. Available from: http://annals.org/article.aspx?doi¼10.7326/M16-0577 [cited 30 July 2017].

362

Chemoprevention Trials

Blot, W.J., Li, J.Y., Taylor, P.R., et al., 1993. Nutrition intervention trials in Linxian, China: Supplementation with specific vitamin/mineral combinations, cancer incidence, and disease-specific mortality in the general population [Internet]. Journal of the National Cancer Institute 85, 1483–1492. Available from: http://www.ncbi.nlm.nih.gov/pubmed/ 8360931 [cited 25 June 2017]. Bolla, M., Lefur, R., Ton Van, J., et al., 1994. Prevention of second primary tumours with etretinate in squamous cell carcinoma of the oral cavity and oropharynx. Results of a multicentric double-blind randomised study [Internet]. European Journal of Cancer 30A, 767–772. Available from: http://www.ncbi.nlm.nih.gov/pubmed/7917535 [cited 24 June 2017]. Burn, J., Bishop, D.T., Mecklin, J.-P., et al., 2008. Effect of aspirin or resistant starch on colorectal neoplasia in the Lynch syndrome [Internet]. The New England Journal of Medicine 359, 2567–2578. Available from: http://www.nejm.org/doi/abs/10.1056/NEJMoa0801297 [cited 24 June 2017]. Burn J, Bishop DT, Chapman PD, et al.: A randomized placebo-controlled prevention trial of aspirin and/or resistant starch in young people with familial adenomatous polyposis [Internet]. Cancer Prevention Research (Philadelphia, Pa.) 4:655–65, 2011 Available from: http://cancerpreventionresearch.aacrjournals.org/cgi/doi/10.1158/1940-6207.CAPR11-0106 [cited 24 June 2017] Burn, J., Gerdes, A.-M., Macrae, F., et al., 2011. Long-term effect of aspirin on cancer risk in carriers of hereditary colorectal cancer: An analysis from the CAPP2 randomised controlled trial [Internet]. Lancet (London, England) 378, 2081–2087. Available from: http://linkinghub.elsevier.com/retrieve/pii/S0140673611610490 [cited 24 June 2017]. Caruso, C., Lio, D., Cavallone, L., et al., 2004. Aging, longevity, inflammation, and cancer [Internet]. Annals of the New York Academy of Sciences 1028, 1–13. Available from: http://www.ncbi.nlm.nih.gov/pubmed/15915584 [cited 2017 June 2019]. Chan, A.T., Giovannucci, E.L., Meyerhardt, J.A., et al., 2005. Long-term use of aspirin and nonsteroidal anti-inflammatory drugs and risk of colorectal cancer. [Internet]. JAMA 294, 914–923. Available from: http://jama.jamanetwork.com/article.aspx?doi¼10.1001/jama.294.8.914 [cited 24 June 2017]. Chang, M.-H., You, S.-L., Chen, C.-J., et al., 2009. Decreased incidence of hepatocellular carcinoma in hepatitis B vaccinees: A 20-year follow-up study [Internet]. Journal of the National Cancer Institute 101, 1348–1355. Available from: http://www.ncbi.nlm.nih.gov/pubmed/19759364 [cited 19 June 2017]. Chaturvedi, A.K., Engels, E.A., Pfeiffer, R.M., et al., 2011. Human papillomavirus and rising oropharyngeal cancer incidence in the United States [Internet]. Journal of Clinical Oncology 29, 4294–4301. Available from: http://www.ncbi.nlm.nih.gov/pubmed/21969503 [cited 30 July 2017]. Chen, A.C., Martin, A.J., Choy, B., et al., 2015. A phase 3 randomized trial of nicotinamide for skin-cancer chemoprevention [Internet]. The New England Journal of Medicine 373, 1618–1626. Available from: http://www.nejm.org/doi/10.1056/NEJMoa1506197 [cited 25 June 2017]. Chhabra, G., Ndiaye, M.A., Garcia-Peterson, L.M., et al., 2017. Melanoma chemoprevention: Current status and future prospects [Internet]. Photochemistry and Photobiology. Available from: http://www.ncbi.nlm.nih.gov/pubmed/28295364 [cited 25 June 2017]. Chou, R., Dana, T., Bougatsos, C., et al., 2014. Screening for hepatitis B virus infection in adolescents and adults: A systematic review to update the U.S. preventive services task force recommendation [Internet]. Annals of Internal Medicine 161, 31. Available from: http://www.ncbi.nlm.nih.gov/pubmed/24861032 [cited 25 June 2017]. Chu, K.-J., ECH, L., Yao, X.-P., et al., 2010. Vitamin analogues in chemoprevention of hepatocellular carcinoma after resection or ablationdA systematic review and meta-analysis [Internet]. Asian Journal of Surgery 33, 120–126. Available from: http://www.ncbi.nlm.nih.gov/pubmed/21163409 [cited 25 June 2017]. Ciriello, G., Gatza, M.L., Beck, A.H., et al., 2015. Comprehensive molecular portraits of invasive lobular breast cancer [Internet]. Cell 163, 506–519. Available from: http://www. ncbi.nlm.nih.gov/pubmed/26451490 [cited 25 June 2017]. Clark, L.C., Combs, G.F., Turnbull, B.W., et al., 1996. Effects of selenium supplementation for cancer prevention in patients with carcinoma of the skin. A randomized controlled trial. Nutritional Prevention of Cancer Study Group [Internet]. JAMA 276, 1957–1963. Available from: http://www.ncbi.nlm.nih.gov/pubmed/8971064 [cited 24 June 2017]. Cole BF, Logan RF, Halabi S, et al.: Aspirin for the chemoprevention of colorectal adenomas: Meta-analysis of the randomized trials [Internet]. Journal of the National Cancer Institute 101:256–66, 2009 Available from: https://academic.oup.com/jnci/article-lookup/doi/10.1093/jnci/djn485 [cited 24 June 2017] Cook, N.R., Lee, I.-M., Zhang, S.M., et al., 2013. Alternate-day, low-dose aspirin and cancer risk: Long-term observational follow-up of a randomized trial [Internet]. Annals of Internal Medicine 159, 77–85. Available from: http://annals.org/article.aspx?doi¼10.7326/0003-4819-159-2-201307160-00002 [cited 24 June 2017]. Cuzick, J., Sestak, I., Bonanni, B., et al., 2013. Selective oestrogen receptor modulators in prevention of breast cancer: An updated meta-analysis of individual participant data [Internet]. Lancet (London, England) 381, 1827–1834. Available from: http://linkinghub.elsevier.com/retrieve/pii/S0140673613601403 [cited 25 June 2017]. Cuzick, J., Sestak, I., Forbes, J.F., et al., 2014. Anastrozole for prevention of breast cancer in high-risk postmenopausal women (IBIS-II): An international, double-blind, randomised placebo-controlled trial [Internet]. Lancet 383, 1041–1048. Available from: http://www.ncbi.nlm.nih.gov/pubmed/24333009 [cited 25 June 2017]. Decensi, A., Torrisi, R., Bruno, S., et al., 2000. Randomized trial of fenretinide in superficial bladder cancer using DNA flow cytometry as an intermediate end point [Internet]. Cancer Epidemiology, Biomarkers & Prevention 9, 1071–1078. Available from: http://www.ncbi.nlm.nih.gov/pubmed/11045790 [cited 25 June 2017]. Drahos, J., Xiao, Q., Risch, H.A., et al., 2016. Age-specific risk factor profiles of adenocarcinomas of the esophagus: A pooled analysis from the international BEACON consortium [Internet]. International Journal of Cancer 138, 55–64. Available from: http://doi.wiley.com/10.1002/ijc.29688 [cited 25 June 2017]. Early Breast Cancer Trialists’ Collaborative Group (EBCTCG), Davies, C., Godwin, J., et al., 2011. Relevance of breast cancer hormone receptors and other factors to the efficacy of adjuvant tamoxifen: Patient-level meta-analysis of randomised trials [Internet]. Lancet (London, England) 378, 771–784. Available from: http://linkinghub.elsevier.com/retrieve/ pii/S0140673611609938 [cited 25 June 2017]. Early Breast Cancer Trialists’ Collaborative Group (EBCTCG), Dowsett, M., Forbes, J.F., et al., 2015. Aromatase inhibitors versus tamoxifen in early breast cancer: Patient-level metaanalysis of the randomised trials [Internet]. Lancet 386, 1341–1352. Available from: http://www.ncbi.nlm.nih.gov/pubmed/26211827 [cited 25 June 2017]. Elmets, C.A., Viner, J.L., Pentland, A.P., et al., 2010. Chemoprevention of nonmelanoma skin cancer with celecoxib: A randomized, double-blind, placebo-controlled trial [Internet]. Journal of the National Cancer Institute 102, 1835–1844. Available from: https://academic.oup.com/jnci/article-lookup/doi/10.1093/jnci/djq442 [cited 25 June 2017]. Fabian, C.J., Kimler, B.F., 2016. Incorporating biomarkers in studies of chemoprevention [Internet]. Advances in Experimental Medicine and Biology 69–94. Available from: http:// www.ncbi.nlm.nih.gov/pubmed/26987531 [cited 25 June 2017]. Fearon, E.R., Vogelstein, B., 1990. A genetic model for colorectal tumorigenesis [Internet]. Cell 61, 759–767. Available from: http://www.ncbi.nlm.nih.gov/pubmed/2188735 [cited 2017 June 2019]. Fernandez Figueras, M.T., 2017. From actinic keratosis to squamous cell carcinoma: Pathophysiology revisited [Internet]. Journal of the European Academy of Dermatology and Venereology 31, 5–7. Available from: http://www.ncbi.nlm.nih.gov/pubmed/28263020 [cited 25 June 2017]. Forastiere, A., Koch, W., Trotti, A., et al., 2001. Head and neck cancer [Internet] The New England Journal of Medicine 345, 1890–1900. Available from: http://www.nejm.org/doi/ abs/10.1056/NEJMra001375 [cited 30 July 2017]. Frieling, U.M., Schaumberg, D.A., Kupper, T.S., et al., 2000. A randomized, 12-year primary-prevention trial of beta carotene supplementation for nonmelanoma skin cancer in the physician’s health study [Internet]. Archives of Dermatology 136, 179–184. Available from: http://www.ncbi.nlm.nih.gov/pubmed/10677093 [cited 25 June 2017]. FUTURE II Study Group, 2007. Quadrivalent vaccine against human papillomavirus to prevent high-grade cervical lesions [Internet]. The New England Journal of Medicine 356, 1915–1927. Available from: http://www.nejm.org/doi/abs/10.1056/NEJMoa061741 [cited 25 June 2017]. Gann, P.H., Deaton, R., Amatya, A., et al., 2013. Development of a nuclear morphometric signature for prostate cancer risk in negative biopsies [Internet]. PLoS ONE 8, e69457. Available from: http://www.ncbi.nlm.nih.gov/pubmed/23922715 [cited 25 June 2017]. Garland, S.M., Kjaer, S.K., Muñoz, N., et al., 2016. Impact and effectiveness of the quadrivalent human papillomavirus vaccine: A systematic review of 10 years of real-world experience [Internet]. Clinical Infectious Diseases 63, 519–527. Available from: http://www.ncbi.nlm.nih.gov/pubmed/27230391 [cited 19 June 2017. George, R., Weightman, W., Russ, G.R., et al., 2002. Acitretin for chemoprevention of non-melanoma skin cancers in renal transplant recipients [Internet]. The Australasian Journal of Dermatology 43, 269–273. Available from: http://www.ncbi.nlm.nih.gov/pubmed/12423433 [cited 25 June 2017]. Giardiello, F.M., Yang, V.W., Hylind, L.M., et al., 2002. Primary chemoprevention of familial adenomatous polyposis with sulindac [Internet]. The New England Journal of Medicine 346, 1054–1059. Available from: http://www.nejm.org/doi/abs/10.1056/NEJMoa012015 [cited 24 June 2017].

Chemoprevention Trials

363

Gillison, M.L., Broutian, T., RKL, P., et al., 2012. Prevalence of oral HPV infection in the United States, 2009–2010 [Internet]. JAMA 307, 693. Available from: http://www.ncbi.nlm. nih.gov/pubmed/22282321 [cited 30 July 2017]. Giuliano, A.R., Palefsky, J.M., Goldstone, S., et al., 2011. Efficacy of quadrivalent HPV vaccine against HPV Infection and disease in males [Internet]. The New England Journal of Medicine 364, 401–411. Available from: http://www.nejm.org/doi/abs/10.1056/NEJMoa0909537 [cited 25 June 2017]. Goff, D.C., Lloyd-Jones, D.M., Bennett, G., et al., 2013. 2013 ACC/AHA guideline on the assessment of cardiovascular risk [Internet], Circulation. Available from: http://circ. ahajournals.org/content/early/2013/11/11/01.cir.0000437741.48606.98 [cited 30 July 2017]. Goossens, M.E., Zeegers, M.P., van Poppel, H., et al., 2016. Phase III randomised chemoprevention study with selenium on the recurrence of non-invasive urothelial carcinoma. The SELEnium and BLAdder cancer Trial [Internet]. European Journal of Cancer 69, 9–18. Available from: http://linkinghub.elsevier.com/retrieve/pii/S0959804916324522 [cited 25 June 2017]. Goss PE, Ingle JN, Ales-Martinez JE, et al.: Exemestane for breast-cancer prevention in postmenopausal women [Internet]. The New England Journal of Medicine 364:2381–2391, 2011 Available from: http://www.ncbi.nlm.nih.gov/pubmed/21639806 [cited 25 June 2017] Grau, M.V., Baron, J.A., Sandler, R.S., et al., 2003. Vitamin D, calcium supplementation, and colorectal adenomas: Results of a randomized trial [Internet]. Journal of the National Cancer Institute 95, 1765–1771. Available from: http://www.ncbi.nlm.nih.gov/pubmed/14652238 [cited 24 June 2017]. Greenberg, E.R., Baron, J.A., Stukel, T.A., et al., 1990. A clinical trial of beta carotene to prevent basal-cell and squamous-cell cancers of the skin [Internet]. The New England Journal of Medicine 323, 789–795. Available from: http://www.ncbi.nlm.nih.gov/pubmed/2202901 [cited 25 June 2017]. de Gruijl, F.R., van Kranen, H.J., Mullenders, L.H., 2001. UV-induced DNA damage, repair, mutations and oncogenic pathways in skin cancer [Internet]. Journal of Photochemistry and Photobiology. B 63, 19–27. Available from: http://www.ncbi.nlm.nih.gov/pubmed/11684448 [cited 25 June 2017]. Gupta, A.K., Paquet, M., Villanueva, E., et al., 2012. Interventions for actinic keratoses [Internet]. Cochrane Database of Systematic Reviews 12, CD004415. Available from: http:// doi.wiley.com/10.1002/14651858.CD004415.pub2 [cited 25 June 2017]. Hanahan, D., Weinberg, R., 2011. Hallmarks of cancer: The next generation [Internet]. Cell 144, 646–674. Available from: http://www.ncbi.nlm.nih.gov/pubmed/21376230 [cited 2017 June 2019]. Haque, T., Rahman, K.M., Thurston, D.E., et al., 2015. Topical therapies for skin cancer and actinic keratosis [Internet]. European Journal of Pharmaceutical Sciences 77, 279–289. Available from: http://www.ncbi.nlm.nih.gov/pubmed/26091570 [cited 25 June 2017]. Heath, E.I., Canto, M.I., Piantadosi, S., et al., 2007. Secondary chemoprevention of Barrett’s esophagus with celecoxib: Results of a randomized trial [Internet]. Journal of the National Cancer Institute 99, 545–557. Available from: http://www.ncbi.nlm.nih.gov/pubmed/17405999 [cited 25 June 2017]. Hennekens, C.H., Buring, J.E., Manson, J.E., et al., 1996. Lack of effect of long-term supplementation with beta carotene on the incidence of malignant neoplasms and cardiovascular disease [Internet]. The New England Journal of Medicine 334, 1145–1149. Available from: http://www.nejm.org/doi/abs/10.1056/NEJM199605023341801 [cited 24 June 2017]. Herrero, R., Quint, W., Hildesheim, A., et al., 2013. Reduced prevalence of oral human papillomavirus (HPV) 4 years after bivalent HPV vaccination in a randomized clinical trial in Costa Rica [Internet]. PLoS ONE 8, e68329. Available from: http://dx.plos.org/10.1371/journal.pone.0068329 [cited 24 June 2017]. Hildesheim, A., Wacholder, S., Catteau, G., et al., 2014. Efficacy of the HPV-16/18 vaccine: Final according to protocol results from the blinded phase of the randomized Costa Rica HPV-16/18 vaccine trial [Internet]. Vaccine 32, 5087–5097. Available from: http://linkinghub.elsevier.com/retrieve/pii/S0264410X14008275 [cited 25 June 2017]. Hong, W.K., Endicott, J., Itri, L.M., et al., 1986. 13-cis-Retinoic acid in the treatment of oral leukoplakia [Internet]. The New England Journal of Medicine 315, 1501–1505. Available from: http://www.nejm.org/doi/abs/10.1056/NEJM198612113152401 [cited 24 June 2017]. Hong, W.K., Lippman, S.M., Itri, L.M., et al., 1990. Prevention of second primary tumors with isotretinoin in squamous-cell carcinoma of the head and neck [Internet]. The New England Journal of Medicine 323, 795–801. Available from: http://www.nejm.org/doi/abs/10.1056/NEJM199009203231205 [cited 24 June 2017]. Jacobi, N., Seeboeck, R., Hofmann, E., et al., 2017. ErbB family signalling: A paradigm for oncogene addiction and personalized oncology [Internet]. Cancers (Basel) 9, 33. Available from: http://www.ncbi.nlm.nih.gov/pubmed/28417948 [cited 2017 June 2019]. Jain, M., Arvanitis, C., Chu, K., et al., 2002. Sustained loss of a neoplastic phenotype by brief inactivation of MYC [Internet]. Science 297 (80), 102–104. Available from: http:// www.ncbi.nlm.nih.gov/pubmed/12098700 [cited 2017 June 2019]. Janakiram, N.B., Rao, C.V., 2014. The role of inflammation in colon cancer [Internet]. In: Advances in experimental medicine and biology, pp. 25–52. Available from: http://www. ncbi.nlm.nih.gov/pubmed/24818718 [cited 24 June 2017]. Jemal, A., Thun, M.J., Ries, L.A.G., et al., 2008. Annual report to the Nation on the Status of Cancer, 1975–2005, featuring trends in lung cancer, tobacco use, and tobacco control [Internet]. Journal of the National Cancer Institute 100, 1672–1694. Available from: http://www.ncbi.nlm.nih.gov/pubmed/19033571 [cited 2017 June 2019]. Joura, E.A., Giuliano, A.R., Iversen, O.-E., et al., 2015. A 9-valent HPV vaccine against infection and intraepithelial neoplasia in women [Internet]. The New England Journal of Medicine 372, 711–723. Available from: http://www.nejm.org/doi/10.1056/NEJMoa1405044 [cited 25 June 2017]. Kadakia, K.C., Barton, D.L., Loprinzi, C.L., et al., 2012. Randomized controlled trial of acitretin versus placebo in patients at high-risk for basal cell or squamous cell carcinoma of the skin (North Central Cancer Treatment Group Study 969251). Cancer 118, 2128–2137. Kamangar, F., Qiao, Y.-L., Yu, B., et al., 2006. Lung cancer chemoprevention: A randomized, double-blind trial in Linxian, China [Internet]. Cancer Epidemiology, Biomarkers & Prevention 15, 1562–1564. Available from: http://cebp.aacrjournals.org/cgi/doi/10.1158/1055-9965.EPI-06-0316 [cited 24 June 2017. Karp, D.D., Lee, S.J., Keller, S.M., et al., 2013. Randomized, double-blind, placebo-controlled, phase III chemoprevention trial of selenium supplementation in patients with resected stage I non-small-cell lung cancer: ECOG 5597 [Internet]. Journal of Clinical Oncology 31, 4179–4187. Available from: http://ascopubs.org/doi/10.1200/JCO.2013.49.2173 [cited 24 June 2017]. Kelloff, G.J., Lippman, S.M., Dannenberg, A.J., et al., 2006. Progress in chemoprevention drug development: The promise of molecular biomarkers for prevention of intraepithelial neoplasia and cancerdA plan to move forward [Internet]. Clinical Cancer Research 12, 3661–3697. Available from: http://clincancerres.aacrjournals.org/cgi/doi/10.1158/ 1078-0432.CCR-06-1104 [cited 2017 June 2019]. Kelly, J., Maynard, L., Mostafid, H., et al., 2015. Celecoxib for the treatment of non-muscle invasive bladder cancer (NMIBC): Results of the randomised BOXIT trial (CRUK/07/004) [Internet]. BJU International 115, 11–13. Available from: http://doi.wiley.com/10.1111/bju.13135 [cited 25 June 2017]. Khuri, F.R., Lee, J.J., Lippman, S.M., et al., 2006. Randomized phase III trial of low-dose isotretinoin for prevention of second primary tumors in stage I and II head and neck cancer patients [Internet] Journal of the National Cancer Institute 98, 441–450. Available from: https://academic.oup.com/jnci/article-lookup/doi/10.1093/jnci/djj091 [cited 24 June 2017]. Klein, E.A., Thompson, I.M., Tangen, C.M., et al., 2011. Vitamin E and the risk of prostate cancer: The selenium and vitamin E cancer prevention trial (SELECT) [Internet]. JAMA 306, 1549–1556. Available from: http://jama.jamanetwork.com/article.aspx?doi¼10.1001/jama.2011.1437 [cited 25 June 2017]. Kristal, A.R., Darke, A.K., Morris, J.S., et al., 2014. Baseline selenium status and effects of selenium and vitamin E supplementation on prostate cancer risk [Internet]. Journal of the National Cancer Institute 106, djt456. Available from: http://www.ncbi.nlm.nih.gov/pubmed/24563519 [cited 25 June 2017]. Lavasani, S., Chlebowski, R.T., Prentice, R.L., et al., 2015. Estrogen and colorectal cancer incidence and mortality [Internet]. Cancer 121, 3261–3271. Available from: http://doi. wiley.com/10.1002/cncr.29464 [cited 25 June 2017]. Lee, J.J., Wu, X., MAT, H., et al., 2011. Global assessment of genetic variation influencing response to retinoid chemoprevention in head and neck cancer patients [Internet]. Cancer Prevention Research (Philadelphia, Pa.) 4, 185–193. Available from: http://cancerpreventionresearch.aacrjournals.org/cgi/doi/10.1158/1940-6207.CAPR-10-0125 [cited 24 June 2017]. Leeman-Neill, R.J., Seethala, R.R., Singh, S.V., et al., 2011. Inhibition of EGFR-STAT3 signaling with erlotinib prevents carcinogenesis in a chemically-induced mouse model of oral squamous cell carcinoma [Internet]. Cancer Prevention Research 4, 230–237. Available from: http://www.ncbi.nlm.nih.gov/pubmed/21163936 [cited 24 June 2017].

364

Chemoprevention Trials

Levine, N., Moon, T.E., Cartmel, B., et al., 1997. Trial of retinol and isotretinoin in skin cancer prevention: A randomized, double-blind, controlled trial. Southwest Skin Cancer Prevention Study Group [Internet]. Cancer Epidemiology, Biomarkers & Prevention 6, 957–961. Available from: http://www.ncbi.nlm.nih.gov/pubmed/9367070 [cited 25 June 2017]. Limburg, P.J., Wei, W., Ahnen, D.J., et al., 2005. Randomized, placebo-controlled, esophageal squamous cell cancer chemoprevention trial of selenomethionine and celecoxib [Internet]. Gastroenterology 129, 863–873. Available from: http://www.ncbi.nlm.nih.gov/pubmed/16143126 [cited 25 June 2017]. Lippman, S.M., Batsakis, J.G., Toth, B.B., et al., 1993. Comparison of low-dose isotretinoin with beta carotene to prevent oral carcinogenesis [Internet]. The New England Journal of Medicine 328, 15–20. Available from: http://www.ncbi.nlm.nih.gov/pubmed/8416267 [cited 24 June 2017]. Lippman, S.M., Lee, J.J., Karp, D.D., et al., 2001. Randomized phase III intergroup trial of isotretinoin to prevent second primary tumors in stage I non-small-cell lung cancer [Internet]. Journal of the National Cancer Institute 93, 605–618. Available from: http://www.ncbi.nlm.nih.gov/pubmed/11309437 [cited 24 June 2017. Ma, J.-L., Zhang, L., Brown, L.M., et al., 2012. Fifteen-year effects of Helicobacter pylori, garlic, and vitamin treatments on gastric cancer incidence and mortality [Internet]. Journal of the National Cancer Institute 104, 488–492. Available from: http://www.ncbi.nlm.nih.gov/pubmed/22271764 [cited 25 June 2017]. Mao, L., Lee, J.S., Fan, Y.H., et al., 1996. Frequent microsatellite alterations at chromosomes 9p21 and 3p14 in oral premalignant lesions and their value in cancer risk assessment [Internet]. Nature Medicine 2, 682–685. Available from: http://www.ncbi.nlm.nih.gov/pubmed/8640560 [cited 24 June 2017]. Messing, E., Kim, K.M., Sharkey, F., et al., 2006. Randomized prospective phase III trial of difluoromethylornithine vs placebo in preventing recurrence of completely resected low risk superficial bladder cancer [Internet]. The Journal of Urology 176, 500–504. Available from: http://linkinghub.elsevier.com/retrieve/pii/S0022534706008305 [cited 25 June 2017]. Meyskens, F.L., 1992. Strategies for prevention of cancer in humans. [Internet]. Oncology (Williston Park) 6, 15–24. Available from: http://www.ncbi.nlm.nih.gov/pubmed/1532731 [cited 2017 June 2019]. Meyskens, F.L., Mukhtar, H., Rock, C.L., et al., 2016. Cancer prevention: Obstacles, challenges, and the road ahead [Internet]. Journal of the National Cancer Institute 108. Available from: http://www.ncbi.nlm.nih.gov/pubmed/26547931 [cited 19 June 2017]. Moon, T.E., Levine, N., Cartmel, B., et al., 1997. Effect of retinol in preventing squamous cell skin cancer in moderate-risk subjects: A randomized, double-blind, controlled trial. Southwest Skin Cancer Prevention Study Group [Internet]. Cancer Epidemiology, Biomarkers & Prevention 6, 949–956. Available from: http://www.ncbi.nlm.nih.gov/pubmed/ 9367069 [cited 25 June 2017]. Mootha VK, Lindgren CM, Eriksson K-F, et al.: PGC-1alpha-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes [Internet]. Nature Genetics 34:267–73, 2003 Available from: http://www.nature.com/doifinder/10.1038/ng1180 [cited 24 June 2017] Muto, Y., Moriwaki, H., Ninomiya, M., et al., 1996. Prevention of second primary tumors by an acyclic retinoid, polyprenoic acid, in patients with hepatocellular carcinoma [Internet] The New England Journal of Medicine 334, 1561–1568. Available from: http://www.ncbi.nlm.nih.gov/pubmed/8628336 [cited 25 June 2017]. Nakagawa, S., Wei, L., Song, W.M., et al., 2016. Molecular liver cancer prevention in cirrhosis by organ transcriptome analysis and lysophosphatidic acid pathway inhibition [Internet]. Cancer Cell 30, 879–890. Available from: http://www.ncbi.nlm.nih.gov/pubmed/27960085 [cited 25 June 2017]. Nelson, H.D., Smith, M.E.B., Griffin, J.C., et al., 2013. Use of medications to reduce risk for primary breast cancer: A systematic review for the U.S. Preventive Services Task Force [Internet]. Annals of Internal Medicine 158, 604–614. Available from: http://annals.org/article.aspx?doi¼10.7326/0003-4819-158-8-201304160-00005 [cited 25 June 2017]. Newling, D.W., Robinson, M.R., Smith, P.H., et al., 1995. Tryptophan metabolites, pyridoxine (vitamin B6) and their influence on the recurrence rate of superficial bladder cancer. Results of a prospective, randomised phase III study performed by the EORTC GU Group. EORTC Genito-Urinary Tract Cancer Cooperative Group [Internet]. European Urology 27, 110–116. Available from: http://www.ncbi.nlm.nih.gov/pubmed/7744151 [cited 25 June 2017]. Omenn, G.S., Goodman, G.E., Thornquist, M.D., et al., 1996. Effects of a combination of beta carotene and vitamin a on lung cancer and cardiovascular disease [Internet]. The New England Journal of Medicine 334, 1150–1155. Available from: http://www.ncbi.nlm.nih.gov/pubmed/8602180 [cited 24 June 2017]. Otley, C.C., Stasko, T., Tope, W.D., et al., 2006. Chemoprevention of nonmelanoma skin cancer with systemic retinoids: Practical dosing and management of adverse effects [Internet]. Dermatologic Surgery 32, 562–568. Available from: http://doi.wiley.com/10.1111/j.1524-4725.2006.32115.x [cited 25 June 2017]. Overholt, B.F., Lightdale, C.J., Wang, K.K., et al., 2005. Photodynamic therapy with porfimer sodium for ablation of high-grade dysplasia in Barrett’s esophagus: International, partially blinded, randomized phase III trial [Internet]. Gastrointestinal Endoscopy 62, 488–498. Available from: http://linkinghub.elsevier.com/retrieve/pii/S0016510705023497 [cited 25 June 2017]. Papadimitrakopoulou, V.A., Lee, J.J., William, W.N., et al., 2009. Randomized trial of 13-cis retinoic acid compared with retinyl palmitate with or without beta-carotene in oral premalignancy [Internet]. Journal of Clinical Oncology 27, 599–604. Available from: http://www.ncbi.nlm.nih.gov/pubmed/19075276 [cited 24 June 2017]. Peirson, L., Fitzpatrick-Lewis, D., Ciliska, D., et al., 2013. Screening for cervical cancer: A systematic review and meta-analysis [Internet]. Systematic Reviews 2, 35. Available from: http://systematicreviewsjournal.biomedcentral.com/articles/10.1186/2046-4053-2-35 [cited 25 June 2017]. Pence, B.C., 1993. Role of calcium in colon cancer prevention: Experimental and clinical studies [Internet]. Mutation Research 290, 87–95. Available from: http://www.ncbi.nlm.nih. gov/pubmed/7694103 [cited 24 June 2017]. Philip, M., Rowley, D.A., Schreiber, H., 2004. Inflammation as a tumor promoter in cancer induction [Internet]. Seminars in Cancer Biology 14, 433–439. Available from: http:// linkinghub.elsevier.com/retrieve/pii/S1044579X04000495 [cited 2017 June 2019]. Qiao, Y.-L., Dawsey, S.M., Kamangar, F., et al., 2009. Total and cancer mortality after supplementation with vitamins and minerals: Follow-up of the linxian general population nutrition intervention trial [Internet]. Journal of the National Cancer Institute 101, 507–518. Available from: http://www.ncbi.nlm.nih.gov/pubmed/19318634 [cited 25 June 2017]. Rockhill, B., Spiegelman, D., Byrne, C., et al., 2001. Validation of the Gail et al. model of breast cancer risk prediction and implications for chemoprevention [Internet]. Journal of the National Cancer Institute 93, 358–366. Available from: http://www.ncbi.nlm.nih.gov/pubmed/11238697 [cited 30 July 2017]. Rothwell, P.M., Wilson, M., Elwin, C.-E., et al., 2010. Long-term effect of aspirin on colorectal cancer incidence and mortality: 20-year follow-up of five randomised trials [Internet]. Lancet (London, England) 376, 1741–1750. Available from: http://linkinghub.elsevier.com/retrieve/pii/S0140673610615437 [cited 24 June 2017]. Rothwell PM, Fowkes FGR, Belch JFF, et al.: Effect of daily aspirin on long-term risk of death due to cancer: Analysis of individual patient data from randomised trials [Internet]. Lancet (London, England) 377:31–41, 2011 Available from: http://linkinghub.elsevier.com/retrieve/pii/S0140673610621101 [cited 25 June 2017] Sabichi, A.L., Lerner, S.P., Atkinson, E.N., et al., 2008. Phase III prevention trial of fenretinide in patients with resected non-muscle-invasive bladder cancer [Internet]. Clinical Cancer Research 14, 224–229. Available from: http://clincancerres.aacrjournals.org/cgi/doi/10.1158/1078-0432.CCR-07-0733 [cited 25 June 2017]. Sabichi, A.L., Lee, J.J., Grossman, H.B., et al., 2011. A randomized controlled trial of celecoxib to prevent recurrence of nonmuscle-invasive bladder cancer [Internet]. Cancer Prevention Research (Philadelphia, Pa.) 4, 1580–1589. Available from: http://www.ncbi.nlm.nih.gov/pubmed/21881030 [cited 25 June 2017]. Saintigny, P., Zhang, L., Fan, Y.-H., et al., 2011. Gene expression profiling predicts the development of oral cancer [Internet]. Cancer Prevention Research (Philadelphia, Pa.) 4, 218–229. Available from: http://cancerpreventionresearch.aacrjournals.org/cgi/doi/10.1158/1940-6207.CAPR-10-0155 [cited 24 June 2017]. Schatzkin, A., Lanza, E., Corle, D., et al., 2000. Lack of effect of a low-fat, high-fiber diet on the recurrence of colorectal adenomas [Internet]. The New England Journal of Medicine 342, 1149–1155. Available from: http://www.ncbi.nlm.nih.gov/pubmed/10770979 [cited 24 June 2017]. de Sévaux, R.G.L., Smit, J.V., de Jong, E.M.G.J., et al., 2003. Acitretin treatment of premalignant and malignant skin disorders in renal transplant recipients: Clinical effects of a randomized trial comparing two doses of acitretin [Internet]. Journal of the American Academy of Dermatology 49, 407–412. Available from: http://www.ncbi.nlm.nih.gov/ pubmed/12963902 [cited 25 June 2017]. Shacter, E., Weitzman, S.A., 2002. Chronic inflammation and cancer [Internet]. Oncology (Williston Park) 16, 217–26, 229–232. Available from: http://www.ncbi.nlm.nih.gov/ pubmed/11866137 [cited 30 July 2017].

Chemoprevention Trials

365

Shaheen, N.J., Sharma, P., Overholt, B.F., et al., 2009. Radiofrequency ablation in Barrett’s esophagus with dysplasia [Internet]. The New England Journal of Medicine 360, 2277– 2288. Available from: http://www.nejm.org/doi/abs/10.1056/NEJMoa0808145 [cited 25 June 2017]. Shoemaker, R.H., Suen, C.S., Holmes, C.A., et al., 2016. The National Cancer Institute’s PREVENT cancer preclinical drug development program: Overview, current projects, animal models, agent development strategies, and molecular targets [Internet]. Seminars in Oncology 43, 189–197. Available from: http://www.ncbi.nlm.nih.gov/pubmed/26970137 [cited 19 June 2017]. Shureiqi, I., Reddy, P., Brenner, D.E., 2000. Chemoprevention: General perspective [Internet]. Critical Reviews in Oncology/Hematology 33, 157–167. Available from: http://www. ncbi.nlm.nih.gov/pubmed/10789490 [cited 19 June 2017]. Siegel, R.L., Miller, K.D., Jemal, A., 2017. Cancer statistics, 2017 [Internet]. CA: a Cancer Journal for Clinicians 67, 7–30. Available from: http://www.ncbi.nlm.nih.gov/pubmed/ 28055103 [cited 2017 June 2019]. Simon, M.S., Chlebowski, R.T., Wactawski-Wende, J., et al., 2012. Estrogen plus progestin and colorectal cancer incidence and mortality [Internet]. Journal of Clinical Oncology 30, 3983–3990. Available from: http://ascopubs.org/doi/10.1200/JCO.2012.42.7732 [cited 24 June 2017]. Slaughter, D.P., Southwick, H.W., Smejkal, W., 1953. Field cancerization in oral stratified squamous epithelium; clinical implications of multicentric origin [Internet]. Cancer 6, 963– 968. Available from: http://www.ncbi.nlm.nih.gov/pubmed/13094644 [cited 24 June 2017]. Smith, B.D., Morgan, R.L., Beckett, G.A., et al., 2012. Recommendations for the identification of chronic hepatitis C virus infection among persons born during 1945–1965 [Internet]. Morbidity and Mortality Weekly Report 61, 1–32. Available from: http://www.ncbi.nlm.nih.gov/pubmed/22895429 [cited 25 June 2017]. Spira, A., Yurgelun, M.B., Alexandrov, L., et al., 2017. Precancer atlas to drive precision prevention trials [Internet]. Cancer Research 77, 1510–1541. Available from: http://www. ncbi.nlm.nih.gov/pubmed/28373404 [cited 25 June 2017]. Steinbach, G., Lynch, P.M., Phillips, R.K., et al., 2000. The effect of celecoxib, a cyclooxygenase-2 inhibitor, in familial adenomatous polyposis [Internet]. The New England Journal of Medicine 342, 1946–1952. Available from: http://www.nejm.org/doi/abs/10.1056/NEJM200006293422603 [cited 24 June 2017]. Steinmaus, C.M., Nuñez, S., Smith, A.H., 2000. Diet and bladder cancer: A meta-analysis of six dietary variables [Internet]. American Journal of Epidemiology 151, 693–702. Available from: http://www.ncbi.nlm.nih.gov/pubmed/10752797 [cited 25 June 2017]. Studer, U.E., Jenzer, S., Biedermann, C., et al., 1995. Adjuvant treatment with a vitamin A analogue (etretinate) after transurethral resection of superficial bladder tumors. Final analysis of a prospective, randomized multicenter trial in Switzerland [Internet]. European Urology 28, 284–290. Available from: http://www.ncbi.nlm.nih.gov/pubmed/8575494 [cited 25 June 2017]. Stürmer, T., Buring, J.E., Lee, I.-M., et al., 2006. Colorectal cancer after start of nonsteroidal anti-inflammatory drug use [Internet]. The American Journal of Medicine 119, 494– 502. Available from: http://linkinghub.elsevier.com/retrieve/pii/S0002934305010740 [cited 24 June 2017]. Szabo, E., Mao, J.T., Lam, S., et al., 2013. Chemoprevention of lung cancer: Diagnosis and management of lung cancer, 3rd ed: American College of Chest Physicians evidence-based clinical practice guidelines [Internet]. Chest 143, e40S–e60S. Available from: http://linkinghub.elsevier.com/retrieve/pii/S0012369213602883 [cited 24 June 2017]. Takai, K., Okuno, M., Yasuda, I., et al., 2005. Prevention of second primary tumors by an acyclic retinoid in patients with hepatocellular carcinoma [Internet]. Intervirology 48, 39– 45. Available from: http://www.ncbi.nlm.nih.gov/pubmed/15785088 [cited 25 June 2017]. Tang, J.Y., Mackay-Wiggan, J.M., Aszterbaum, M., et al., 2012. Inhibiting the hedgehog pathway in patients with the basal-cell nevus syndrome [Internet]. The New England Journal of Medicine 366, 2180–2188. Available from: http://www.nejm.org/doi/abs/10.1056/NEJMoa1113538 [cited 25 June 2017]. Tangrea, J.A., Edwards, B.K., Taylor, P.R., et al., 1992. Long-term therapy with low-dose isotretinoin for prevention of basal cell carcinoma: A multicenter clinical trial. IsotretinoinBasal Cell Carcinoma Study Group [Internet]. Journal of the National Cancer Institute 84, 328–332. Available from: http://www.ncbi.nlm.nih.gov/pubmed/1738183 [cited 25 June 2017]. Taoudi Benchekroun, M., Saintigny, P., Thomas, S.M., et al., 2010. Epidermal growth factor receptor expression and gene copy number in the risk of oral cancer [Internet]. Cancer Prevention Research (Philadelphia, Pa.) 3, 800–809. Available from: http://cancerpreventionresearch.aacrjournals.org/cgi/doi/10.1158/1940-6207.CAPR-09-0163 [cited 24 June 2017]. Thompson, I.M., Goodman, P.J., Tangen, C.M., et al., 2003. The influence of finasteride on the development of prostate cancer [Internet]. The New England Journal of Medicine 349, 215–224. Available from: http://www.nejm.org/doi/abs/10.1056/NEJMoa030660 [cited 25 June 2017]. Thompson, I.M., Goodman, P.J., Tangen, C.M., et al., 2013. Long-term survival of participants in the prostate cancer prevention trial [Internet]. The New England Journal of Medicine 369, 603–610. Available from: http://www.nejm.org/doi/10.1056/NEJMoa1215932 [cited 25 June 2017]. Tufaro, A.P., Azoury, S.C., Crompton, J.G., et al., 2015. Rising incidence and aggressive nature of cutaneous malignancies after transplantation: An update on epidemiology, risk factors, management and surveillance [Internet]. Surgical Oncology 24, 345–352. Available from: http://linkinghub.elsevier.com/retrieve/pii/S0960740415300293 [cited 25 June 2017]. Uray, I.P., Dmitrovsky, E., Brown, P.H., 2016. Retinoids and rexinoids in cancer prevention: From laboratory to clinic [Internet]. Seminars in Oncology 43, 49–64. Available from: http://www.ncbi.nlm.nih.gov/pubmed/26970124 [cited 24 June 2017]. de Villiers, E.-M., Fauquet, C., Broker, T.R., et al., 2004. Classification of papillomaviruses [Internet]. Virology 324, 17–27. Available from: http://linkinghub.elsevier.com/retrieve/pii/ S004268220400220X [cited 25 June 2017]. Vogel, V.G., Costantino, J.P., Wickerham, D.L., et al., 2010. Update of the National Surgical Adjuvant Breast and Bowel Project Study of Tamoxifen and Raloxifene (STAR) P-2 trial: Preventing breast cancer [Internet]. Cancer Prevention Research (Philadelphia, Pa.) 3, 696–706. Available from: http://cancerpreventionresearch.aacrjournals.org/cgi/doi/10. 1158/1940-6207.CAPR-10-0076 [cited 25 June 2017]. Vokes, E.E., Weichselbaum, R.R., Lippman, S.M., et al., 1993. Head and neck cancer [Internet]. The New England Journal of Medicine 328, 184–194. Available from: http://www. nejm.org/doi/abs/10.1056/NEJM199301213280306 [cited 24 June 2017]. Wang, J.-B., Abnet, C.C., Fan, J.-H., et al., 2013. The randomized Linxian Dysplasia Nutrition Intervention Trial after 26 years of follow-up: No effect of multivitamin supplementation on mortality [Internet]. JAMA Internal Medicine 173, 1259–1261. Available from: http://archinte.jamanetwork.com/article.aspx?doi¼10.1001/jamainternmed.2013.6066 [cited 25 June 2017]. Wattenberg, L.W., 1985. Chemoprevention of cancer. [Internet]. Cancer Research 45, 1–8. Available from: http://www.ncbi.nlm.nih.gov/pubmed/3880665 [cited 2017 June 2019]. Weinstein, I.B., 2002. Cancer. Addiction to oncogenesdThe Achilles heal of cancer [Internet]. Science 297, 63–64. Available from: http://www.sciencemag.org/cgi/doi/10.1126/ science.1073096 [cited 2017 June 2019]. Weinstein, I.B., Joe, A., Felsher, D., 2008. Oncogene addiction [Internet]. Cancer Research 68, 3077–3080. Available from: http://www.ncbi.nlm.nih.gov/pubmed/18451130 [cited 2017 June 2019]. William, W.N., Papadimitrakopoulou, V., Lee, J.J., et al., 2016. Erlotinib and the risk of oral cancer [Internet]. JAMA Oncology 2, 209. Available from: http://oncology.jamanetwork. com/article.aspx?doi¼10.1001/jamaoncol.2015.4364 [cited 2017 June 2019]. Wong, B.C.Y., Zhang, L., Ma, J., et al., 2012. Effects of selective COX-2 inhibitor and Helicobacter pylori eradication on precancerous gastric lesions [Internet]. Gut 61, 812–818. Available from: http://gut.bmj.com/lookup/doi/10.1136/gutjnl-2011-300154 [cited 25 June 2017]. Xiong, D., Pan, J., Zhang, Q., et al., 2017. Bronchial airway gene expression signatures in mouse lung squamous cell carcinoma and their modulation by cancer chemopreventive agents [Internet]. Oncotarget 8, 18885–18900. Available from: http://www.ncbi.nlm.nih.gov/pubmed/27935865 [cited 24 June 2017]. van Zandwijk, N., Dalesio, O., Pastorino, U., et al., 2000. EUROSCAN, a randomized trial of vitamin A and N-acetylcysteine in patients with head and neck cancer or lung cancer. For the EUropean Organization for Research and Treatment of Cancer Head and Neck and Lung Cancer Cooperative Groups [Internet]. Journal of the National Cancer Institute 92, 977–986. Available from: http://www.ncbi.nlm.nih.gov/pubmed/10861309 [cited 24 June 2017].

366

Chemoprevention Trials

Zhai, R., Chen, F., Liu, G., et al., 2010. Interactions among genetic variants in apoptosis pathway genes, reflux symptoms, body mass index, and smoking indicate two distinct etiologic patterns of esophageal adenocarcinoma [Internet]. Journal of Clinical Oncology 28, 2445–2451. Available from: http://ascopubs.org/doi/10.1200/JCO.2009.26.2790 [cited 25 June 2017].

Further Reading Rashid, S., 2017. Cancer and chemoprevention: An overview. Springier Publishing. Cho, W.C. (Ed.), 2013. Cancer chemoprevention and treatment by diet therapy. Springer Publishing. Drew, D.A., Cao, Y., Chan, A.T., 2016. Aspirin and colorectal cancer: The promise of precision chemoprevention. Nature Reviews. Cancer 16, 173–186.

Cholangiocarcinoma: Diagnosis and Treatment ska, Science to the Point, Archamps Technopole, Archamps, France Katarzyna Szyman © 2019 Elsevier Inc. All rights reserved.

Abbreviations 18

F-FDG PET/CT 2-Deoxy-2-[fluorine-18]fluoro-D-glucose PET integrated with CT ALP Alkaline phosphatase CA 19-9 Carbohydrate antigen 19-9 CC Cholangiocarcinoma CT Computed tomography CEA Carcinoembryonic antigen EUS Endoscopic ultrasound scan ICC Intrahepatic cholangiocarcinoma ICD-O International Classification of Diseases for Oncology FISH Fluorescence in situ hybridization FNA Fine-needle aspiration FLR Future remnant liver HBS Hepatobiliary scintigraphy IDUS Intra-ductal ultrasound MDCT Multidetector computed tomography MRCP Magnetic resonance cholangiopancreatography MRI Magnetic resonance imaging OLT Orthotopic liver transplantation PET Positron emission tomography PTC Percutaneous transhepatic cholangiography

Definition and Classification Cholangiocarcinoma (CC; ICD-O code: 8160/3; also called bile duct carcinoma and bile duct adenocarcinoma) is a term which encompasses a heterogeneous group of malignant tumors with pathological features of biliary tract differentiation, distinct from gallbladder cancer. It is presumed to arise from the intra- or extrahepatic biliary epithelium (cholangiocytes), although some studies have suggested that it may also arise directly from transdifferentiation of hepatocytes. Cholangiocarcinoma is classified into subtypes based on its anatomical origin, with, however, some debate and even confusion in the nomenclature. Traditionally, cholangiocarcinomas were divided into intrahepatic and extrahepatic. However, the boundary between intra- and extrahepatic biliary tree has been somewhat confused in the literature. Still, while the term “intrahepatic cholangiocarcinoma” seems to be well accepted, the use of the term “extrahepatic cholangiocarcinoma” and the nomenclature of the subtypes have been extensively discussed and revised in the past decade, and different terms referring to the same entities can be encountered in both clinical guidelines and in scientific literature. Intrahepatic cholangiocarcinoma (ICC) is an intrahepatic malignancy with biliary epithelial differentiation. ICC can arise in any portion of the intrahepatic biliary tree, from the segmental and area ducts and their major branches to the smallest bile ducts and ductules. ICC arising from the small intrahepatic bile ducts is often called peripheral ICC but the International Classification of Diseases for Oncology (ICD-O) discourages the use of this term. CC originating from major intrahepatic ducts, including the hilum, has been called hilar CC. However, due to symptoms which are similar to those of extrahepatic CC, it has been classified as an extrahepatic lesion. Hilar tumors arising near the junction of the right and left hepatic duct are traditionally called Klatskin tumors. Current guidelines recommend classifying extrahepatic CC into two subtypes: perihilar and distal CC. However, the term “hilar cholangiocarcinoma” is still widely used. Moreover, one should bear in mind that distinguishing hilar and perihilar CC at advanced stages is usually controversial. All in all, the exact meaning (anatomical scope) of these two terms differs between different specialists.

Presentation and Diagnosis Early CC symptoms are not specific and so most patients present with advanced disease, with extensive hepatic artery and/or portal vein infiltration by tumor and/or with distant metastases, which makes the definitive treatment impossible. A majority of them

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65306-0

367

368

Cholangiocarcinoma: Diagnosis and Treatment

present when jaundice or the sequelae of biliary obstruction appear. The degree of jaundice depends on the extent of biliary involvement. Abdominal pain, weight loss, fever, malaise, and/or hepatomegaly occur in half of the cases, usually in those with the advanced disease. Patients with unrelieved obstruction of large intrahepatic bile ducts may die from complications, such as liver failure or sepsis. It has been proposed that CC can be diagnosed based on a combination of clinical presentation, blood serum testing, and imaging. However, as many of the criteria lack specificity, a pathological examination is recommended for definitive diagnosis in most patients. The initial workup in case of suspected CC comprises blood serum testing. Elevated serum bilirubin and alkaline phosphatase (ALP) levels are common in CC patients. However, incomplete obstruction of the right or left hepatic ducts may lead to isolated elevation of ALP. Increased liver transaminase levels and prothrombin time are detected in case of a long-standing biliary obstruction but also cholestatic hepatocellular injury. Increased serum carbohydrate antigen 19-9 (CA 19-9) and carcinoembryonic antigen (CEA) levels are very common (up to 90%) but they are not specific to CC, either. High-quality cross-sectional imaging plays a key role in the diagnostic process as well as in the preoperative workup. Highresolution contrast-enhanced multidetector computed tomography (MDCT) and magnetic resonance imaging/magnetic resonance cholangiopancreatography (MRI/MRCP) are considered to be the most accurate. For tissue diagnosis based on brush cytology, both endoscopic retrograde cholangiopancreatography (ERCP) and percutaneous transhepatic cholangiography (PTC) have similar sensitivity and specificity. ERCP with brush cytology is particularly specific for the diagnosis of distal CC and has been the standard diagnostic method for years. However, it has a low sensitivity for some cellular patterns. Various techniques have been investigated to increase sensitivity of cytological testing. In particular, fluorescence in situ hybridization (FISH) has been reported to increase the detection sensitivity to up to 93%. Endoscopic methods, such as endoscopic ultrasound scan (EUS), intra-ductal ultrasound (IDUS), and cholangioscopy, are used selectively in diagnosing and staging. Cholangioscopy offers direct visualization of biliary strictures and enables targeted biopsies, which increases both diagnostic sensitivity and specificity. IDUS enables detailed imaging of the bile ducts and periductal tissue and has been reported to improve diagnostic accuracy of ERCP. However, stents that are often required to drain obstructed bile ducts make the interpretation of IDUS results difficult. If this is the case, the use of EUS in combination with fine-needle aspiration (FNA) may be preferable. The latter is also useful to detect suspicious regional lymph node involvement. Radiological analyses are necessary for assessing the disease extent (locoregional or distant spread), staging, and resectability. CCs have a locally aggressive behavior, presenting mainly with infiltration of contiguous structures (liver, hepatic artery and portal vein) and nodal involvement (up to 30% at diagnosis), microscopically invasive spread with neural, perineural and lymphatic involvement, and subepithelial extension. Intrahepatic CCs can be divided into mass-forming, periductal infiltrating, and intraductal growth types. Perihilar CCs can have exophytic (mass-forming) or intraductal macroscopic growth patterns. Over 90% of CCs are well differentiated and mucin-producing adenocarcinomas.

Epidemiology and Risk Factors Burden Globally, cholangiocarcinoma is a very rare tumor, accounting for 3% of all gastrointestinal malignancies and less than 1% of all cancers worldwide. However, intrahepatic CC is the second most common primary liver cancer (after hepatocellular carcinoma), with very high mortality rates. The exact incidence of different subtypes is difficult to estimate due to changing nomenclature and coding issues. However, a majority of CCs arise from the intrahepatic upper third of the biliary tract, including the hilum (50%–70% of all tumors). In countries for which incidence trends have been studies, the rates are reported to be rising. The overall CC incidence shows striking geographical differences, with the highest incidence rates in northeastern Thailand where parasitic infections with endemic liver flukes lead to infestation of the billiary tree. CC occurs with roughly the same incidence in men and women, in most cases after 60 years of age.

Etiology and Risk Factors Infections with liver flukes (Clonorchis sinensis and Opisthorchis viverini), common in such areas of high CC incidence as Thailand (O. viverini) and China (C. sinensis), have been shown to be associated with an increased risk of CC, with a high proportion of CC patients being infected. Chronic infection with these parasites causes severe inflammation damage. Moreover, infections with hepatitis B or C virus have been shown to increase the risk of developing intrahepatic CC by contributing to the development of liver cirrhosis. Several other medical conditions of the gastrointestinal tract have been associated with an increased risk of CC, such as inflammatory bowel disease (in particular chronic ulcerative colitis), primary sclerosing cholangitis, choledocholithiasis, and biliary stone disease. Elevated CC incidence rates have also been reported in patients with congenital anomalies of the intrahepatic and extrahepatic bile ducts (e.g. cysts, congenital dilation of the bile ducts, choledochal cyst, Caroli disease, congenital hepatic fibrosis, polycystic disease, abnormal pancreaticocholedochal junction).

Cholangiocarcinoma: Diagnosis and Treatment

369

A history of exposure to thorotrast, an alpha-particle emitter that was used as a radiopaque intra-arterial contrast medium between 1930 and 1955 is also a causative factor of CC, with latency period that may range from 25 to 48 years. Finally, it has also been suggested that tobacco smoking, alcohol consumption and diabetes may contribute to the risk of intrahepatic CC.

Pathology and Genetics The genetic background of CC is poorly defined. Gains and losses in chromosomal regions containing oncogenes and tumor suppressor genes (e.g., EGFR, ERBB2, MAP2K2/MEK) have been linked to the development of CC, while somatic mutations are relatively rare. Mutations of KRAS, TP53, and IDH1 seem to be the most common genetic alterations in intrahepatic CC. The incidence of KRAS mutations shows a striking geographical variation, ranging from 4% in Thai patients to as much as 100% in British patients, while Japanese have an intermediate prevalence (about 60%). This may reflect the prevalence of liver flukes. Indeed, the mutational profiles of CCs associated with O. viverini infection have been shown to differ from those of tumors not associated with these infections. Moreover, hotspot KRAS mutations have been shown to be associated with intrahepatic CC initiation in mouse models and with a proliferation subgroup of intrahepatic patients with poor prognosis. TP53 mutations do not have any particular hotspot and coexist with KRAS mutations in a small proportion of tumors. Mutations in both IDH1 and 2 have been frequently reported in CC, with some studies suggesting that they are more common in intrahepatic than in extrahepatic tumors. Two identified hotspots: R132 IDH1 and R172 IDH2, promote cholangiocarcinogenesis through repression of hepatocyte differentiation. IDH and KRAS mutations as well IDH and TP53 mutations seem to be mutually exclusive. Other most frequently altered genes include ARID1A, BAP1, KDM6B, and SETD2, interestingly all involved in chromatin remodeling (Table 1). A recent comprehensive analysis of nearly 500 samples has shown that grouping intrahepatic CC patients by classifier mutations in the three most recurrently mutated genes (TP53-mutant, KRAS-mutant, IDH1-mutant, wild-type in the three genes (called “undetermined”)) reveals distinct molecular and pharmacogenomic profiles. In TP53-driven tumors, G to A transversions were the most prevalent, including the ApT > NpG associated with aristolochic acid. R249S was enriched but the aflatoxin signature was not enriched like in hepatocellular carcinoma and, compared to other mutants, R249S was not associated with the HBV status or overall survival. C to T transitions were most prevalent in KRAS-mutant tumors, with a preference for an upstream purine (G or C), suggesting a role of the APOBEC signature. IDH1-mutant tumors were significantly enriched for C to A transversions. These mutations are related to oxidative stress and wild-type IDH1 has been shown to protect cells from oxidative stress-induced damage. TP53driven tumors were significantly enriched for PTEN, RB1, and LATS2 mutations, whereas BLAF-1 alterations were prevalent in IDH1-mutant tumors, SMAD4 in the KRAS group, and KDM6B in undetermined tumors. Out of previously identified genes, ARID1 alterations were present in all the groups except for TP53-driven tumors but they were significantly associated only with undetermined tumors. BAP1 mutations were significantly associated with IDH1-mutant tumors. Interestingly, no chromatin Table 1

Gene loci that are most frequently altered in cholangiocarcinoma

Pathway

Gene (locus)

Alteration type

Clinical significance and other comments

P13K/PTEN/Akt (control of cell metabolism) p53 signaling (cell-cycle control)

KRAS (12p12.1) PTEN (10q23) TP53 (17p13.1)

Point mutations (hotspots) Point mutations and deletions Point mutations (scattered)

Tyrosine kinase receptors

RB1 (13q14.2) FGFR2 (10q26.13)

Mutations Fusions

Citrate cycle

IDH1 (2q34) and IDH2 (15q26.1)

Point mutations (hotspots)

Chromatin remodeling

ARID1A (1p36.11)

Point mutations, deletions, and insertions Point mutations, deletions, and insertions Point mutations and deletions Point mutations Point mutations Point mutations and insertions

In up to 12% of intrahepatic CCs Prevalent in TP53-driven intrahepatic tumors In up to 20% of intrahepatic CCs; TP53-driven molecular signature associated with an in silico sensitivity to some RNA synthesis and mTOR inhibitors Prevalent in TP53-driven intrahepatic tumors In 6%–15% of CC cases; associated with shorter overall survival In up to 12% of intrahepatic CCs; also reported in extrahepatic tumors; mutations associated with sensitivity to dasatinib and PARP inhibitors; IDH1driven molecular signature associated with an in silico sensitivity to microtubule modulators (e.g. taxanes)

BAP1 (3p21.1)

Hippo signaling TGF-ß signaling

KDM6B (17p13.1) SETD2 (3p21.31) LATS2 (13q12.11) SMAD4 (18q21.2)

Prevalent in IDH1-mutant intrahepatic tumors

Prevalent in TP53-driven intrahepatic tumors Prevalent in KRAS-driven intrahepatic tumors

370

Cholangiocarcinoma: Diagnosis and Treatment

remodeling genes were associated with the TP53 group. Both TP53- and KRAS-driven profiles were associated with a worse overall survival and shorter recurrence-free time. The four molecular signatures were associated with an in silico sensitivity to different groups of anticancer drugs (Table 1). Up to two thirds of all biliary tract carcinomas overexpress ERBB2, while membranous expression of E-cadherin, alpha-catenin, and beta-catenin is downregulated in a majority of intrahepatic CCs and this decreased expression has been shown to correlate with a high grade of intrahepatic CC. Moreover, bile acids have been demonstrated to further activate EGFR and increase the expression of oxidative agents. MET overexpression has also been shown in intrahepatic CCs, correlating with tumor grade (low expression in poorly differentiated tumors) and increased proliferation indices. Furthermore, bcl-2, an antiapoptotic protein, has been reported to be overexpressed in intrahepatic CC, while telomerase activity is detected in up to 100% of cases. Chronic inflammation which is thought to contribute to tumor development is associated with the release of inflammatory cytokines that induce oxidative stress and DNA damage. These cytokines may also create an immunesuppressive environment that promotes tumor cell survival and proliferation by blocking apoptosis which would normally be induced by DNA damage.

Biomarkers Early detection of CC is extremely difficult and no reliable screening biomarkers have been identified. Most CC patients present with advanced disease and have a median survival of less than one year despite treatment with current standard chemotherapy. Even patients who undergo apparently curative resection have poor outcomes due to a high rate of tumor recurrence. The most reliable (statistically significant) prognostic markers include several clinicopathological criteria. Cachexia, poor performance status, serum bilirubin of 9 mg/dL or greater, multifocal disease, hilar or proximal sites, high tumor grade, sclerotic histology, liver invasion, lymph node involvement, and advanced stage have been shown to be significantly associated with a poor outcome. Liver cirrhosis has been associated with an increased risk of recurrence and a reduced survival following resection in intrahepatic CC patients. The limited ability to reliably acquire tissue samples has resulted in an ongoing quest for CC blood serum biomarkers. In addition to liver-specific testing (see “Presentation and diagnosis”), serum CA 19-9 and CEA levels are used as a part of the diagnostic workup, with increased levels suggesting a gastrointestinal malignancy. However, these two markers are not specific to CC and their sensitivity is also variable. They might be more useful in surveillance for posttreatment recurrence but identifying more reliable and more specific biomarkers remains an urgent need. Serum IgG4 levels may be used to rule out autoimmune cholangitis during the diagnostic workup of perihilar CC. However, defining clear cut-offs is an on-going challenge. Recently, a new test measuring the IgG4/IgG RNA ratio has been developed. It can, reportedly, distinguish IgG24-associated cholangitis from primary cholangitis and CC. However, its value still needs to be clinically evaluated. Cytokeratin- and mucin-based biomarkers (in particular CYFRA 21-1 and MUC-5) have also been reported to be of potential diagnostic value but they need to be evaluated in larger cohorts.

Management and Therapy Complete tumor resection with negative margins is the only curative procedure for CC. However, definitive surgery can only be applied to selected patients with well-localized lesions in whom negative microscopic margins can be obtained (“R0 resection”) while maintaining adequate liver remnant, vascular inflow/outflow, and biliary drainage. Correct determination of the tumor resectability is a major challenge of the preoperative workup, and requires multidisciplinary and multimodality evaluation. Imaging plays a decisive role in evaluating tumor resectability. However, it can also lead to confusion due to overlapping appearances with other hepatobiliary diseases, including benign lesions. CT and MRI are commonly used in various combinations with cholangiographic analyses in preoperative planning. When well performed, the accuracy of MRI and CT in predicting resectability exceeds 75%. [18F]-FDG PET-CT seems to have no additional value. Defining anatomic location and ductal extent of the tumor is an important step in preoperative planning. The Bismuth-Corlette classification is probably the most widely adopted staging system. It classifies perihilar CCs into four subtypes based on their anatomic location within the biliary tree. However, it is mainly informative for surgeons for planning resection type and not to determine resectability since other parameters, such as distant metastases and vascular involvement, are not included. For optimal determination of resectability, patients with potentially resectable CC may undergo staging laparoscopy. However, recent studies question the overall diagnostic yield and accuracy of staging laparoscopy, mainly due to improved imaging techniques. Since extended liver resections are often required, it is critical to assess the future remnant liver (FRL) preoperatively. CTvolumetric analysis is the standard technique used to this effect. However, as liver volume is not synonymous with liver function and function is not homogeneously distributed in the liver, FRL function assessment is additionally performed. 99mTc-mebrofenin hepatobiliary scintigraphy (HBS) is a validated quantitative dynamic liver function test, with mebrofenin, an iminodiacetic derivative, used as a tracer. The presence of severe underlying medical comorbidities, distant metastasis, involvement of major vascular structures not amenable to reconstruction, bilateral segmental ductal involvement, unilateral segmental ductal extension with contralateral

Cholangiocarcinoma: Diagnosis and Treatment

371

vascular inflow involvement, and calculated inadequate future liver remnant (FLR) are generally considered as contraindications to surgical resection. The reported resectability rates show much variability between centers, ranging from 20% to 90%. Many CC patients present with obstructive jaundice which has a negative effect on liver function, increases the risk of biliary infection and impairs cellular immunity, and may increase the in-hospital mortality by 10%. To improve patient outcomes, preoperative biliary drainage (PBD; biliary stenting) is recommended by many surgical teams. It reduces jaundice, improves liver function and the patient’s condition, at the same time improving the ability of the liver to regenerate postoperatively. The impact of these effects is particularly high in patients with an insufficient FLR and preoperative drainage has been shown to improve outcomes especially in patients requiring extended resections. However, the drainage itself may induce severe complications and the optimal drainage method is still a subject of debate. Percutaneous transhepatic biliary drainage (PTBD) had been widely used and has an advantage of a direct access to the biliary duct. However, it has been shown to have no positive effect on postoperative morbidity and mortality and to increase potential risks. Currently, endoscopic biliary drainage (EBD) is the most commonly used in Western countries as having fewer complications and better outcomes, while nasobiliary drainage (ENBD) is favored in many Asian countries. All in all, each method comes with its own set of complications and the potential risks for the patient need to be carefully evaluated against potential benefits when taking a decision whether to perform preoperative drainage at all, with many specialists advocating that it is only recommended if the FLR is small. At the same time, more studies are needed to determine the most optimal drainage method. To avoid the postoperative liver dysfunction resulting from extended hepatic resection (insufficient FRL volume or function), portal vein embolization (PVE) may be used. The use of preoperative radiotherapy has also been proposed in order to prevent seeding metastases following biliary drainage. It has also been reported to have an effect on the tumor size and that it may relieve jaundice in patients without biliary stenting. However, its benefits remain controversial. R0 resection requires an aggressive surgical approach combining bile duct resection with extended liver resection frequently accompanied with vascular reconstructions, and regional lymphadenectomy. These extended resections are associated with higher morbidity and mortality rates than experienced in liver resections without bile duct resection, probably because of the sequelae of obstructive jaundice. Overall, up to 50% of CC patients suffer from severe postoperative complications. Liver failure is a major complication after extensive hepatectomy and is a major cause of mortality in patients with perihilar CC. Other complications include, among others, biliary leakage, infections, bleeding complications, and cholangitis. The 30-day operative mortality rate may be as high as 25%. Surgical biliary tract bypass may be performed in patients whose tumors are found to be unresectable at operation. Liver transplantation in CC patients remains controversial. Theoretically, orthotopic liver transplantation (OLT) offers the advantage of resecting all of the structures that may be affected by the tumor, for example the portal vein, bilateral hepatic ducts and atrophic liver lobes. However, it is often not considered appropriate because of the high incidence of local recurrence and no standardized criteria to select patients for transplantation. Generally, transplantation is sometimes considered for selected patients with localized disease. Chemoradiation prior to transplantation has been shown to improve survival, according to some opinions enough to make the transplant meaningful. Systematic studies on the recurrence rates are scarce and rates from 50% to 100% have been reported. Patients with recurrent disease are not candidates for curative therapy and can only receive adjuvant therapy to improve long-term outcome. Adjuvant therapy has been advocated to reduce the high incidence of local recurrence, but it does not appear to improve survival after curative resection. The role of adjuvant radiotherapy remains unclear. Cholangiocarcinoma is radiosensitive but bile duct tolerance to radiation is limited, with biliary and duodenal stenosis being frequently induced complications. Chemotherapy, as either first or second-line is of potential benefit. Different regimens in different clinical settings have been and are still being tested in clinical trials. The most recent (February 2018) US National Comprehensive Cancer Network (NCCN) guidelines recommend using fluoripyrimidine- or gemcitabine-based regimens in post-resection CC patients (both intrahepatic and extrahepatic). One of these two regimens, or the gemcitabine/cisplatin combination may be used in unresectable extrahepatic CC patients. The latter is also the treatment of choice for palliative treatment in case of metastatic extrahepatic disease.

See also: Hepatocellular Carcinoma: Pathology and Genetics.

Further Reading Bridgewater, J., et al., 2014. Guidelines for the diagnosis and management of intrahepatic cholangiocarcinoma. Journal of Hepatology 60 (6), 1268–1289. Chmielowski, B., Territo, M., 2017. Manual of clinical oncology, 8th edn. Wolters Kluwer. Ganeshan, et al., 2012. Extrahepatic biliary cancer: New staging classification. World Journal of Radiology 4 (8), 345–352. Lewis, H.L., et al., 2017. Current management of perihilar cholangiocarcinoma and future perspectives. Chirurgia (Bucur). 112 (3), 193–207. International Agency for Research on Cancer, 2010. In: Bosman, F.T., Carneiro, F., Hruban, R.H., Theise, N.D. (Eds.), WHO classification of tumors of the digestive system, 4th edn. Lyon, France. National Comprehensive Cancer Network (NCCN). (2018) NCCN clinical practice guidelines in oncology. Hepatobilliary Cancers. Version 1.2018. Nepal, C., et al., 2017. Genomic perturbations reveal distinct regulatory networks in intrahepatic cholangiocarcinoma. Hepatology. https://doi.org/10.1002/hep.29764. Rassam, F., et al., 2018. Modern work-up and extended resection in perihilar cholangiocarcinoma: The AMC experience. Langenbeck’s Archives of Surgery 403 (3), 289–307. Zhang, W., Yan, L.-N., 2014. Perihilar cholangiocarcinoma: Current therapy. World Journal of Gastrointestinal Pathophysiology 5 (3), 344–354.

Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate Daniel Jeffery, Katrina Podsypanina, Tejas Yadav, and Genevie`ve Almouzni, Curie Institute, PSL Research University, CNRS, Equipe Labellisée Ligue contre le Cancer, Paris, France; and Sorbonne University, UPMC Univ Paris 06, CNRS, Paris, France © 2019 Elsevier Inc. All rights reserved.

Glossary Centromere Chromosomal region of constriction joining two sister chromatids to which the microtubules of the spindle attach during cell division, via the kinetochore. Coined in 1936 by Cyril Darlington. From Greek centro- for “center” and -meros for “part.” Chromatin Macromolecular complex of DNA, protein and RNA that makes up the chromosomes of eukaryotic cells. Coined in 1879 by Walther Flemming to describe protoplasm in cell nuclei. From Latinized form of Greek khromat- for “color”, and the chemical suffix -in. Epigenetics Coined by Conrad Waddington in 1942 in reference to Aristotle’s theory of epigenesis, proposing that embryos develop progressively from an undifferentiated egg cell. From Greek epi-, meaning “over, outside of, around,” and -genetics, the study of inherited traits. Epigenetic features refer to the “structural adaptation of chromosomal regions to register, signal or perpetuate altered activity states” Bird (2007). Epigenome The ensemble of modifications that govern the regulation of the genome. Histone chaperone Histone chaperones are defined mainly as factors that associate with histones and stimulate a reaction involving histone transfer without being part of the final product. In 1978 the term “molecular chaperone” by Ron Laskey was applied to nucleoplasmin, a protein enabling the storage of histones H3–H4 in Xenopus oocytes. Later, it has been extended to all factors that prevent incorrect interactions between histones and DNA and facilitate nucleosome assembly. Histone Five types of basic proteins found in chromatin, namely H1, H2A, H2B, H3, and H4 and their variants. Coined in 1884 by Albrecht Kossel. From Greek histos-, meaning “warp, web”, and -one, meaning “weaker”. Kinetochore complex of proteins associated with the centromere to which the microtubules of the spindle attach during cell division. Coined in 1934 by L.W. Sharp from the Greek kinisis- for “movement” and -choros for “place, region”. Nucleosome Fundamental repeating subunit of chromatin, made up of DNA wrapped around a single hetero-complex of histones. Coined in 1975 by Pierre Oudet to reflect their nuclear origin and in reference to the “nu” bodies described by Olins and Olins (1974). Oncogene A gene with the potential to transform a cell into a tumor cell. Coined in 1969 by Robert Huebner and George Todaro. From Latinized form of Greek onco- meaning “tumor, mass” and -gene for “to produce”.

Nomenclature 2-HG 2-hydroxyglutarate 5caC 5-carboxylcytosine 5fC 5-formylcytosine 5hmC 5-hydroxymethylcytosine 5hmU 5-hydroxymethyluracil 5mC 5-methylcytosine 8oxoG 8-oxoguanine ADAADi Active DNA-dependent ATPase A domain inhibitor ADS Alternative DNA structures ALT Alternative lengthening of telomeres AML Acute myeloid leukemia ARCH Age-related clonal hematopoiesis BET Bromodomain and extraterminal domain–containing CCAN Constitutive-centromeric associated network DIPG Diffuse intrinsic pontine glioma DNMT DNA-methyltransferase DSB DNA double-strand break DSC DNA synthesis coupled DSI DNA synthesis independent

372

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65276-5

Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate

373

HAT Histone acetyl-transferase HDAC Histone deacetylase HDM Histone demethylase HMT Histone methyl-transferase HR Homologous recombination ICB Immune checkpoint blockade IDH Isocitrate dehydrogenase iPSC Induced pluripotent stem cell KMT Lysine methyltransferase MEF Mouse embryonic fibroblast NMC NUT-midline carcinoma panNET Pancreatic neuroendocrine tumors PRC2 Polycomb-repressive complex-2 PTM Posttranslational modification SAM S-adenosyl methionine TCGA The Cancer Genome Atlas TKI Tyrosine kinase inhibitor TSS Transcription start site a-KG Alpha-ketoglutarate

Introduction The Hierarchy of Chromatin Organization: A Versatile Modular Structure Within the nucleus, eukaryotic genomes are organized as nucleoprotein complexes comprising DNA, RNA, and proteins. This complex, called chromatin, follows a hierarchical organization that ranges from the basic unit, the nucleosome, up to higherlevel domains in the nuclear space in interphase and culminates with chromosome compaction in mitosis. Its repeated module, the nucleosome, comprises
147 bp of double-stranded DNA wrapped around an octamer of histone proteins plus intervening linker DNA. It thus forms periodic arrays giving rise to the nucleofilament or “beads-on-a-string” characteristic observed in early electron microscopy experiments (Fig. 1A). Transcription factors, or various chromatin-associated proteins and RNA punctuate this organization and help in defining specific domains. The nucleofilament undergoes further coiling to give rise to higherorder chromatin structures (Fig. 1B). Notably, DNA can be modified, distinct histone variants can form specific core particles and histones can be modified. These features of chromatin organization in space and time thus define, for a given cell type, an epigenome that corresponds to a cell fate decision. Here, we will refer to epigenetic features as to the “structural adaptation of chromosomal regions to register, signal or perpetuate altered activity states” as defined by A. Bird in 2007. Notably, distinct nucleosome features such as histone composition and modifications along with specific positioning characterize regulatory regions of the genome, such as promoters, enhancers, coding regions, replication origins, telomeres, and centromeres. This versatility of chromatin in space and time allows switching and reversibility, a general characteristic of the modular protein–DNA interactions that constitute chromatin. Thus, chromatin functions as a modular molecular scaffold with epigenetic features that can change and adapt according to cellular states during development and lifetime, and in response to various environmental stimuli.

Epigenetic Regulators: Shaping Chromatin in the Nucleus Specific chromatin factors, acting at various organizational levels, exert spatial and temporal control over gene expression and are referred to as epigenetic regulators (Fig. 1C). DNA modifications: Covalent modifications of DNA bases, such as 5-methylcytosine (5mC) and 5-hydroxymethylcytosine (5hmC), play central roles in epigenetic regulation of gene expression. DNA-methyltransferases (DNMTs) are enzymes that catalyze the covalent addition of a methyl group to a cytosine base of DNA. DNA methylation is commonly associated with promoter silencing and transcriptional repression. On the other hand, 10–11 translocation (TET) enzymes, TET1, 2, and 3, sequentially convert 5mC to 5hmC, then 5-formylcytosine (5fC), and 5-carboxylcytosine (5caC). Several additional DNA adducts, such as 8oxoguanine (8oxoG) and 5-hydroxymethyluracil (5hmU), important for epigenetic regulation should also be considered and the discovery of novel modifications is growing. Histone choice and chaperones: As key components of the nucleosome, the core histone proteins (H3, H4, H2A, and H2B), along with the linker histone H1, serve as the building blocks of the epigenome. There are multiple forms of the four core histones, called histone variants, originally characterized based on migration properties using Triton Acid Urea gel electrophoresis by S. Franklin and A. Zweidler in 1977. The differences in amino acid sequence for the core histones from their variants range from single residue

374

Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate

(A)

(B) Nuclear position

Nucleus

Higher-order chromatin

Structural RNA

Nucleosome

Histone tails

DNA

Histone variants

(C)

ADP + Pi

ATP

V.

P

IIIa.

IIIb.

Me

II.

IV. I.

Ac

I. DNA modifications: DNMTs and TET enzymes II. Histone chaperones: CAF-1, HIRA, DAXX, HJURP, ASF1, NASP, MCM2, FACT, NAP1 and histone variant choice: H3.1/2, H3.3, CENP-A, H2A III. Histone modifiers: a. Writers: HATs, HMTs, Kinases, Ubiquitin Ligases b. Erasers: HDACs, HDMs, Phosphatases, Deubiquitinases IV. Histone readers: Proteins containing following domains: Bromodomain (read Lysine acetylation) Chromodomain (read lysine methylation), PHD finger (read lysine methylation and acetylation) WD-40 repeats (read lysine methylation) V. Remodeling factors: SWI/SNF, ISWI, CHD, INO80 families

Fig. 1 Chromatin structure and organization: key regulators. (A) Electron micrographs. Left panel: low ionic-strength chromatin spread, the “beads on a string.” Scale bar: 30 nm; Right panel: isolated mononucleosomes derived from nuclease-digested chromatin. Scale bar: 10 nm. (B) The hierarchy of chromatin organizationdnumerous epigenetic factors help organize chromatin in the 4D nucleus: along with the linker histone H1, DNA wrapped around one (H3–H4)2 tetramer capped by two H2A–H2B dimers forms the nucleosomedthe fundamental repeating unit of chromatin. DNA and histone tails can be modified. The presence of histone variants adds further complexity. Arrays of nucleosomes fold into higher-order chromatin structures, potentially guided by noncoding RNA. The nuclear localization of a given chromosomal domain represents an additional level of regulatory information. (C) Key chromatin regulators act at each hierarchical level. (A) From Olins, D. E., Olins, A. L. (2003). Chromatin history: Our view from the bridge. Nature Reviews Molecular Cell Biology 4, 809–814dwith permission; (B and C) Modified from Probst, A. V., Dunleavy, E., Almouzni, G. (2009). Epigenetic inheritance during the cell cycle. Nature Reviews Molecular Cell Biology 10, 192–206.

alterations to greater than 50% divergence. Within the nucleus, the spatiotemporal dynamics of these various histone bricks are controlled by a specialized group of proteins called histone chaperones (Fig. 2). Thus, histones are never found alone as histone chaperones escort them throughout their cellular life. Histone modificationsdwriters, erasers, and readers: While DNA is wrapped around the core histones, the N-terminal tails of histones remain exposed and undergo a variety of posttranslational modifications (PTMs). Individual and combined PTMs constitute what has been called a “histone code” with direct effects on the binding affinity of chromatin-interacting factors as well as DNA accessibility to various cell machineries. To place modifications, a series of enzymes have been identified as “writers,” such as histone acetyl- and methyl-transferases (HATs and HMTs), including EZH2, the catalytic subunit of PRC2 (Polycomb repressive complex 2). To remove modifications, a series of enzymes have been identified as “erasers,” such as histone deacetylases (HDACs) and demethylases (HDMs) (Fig. 1C). “Readers” that recognize specific chromatin PTMs comprise bromodomain and extraterminal domain–containing proteins (BET), chromodomain-, PHD finger- and WD40-containing complexes. They further recruit or stabilize factors at precise genomic loci to regulate nuclear functions. While a number of modifications are imposed on the nucleosome particle, several of them are also placed on histones before deposition. The dynamics of histone deposition and modifications should also be considered.

Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate

375

Fig. 2 Histone chaperones dedicated to distinct histone variants help their incorporation into chromatin at particular genomics locations and precise times during the cell cycle. Modified from Sitbon, D., Podsypanina, K., Yadav, T., Almouzni, G. (2017). Shaping chromatin in the Nucleus: The bricks and the architects. Cold Spring Harbor symposia on quantitative biology.

Chromatin remodelers and higher order chromatin: Chromatin remodelers are large, multisubunit, ATP-dependent molecular motors that regulate nucleosome structure, positioning and organization (Fig. 1C). These helicases share a common ATPase domain and can be classified into four subgroups based on their divergent domains: SWI/SNF, INO80, ISWI, and Mi2-CHDNuRD complexes. Factors contributing to higher order structure include transcription factors, insulators, heterochromatinassociated proteins such as HP1, and noncoding RNA, while several others remain to be identified. While we describe the various factors individually, they are often found in combination, for example, a remodeler as a subunit of a larger complex. Also, note that each of the modifiers or remodeler complexes often contain a histone chaperone as a subunit or are associated with one of them.

Connections Between Epigenetic Regulators and Cancer Biology Importantly, the relationships between epigenetic regulators, nuclear architecture and cancer biology have been stressed in a series of reviews. Indeed, several chromatin factors, such as histone modifiers and chromatin remodeling enzymes are disrupted or altered in cancers, including histone variants and chaperones. With the Cancer Genome Atlas (TCGA) reporting complete genomic and transcriptional data from several thousands of cancer patients and enabling the profiling of somatic tumor mutations, a series of potential chromatin-associated drivers of cancer have been identified. Nearly a quarter of the top 25 mutated genes, ranked by the frequency of alteration across all tumor types, encode chromatin-modifying enzymesdKMT2C/MLL3, KMT2D/MLL2 (Lysine methyltransferases 2C, 2D), ARID1A (AT-Rich Interaction Domain 1A), PBRM1 (Polybromo 1), SETD2 (SET Domain Containing 2), CREBBP (CREB Binding Protein), and SMARCA4/BRG1 (SWI/SNF-related Matrix-Associated Regulator of Chromatin A4). Indeed, one or more of the 12–15 subunits of SWI/SNF chromatin remodeling complexes are mutated in nearly 20% of all solid malignancies and, hence, represent attractive therapeutic targets. Importantly, individual SWI/SNF mutations vary in frequency in different cancer types, based on TCGA data. In addition to cancer development and maintenance, epigenetic factors have also been shown to correlate with patient response and clinical outcomes to current treatments. For example, specific chromatin regulators, including SWI/SNF chromatin remodelers, can serve as biomarkers for drug response (e.g., SMARCB4 for docetaxel response or worse prognosis) and the histone chaperone CAF-1 holds prognostic value in renal, endometrial, and cervical carcinomas. Genetic evidence supports a causal role for histone H3 mutations in aggressive brain cancers in children. Specifically, point mutations of H3 lysine 27 to methionine/isoleucine (H3K27M/I) or glycine 34 to arginine or valine (H3G34R/V) lead to critical changes in the histone methylation status and are found in 35%–80% of pediatric gliomas, depending on the subtype. Recently, recurrent mutations of histone H3 lysine 36 to methionine/isoleucine (H3K36M/I) or glycine 34 to tryptophan/leucine (H3G34W/L) have also been reported in specific bone cancers. Several instances of H1 loss-of-function mutations have been found in diffuse large B-cell/follicular. Although these particular histone mutations are linked to a subset of cancers, changes in histone levels are commonly observed in nearly all types of tumor. So far, epidrugs have been used mainly in combinations with other cytotoxic treatments with the goal of sensitizing cancer cells as an end point (Table 1). Dynamic changes to the chromatin landscape over time can gradually or acutely influence specific

376

Epidrugs in ongoing clinical trials for cancer treatment (December 2017)

Target a HMT inhibitors DOT-1L EED (PRC2 component) EZH2

Epidrug b EPZ-5676 MAK683 CPI-1205 Tazemetostat

HDM inhibitors LSD1

GSK2879552

LSD1 (off-target)

IMG-7289 INCB059872 Tranylcypromine

DNMT inhibitors pan DNMT

5-Azacytidine

National clinical trial (NCT) identifier c

Current status d

NCT02141828 NCT02900651 NCT02395601 NCT02889523 NCT02601937 NCT03213665, NCT03155620 NCT02860286 NCT02601950

Phase 1 Phase 1 Phase 1 Phase 1/2 Phase 1 Phase 2

NCT02929498 NCT02177812 NCT02842827 NCT02712905 NCT02717884, NCT02273102 EMEA/H/C/000978 NDA 050794 NCT01386346 NCT00336063 NCT02940483, NCT03206021 NCT01155583 NCT02959437 NCT02788201 NCT02374099 NCT02260440 NCT02828358 NCT02900560 NCT03264404, NCT01845805 NCT02178072 NCT01281124

Delivery e

Cancer type f

Molecular selection (biomarker testing) for patient stratification g

Monotherapy Monotherapy Monotherapy Combination Monotherapy Both

Pediatric leukemia DLBCL/solid tumors DLBCL DLBCL Pediatric sarcoma Mixed

MLL-rearranged No No No SMARCB1-negative or SSX-SS18 fusion EZH2, SMARCB1, or SMARCA4 mutation

Monotherapy

Mesothelioma Mixed

BAP1-deficient SMARCB1-negative/EZH2 GOF mutation

Phase 2 Phase 1 Phase 1 Phase 1/2 Phase 1/2

Both Monotherapy Combination Combination Combination

MDS AML AML/MDS AML/MDS/solid tumors AML/MDS

No No No No No

EMA approved (1st line) FDA approved (1st line) Phase 1 Phase 1 Phase 1

Monotherapy Monotherapy Combination Combination Monotherapy

MDS/AML/CML MDS Esophageal Nasal/nasopharyngeal Pediatric brain

No

Phase 1/2 Phase 1/2 Phase 2 Phase 2 Phase 2 Phase 2 Phase 2 Phase 2

Combination Combination ND Combination Combination Combination Combination Combination

MM Solid tumors Urinary/bladder Breast Colorectal Infant ALL Ovarian Pancreatic

Phase 2 Phase 2

Monotherapy Monotherapy

Head/neck Lung

Testing EBV promoter meth., histone acetyl. No

COXEN model No KMT2A rearrangement; testing LINE-1 meth. No HPV Testing DNA meth., mir29 expression

Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate

Table 1

pan DNMT

5-Azacytidine

NCT01522976 NCT02319135, NCT03151408 NCT02951156 NCT01566695 EMEA/H/C/002221 NDA 21-790 NCT02959164 NCT02951728 NCT03250962 NCT01876641 NCT03346642 NCT02788201 NCT02634827 NCT02957968 NCT02664181 NCT02159820 NCT01882660 NCT03026842 NCT01317953

Phase 2/3 Phase 3

Both Combination

MDS/CML AML

Phase 3 Phase 3 EMA approved FDA approved Phase 1 Phase 1/2 Phase 1/2 Phase 1/2 Phase 1/2 Phase 2 Phase 2 Phase 2 Phase 2 Phase 2/3 Pre-operative Phase 4 Phase 1

Combination Combination Monotherapy Monotherapy Combination Combination Combination Combination Combination ND Combination Combination Combination Combination Monotherapy Monotherapy Monotherapy

DLBL MDS AML MDS Pancreatic/sarcoma DLBCL Hodgkin Lymphoma Melanoma Large B-cell Lymphoma Urinary/Bladder AML Breast Lung Ovarian Colorectal AML Lung

NCT02446652

Phase 3

Combination

Cervix

No

Phase 1 Phase 1 Phase 1 Phase 2 Phase 1 Phase 1/2 Phase 1

Monotherapy Monotherapy Combination Combination Combination Monotherapy Monotherapy

Mixed Nut-midline Carcinoma Prostate Breast Mixed Mixed Mixed

No BRD4-NUT translocation No ERþ No No

CPI-0610 CPI-0610 CPI-0610 GS-5829 GS-5829 GSK2820151 RO6870810 ZEN003694 ABBV-075 ODM-207 SF1126

NCT03205176 NCT01587703 NCT03150056 NCT02964507 NCT02711137 NCT02419417 NCT03220347; NCT01949883 NCT02158858 NCT02157636 NCT02986919 NCT02983604 NCT02607228 NCT02630251 NCT03068351 NCT02711956 NCT02391480 NCT03035591 NCT03059147

Phase 1 Phase 1 Phase 2 Phase 1 Phase 1 Phase 1 Phase 1 Phase 1 Phase 1 Phase 1/2 Phase 1

Monotherapy Monotherapy Monotherapy Combination Combination Monotherapy Both Combination Combination Monotherapy Monotherapy

AML/MDS MM MPNST Breast CRPC Solid tumors MM CRPC Mixed Solid tumors Liver

No No

Quisinostat

NCT02948075

Phase 2

Combination

Ovarian

No

Decitabine

BET inhibitors BRD2-4

BRD2-4/T

BRD4 PI3K and BRD4 HDAC inhibitors HDAC1

AZD5153 I-BET762 I-BET762 I-BET762 INCB054329 BMS-986158 CC-90010

No

COXEN model FLT3 mutation HER2No

Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate

Epigallocatechin-3gallate Hydralazine

Testing cytogenetic abnormalities No

t8;21 No

ERþ HER2- breast cancer No

377

(Continued)

Epidrug b

HDAC1/2

Romidepsin

HDAC3/4

Tasquinimod

HDAC6

ACY-241 Resminostat Ricolinostat

pan HDAC

Abexinostat AR-42 Belinostat

Entinostat

Givinostat Mocetinostat Panobinostat

National clinical trial (NCT) identifier c

Current status d

Delivery e

Cancer type f

Molecular selection (biomarker testing) for patient stratification g

022393/S-004 NCT02512172 NCT02393794 NCT02788201 NCT01796002 NCT01743469

FDA approved (2nd line) Phase 1 Phase 1/2 Phase 2 Phase 3 Phase 2

Monotherapy Both Combination ND Combination Monotherapy

NCT02635061 NCT02400788 NCT02632071 NCT02091063 NCT01543763 NCT02282917 NDA 206256 NCT02137759 NCT01686165 NCT02737046 NCT02780804 NCT02915523

Phase 1 Phase 1/2 Phase 1 Phase 1/2 Phase 1 Phase 0 FDA approved (2nd line) Phase 2 Phase 2 Phase 2 Phase 1 Phase 1/2

Combination Combination Combination Monotherapy Combination Monotherapy Both Combination Combination Combination Monotherapy Combination

NCT01038778, NCT03024437 NCT02708680 NCT01305499 NCT03018249 NCT01928576 NCT03179930 NCT02115282 NCT01761968 NCT02282358 NCT02805660 NCT02954991 NDA 205353 NDA 205353 NCT02032810

Phase 1/2

Combination

CTCL/PTCL Colorectal Breast Urinary/bladder PTCL Lung/ovarian/kidney/ gastric Lung Liver Breast Lymphoma Mixed Meningioma/schwannoma PTCL Brain DLBL T-cell leukemia/lymphoma Pediatric brain Ovarian/peritoneal/ fallopian Renal

Phase 1/2 Phase 2 Phase 2 Phase 2 Phase 2 Phase 3 Phase 2 Phase 1/2 Phase 1/2 Phase 2 FDA & EMA approved (3rd line) FDA & EMA approved (3rd line) Phase 1

Combination Combination Combination Combination Combination Combination Monotherapy Monotherapy Combination Combination Combination Combination Combination

Breast AML Endometrial Lung Lymphoma Breast Bone marrow/blood Lymphoma Solid tumors Lung MM MM Skin

No Microsatellite stable ER-/PR-/HER2COXEN model No No No No No No No No MRSI-based No No Testing lymphocyte levels No Testing PR, Ki67, and p21 levels No ERþ/PRþ/HER2-; testing lysine acetyl No Acetyltransferase mutations (CREBBP or EP300) No No

Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate

Target a

Epidrugs in ongoing clinical trials for cancer treatment (December 2017)dcont'd

378

Table 1

pan HDAC

Panobinostat

Pracinostat SAHA (Vorinostat)

Vorinostat HAT inhibitors p300/CREBBP

Curcumin

Monotherapy Monotherapy Both Combination Monotherapy Combination

Pediatric brain MDS/AML DLBL MM Lymphoma MDS

Phase 3 FDA approved (2nd line) Phase 1 Phase 1 Phase 1 Phase 1 Phase 1/2 Phase 1/2 Phase 1/2 Phase 1/2

Combination Monotherapy Combination Combination Combination Combination Combination Combination Combination Combination

AML CTCL Kidney/urinary/bladder Liver Nasal/nasopharyngeal Pancreatic Adult brain Head/neck Lung Sarcomas

Phase 1/2 Phase 2 Phase 2 Phase 2 Phase 2 Phase 2 Phase 2

Combination ND Combination Combination Combination Monotherapy Monotherapy

Skin Urinary/bladder Breast CNS Colorectal Lymphoma Mouth

NCT01802333 NCT01554852 NCT02759601

Phase 3 Phase 3 Phase 1/2

Combination Combination Monotherapy

AML MM Liver

NCT02520115 NCT01898104 NCT01622439 NCT00879437, NCT01817751 NCT02124174, NCT01342692 NCT02068586 NCT02068590 NCT01045538 NCT01064921

Phase 1 Phase 1/2 Phase 1/2 Phase 2

Combination Combination Combination Combination

Ovarian Colorectal DLBL Adult/pediatric brain

Phase 2

Combination

AML/MDS

Phase 2 Phase 3 Phase 1/2 Phase 1

Monotherapy Combination Combination Combination

Melanoma Head/neck Gastric Head/neck

NCT03072992 NCT03192059

Phase 2 Phase 2

Combination Combination

Breast Cervical/endometrial/ uterine

Testing H3K27 mutations No Testing biomarkers No Testing SNPs, cytokines, lymphocytes NO No Testing EBV promoter meth., histone acetyl. No

BRAF V600 mutant COXEN model No

Testing HR23B expression, aneuploidy, polyploidy, whole exome, CGH array No No Folate receptor induction No

No No

379

Phase 1 Phase 1/2 Phase 2 Phase 2 Phase 2 Phase 2

Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate

Tefinostat (CHR2845) Valproic acid

NCT02717455 NCT01451268 NCT01238692 NCT02506959 NCT01261247 NCT03151304, NCT01873703 NCT03151408 NDA 021991 NCT02619253 NCT01075113 NCT00336063 NCT02349867 NCT00555399 NCT02538510 NCT02638090 NCT01294670, NCT01879085 NCT02836548 NCT02788201 NCT02395627 NCT02035137 NCT02316340 NCT00875056 NCT01175980

(Continued)

Epidrugs in ongoing clinical trials for cancer treatment (December 2017)dcont'd

p300/CREBBP

IDH inhibitors IDH1

IDH1/2 IDH2

Epidrug b Curcumin

National clinical trial (NCT) identifier c NCT02439385, NCT02439385 NCT02100423 NCT02064673

Ivosidenib (AG-120) NCT02074839, NCT03245424 NCT03343197 NCT02989857 NCT03173248 BAY1436032 NCT02746081 NCT03127735 IDH305 NCT02381886 Vorasidenib (AG-881) NCT02492737 NCT03343197 Enasidenib (AG-221) NDA 209606 NCT02577406

Current status d

Delivery e

Cancer type f

Molecular selection (biomarker testing) for patient stratification g

Phase 2

Combination

Colorectal

No

Phase 2 Phase 3

Combination Monotherapy

Leukemia/lymphoma Prostate

Phase 1

Monotherapy

AML

IDH1 mutation

Phase 1 Phase 3 Phase 3 Phase 1 Phase 1 Phase 1 Phase 1 Phase 1 FDA approved (2nd/3rd line) Phase 3

Monotherapy Monotherapy Combination Monotherapy Monotherapy Monotherapy Monotherapy Monotherapy Monotherapy Monotherapy

Brain Cholangiocarcinoma AML/MDS Solid tumor AML Mixed AML/MDS Brain AML Leukemia

IDH1 or IDH2 mutation IDH1 mutation IDH1-R132X mutation IDH1-R132 mutation IDH1 or IDH2 mutation IDH2 mutation

Abbreviations: acetyl, acetylation; ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; CGH, comparative genomic hybridization; CML, chronic myeloid leukemia; CNS, central nervous system; COXEN, co-expression extrapolation; CRPC, castration resistant prostate cancer; CTCL, cutaneous T-cell lymphoma; DLBL, diffuse large B-cell lymphoma; DNMT, DNA methyl-transferase; EMA, European Medicines Agency; ER, estrogen receptor; FDA, Food & Drug Administration (USA); GOF, gain of function; HAT, histone acetyl-transferase; HDAC, histone deacetylase; HDM, histone demethylase; HMT, histone methyl-transferase; MDS, myelodysplastic syndromes; meth, methylation; mixed, solid and hematologic malignancies; MM, multiple melanoma; MPNST, malignant peripheral nerve sheath tumors; MRSI, magnetic resonance spectroscopic imaging; PR, progesterone receptor; PTCL, peripheral T-cell lymphoma. a Target: the epigenetic regulator (molecule(s) or complex(es)) directly affected by the drug. b Epidrugs: all anticancer drugs in the list were selected from the human epigenetic drug database (hedds.org: December 2017) and others with reported epigenetic effects in current literature. c National Clinical Trial (NCT identifier): unique reference from an online database (www.ClinicalTrials.gov) providing details about each trial (start date, study design, current status, etc.). d Current status: the table reports the most advanced clinical trial per cancer type. e Delivery: the administration of the drug as a single agent or in combination with other drugs. f Cancer type: refers to the cancers where the epidrug is tested against a specific tissue/organ tumor, excluding trials for mixed cancer types (except if they use molecular stratification of patients or are the only representative trial for that epidrug). g Molecular selection (biomarker testing) for patient stratification: details about mutations, molecular markers or other physiological features used to select target patients.

Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate

Target a

380

Table 1

Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate

381

genome functions. Importantly, chromatin regulators can also affect the tumor microenvironment and immune system. Understanding the combination of effects and the dynamics of disease progression is of paramount importance. By interfering with specific epigenetic factors with known effects in space and time, in either the cancer cells or cells of the microenvironment, we may be able to alter precise chromosomal territories at defined moments in the cell cycle to adapt the treatment strategy to the type of tumor. This capacity to selectively manipulate genome regulation may provide us with the ability to combat some of the fundamental difficulties of cancer treatment.

Epigenetic Links to the Roadblocks of Cancer Treatment In the past four decades, advances in the treatment of malignant disease have resulted in growing improvement in overall clinical outcomes. Nevertheless, fundamental problems persist which result in drug resistance, disease recurrence and, ultimately, patient mortality. To name the most critical ones, the roadblocks to patient cure include: clonogenic potential of cancer cells, tumor evolution, tumor microenvironment, inefficient immune surveillance, tumor heterogeneity, and drug-related side effects (Fig. 3A–F). Since multiple chromatin components have been shown to underpin one or several aspects of these pervasive complications, epigenetic targeting represents an important avenue to explore. Clonogenic potential (Fig. 3A) represents the ability of a single cancer cell to reconstitute the entire tumor as if it were a stem cell. Theoretically, even a single remaining malignant cell following treatment can lead to disease recurrence if it harbors this clonogenic capacity. In fact, the principles of combining chemo- and radiotherapy conceived to combat minimal residual disease are still applied today. However, the “stemness” of residual cancer cells is reminiscent of the pluripotency displayed by normal stem cells. The differences between stem and differentiated cells, supported by changes to chromatin architecture during stem cell differentiation, has to be taken into account for rational drug design. In particular, it is intriguing that the plasticity of transitions between

(B)

Tumor evolution

(A)

(C)

Tumor micro-environment

Clonogenic potential Chromatin remodeling

Histone modification

Higher-order chromatin Histone variant choice

(F)

DNA modifications

Histone chaperones

(D)

Inefficient immune surveillance

Drug-related side-effects

Tumor heterogeneity (E)

Fig. 3 Epigenetic Links to the Roadblocks of Cancer treatment. Starting from the scheme depicting the hallmarks of cancer, we place the building blocks of chromatin, the nucleosome, and its versatile forms as central players underpinning different roadblocks of cancer treatment (A–F). See section Epigenetic links to the Roadblocks of Cancer Treatment in text for details. Original artwork based on Hanahan, D. and Weinberg, R.A. (2000). The hallmarks of cancer. Cell 100, 57–70.

382

Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate

differentiated and undifferentiated states shows direct ties to the histone chaperone CAF-1, a chromatin factor found overexpressed in series of aggressive cancers. Indeed, CAF-1 knockdown combined with the modulation of transcription factors improves reprogramming of mouse fibroblasts to induced pluripotent stem cells (iPSCs) by several orders of magnitude and in embryonic stem cells it can enhance totipotency. Therefore, modulation of the expression of CAF-1 may have opposing effects on tumor growth and clonogenicity. Progress in understanding parameters involved in control of stemness at the chromatin level will help shed further light on the problem of residual disease in cancer. Tumor evolution (Fig. 3B) occurs as the disease progresses from less aggressive to more malignant stages. Cancer cells divide in the context of genetic and epigenetic instability. As a result, a tumor is equipped with a vast repertoire of divergent subclones of varying fitness. Continuous tumor growth creates bottlenecks for survival, resulting in selection of individual subclones. As selective pressure mounts during metastasis or administration of treatment, (epi)genetic plasticity provides additional adaptability to the evolving tumor. This form of drug escape may be overcome by modulating the (epi)genome. For example, cancer cells resistant to tyrosine kinase inhibitors (TKI), such as gefitinib, show decreased levels of H3K4me2/3, correlated with poor prognosis. In turn, TKI-resistant tumors become susceptible to KDM5A (H3K4 demethylase) silencing or inhibition of KDM5A-associated HDACs, while the parental tumor cell lines remained unaffected. These results suggest that epigenetic factors such as H3K4 demethylase KDM5A likely underlie nongenetic tumor heterogeneity leading to resistance. This example underlines the capacity to exploit chromatin plasticity for overcoming drug resistance linked to (epi)genetic instability and malignant evolution. The tumor microenvironment (Fig. 3C), a complex network of cells and extra-cellular factors surrounding a developing tumor, functions as the “ecosystem” for an expanding population of cancer cells. These nonmalignant tissues can be co-opted by the tumor to sustain its growth and can also physically block the access of anticancer drugs or the immune system to the tumor. In some instances, marked changes in expression of chromatin factors, such as histone PTMs and DNMTs, have been reported in cancerassociated fibroblasts in contrast to their normal counterparts, despite their nonmalignant genetic status. In breast cancers, these epigenetic changes have been shown to promote tumor invasiveness and malignancy. Increased understanding of epigenetic changes in cancer-associated stroma provides clues for how to overcome this roadblock to cancer treatment. However, the efficacy of currently available epigenetic therapies has not been specifically evaluated in cancer stroma. Describing the metabolic and epigenetic pathways connecting immune, stromal, and cancer cells could provide necessary clues to combat drug resistance linked to microenvironment. Inefficient immune surveillance (Fig. 3D) is a failure of the immune system to recognize malignant tissue as abnormal or pathological. In the absence of immune recognition, the patient may not present with overt clinical symptoms and there is a paucity of tumor-specific antigens that can generate a proficient antitumor response. Recent experiments studying immune checkpoint blockade (ICB) indicated that tumor-specific effector T-cells are often kept inactive by de novo methylation programs that hinder clonal expansion and diversity of effector T-cells. In this case, an epigenetic strategy to reactivate the T-cell antitumor response could involve reversal of these DNA methylation changes. Indeed, examples of exposure to DNA-demethylating agents, such as decitabine, prior to ICB showed restoration of antitumor CD8 T-cell function, highlighting a role for epigenetic reprogramming in immune surveillance. Other methods could also participate in controlling T cell fate and T cell differentiation. This connection between the immune system and epigenetic regulators offers a number of possibilities that are just beginning to be explored. Tumor heterogeneity (Fig. 3E) originates from the combined effects of: (1) genetic instability and selection of divergent mutational clones and (2) hierarchical development of a cancer cell population. Epigenetic pathways influence both of these causative factors. For example, aberrant DNA methylation is associated with an increased frequency of cytosine to thymidine (C > T) transitions, initiated by the spontaneous deamination of methylated cytosines, a mutational signature found in all cancer types and most cancer samples. Also, tumors develop as populations of mixed subclones at various stages of lineage commitment. As already mentioned above, factors like the histone chaperone CAF-1 and linker histone H1 are known to regulate lineage specification and cell fate determination. Indeed, cancer-specific chromatin patterns are reminiscent of those in embryonic stem cells and some adult stem cells. This example highlights the ability to manipulate chromatin factors, involved in cell fate decisions, to influence tumor heterogeneity. Drug-related side effects (Fig. 3F) are commonly associated with cancer therapies, to the extent where patients may even perceive the treatment as being worse than the disease itself. Due to the fact that cancer originates from the host’s own cells, classical anticancer drugs target properties that are shared by both tumor and normal cells. Therefore, the therapeutic window can be very narrow and epigenetic manipulation can provide interesting means to help direct a more selective targeting and limit drugrelated side effects. For example, the use of the DNMT inhibitor, decitabine, which led to the preferential de-repression of gene promoters silenced only in cancer cells, has been reported as a disease-selective means to avoid affecting noncancer cells. Thus, targeting epigenetic pathways in the future could improve upon the selectivity and specificity of currently available anticancer drugs.

Chromatin Regulators in Cancer: Epi(genetic) Drug Targets DNA- and Histone-Modifying Enzymes Drugs inhibiting the enzymatic activities of DNA-modifying enzymes and PTM writers, readers and erasers that can provide cytotoxic effects with a strong potential in cancer therapy are listed (Table 1). Alterations in global DNA methylation levels are linked to aberrant changes in gene expression of oncogenes as well as tumor-suppressor genes. DNMT inhibitors, azacitidine and decitabine,

Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate

383

are cytosine-analogs that cannot be methylated resulting in the sequestration of DNMTs and subsequent global DNA hypomethylation. At high doses, DNMT inhibitors impair DNA synthesis or translation of proteins causing cell death. Gain-of-function mutations in genes encoding isocitrate dehydrogenases (IDHs) are commonly associated with AML, breast and brain cancers. IDH mutations affect both DNA and histone methylation. IDH1 and IDH2 catalyze the production of alphaketoglutarate (a-KG), a key cofactor in TET-mediated conversion of 5mC to 5hmC and the HMT function of the Jumonji-C domain-containing family of enzymes. Mutations altering R132 (IDH1), R140 or R172 (IDH2) enable them to convert a-KG to 2-hydroxyglutarate (2-HG), which competitively interferes with a-KG-dependent epigenetic reactions such as DNA modification and histone-methylation. Thus, therapies targeting these specific chromatin-associated mutations are currently being tested in several clinical trials (Table 1). At the histone level, the presence of acetyl group modifications on the histone tails has generally been correlated with gene activation. Different classes of HDACs catalyze the removal of acetyl groups from a variety of histone and nonhistone substrates. HDAC inhibitors, such as vorinostat, interfere with the active site of the enzyme resulting in hyperacetylated substrates and associated changes in gene expression and posttranslational regulation of nonhistone proteins. However, DNMT and HDAC inhibitors as mono-therapies have demonstrated limited clinical efficacy, especially against solid tumors. It remains unclear if this poor success was due to a lack of informed patient selection based on predictive biomarkers. This highlighted the need to choose patients by careful molecular stratification in future trials. Careful selection of patient populations for drug testing showed promise in early clinical trials evaluating a new generation of epigenetic drugs, such as targeted inhibitors of BRD4 and its homologs (BET family proteins) or EZH2. BRD4 is a bromodomaincontaining reader of acetylated-histones and a transcriptional coactivator, also implicated in DNA replication and repair. BRD4 fusion with NUTM1 encodes an oncogenic protein that defines highly aggressive NUT-midline carcinoma (NMC). This fusion leads to oncogenesis by aberrant recruitment of HATs and transcriptional activation of oncogenes, including MYC. As well, several lines of evidence demonstrate tumor cell addiction to functional BRD4 activity in acute myeloid leukemia (AML), which harbor translocations or mutations for the gene encoding the MLL1 histone lysine methyltransferase. Another epigenetic target includes the histone methyltransferase EZH2, catalytic subunit of the polycomb-repressive complex 2 (PRC2), which promotes transcriptional silencing and facultative heterochromatin formation. At the molecular level, EZH2 methylates lysine 27 of histone H3 (H3K27) using Sadenosyl methionine (SAM) as a cofactor. EZH2 gain-of-function mutations are associated with non-Hodgkins Lymphoma. At the same time, EZH2 inactivating mutations have also shown oncogenic properties. Thus, understanding the mechanisms of PRC2-mediated gene regulation may help treat patients with EZH2 mutant cancers. BET and EZH2 inhibitors are now in Phase II trials in molecularly selected patients, based on criteria described in Table 1. Current small-molecule BET inhibitors can be broadly classified as benzodiazepine-derivatives, namely JQ1 and its analog OTX015/MK-8625 (Table 2), and quinoline class drugs such as I-BET151 and I-BET762. BET inhibitors mimic acetyl-lysines and serve as competitive inhibitors of BET protein binding to chromatin. EZH2 inhibitors, such as tazemetostat, work as SAMcompetitive small molecules that block HMT activity of the PRC2 complex. Moreover, PRC2 shows epigenetic antagonism with specific chromatin remodeling complexes. Hence, these interrelationships suggest that chromatin remodelers may also serve as potent drug targets.

Chromatin Remodeling Enzymes ATP-dependent chromatin remodeling complexes such as SWI/SNF facilitate the exchange or displacement of nucleosomes. Although at least one alteration in SWI/SNF complexes is directly linked to oncogenesis, few therapeutic agents have been characterized against individual subunits. Currently, ADAADi (Active DNA-dependent ATPase A Domain inhibitor) has shown an ability to interfere with SMARCA4/BRG1, a core catalytic ATPase of the SWI/SNF chromatin remodeling complex. Previously, depletion of SMARCA4 was found to enhance sensitivity toward platinum-based chemotherapy and docetaxel. Despite their overlapping

Table 2

Epigenetic drugs outside of current clinical trials for cancer treatment

Status

Class of epidrugs

Epidrug

Ongoing and upcoming (noncancer)

DNMTi BETi HDACi DNMTi BETi HDACi HMTi

Procainamide RVX-208 Resveratrol, sodium butyrate Zebularine OTX015/MK-8625 CHR-3996, pivanex 3-Deazaneplanocin A, BIX-01294, CPI360, EI1, EPZ004777, GSK126, GSK343, and UNC0638 Bizine, clorgyline, GSK-J4, JIB-04, KDM5-C70, and pargyline CPI203, I-BET151, and JQ1 Trichostatin A, CG-1521, HC-toxin, ITF-A, and ITF-B C646

Previous clinical trials for cancer treatment Preclinical only

HDMi BETi HDACi HATi

384

Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate

molecular functions, various SWI/SNF complexes display heterogeneity in their subunit composition. Synthetic lethal relationships have been reported between independent catalytic subunits such as SMARCA2 and SMARCA4. The loss of either gene is dispensable for viability while their combined loss results in cell death. Interestingly, isolated inactivation of SMARCA2 or SMARCA4 increases cisplatin sensitivity. Similarly, the structural subunit ARID1B is essential for survival of tumor cells lacking its mutually exclusive partner ARID1A. Thus, synthetic interdependencies of intracomplex subunits within chromatin remodeler assemblies provide druggable targets for anticancer therapy. Importantly, SWI/SNF complexes display interdependent relationships with several pathways involving PRC2, PARP (important for single-strand DNA break repair), and PI3K/AKT (cell cycle regulators). This genetic crosstalk provides additional vulnerabilities in SWI/SNF-mutant tumors. The SWI/SNF chromatin remodeler and PRC2 repressive complexes show opposite effects on target gene expression, such that tumors defective for SWI/SNF subunits display increased PRC2 activation. This results in a PRC2 dependency that is only present when SWI/SNF is compromised. Inhibiting the EZH2 subunit of the PRC2 complex with tazemetostat selectively increased efficacy in patients negative for SMARCA4 or the core SMARCB1/INI1 subunit, in Phase I clinical trials. Although there are currently no ARID1B inhibitors, in preclinical studies ARID1A-deficient cancer cells have shown increased sensitivity to EZH2 inhibitors as well as PARP and PI3K/AKT inhibition. In addition, ARID1A and p53 interact with each other and act as codependent tumor suppressors in certain cancers, providing another example of SWI/SNF interplay with cell cycle regulation. The increased efficacy brought about by the molecular selection of patients for treatment with epigenetic drugs suggests that educated selection of patients may be essential in the future. Indeed, one reason that the first generation of epigenetic therapies showed limited clinical effects is likely the absence of rational prescreening of patients. However, it is important to consider that while this stratification may need to be tumor-type specific, the characterization of additional biomarkers for these interdependent factors could have widespread applicability. Careful molecular selection would be especially helpful for those situations where the epigenetic targets were deemed undruggable but have demonstrated an interdependent relationship with a different factor for which drugs are already available. This also highlights the opportunity represented in characterizing new vulnerabilities in chromatin regulators that have been shown to play a role in DNA replication and repair or cell cycle progression.

New Potential Targets in Anticancer Therapy: Histones and Chaperones As the foundation of chromatin structure, nucleosome composition and loading influences many aspects of cancer progression. A wealth of knowledge has identified histone variant-chaperone partnerships critical for DNA replication and repair, gene expression, cell proliferation, and differentiation (Fig. 2).

Histone Variants To date, eight histone H3 variants have been identified in humans and the identification of homologous variants across species reveals a significant degree of evolutionary conservation. The most widely studied H3 variants (Fig. 2) can be classified based on two modes of incorporation into chromatin: DNA-synthesis coupled (DSC) or DNA-synthesis independent (DSI) pathways. In humans, the highest expression of the replicative variants H3.1 and H3.2 occurs during S phase of the cell cycle, facilitating deposition via the DSC pathway during DNA replication or DNA break repair involving DNA synthesis. Meanwhile, the DSI pathway involves incorporation of the “replacement” variant H3.3, expressed throughout all cell cycle phases. Importantly, the H3.3 variant also plays a significant role in transcription. Expression of CenH3 (histone H3-like centromeric protein A), called CENP-A in humans, is also independent of DNA synthesis. It is incorporated at centromeric sites during late mitosis (M)/early G1 phase while its expression peaks during the G2/M phases. CENP-A is the most divergent H3 variant and has a unique function, acting as the foundation for the constitutive-centromeric associated network (CCAN), which nucleates the kinetochore, and epigenetically determines the sites at which spindle microtubules attach during cell division. Without CENP-A this scaffold for cell division is unable to form and cells fail to segregate their chromosomes appropriately, resulting in cell death. CENP-A acts as one of many examples where specific histone variants play a significant role in the epigenetic control of cellular functions. The H2A family also contains several different variants found in humans, including H2A.X, H2A.Z, macroH2A1, macroH2A2, H2A.F/Z, and H2A.Bbd. Briefly, H2A.Z is expressed throughout all cell cycle phases and preferentially incorporated at gene promoters and regulatory regions to promote transcription. Two non-allelic genes, H2AFZ and H2AFV, encode two distinct H2A.Z isoforms: H2A.Z.1 and H2A.Z.2, respectively. H2A.X, synthesized in both S and G1 phases, is a variant directly involved in the DNA double-strand break (DSB) repair pathway as well as several distinct functions involving sex chromosome inactivation, stem cell development, and maintenance of cellular senescence. MacroH2A1 is commonly associated with transcriptionally silent heterochromatin but at a subset of genes it is found near transcription start sites (TSSs) and CTCF-binding sites. MacroH2A1.1 and macroH2A1.2 are two alternatively spliced isoforms originating from the H2AFY gene. MacroH2A1.1 is expressed in differentiated cells, while macroH2A1.2 is found in embryonic stem cells and the early embryo.

Histone Chaperones Histone chaperones escort histones throughout their cellular life. Some of them show very little selectivity and represent “casual” histone partners. In contrast, others associate with and are “dedicated” to a specific variant. They function as a network to ensure the

Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate

385

proper supply of histones in tune with the cellular demand. Dedicated chaperones play a crucial role for the incorporation of distinct histone variants into chromatin at particular locations and times throughout the cell cycle. The histone chaperone CAF1, dedicated to the replicative H3 variants, deposits histones onto newly synthesized DNA through its association with PCNA, an accessory factor of DNA polymerase. The HIRA complex promotes incorporation of the H3.3 variant, independently of DNA synthesis, at gene bodies, promoters and regulatory elements. DAXX/ATRX ensures accumulation of H3.3 at peri-centromeric heterochromatin and telomeres. Finally, a dedicated chaperone HJURP ensures the timely incorporation of de novo CENP-A at centromeres in late mitosis/G1 in mammals. The chaperone Asf1 is considered to act as a hand-over chaperone, upstream of other chaperones involved more directly in histone deposition. Asf1 deals with both parental and newly synthesized histones in the DSC and DSI pathways. NASP is another histone chaperone, with lower specificity, that acts as the emergency reservoir for soluble histones H3. Interestingly, MCM2, a DNA replication licensing factor and a component of the DNA helicase at the replication fork, also executes a histone chaperone function by favoring parental histone recycling. FACT is a histone chaperone that acts as a multifunctional heterodimer facilitating transcription, DNA replication and repair. FACT helps deposit the H2A–H2B dimer and (H3–H4)2 tetramer onto DNA and also disrupts core histone–histone and histone–DNA interactions. A coordinated network of chaperones, along with other chromatin regulators, regulates chromatin dynamics during critical cellular processes such as cell differentiation and organism development and is disrupted in various diseases. Importantly, a number of histone chaperones and variants are overexpressed or mutated in different cancer types. Among histone chaperones: CAF-1 is overexpressed in a series of aggressive cancers. ASF1b, one of the two human paralogs of the Asf1 chaperone, is overexpressed in a variety of tumor types. Considering histone variants: the increase in levels of CENP-A or its partner histone chaperone HJURP proves to be a key feature of tumor progression and poor prognosis. Perturbation of these factors to limit tumor growth represents opportunities that remain to be explored.

Changing Chromatin Factor RelationshipsdA Case of Cellular Adaptability Experimental CENP-A overexpression is associated with altered genomic localization to the chromosome arms in addition to centromeric incorporation. Importantly, CENP-A and HJURP levels are higher in p53-deficient transformed cells compared to p53 wild-type tumors. Ectopic CENP-A deposition is carried out by DAXX instead of its cognate chaperone HJURP. Although DAXX acts as a casual chaperone for CENP-A in this scenario, it deposits it at locations normally occupied by its variant H3.3 (Fig. 2), namely sites of high histone turnover along the chromosome arms. As previously proposed, this inherent flexibility in chaperone-variant interactions could represent adaptability in the face of transformation-induced stress and cytotoxic chemoand radiotherapies.

Context of Dosage Perturbation Matters In p53-null mouse embryonic fibroblasts (MEFs), exogenous induction of HJURP or CENP-A alone does not act as a driver of transformation. Strikingly though, increased CENP-A and HJURP expression plays a critical role in the maintenance of the transformed state in p53-null cells. Experimental downregulation of HJURP in fully transformed p53-null cells results in cell death and loss of tumor viability in vivo while p53-proficient cells respond by growth arrest. This “epigenetic” addiction for HJURP in the context of p53-loss suggests that histone variants and their chaperones might act as crucial navigators of the neoplastic process instead of serving as classical drivers or simply passengers.

Targeting Histone Chaperones: The Architects of Chromatin Given the epigenetic addiction to HJURP in p53-deficient transformed cells, interfering with de novo CENP-A deposition in cycling cells could represent an ideal pharmacological strategy. Since p53-proficient cells tolerated HJURP downregulation, it suggests that there is a therapeutic window for inhibitors of HJURP or its interaction with CENP-A in p53-null tumoral cells compared to nontumoral quiescent cells (Fig. 4). Essentially, p53-HJURP codependence serves as an example of synthetic lethality due to epigenetic addiction. Several approaches to block HJURP can be envisaged to treat p53-deficient tumors. These include interfering with: (1) (2) (3) (4)

The N-terminal domain of HJURP that mediates protein–protein interactions with the CENP-A. Dimerization of HJURP, important for its activity, and potentially required to accommodate a CENP-A-H4 tetramer. Phosphorylation and DNA-binding of HJURP that is essential for timely CENP-A deposition. Complex formation with partners CENP-C, a key centromeric factor, or the MIS18 complex (consisting of MIS18BP1, MIS18a, and MIS18b), that are required for proper CENP-A incorporation. (5) PLK1 kinase-dependent phosphorylation and recruitment of MIS18. Beyond the example of HJURP, additional histone chaperones would also be worth exploring in future epigenetic therapies. For example, CAF-1, a reliable proliferative marker, acts as a molecular barrier to reprogramming of cell fate. The dual role of CAF-1 in maintaining stemness and promoting cell division may make it an attractive complementary target to existing cytotoxic therapies. The dynamics of CAF-1 posttranslational modifications could provide a means to target its proliferative role in tumors. For example,

Tumor progression

HJURP/CENP-A levels

WT

1st hit: loss of p53

2 nd

hit: oncogenic signal

Ablation of HJURP function

Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate

increasing HJURP dep endency

386

low CENP-A detection slow cell cycles = slow CENP-A depletion

p53 Genome integrity maintained

p53 critically low Centromere CENP-A goes dysfunction undetected Aneuploidy rapid cell cycles = rapid CENP-A depletion

Apoptosis

Fig. 4 The unbalance between histone chaperone and variants in p53-deficient tumors: an Achilles’ heel in cancer cells. HJURP sustains cellular transformation in cells lacking p53. HJURP and CENP-A levels are increased in cells that lose p53 expression and increase even further following oncogenic transformation. When HJURP is depleted in wild-type cells with functional p53, p53 senses gradual CENP-A depletion and induces cell cycle arrest in order to maintain genome integrity. In cells lacking p53, HJURP depletion results in rapid CENP-A loss at centromeres, leading to centromere dysfunction, aneuploidy, and p53-independant apoptosis. From Filipescu, D., Naughtin, M., Podsypanina, K., Lejour, V., Wilson, L., Gurard-Levin, Z. A., Orsi, G. A., Simeonova, I., Toufektchan, E., Attardi, L. D. et al. (2017). Essential role for centromeric factors following p53 loss and oncogenic transformation. Genes & Development 31, 463–480.

phosphorylation of the large subunit of CAF-1 (p150) by the replication kinase Cdc7-Dbf4 is important for stabilizing its monomeric state and promotes CAF-1 interaction with PCNA. By blocking this modification, CAF-1 function could be modulated during hyperproliferation. While hyperphosphorylation of the p60 subunit keeps CAF-1 inactive during mitosis, in interphase active CAF-1 comprises both hypophosphorylated and/or phosphorylated forms of p60. The phosphorylation of the p60 subunit is regulated by cyclin/Cdk kinase complexes and PP1 phosphatase. Importantly, the phosphorylated form of p60 is preferentially recruited to sites of UV-induced DNA damage. Therefore, therapeutic interference with the regulation of CAF-1 phosphorylation could hinder its nucleosome assembly activity during S-phase and at specific sites of DNA damage in combination with epidrug-treatments. The proliferation-dependent isoform of the Asf1 histone chaperone, ASF1b, is prognostic of breast cancer outcomes. Thus, interfering with the histone supply function of Asf1 could also be promising. Like aforementioned chaperones, phosphorylation of free Asf1 by Tousled-like kinases (TLK) has been suggested to play a role in S phase progression by regulating Asf1-mediated histone delivery at sites of chromatin assembly. Inhibition of these kinases could potentially interrupt Asf1 activity in targeted cancer types. Furthermore, depletion of Asf1 (both a and b) activates the telomerase-independent telomere maintenance pathway ALT (alternative lengthening of telomeres) in primary and cancer cells. This suggests restoration of Asf1 function could provide a means to suppress the ALT pathway used by immortal cells to maintain their telomeres. Loss-of-function mutations in the DAXX and ATRX genes are frequently reported in pancreatic neuroendocrine tumors (panNETs) and are also invariably linked to ALT. The ALT pathway is dependent on homologous recombination (HR). Thus, inhibitors against protein kinase ATR, a key regulator of HR, could prove to be effective in ALT-positive tumors carrying mutations in ATRX and/or ASF1. Considering another histone chaperone complex, FACT is overexpressed in nearly 20% of all breast cancers. FACT recognizes and binds to alternative DNA structures (ADS) leading to p53-activation. As such, curaxin, a FACT-complex anticancer drug that promotes ADS formation, could prove to be efficacious in treating p53-proficient cancers. Additional histone chaperones, such as MCM2, NASP, HIRA, etc., need to be examined for their roles in oncogenesis and tumor development and, thus, their suitability as potential epigenetic targets.

Targeting the Histone Variants: Core Components of Chromatin Elevated CENP-A levels and recurrent H3.3/H3.1 K27, K36, and G34 mutations are examples of histone variants associated with particular types of cancer. Mutations of H3K27 residues in different histone H3 variants prevent methylation by PRC2 and have been strongly linked with pediatric brain tumors. A recent study, using model and primary patient-derived diffuse intrinsic pontine glioma (DIPG) cell lines, reported PRC2 dependency in H3K27M tumors. Notably, small-molecule EZH2 inhibitors were able to stop these tumor cells from growing. Thus, promising therapeutic opportunities can be envisaged using polycomb complex inhibitors. In addition, a 2014 screen by K. Funato and colleagues demonstrated that inhibitors against Menin, a transcriptional regulator and a member of the trithorax family histone methyltransferase complex, slowed brain tumor growth in mice. These findings highlight Menin as a potential target in patients with gliomas harboring the H3.3K27M mutation. In another study the same year by Hasizume et al. showed that GSKJ4, an inhibitor of H3K27 histone demethylation, extends survival in mice grafted with human H3K27M mutant brain tumors. Thus, pharmacological inhibition of histone K27 demethylation may hold promise in treating K27M pediatric gliomas. Besides these histone mutants, variants of histone H2A should also be considered. For example, increased expression of H2A.Z isoforms confers poor prognosis in metastatic melanoma. Accordingly, H2A.Z.2 knockdown sensitized melanoma cells to treatment

Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate

387

with doxorubicin, JQ1 or PD325901 (inhibitor of MEK, a MAPK kinase). Expression of macroH2A1.1 decreases in several cancer types and the low macroH2A1.1 to macroH2A1.2 ratio negatively affects patient outcomes. For example, loss of macroH2A has been associated with progression of and osteosarcoma. This is likely due to the antiproliferative roles ascribed to macroH2A1.1 compared to the pro-proliferative role of macroH2A1.2. As well, previous studies have reported that macroH2A1.1 expression reduces PARP-1 protein levels and suppresses growth of lung and cervical cancer cells. This raises the possibility that PARP-1 inhibitors could have therapeutic value in cancers that display decreased macroH2A1.1 expression. In addition, phosphorylation of H2A.X is one of the first events after the occurrence of DSBs, which activates the DNA damage repair (DDR) response. Therefore, interfering with H2A.X could be highly complementary to existing cytotoxic therapies. Lastly, H2A.Z is overexpressed in hormoneresistant cancers. In prostate cancer cell lines, H2A.Z was shown to be downregulated upon overexpression of the histone deacetylase SIRT1 (Sirtuin-1). Altogether, different histone variants display critical associations with clinical responses and disease progression in specific cancer types. In the future, the connection between histone variants and resistance to anticancer drugs will have to be extensively pursued.

Conclusions and Perspectives With the advent of epigenetic targeting strategies and advanced knowledge in cancer biology, new therapeutic opportunities and challenges lie ahead. The merits of early pan-cancer interventions against HDACs and DNMTs were overshadowed by the absence of mechanism-based patient stratification criteria. In recent trials using molecularly chosen populations, second-generation drugs against carefully selected targets, EZH2 and BET family proteins, proved more effective. Therefore, a compelling argument can be made for selecting the right tumors in an effort to move away from the pan-cancer type treatments of the past. Understanding how targeting the epigenome could maximally reinforce our current therapies against the pervasive roadblocks of cancer still remains a challenge and there is a need to integrate not only the tumor but also its ecosystem in our therapeutic decisions. Treating carefully selected patient cohorts may also prove to be inadequate unless specific molecular biomarkers are used to choose these patients. For the development of next generation epigenetic anticancer therapies, there is a pressing need not only to expand our arsenal of epigenetic targets but also to consider novel epigenetic-oncogene interactions (both synergistic and antagonistic). Histone variantchaperone interplay represents one such druggable epigenetic relationship that needs to be considered. Indeed, histone chaperones could be promising anticancer targets for future pharmacological and clinical efforts. Not only the direct histone variant-chaperone interactions but also the protein–peptide, protein–nucleic acid, and enzymatic control that regulates these interactions in space and time could be targeted. Based on the encouraging example of HJURP-addiction in p53-null tumors, it could be of significant value to characterize additional histone chaperones for interdependencies with tumor suppressor genes and oncogenes. As appears to be the case for EZH2 where both up- and downregulation can have tumor-promoting effects, an important consideration for drug development will be the dosage of chromatin components. Moreover, histone variant choice underlies a key question: is it the means of histone incorporation by a histone chaperone (e.g., HJURP vs. DAXX in the case of CENP-A) and/or the choice of histone variant itself that matters for tumor growth and survival? A detailed examination of the fine-tuned plasticity in chaperone-variant interactions will be necessary to delineate contributions to beneficial adaptation during disease.

Prospective Vision It has become increasingly evident that the effective treatment of cancer will require the combination of different approaches that counter the different hallmarks of cancer and overcome the roadblocks of current anticancer agents. In parallel, efforts should be made to assess large-scale, tissue-specific genomic, and epigenomic profiling to identify exceptional responders or entire subsets of patients where monotherapy against epigenetic factors will prove to be efficacious (e.g., ARID1B inhibition in ARID1A-deficient tumors). In cases of precancerous conditions, such as Age-related Clonal Hematopoiesis (ARCH), modulating the epigenome could even be considered to prevent progression to malignant neoplasia. Conventional epigenetic drugs have aimed to block the catalytic activity of enzymatic chromatin modifiers and remodelers. However, noncatalytic domains may also play an important role. For example, compounds that compromise noncatalytic roles of PRC2 subunits, EZH2, SUZ12, and EED, may be required to obtain optimal antitumor benefits in SWI/SNF-defective cancers. Related to this, the “interfacial inhibitor concept” proposes that small-molecules can blockade interacting surfaces within large, macromolecular assemblies, as with topoisomerase enzymes, and unhinge the functional complex, effectively hindering its molecular activity. Similarly, histone chaperones are nonenzymatic factors that could be targeted by interfering with their regulation. Epigenetic drugs could limit disease progression by crippling the oncogenic machinery of tumors and, potentially, also prevent recurrence in tumors that exploit chromatin pathways to develop resistance to existing therapies. This idea stems from the assumption that the epigenetic modifications responsible for tumor adaptation would remain reversible and sensitive to external manipulation, contrary to genetic changes. For example, loss of DNA methylation before immune checkpoint blockade can boost T cell renewal and enhances disease clearance in mouse models. This approach offers the promise of powerful treatments for chronic infections and cancer. More generally, the therapeutic promise of epigenome modulation could play a role beyond cancer with potential to treat other pathologies linked to defects in DNA metabolism or immunological disorders.

388

Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate

Acknowledgments We would like to thank Dr. Sophie Postel-Vinay for contributions to the table summarizing epigenetic drugs in ongoing clinical trials and Dr. Iva Simeonova for critical reading of the manuscript.

References Bird, A., 2007. Perceptions of epigenetics. Nature 447, 396–398. Olins, A., Olins, D., 1974. Spheroid chromatin units (n bodies). Science 183, 330–333.

Further Reading Alexandrov, L.B., Jones, P.H., Wedge, D.C., Sale, J.E., Campbell, P.J., Nik-Zainal, S., Stratton, M.R., 2015. Clock-like mutational processes in human somatic cells. Nature Genetics 47, 1402–1407. Attwood, K., Fleyshman, D., Prendergast, L., Paszkiewicz, G., Omilian, A.R., Bshara, W., Gurova, K., 2017. Prognostic value of histone chaperone FACT subunits expression in breast cancer. Breast Cancer (Dove Medical Press) 9, 301–311. Barone, T.A., Burkhart, C.A., Safina, A., Haderski, G., Gurova, K.V., Purmal, A.A., Gudkov, A.V., Plunkett, R.J., 2017. Anticancer drug candidate CBL0137, which inhibits histone chaperone FACT, is efficacious in preclinical orthotopic models of temozolomide-responsive and -resistant glioblastoma. Neuro-Oncology 19, 186–196. Burgess, R.J., Zhang, Z., 2013. Histone chaperones in nucleosome assembly and human disease. Nature Structural & Molecular Biology 20, 14–22. Cheloufi, S., Elling, U., Hopfgartner, B., Jung, Y.L., Murn, J., Ninova, M., Hubmann, M., Badeaux, A.I., Euong Ang, C., Tenen, D., et al., 2015. The histone chaperone CAF-1 safeguards somatic cell identity. Nature 528, 218–224. Filipescu, D., Naughtin, M., Podsypanina, K., Lejour, V., Wilson, L., Gurard-Levin, Z.A., Orsi, G.A., Simeonova, I., Toufektchan, E., Attardi, L.D., et al., 2017. Essential role for centromeric factors following p53 loss and oncogenic transformation. Genes & Development 31, 463–480. Filippakopoulos, P., Qi, J., Picaud, S., Shen, Y., Smith, W.B., Fedorov, O., Morse, E.M., Keates, T., Hickman, T.T., Felletar, I., et al., 2010. Selective inhibition of BET bromodomains. Nature 468, 1067–1073. Flynn, R.L., Cox, K.E., Jeitany, M., Wakimoto, H., Bryll, A.R., Ganem, N.J., Bersani, F., Pineda, J.R., Suva, M.L., Benes, C.H., et al., 2015. Alternative lengthening of telomeres renders cancer cells hypersensitive to ATR inhibitors. Science 347, 273–277. Gentric, G., Mieulet, V., Mechta-Grigoriou, F., 2017. Heterogeneity in Cancer metabolism: New concepts in an old field. Antioxidants & Redox Signaling 26, 462–485. Ghoneim, H.E., Fan, Y., Moustaki, A., Abdelsamed, H.A., Dash, P., Dogra, P., Carter, R., Awad, W., Neale, G., Thomas, P.G., et al., 2017. De novo epigenetic programs inhibit PD-1 blockade-mediated T cell rejuvenation. Cell 170, 142–157.e119. Gurard-Levin, Z.A., Quivy, J.P., Almouzni, G., 2014. Histone chaperones: Assisting histone traffic and nucleosome dynamics. Annual Review of Biochemistry 83, 487–517. Hammond, C.M., Stromme, C.B., Huang, H., Patel, D.J., Groth, A., 2017. Histone chaperone networks shaping chromatin function. Nature Reviews. Molecular Cell Biology 18, 141–158. Ishiuchi, T., Enriquez-Gasca, R., Mizutani, E., Boskovic, A., Ziegler-Birling, C., Rodriguez-Terrones, D., Wakayama, T., Vaquerizas, J.M., Torres-Padilla, M.E., 2015. Early embryonic-like cells are induced by downregulating replication-dependent chromatin assembly. Nature Structural & Molecular Biology 22, 662–671. Kim, K.H., Roberts, C.W., 2016. Targeting EZH2 in cancer. Nature Medicine 22, 128–134. Kim, Y., Kim, H., Park, H., Park, D., Lee, H., Lee, Y.S., Choe, J., Kim, Y.M., Jeoung, D., 2014. miR-326-histone deacetylase-3 feedback loop regulates the invasion and tumorigenic and angiogenic response to anti-cancer drugs. The Journal of Biological Chemistry 289, 28019–28039. Lacoste, N., Woolfe, A., Tachiwana, H., Garea, A.V., Barth, T., Cantaloube, S., Kurumizaka, H., Imhof, A., Almouzni, G., 2014. Mislocalization of the centromeric histone variant CenH3/CENP-A in human cells depends on the chaperone DAXX. Molecular Cell 53, 631–644. Maze, I., Noh, K.M., Soshnev, A.A., Allis, C.D., 2014. Every amino acid matters: Essential contributions of histone variants to mammalian development and disease. Nature Reviews. Genetics 15, 259–271. Mohammad, F., Weissmann, S., Leblanc, B., Pandey, D.P., Hojfeldt, J.W., Comet, I., Zheng, C., Johansen, J.V., Rapin, N., Porse, B.T., et al., 2017. EZH2 is a potential therapeutic target for H3K27M-mutant pediatric gliomas. Nature Medicine 23, 483–492. Monteiro, F.L., Baptista, T., Amado, F., Vitorino, R., Jeronimo, C., Helguero, L.A., 2014. Expression and functionality of histone H2A variants in cancer. Oncotarget 5, 3428–3443. Morel, D., Almouzni, G., Soria, J.C., Postel-Vinay, S., 2017. Targeting chromatin defects in selected solid tumors based on oncogene addiction, synthetic lethality and epigenetic antagonism. Annals of Oncology 28, 254–269. Muller, S., Almouzni, G., 2017. Chromatin dynamics during the cell cycle at centromeres. Nature Reviews. Genetics 18, 192–208. Narlikar, G.J., Sundaramoorthy, R., Owen-Hughes, T., 2013. Mechanisms and functions of ATP-dependent chromatin-remodeling enzymes. Cell 154, 490–503. Safina, A., Cheney, P., Pal, M., Brodsky, L., Ivanov, A., Kirsanov, K., Lesovaya, E., Naberezhnov, D., Nesher, E., Koman, I., et al., 2017. FACT is a sensor of DNA torsional stress in eukaryotic cells. Nucleic Acids Research 45, 1925–1945. Sato, T., Issa, J.J., Kropf, P., 2017. DNA Hypomethylating drugs in Cancer therapy. Cold Spring Harbor Perspectives in Medicine 7, a026948. Siegel, R.L., Miller, K.D., Jemal, A., 2017. Cancer statistics, 2017. CA: A Cancer Journal for Clinicians 67, 7–30. Vardabasso, C., Gaspar-Maia, A., Hasson, D., Punzeler, S., Valle-Garcia, D., Straub, T., Keilhauer, E.C., Strub, T., Dong, J., Panda, T., et al., 2015. Histone variant H2A.Z.2 mediates proliferation and drug sensitivity of malignant melanoma. Molecular Cell 59, 75–88. Vardabasso, C., Hasson, D., Ratnakumar, K., Chung, C.Y., Duarte, L.F., Bernstein, E., 2014. Histone variants: Emerging players in cancer biology. Cellular and Molecular Life Sciences 71, 379–404. Wang, C.Y., Filippakopoulos, P., 2015. Beating the odds: BETs in disease. Trends in Biochemical Sciences 40, 468–479. Weinberg, D.N., Allis, C.D., Lu, C., 2017. Oncogenic mechanisms of histone H3 mutations. Cold Spring Harbor Perspectives in Medicine 7, a026443. Wijdeven, R.H., Pang, B., van der Zanden, S.Y., Qiao, X., Blomen, V., Hoogstraat, M., Lips, E.H., Janssen, L., Wessels, L., Brummelkamp, T.R., et al., 2015. Genome-wide identification and characterization of novel factors conferring resistance to topoisomerase II poisons in Cancer. Cancer Research 75, 4176–4187. Wilson, B.G., Wang, X., Shen, X., McKenna, E.S., Lemieux, M.E., Cho, Y.J., Koellhoffer, E.C., Pomeroy, S.L., Orkin, S.H., Roberts, C.W., 2010. Epigenetic antagonism between polycomb and SWI/SNF complexes during oncogenic transformation. Cancer Cell 18, 316–328. Wilting, R.H., Dannenberg, J.H., 2012. Epigenetic mechanisms in tumorigenesis, tumor cell heterogeneity and drug resistance. Drug Resistance Updates 15, 21–38. Wu, Q., Sharma, S., Cui, H., LeBlanc, S.E., Zhang, H., Muthuswami, R., Nickerson, J.A., Imbalzano, A.N., 2016. Targeting the chromatin remodeling enzyme BRG1 increases the efficacy of chemotherapy drugs in breast cancer cells. Oncotarget 7, 27158–27175. Xu, W., Yang, H., Liu, Y., Yang, Y., Wang, P., Kim, S.H., Ito, S., Yang, C., Wang, P., Xiao, M.T., et al., 2011. Oncometabolite 2-hydroxyglutarate is a competitive inhibitor of alphaketoglutarate-dependent dioxygenases. Cancer Cell 19, 17–30.

Chromosome Rearrangements and Translocations Stefan K Bohlander, Purvi M Kakadiya, and Alix Coysh, University of Auckland, Auckland, New Zealand © 2019 Elsevier Inc. All rights reserved.

Glossary Chromosome inversion A chromosome segment reversed end-to-end; developed when a single chromosome sustains at least two double strand breaks, followed by rearrangement and repair within itself. Chromosome inversions can either be paracentric (involving parts of a chromosome on one side of the centromere) or pericentric (involving parts of a chromosome either side of the centromere). Inversions can also either be unbalanced or balanced, dependent upon whether there is or is not a gain or loss of genomic content following chromosomal rearrangement. Chromosome translocation The exchange of chromosomal material when breakage, rearrangement and repair occur between chromosomes (usually involving distinct chromosomes, but sometimes involving different loci within homologous chromosomes). Theoretically, translocations can occur between any loci on any two chromosomes, but many random exchanges produce translocations that are mitotically lethal because they are dicentric (have two centromeres) and cause chromosome breakage, or mitotically unstable because they are acentric (have zero centromeres) and are often lost. A translocation is reciprocal if the exchange of material is bidirectional and reciprocal if it is unidirectional. Translocations can also either be unbalanced or balanced, dependent upon whether there is or is not a gain or loss of genomic material following chromosomal rearrangement. Chromothripsis The phenomenon where hundreds or thousands of clustered chromosome breaks occur simultaneously causing extremely complex structural rearrangements. Copy-number alteration Duplications or deletions of chromosome sections resulting in the gain or loss of genetic material, respectively. Copy-number alterations can develop due to errors during both DNA replication and DNA repair, and the specific mechanisms differ dependent on the length of the duplication/deletion. Copy-number alterations can be the result of unbalanced chromosome inversions or translocations. Cytogenetics A branch of genetics that studies the number, structure, and function of chromosomes; particularly focused on how chromosomes relate to cell behaviors during mitosis and meiosis. Structural rearrangements A chromosomal abnormality resulting in a change to the normal chromosome structure, including translocations, inversions, duplications, deletions, and chromothriptic events.

Abbreviations AEJ Alternative end joining ALL Acute lymphoblastic leukemia AML Acute myeloid leukemia BFB Breakage fusion bridge BIR Breakage induced replication C-NHEJ Classical nonhomologous end joining CGH Comparative genomic hybridization CML Chronic myeloid leukemia CNA Copy-number alteration DSB Double strand break FISH Fluorescent in situ hybridization HR Homologous recombination IR Ionizing radiation MRN MRE11/Rad50/NBN NHEJ Nonhomologous end joining PML Promyelocytic leukemia SCID Severe combined immunodeficiency syndrome SNP Single nucleotide polymorphism SSA Single strand annealing V(D)J Variable diversity joining WGS Whole-genome sequencing

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65105-X

389

390

Chromosome Rearrangements and Translocations

Gene Names ABL1 ABL Proto-oncogene 1 nonreceptor tyrosine kinase AID Activation induced deaminase ALK ALK receptor tyrosine kinase BCL2 BCL2 apoptosis regulator BCR BCR RhoGEF and GTPase activating protein BLM Bloom syndrome RECQ like helicase BRCA1 Breast cancer 1 CBFB Core-binding factor beta BRCA2 Breast cancer 2 CDK4 Cyclin dependent kinase 4 DNA-PKcs DNA-dependent protein kinase catalytic subunit EML4 Echinoderm microtubule associated protein like 4 ERCC1 ERCC excision repair 1 endonuclease noncatalytic subunit ETO Eight twenty one protein ETV6 ETS variant 6 FEN1 Flap structure-specific endonuclease 1 FOXO1 Forkhead box O1 FOXO3 Forkhead box O3 FOXO4 Forkhead box O4 HMGA2 High mobility group AT-hook 2 IGH Immunoglobulin H INPP5D Inositol polyphosphate-5-phosphatase D Ku70 70 kDa subunit of Ku Antigen Ku80 86 kDa subunit of Ku Antigen LigIII DNA Ligase 3 LigIV DNA Ligase 4 MDM2 MDM2 Proto-oncogene MLL Myeloid/lymphoid mixed-lineage leukemia MRE11 MRE11 homolog double strand break repair nuclease MYC MYC Proto-oncogene bHLH transcription factor NBN Nibrin (also known as NBS1) PARP1 Poly(ADP-ribose) polymerase 1 PAX3 Paired box 3 PLZF Promyelocytic leukemia zinc finger PML Promyelocytic leukemia RB1 Retinoblastoma protein RAD1 RAD1 checkpoint DNA exonuclease RAD50 RAD50 double strand break repair protein RAD51 RAD51 recombinase RAD52 RAD52 homolog DNA repair protein RAG Recombination activating gene RARA Retinoic acid receptor alpha RB1 RB transcriptional corepressor 1 RECQL4 RECQ-like helicase 4 RPA Replication protein A RUNX1 Runt-related transcription factor 1 (aka AML1: Acute myeloid leukemia 1 protein) TGFBR2 Tumor growth factor beta receptor 2 TP53 Tumor protein 53 WRN Werner syndrome RECQ like helicase XRCC1 X-ray repair cross complementing 1 XRCC2 X-ray repair cross complementing 2 XRCC3 X-ray repair cross complementing 3 XRCC4 X-ray repair cross complementing 4

Chromosome Rearrangements and Translocations

391

Units

bp Base pairs kbp Kilo base pairs Mbp Mega base pair

Introduction to Structural Rearrangements in Cancers Cancers and Genome Instability Cancers are complex diseases that arise as a result of genetic, epigenetic and environmental influences, or a combination thereof, which cause genetic or epigenetic aberrations in cells. When these alterations are pro-tumorigenic (i.e., they provide a survival advantage by fulfilling one of the 10 hallmarks of cancer, they are positively selected for in a manner analogous to Darwinian evolution at the cellular level. Therefore, pro-tumorigenic lesions accumulate over time driving oncogenesis and may eventually lead to cancer. It is possible for such genetic and epigenetic changes to be acquired in cells because there is a very low level of genome instability in all cells. This instability exists because there is a significant energetic cost to maintaining a stable genome, alongside the fact that it is optimal to have high genome stability for proper functioning and faithful reproduction, but to also have slight genome instability to allow evolutionary changes to occur. Furthermore, genome instability can be increased by the acquisition of mutations followed by selection and clonal expansion of the fittest cells, and in this way a pro-tumorigenic cycle is established.

Structural Rearrangements Many different DNA alterations can contribute to tumorigenesis, where one class is termed “structural rearrangements.” These are genomic abnormalities where the structure of a chromosome is altered and they include, translocations (rearrangements of chromosome segments usually between nonhomologous chromosomes), inversions (reversal of the direction of a chromosome segment within a single chromosome) and copy-number alterations (CNAs) (deletions or duplications of chromosome segments). In addition, although extremely rare there are also complex structural rearrangements that involve more than two DNA breakpoints and the exchange of chromosomal DNA between two or more chromosomes. The most severe complex structural rearrangement is chromothripsis, where hundreds or thousands of chromosomal rearrangements develop in a localized genomic region during a single event. Structural rearrangements can be respectively described as either unbalanced or balanced, dependent on whether the rearrangement results, or does not result, in the gain or loss of genomic content. For translocations, the descriptors nonreciprocal and reciprocal can be used, respectively referring to whether the translocation of chromosomal material is one-way or bi-directional between two chromosomes. Balanced rearrangements are more common than unbalanced and reciprocal rearrangements are more common than nonreciprocal. Furthermore, it is more common for structural rearrangements to develop within a single chromosome than between chromosomes and between loci with closer spatial proximity in the interphase nucleus.

Structural Rearrangements in Cancer Research and Clinics Structural rearrangements are mainly identified by cytogenetics, which is concerned with the number, structure, and function of chromosomes as well as how chromosomal variations may affect cell function in health and disease. Studying structural rearrangements has been highly important in cancer research, and in fact, it was the discovery of recurring translocations in chronic myeloid leukemia (CML) and acute myeloid leukemia (AML) patients that was the first evidence to conclusively demonstrate cancers as genetic disorders. Furthermore, the discovery that many nonrandom chromosomal aberrations are frequently associated with driving specific cancers, has shown the importance of structural rearrangements in tumorigenesis, as well as elucidating mechanisms through which they may be generated (i.e., errors during DNA repair or replication) or through which they may function (i.e., fusion proteins and deregulation of gene expression). Moreover, it has enabled structural rearrangements to be important diagnostic, prognostic and disease management indicators, as well as identifying novel targets for cancer therapies. This is particularly true for blood cancers, lymphomas, leukemias and myelomas, where the distinction between critical and irrelevant structural rearrangements is more clear, and consequently it has become routine in clinical practice to perform chromosomal analysis for blood cancer patients. In contrast, it has only become common practice to assess structural rearrangements in very few solid tumors, namely some sarcomas, prostate cancers and lung cancers where often only certain rearrangements are searched for. Hence, the field of cytogenetics and structural rearrangements is much more advanced for hematological malignancies than solid tumors, and thus this article will largely use structural rearrangements in blood cancers as examples, in particular frequently recurring balanced chromosome translocations which are most well researched. This article will focus on the origin of structural rearrangements, their functional consequences toward tumorigenesis, the methods that enable their detection, and the research and medical advancements that come from studying them.

392

Chromosome Rearrangements and Translocations

The Origin of Structural Rearrangements in Cancer Like other forms of genome aberrations, structural rearrangements develop due to errors during DNA replication and DNA repair. There is a multitude of different mechanisms that can result in structural rearrangements, however the most important mechanisms are those that involve the acquisition of DNA double strand breaks (DSBs). This is because faulty DSB repair is a common cause of all types of structural rearrangements, and in fact, the incorrect repair of two DSBs which happen at close spatial and temporal proximity in the interphase nucleus is the only means through which chromosome inversions and translocations can develop, as well as being the major cause of cancer associated CNAs.

Sources of DSBs For DNA DSBs to develop, the phosphate backbones of two complementary DNA strands need to break simultaneously. DSBs can originate from both endogenous and exogenous causes. Firstly, many DSBs occur spontaneously due to a variety of endogenous factors: for example, natural metabolite by-products in cells (e.g., reactive oxygen species), chromosome nondisjunction during cell division, excision of transposable elements, physical or mechanical stress on DNA (e.g., mitotic spindle stress on dicentric chromosomes), and inadvertent action of nuclear enzymes of lymphoid cells (e.g., RAG complex and AID). In addition, intentional cellular processes are an endogenous source of sitespecific DSBs, where examples of such processes include homologous recombination to produce gametes, V(D)J recombination in lymphocytes to assemble immunoglobulin antigen receptor genes and T cell receptor genes, and chromatin condensation and expansion via topoisomerases and helicases. Moreover, DSBs can also result from natural DNA impediments that cause DNA replication to pause or stop (e.g., unusual DNA and chromatin structures, transcription machinery, or collisions with DNA binding proteins. In contrast, major exogenous sources of DSBs include ionizing radiation (IR), cigarette smoke, and anticancer drugs (e.g., crosslinking agents like Cisplatin and Mitomycin C, alkylating agents like Methyl methanosulfonate and Temozolomide), topoisomerase inhibitors like Etoposide, replication inhibitors like Aphidicolin and hydroxyurea, and radiomimetic compounds like bleomycin and phleomycin). In addition to all of the above, which are direct sources of DNA strand breaks, it is important to note that DSBs can also result when other forms of DNA damage are not repaired properly, such as base mismatches, uracil bases, abasic sites, pyrimidine dimers, and chemically modified bases (e.g., alkyl groups and O-6-methylguanine) or when some DNA lesions are encountered during DNA replication, for example strand cross-links and damaged nucleotides.

Detection and Response to DSBs Once a DSB is generated, the lesion has to be detected rapidly before it is passed on to daughter cells, because otherwise it could have serious consequences including blood disorders and a range of cancers. Following detection, the cell needs to decide whether the damage can be repaired or whether it is too extensive, because this determines whether the appropriate response would be DNA damage repair, senescence (prevention of further cell cycling in affected cells) or apoptosis (induce programmed cell death in affected cells). It is not yet clear how this decision is made by cells, but it is known that TP53 (Tumor protein p53), RB (Retinoblastoma protein), and other tumor suppressor proteins play an integral role in this process. Furthermore, it is understood that this decision is made very rapidly, typically within a few minutes of DSB generation. DSB recognition and the subsequent cell signaling are initiated by the two broken ends of a double stranded DNA molecule, because unlike telomeric ends of DNA molecules, broken ends are not hidden by telomeric extensions and telomere associated proteins. As a result, broken ends at DSBs are recognized and attract several proteins that orchestrate the appropriate response, by recruiting additional proteins involved in repair, cell senescence or apoptosis (Fig. 1). Altogether, the signaling cascades from DSB recognition through to completion of the DSB response are very complex. This is because: DSB recognition pathways differ dependent on cell type and DSB origin, there is a great number of proteins that can be involved in the recognition, repair, senescence and apoptosis processes, and there are many different repair mechanisms that can resolve DSBs, where the particular proteins that respond to the broken ends determines which repair pathway will take place. Some commonalities amongst all repair mechanisms are that significant modifications occur to chromatin, histones and other proteins both at the site of the DSB as well as extending outwards (e.g., poly(ADP-ribosylation), ubiquitination, sumoylation, acetylation, and phosphorylation), and that the gap between the two broken ends is bridged. In some circumstances the DSB may be repaired faithfully without any errors, but in other cases the repair can result in aberrations from the original DNA molecule, with some of the most severe consequences being structural rearrangements.

DSB Repair Mechanisms and Generation of Structural Rearrangements The mechanisms that produce structural rearrangements from DSBs can be divided into two broad classes, homology-based and nonhomology-based, which respectively do and do not require extensive homology to enable DSB repair. For instance, homology-based mechanisms require homologous template sequences approximately 97% similar to one another, whereas nonhomology-based mechanisms might only require a few to zero base pairs of homology. Consequently, homology-based

Chromosome Rearrangements and Translocations

393

Fig. 1 Detection and resolution of DSBs through different repair pathways. The resolution of DSBs involves the detection of the DSB, protein signaling pathways, recruitment of proteins involved in DSB repair and filling in the gap between the two break points. This example shows these steps and depicts both error-prone (red) and relatively error-free (green) pathways. The MRN complex is composed of MRE11, Rad50, and NBN.

mechanisms are typically error-free whereas nonhomology-based mechanisms are more error-prone. Respectively, this means that the DSB repair usually produces a DNA sequence that is either identical or different to the original DNA sequence before the DSB occurred. Therefore, it would be more common for nonhomology-based mechanisms to join DSB ends incorrectly and generate structural rearrangements than it would for canonical homology-based mechanisms. Interestingly, in cancers proteins involved in DNA repair mechanisms are frequently mutated and become faulty. Subsequently increasing genomic instability and the acquisition of structural rearrangements. In fact, the mutation of DNA repair pathways is so essential that a two-step model has been proposed for the predisposition of structural rearrangements to tumorigenesis, whereby there must both be an error in DSB repair mechanisms and a failure to initiate cell cycle arrest in order for cancers to develop.

Homology-based mechanisms Homologous recombination (HR) HR is a DSB repair mechanism that operates during the S and G2 phases of the mitotic cell cycle in response to DNA damage. This is because HR requires a homologous sister chromatid (second intact copy of the affected chromosome) to use as a DNA repair template, and in mitosis these are more accessible during and shortly after nuclear DNA replication. HR also takes place during the meiotic cell cycle where it occurs in prophase 1 until metaphase 1 to generate crossing overs. Because HR is very cell cycle specific, the cyclin-dependent Kinases that mediate cell cycle progression are important regulators of HR. HR begins with the MRE11/RAD50/NBN (MRN) complex binding to the broken ends of the DSB. Here MRE11 excises nucleotides from the 50 ends leaving 30 single strand DNA overhangs at the DSB break points. Next, the 30 overhangs are coated with RPA, and then RPA is replaced by RAD51 in a reaction promoted by mediator proteins, such as XRCC2, XRCC3, RAD52, and BRCA2, amongst others. This produces the Rad51-single stranded DNA-nucleoprotein filament which then invades at the homologous site on the double stranded homologous sister chromatid. At this site a Holliday junction develops and the homologous sister chromatid is used as a template to perform DNA repair. Once the Holliday junction is resolved the DNA repair is complete. Although canonical HR is very efficient, defective HR is known to play a role in the generation of structural rearrangements in disease, such as Downs syndrome (trisomy 21), hereditary breast and ovarian cancers (BRCA1 and BRCA2 mutations), Bloom’s syndrome (BLM mutations), Werner’s syndrome (WRN mutations), and Rothmund–Thompson syndrome (RECQL4 mutations). Mutations in HR genes can contribute to cancers by causing homologous chromatids to misalign during DSB repair, such that the DSB and subsequent Holliday junctions are also misaligned. The subsequent cross over would therefore be unequal as it would occur at nonallelic regions on the homologous chromosomes. This can enable the transfer of genetic material between homologous chromosomes, potentially causing deletion, duplication or inversion of the chromosomal region which may favor cancer development. Secondly, mutations in HR genes can cause nondisjunction to occur where homologous chromatids do not segregate properly during anaphase, resulting in chromosome breakage or aneuploidy (abnormal number of chromosomes in a cell), both of which can contribute to cancer development. For example, mitotic nondisjunction is known to play a role in retinoblastoma when chromosome 13 with wildtype RB1 is lost through nondisjunction and the remaining chromosome 13 has a mutant RB1. Lastly, HR gene mutations can cause HR to be downregulated, consequently increasing the frequency of more error prone DSB repair pathways, which ultimately increases the acquisition of tumorigenic structural rearrangements and genomic instability. Interestingly, the genomic instability caused by mutations in HR genes actually makes cancer cells more susceptible to chemotherapeutic drugs, because these cells have an impaired ability to tolerate the induction of many DSBs at once. Single strand annealing (SSA) SSA is an alternate HR pathway which repairs DSBs between two DNA repeats that are less than 25 kbp (kilo base pairs) apart and oriented in the same direction. Unlike the classical HR pathway described above, SSA is unique in that it does not require a separate

394

Chromosome Rearrangements and Translocations

homologous chromosome or DNA sequence for the DSB repair. Instead, because SSA only occurs in repeat regions, it merely uses one of the unbroken repeats as a homologous template to resolve the DSB. Because the required template for SSA is always present, SSA does not preferentially occur in any particular part of the cell cycle. The first step in SSA is to perform single stranded DNA resection at the site of the DSB, generating complementary 30 overhanging free DNA ends either side of the breakpoint. The overhanging strands are coated by RPA to prevent them adhering to themselves, and then Rad52 assists the alignment of the 30 overhangs so that the complementary repeat sequences can anneal and form a single repeat. After annealing is complete, any remaining single stranded nonhomologous flaps (remnants from the 30 overhangs) are reselected by Rad1 and ERCC1. Lastly, DNA synthesis fills in any gaps and ligation produces a continuous double stranded DNA molecule and DSB repair is complete. Altogether, SSA typically deletes one repeat alongside any sequences between the two repeats, and such deletions cause structural rearrangements and can be tumorigenic. SSA always results in a CNA, and translocations or inversions develop when SSA occurs between two unlinked repeats, but this is highly uncommon due to the requirement for two DSBs to occur at the same time. Breakage induced replication (BIR) template switching BIR template switching occurs when the DNA repair machinery from another DSB repair mechanism unexpectedly switches to its homologous chromosome as a template to complete the DSB repair. Therefore, the repair proteins involved depend on the original DSB repair mechanism prior to template switching. BIR template switching is most common when errors occur during the repair of a single DSB, and BIR has not been shown to preferentially occur during any particular phase of the cell cycle. As in nonallelic HR, because the broken end of the DSB invades a nonallelic region of the homologous chromosome during BIR template switching, it subsequently results in the transfer of genetic material between the respective chromosomes to result in chromosome inversions, translocations and CNAs which can potentially be tumorigenic. Furthermore, in cancers BIR template switching and the subsequent structural rearrangements often increase in frequency due to defective or downregulated cohesins (i.e., because cohesins function to reduce interchromosomal strand invasion).

Nonhomologous-based mechanisms Nonhomologous end joining (NHEJ) Classical and alternate NHEJ pathways are the major repair pathways for DSBs in cells, and they are respectively referred to as Classical-NHEJ (C-NHEJ) and alternative end joining (AEJ). C-NHEJ is more efficient than AEJ and C-NHEJ utilizes all of the classical NHEJ factors whereas AEJ lacks one or more NHEJ components as well as utilizing nonclassical factors. These mechanisms also differ in when they predominantly occur during the cell cycle, although it is interesting that NHEJ mechanisms function at least at low levels throughout all cell cycle stages. C-NHEJ is most prevalent in the G1 phase when cells are growing but not yet ready to replicate their nuclear DNA because it does not require an attack template from a homologous sister chromatid. In contrast, AEJ pathways are backup survival mechanisms that occur when C-NHEJ is defective or less active, and AEJ is most frequent during S phase of the cell cycle. The first step in C-NHEJ is DSB recognition and signaling by the MRN complex, followed by binding of Ku70 and Ku80 to the broken ends to prevent the broken ends drifting apart. Next, the coiled coil domains of Rad50 (component of the MRN complex) extend outwards to connect the broken ends to one another, the MRN complex performs some DNA resection at the broken ends and finally the broken ends are joined together by Lig IV and XRCC4. Other important proteins in C-NHEJ include, DNA-PKcs (functions as a molecular sensor for DNA damage and cooperates with Ku70/Ku80), Cernunnos (may serve as a bridge between XRCC4 and other factors), Artemis nuclease (crucial for processing hairpin structures that form during V(D)J recombination), Metnase (methylates histones to open chromatin to process DNA ends), and BRCA1 (interacts with many DNA repair proteins and is known to be important for certain types of NHEJ, but the specific role of BRCA1 is currently unclear). In contrast to C-NHEJ, AEJ has a requirement for small regions of microhomology between 5 and 25 bp (Base Pairs), it is more heavily dependent on the MRN complex, it lacks several C-NHEJ factors (e.g., Ku70, Ku80, XRCC4 or LigIV) and it involves several other factors (e.g., PARP1, LigIII, FEN1, and XRCC1, amongst others). The first step of AEJ is the recognition and signaling of a DSB by the MRN complex, followed by the identification of microhomologous sequences upstream or downstream of the DSB. The microhomologous regions are then used as the bases for which to align the strands with mismatched ends, and following ligation of the broken DNA strands hanging ends result. Next, any overhanging nucleotides are resected and finally any missing base pairs are filled in (both by the MRN complex). Because AEJ ligates the DNA strands without checking for homology it causes deletions flanking the original DSB, as it removes any bases required to align the two broken DNA strands. Because C-NHEJ and AEJ are both error prone they commonly result in structural rearrangements. CNAs almost always result from these repair mechanisms because small insertions and deletions frequently occur during C-NHEJ and always occur during AEJ. Furthermore, these mechanisms can incorrectly repair DSBs to cause chromosome inversions and translocations, where some models suggest that AEJ does this at a higher frequency, but other systems support that C-NHEJ plays a greater role. Furthermore, if any of the NHEJ factors are mutated the propensity of structural rearrangements increases, and consequently mutations in NHEJ are known to cause several conditions, such as hereditary breast and ovarian cancer syndrome (BRCA1 mutations), LigIV syndrome (LIGIV mutations), Cernunnos-Severe Combined Immunodeficiency Syndrome (SCID) (Cernunnos mutations), Artemis-SCID (Artemis mutations), Ataxia Telangiectasia (ATM mutations), and Fanconi anemia (mutations in multiple genes). Additionally, significant upregulation of NHEJ factors is observed in several cancers and thus thought to play a role in tumorigenesis, for example breast cancers (FEN1 LIGIII and MRE11 genes), prostate cancers (FEN1 and NBS1 genes), lung cancers (FEN1 and

Chromosome Rearrangements and Translocations

395

XRCC1 genes), stomach cancers (FEN1 gene), pancreatic cancers (FEN1 gene) head and neck cancers (NBS1 gene), squamous cell carcinomas of the oral cavity (NBS1 gene), Ewing Sarcoma (PARP1 gene), germ cell tumors (PARP1 gene), tyrosine kinase-activated leukemias (PARP1 gene), CML (LIGIII gene), multiple myeloma (LIGIII gene) and neuroblastomas (FEN1 and PARP1 genes). Lastly, it is also known that because C-NHEJ is crucial for V(D)J recombination, somatic recombination and class switch recombination in lymphocytes, dysfunctional C-NHEJ is a major source of structural rearrangements in lymphoid cancers and other disorders (e.g., Nijmegen breakage Syndrome, Fanconi anemia, and Bloom’s syndrome). Breakage fusion bridge (BFB) cycle The BFB cycle is a mechanism of genomic instability that occurs during anaphase as a result of telomere loss, dicentric chromosome formation, and mechanical and physical mitotic spindle stress. For instance, if a single chromatid lost its telomeric repeats at one end and the DSB was not repaired prior to DNA replication, then the original chromatid plus its duplicate would both lack telomeric repeats. Without telomeres to prevent chromosome fusion, the two sister chromatids would join to form a single dicentric chromosome. During anaphase of the cell cycle, the two centromeres of the dicentric chromosome would be pulled in opposite polar directions to form a bridge shaped structure until the centromeres were pulled so far away from one another that a DSB resulted. If this DSB does not occur at the exact location where the two chromosomes originally fused, then structural rearrangements would result (e.g., a deletion from one chromosome will result in an inverted duplication on the other). Because the BFB cycle produces two chromosomes lacking telomeres at their broken ends the cycle could continue to produce further duplication, deletion, inversion, and translocation of chromosomal material. The BFB cycle has been observed to play a role in the production of tumorigenic structural rearrangements in some cancers, where it produces giant abnormal chromosomes (cancer associated neochromosomes) in which oncogenes are highly amplified (e.g., CDK4, HMGA2, and MDM2). Furthermore, sometimes these neochromosomes can have fragments from many different chromosomes stitched together, due to the BFB cycle occurring following chromothripsis events (thousands of simultaneous DSBs). Although BFB-mediated neochromosomes only occur in approximately 3% of cancers, they are very common in certain rare cancers (e.g., in 90% of parosteal osteosarcomas). Polymerase slippage Unlike the above nonhomologous-based mechanisms, polymerase slippage is a mechanism of replication errors rather than of DSB repair. Polymerase slippage is most predominant when DNA/RNA synthesis is beginning or when it is stalled and resumed (e.g., following the collapse of a replication fork). Polymerase slippage occurs when small scale tandem repeats in the genome disrupt DNA/RNA polymerase unclamping and reclamping during DNA/RNA synthesis. Specifically, such repeats may cause the DNA/RNA polymerase complex to reclamp at the wrong position (i.e., shift a few bases backwards or forwards). Respectively, this can cause duplication or deletion of bases in repeat regions and therefore result in small CNAs (where duplications are more common than deletions). Polymerase slippage can take place in both coding and noncoding regions, and in coding regions it can result in the generation of abnormal proteins. If polymerase slippage occurs in critical conserved regions of tumor suppressor genes or oncogenes, it can play a role in tumorigenesis. For example, it has been shown that RNA polymerase slippage is responsible for the conversion of TGFBR2 between its tumor suppressive form and oncogenic role during colorectal cancer development, by introducing a small frameshift mutation in a microsatellite region that deletes its transmembrane and intracellular kinase domains.

Determining how a structural rearrangement developed It is possible to determine which DNA repair mechanism was likely responsible for the generation of a structural rearrangement in cancer. Firstly, key DNA repair genes could be sequenced to observe if there are any mutations, and if so the mutated repair gene would implicate a specific repair pathway(s) (e.g., LIGIV, Cernunnos, and Artemis mutations correspond to faulty NHEJ). However, if a gene is common to many different repair pathways then its mutation would not be sufficient to implicate a specific pathway (e.g., mutations in the components of the MRN complex). Another approach is to analyze the DNA sequence at the site of a structural rearrangement and obtain clues from any abnormalities at rearrangement boundaries or from the type of rearrangement present. For example, small insertions or deletions at the boundaries of the rearrangement would suggest that NHEJ had taken place, and in lymphocytes if there are heptamer or nonamer consensus sequences close to the breakpoints it would suggests that NHEJ had occurred as part of somatic recombination. Alternatively, multiple rearrangements or aneuploidies would suggest that multiple rounds of the BFB cycle had happened, whereas evidence of chromothripsis would suggests that replication fork stalling followed by BIR Template Switching may have ensued.

Consequences of Structural Rearrangements Possible Outcomes of Incorrect Repair of DSBs Multiple simultaneous DSBs generate free double stranded DNA ends that, regardless of the particular DSB repair mechanism, can either be repaired correctly by rejoining fragments in their original order, or misrepaired by joining fragments together in the incorrect order to produce structural rearrangements. The more DSBs exist locally at any given time, the more DNA free ends will be present (e.g., two DSBs ¼ four free ends, three DSBs ¼ six free ends etc.), and the more complex the potential misrepair outcomes

396

Chromosome Rearrangements and Translocations

can be. This is because with an increasing number of DNA free ends, the possible combinations for how DNA fragments can be incorrectly joined together increases in a nonlinear fashion, and with more than two simultaneous DSBs complex structural rearrangements can result. Interestingly, with multiple DSBs, it is much more common for structural rearrangements to be intrachromosomal rather than interchromosomal due to a range of steric constraints restricting the movement of chromosome ends to a radius of approximately 1 mm (e.g., DNA packaging, chromatin packaging, and nuclear chromosome arrangement). Because it is more likely for two DSBs to occur at the same time in close proximity in the interphase nucleus than three or more DSBs, this article will focus on the outcomes from the misrepair of two concurrent DSBs.

Two DSBs on the same chromosome arm The misrepair of two simultaneous DSBs on the same chromosome arm can produce either a deletion or an inversion (Fig. 2A). A deletion results when the chromosomal fragment flanked by two DSBs joins end-to-end with the fragment itself producing a circular acentric chromosome, and the two remaining chromosomal fragments fuse to produce a chromosome with an interstitial deletion between the location of the DSBs. Because the acentric chromosome lacks a centromere it will be lost during anaphase when chromosomes are pulled to opposite spindle poles. Alternatively, a paracentric inversion results when the chromosomal fragment between the two breakpoints is inverted and rejoined in the opposite direction to its original position.

Two DSBs on different arms of a homologous chromosome The misrepair of two simultaneous DSBs on different arms of a homologous chromosome can also produce either a deletion or an inversion (Fig. 2B). A deletion results when the ends of the centromere-containing fragment fuse to one another to produce a ring chromosome, and the two telomere-containing fragments fuse to one another to produce an acentric chromosome. The acentric chromosome will be lost during subsequent cell divisions causing deletions of the two distal fragments. In contrast, a pericentric inversion results when the two distal fragments fuse to the opposite chromosome positions from their original orientation.

Two DSBs on the same arm of two homologous chromosomes The misrepair of two simultaneous DSBs on the same arm of two different homologous chromosomes will produce translocations that can have either of two outcomes. The first outcome is that a dicentric and acentric chromosome are respectively produced by the two centromere-containing fragments fusing and the two remaining fragments fusing (Fig. 2C). The acentric chromosome will be lost during subsequent cell divisions and result in distal deletions of chromosomal material, and the dicentric chromosome will break during anaphase causing duplications/deletions of chromosomal content if the breakpoint is not exactly where the two chromosome fragments originally fused. The breakage of the dicentric chromosome may also be lethal. Alternatively, the two centromere-containing fragments can fuse to the two distal acentric fragments, but in the incorrect homologous chromosome, producing interstitial deletions or duplications (Fig. 2C).

Two DSBs on two nonhomologous chromosomes The misrepair of two simultaneous DSBs on two nonhomologous chromosomes can produce translocations with two different outcomes. Again, an acentric and dicentric chromosome may each be produced by fusing of the two centromere containing fragments and acentric fragments resulting in a distal deletion and possibly cell death during cell division. Alternatively, a balanced translocation can result when the two acentric fragments fuse to the centromere containing fragments of the wrong nonhomologous chromosome.

Driver and Passenger Structural Rearrangements It is now well understood that chromosome inversions, CNAs and especially translocations can all have functional effects toward tumorigenesis. Such structural rearrangements may either be “driver” or “passenger” in nature; respectively, causing or contributing to cancer development, or being produced as a by-product of cancer-related genome instability. However, it is very difficult to determine the driver or passenger status of a rearrangement. The two main types of driver rearrangements are those that produce a fusion gene and those that lead to the transcriptional deregulation of breakpoint adjacent genes. Passenger rearrangements do not possess tumorigenic functional consequences, and likely candidates are rearrangements that are not recurring, do not produce obvious fusion products and are not in the vicinity of tumor suppressor genes and proto-oncogenes.

Types of tumorigenic functional consequences from structural rearrangements There are many different types of functional consequences that can result from structural rearrangements in cancers, from gene duplications or gene deletions of tumor suppressor genes or oncogenes, through to complex functional consequences where multiple genes can be impacted at once and novel oncoproteins can be generated. This section will focus on the more complicated functional consequences of structural rearrangements with chromosome translocations as examples, however the same principles can apply for inversions, deletions and complicated rearrangements. One complex consequence of structural rearrangements is the production of fusion genes that produce tumor promoting fusion proteins. Fusion genes are generated when the translocation breakpoints occur in two gene loci, typically in introns, that result in a translocation of chromosome material between the two loci. Consequently, fusion genes produce a fusion transcript that contains

Chromosome Rearrangements and Translocations

(A)

397

(B) A D A

C

inversion B

B

B A

Pericentric inversion

A

C

C

E

D

B C

Consequences:

Consequences :

D

D

gene fusion

E

deletion of TSG

D

gene fusion B

deletion of TSG Disruption of TSG A

C

deletion

ring chromosome

Disruption of TSG

A

B

D

E

acentric fragment lost

lost

(C)

A

A

B

B

C

C

E

D

A

A

B

B

D

C

C

E

D

D

E

E

Deletion

Duplication and Translocation

Consequences: gene fusion disruption of TSG

A

E

B

D

C

D E

A

Acentric Fragment

Dicentric Chromosome

B C

Fig. 2 Possible types of structural rearrangements that can result from the misrepair of two simultaneous DSBs depending on the location of the DSBs. (A) exhibits the possible outcomes (paracentric inversion or deletion and an acentric fragment) resulting from the misrepair of two simultaneous DSBs on the same arm of a homologous chromosome. (B) exhibits the possible outcomes (pericentric inversion or ring chromosome and an acentric fragment) from the misrepair of two simultaneous DSBs on different arms of homologous chromosomes. (C) exhibits the possible outcomes (deletion alongside duplication or a dicentric chromosome and an acentric fragment) from the misrepair of two simultaneous DSBs on the same arm of nonhomologous chromosomes (and these possible outcomes are the same as those from the misrepair of two simultaneous DSBs on different arms of nonhomologous chromosomesdshown in Fig. 3). It should be noted that ring chromosomes, dicentric chromosomes and acentric fragments are all frequently mitotically unstable and result in anueploidies.

coding exons from the first gene at the 50 end and coding exons from the second gene at the 30 end. Only in-frame fusion transcripts that do not disturb the reading frames for either gene will result in the translation of chimeric fusion proteins, consisting of an N-terminal protein region from the first gene and a C-terminal protein region from the second. For instance, the t(9;22)(q34;q11) translocation (in this case, the derivative chromosome 22 is also known as the Philadelphia chromosome) result

398

Chromosome Rearrangements and Translocations

Fig. 3 Possible functional consequences that can result from structural rearrangements. Misrepair of two or more simultaneous DSBs results in the production of structural rearrangements which can have functional consequences. This example shows how two simultaneous DSBs on different arms of nonhomologous chromosomes can produce balanced translocations with either a fusion gene and/or deregulated gene expression as functional consequences.

in a fusion protein of N-terminal BCR and the C-terminal ABL1 (BCR/ABL1) in 90% of CML patients as well as in many acute lymphoblastic leukemia (ALL) patients). In addition to a single chimeric fusion protein being produced, two reciprocal fusion proteins can be produced from reciprocal translocations. In this case, one of the fusion proteins may be dominant in driving tumorigenesis and the other reciprocal fusion protein is irrelevant, but in other cases both fusion proteins play a role (e.g., PLZF/RARA and RARA/PLZF frequently found in acute promyelocytic leukemiaPML). Another consequence of a translocation can be the transcriptional deregulation of a gene close to the breakpoint by the juxtaposition of a strong enhancer. Examples of this are the t(8;14)(q24;q32) translocation associated with Burkitt’s lymphomas and the t(14;18)(q32;q21) translocation commonly observed in follicular lymphomas. These translocations both involve the IGH intron enhancer and/or locus control regions. Respectively, these translocations result in the overexpression of MYC and BCL2, causing uncontrolled cell division, increased cell survival, and malignant transformation. It should be noted that in cases where a balanced translocation leads to the formation of a fusion gene there will always be the disruption of one normal allele of the genes involved in the fusion. Thus, if one or both of the genes have a normal function that suppresses growth it can be speculated that in addition to the formation of two reciprocal fusion genes there will be haploinsufficiency of a tumor suppressor gene. This is, for example, the case in the t(12;21)(p13;q22) translocations in ALL that affects ETV6 and RUNX1. RUNX1 and ETV6 are transcription factors that suppress cell division and encourage cell differentiation in hematopoiesis. Their loss-of-function can be tumorigenic in blood cells. In addition, it is suggested that tumor suppressor gene functions may be disrupted by fusion proteins which interfere with protein–protein interactions (e.g., fusions between INPP5d and ABL1 caused by t(2;9)(q37;q34) translocations in ALL by altering the protein–protein interactions of the tumor suppressive INPP5D). Lastly, it should be noted that many of the complex functional consequences of structural rearrangements outlined above may also lead to increased genomic instability (e.g., through upregulation of cell division, production of DSB causing agents like reactive oxygen species, or through knocking out the functions of essential tumor suppressor genes etc.). Altogether, chromosomal translocation and other rearrangements can either produce fusion proteins, result in deregulation of gene expression or both. Interestingly, it is much more common for structural rearrangements resulting in chimeric fusion proteins to occur in myeloid malignancies and some solid tumors, whereas rearrangements resulting in deregulated gene expression are more typical of lymphoid tumors. Furthermore, for lymphoid cancers the rearrangements generally involve one of the immunoglobulin loci or the T cell receptor loci, as a consequence of faulty somatic recombination (Table 1).

Methods for the Detection of Structural Rearrangements in Cancer Structural rearrangements have been detected and studied for a very long time, since the early 1900s, by observing changes to chromosome structure under light microscopes. However, structural rearrangements in human cancers were not detected until the 1960s and 1970s when a recurrent small chromosome was identified in the leukemic cells from CML patients (named the Philadelphia chromosome after the city where it was discovered). Later with the advent of chromosomal banding techniques the Philadelphia chromosome was determined to be the result of a translocation, the t(9;22)(q34;q11) translocation. Since, the methods for detecting structural rearrangements in cancers have been greatly improved and there are many genetic and cytogenetic techniques that detect different types and sizes of structural rearrangements.

Cytogenetic Methods Cytogenetics is a branch of genetics that studies the number, structure and function of chromosomes. Cytogenetics is particularly focused on how chromosomes relate to cell behaviors, especially during mitosis and meiosis.

Chromosome Rearrangements and Translocations Table 1

399

Examples of commonly recurring structural rearrangements in different cancer types

Breakpoints

Functional consequences

Frequently associated cancer(s)

Inv(2)(p21;p23)

Produces an in-frame fusion protein EML4/ALK, constitutive tyrosine kinase signaling and uncontrolled cell proliferation Produces an in-frame fusion protein (INPP5d/ABL1), deregulated gene expression through disruption of protein–protein interactions of INPP5D, and cancer transformation Produces an in-frame fusion protein PAX3/FOXO1, deregulated gene expression through transcriptional activation, increased cell growth, increased cell survival, increased cell motility, and decreased cell differentiation Produce in-frame fusion proteins (FOXO3/MLL and FOXO4/MLL, respectively), deregulated gene expression through gene transactivation, increased cell stemness, and cancer transformation Causes juxtaposition of breakpoint adjacent genes (MYC and IGH), overexpression of MYC, uncontrolled cell proliferation, and increased cell survival Produces an in-frame fusion protein AML1/ETO, aberrant recruitment of epigenetic modifiers that regulate myelomonocytic development. Decreased differentiation and increased uncontrolled cell proliferation Produces an in-frame fusion protein BCR/ABL1, constitutive tyrosine kinase signaling, and uncontrolled cell proliferation Produces an in-frame fusion protein RUNX1/ETV6, haploinsufficiency of both genes, dysregulation of hematopoiesis, increased cell proliferation, and less influence toward cell differentiation Causes juxtaposition of breakpoint adjacent genes (IGH and BCL2), overexpression of BCL2, uncontrolled cell proliferation, and increased cell survival Produces two functional in-frame fusion proteins (PLZF/RARA and RARA/PLZF), deregulated gene expression, decreased granulocyte differentiation and uncontrolled cell proliferation

Nonsmall cell lung cancers

T(2;9)(q37;q34) T(2;13)(q36;q14)

T(6;11)(q21;q23) and T(X;11)(q13;q23) T(8;14)(q24;q32) T(8;21)(q22;q22)

t(9;22)(q34;q11) T(12;21)(p13;q22) T(14;18)(q32;q21) t(15;17)(q22;q21)

ALL Alveolar rhabdomyosarcomas

Alveolar rhabdomyosarcomas and leukemias Burkitt’s lymphomas AML

ALL, AML, and CML Leukemias and lymphomas Follicular lymphomas AML and PML

Eleven examples of structural rearrangements found frequently in cancers, with information on their breakpoints, associated genes, functional consequences (fusion genes and/or deregulated gene expression) and associated cancers (mostly hematological malignancies).

Standard cytogenetic methods Standard cytogenetics uses a variety of chromosomal preparation techniques that determine the karyotype of cells under a light microscope, specifically attention is paid to chromosome number, chromosome length, position of the centromeres and the chromosome banding patterns, as well as any other distinctive characteristics. Karyotyping is achieved through chromosome banding techniques, which use dyes that produce horizontal bands across chromosomes when observed under a light microscope (Fig. 4). The banding patterns for a given chromosome pair is unique, enabling chromosomes of otherwise similar size to be distinguished from one another (Fig. 4). Alterations to the typical banding pattern of a given chromosome can identify chromosome breakpoints or large translocations, inversions and CNAs. Generally, chromosome banding techniques stain metaphase chromosomes, and typically around 20 cells are analyzed per sample. Quinacrine banding was the first chromosome banding technique but it is now not as widely used as Giemsa banding. Reverse banding produces the opposite black and white banding pattern from what is seen using quinacrine and Giemsa banding and it is a useful technique for staining the distal ends of chromosomes. High resolution banding is a method that stains chromosomes at prophase or early metaphase when they are not completely condensed, because this produces many more chromosome bands and therefore allows more detailed karyotyping and the detection of less obvious aberrations. However, high resolution banding cannot be achieved in most tumor samples. There are also banding methods that only stain certain chromosomal regions, for instance, C-banding stains constitutive heterochromatin which is typically localized around the centromere, and nucleolar organizing regions banding stains satellites and stalks of acrocentric chromosomes which have centromeres very near one end of the chromosome. In general standard banding techniques have a resolution of only about 5–10 Mbp (mega base pairs). This means that any structural rearrangements smaller than the resolution limit cannot be detected confidently. Molecular cytogenetic methods Molecular cytogenetics is a field that combines molecular biology and cytogenetics, by utilizing a series of probe hybridization techniques to visualize one or more specific genomic regions (e.g., fluorescent in situ hybridization (FISH), array comparative genomic hybridization (array-CGH), and single nucleotide polymorphism (SNP) arrays).

400

Chromosome Rearrangements and Translocations

1

2

3

6

7

8

13

14

19

4

9

5

10

11

12

15

16

17

18

20

21

22

X/Y

Fig. 4 Karyotype of a healthy human male using Giemsa banding. Twenty-three pairs of Giemsa banded chromosomes from an XY healthy human male, showing the unique black and white banding patterns for each chromosome pair that enables specific chromosomes and chromosome regions to be identified.

FISH uses fluorescently-labeled single stranded nucleotide probes to detect the presence or absence of specific DNA sequences on chromosomes (Fig. 5) or the presence or absence of specific RNA targets in cells. Fluorescence microscopy and computational analysis are used to identify whether any probes hybridized to the sample and if so where they hybridized in the genome or in the cell. Hence, FISH can be used to detect chromosome translocations, chromosome inversions and large CNAs. In oncology, FISH is typically carried out on approximately 100–200 cells to identify tumor cells even in residual disease and to observe the heterogeneity seen in tumors. FISH can be performed with different types of probes, for instance, probes binding the length of a chromosome can detect the gain or loss of whole chromosomes or large chromosomal regions, shorter probes can be used to hybridize more specifically to detect smaller CNAs, inversions or translocations. The extent to which probes overlap can be controlled to alter the resolution at which features can be detected, and different combinations of fluorescent probe colors can produce specific secondary colors. While the resolution of FISH can be as high as a few kbps, but typically is a few 100 kbp, it will only yield information regarding the region that is covered by the probe. Thus a typical FISH experiment will only assay a tiny fraction (less than 0.01%) of the genome and alterations that are outside this region will be invisible to the assay. Like FISH, array-CGH also uses fluorescently-labeled single stranded nucleotides probes and fluorescence microscopy to detect chromosomal changes. However, in this method the tumor DNA itself is fluorescently labeled in one color and, together with a fluorescently labeled reference DNA, which has a different color, hybridized to an array of probes (e.g., bacterial artificial chromosome clones) that represents the whole genome (Fig. 6). Array CGH is only able to detect CNAs and is blind to balanced rearrangements. Array-CGH can detect alterations across an entire genome (with good coverage) and at a much greater resolution than chromosome banding and FISH, as low as 2 kbp. The resolution is dependent on the size and number of the cloned DNA fragments of the array used. Thus, alterations can be detected in greater detail and the aberrations can be mapped directly onto the genomic DNA sequence. SNP (single nucleotide polymorphism) arrays are another type of DNA microarray which were originally designed to detect SNPs in the genome only, but can now be used to detect large chromosome abnormalities through identification of CNAs across genomes. SNP arrays involve the fragmentation and fluorescent labeling of sample DNA, followed by application to an SNP array chip. The SNP array chip has approximately a million different probes attached to it, which are roughly 25 nucleotides in length and are designed to bind to a portion of DNA sequence harboring the polymorphic SNP site. Fluorescence microscopy alongside computational analysis systems then detects, reports and interprets the hybridization signals. In this way, the SNPs present in a sample and the subsequent genotype can be determined, and if the intensity of the fluorescent signal for each of these SNPs is compared between multiple samples then CNAs can be inferred. In oncology, SNP arrays are used to detect thousands of CNAs across the genome, and comparisons between tumor and nontumor samples from the same patient can give an indication as to whether a particular CNA is germline-inherited or somatic in origin, as well as detecting loss of heterozygosity and regions of chromosomes that have been duplicated or deleted. Amazingly, SNP arrays have been reported to be over 99.5% accurate in SNP genotyping and CNA detection, with differences in probe and chip design influencing the accuracy and potential of a given SNP array experiment.

Chromosome Rearrangements and Translocations

401

Fig. 5 FISH used to detect the t(9;22) translocation in AML. A metaphase cell positive for the t(9;22)(q34;q11) translocation in the leukemic cells from an AML patient. The chromosomes can be seen in blue, and the green and red spots on chromosome 22 (upper left) indicate a positive result for the BCR/ABL gene fusion.

Fig. 6 Array-CGH copy number profile of a IMR32 Neuroblastoma cell line. Differences between the reference and test samples indicate copy number alterations in the IMR32 Neuroblastoma cell line, where regions of green indicate deletions, regions of blue indicate duplications and regions of red indicate gains and regions of yellow indicate copy number neutrality. The top left label shows an amplification of the MYCN proto-oncogene, the top right label and the bottom left labels respectively show the gain of chromosome regions 1q and 17q, and the bottom left label indicates a loss of chromosome region 1p.

402

Chromosome Rearrangements and Translocations

Advances in the field of cytogenetics are focused on molecular cytogenetics techniques in particular, including automated analysis systems for calculating the results from FISH and other array methods, and improving the resolution of such methods to detect less conspicuous abnormalities.

Genetic Methods DNA sequencing methods DNA sequencing techniques determine the precise order of nucleotides in a DNA molecule. Whole-genome sequencing (WGS) uses a number of next-generation sequencing (NGS) methods that sequence entire genomes and they may thus be used to identify structural rearrangements across a genome. For the detection of structural rearrangements, the most efficacious WGS NGS approach is to combine mate pair sequencing and short-insert paired-end sequencing. In mate pair sequencing, DNA is sheared into fragments of a desired length, often between 2 and 5 kbp. However this poses an issue because it is not feasible to sequence long fragments over 1 kbp using NGS. To combat this, the long fragments are biotinylated at their ends and circularized, and then the DNA rings are fragmented into approximately 400–600 bp. Next the biotin-containing fragments are extracted and the result is fragments short enough to be sequenced which still contain the two ends of the initial long fragment. Because the distance between these two ends was between 2 and 5 kbp depending on the original fragmentation parameters, if they are sequenced and mapped to a reference genome and found to be much closer together or further apart than they should be or even map to different chromosomes, then this indicates that a structural rearrangement has taken place. In short-insert paired-end sequencing, DNA is fragmented into much shorter insert sizes between 200 and 800 bp, and adapter molecules containing primers are attached to the ends of the fragmented molecules. Because the primers in the adapters are different at each end, this allows bi-directional sequencing of the DNA fragments, which provides greater sequencing accuracy than sequencing from a single end because it generates greater coverage when aligned to a reference genome. Both of these methodologies complement one another in the detection and evaluation of structural rearrangements. This is because mate pair Sequencing enables more comprehensive detection of large structural rearrangements, and paired-end sequencing provides more detailed sequencing data filling in any gaps that may be present in the mate pair sequencing genome coverage, as well as being particularly useful in highly repetitive regions. When the data from each of these DNA sequencing methods is analyzed together it facilitates determination of: the genomic loci of structural rearrangements, the size of the rearrangements, the altered DNA sequence following each rearrangement, the genes involved in each rearrangement, and an estimation of the impact of each rearrangement. Together, this gives WGS methods an advantage of having unprecedented sensitivity and resolution of structural and mutational changes. However, it should be noted that mate-pair sequencing is technically challenging and that WGS for the detection of structural rearrangements in tumor samples is a costly and highly experimental approach at present and far from being used routinely.

When Specific Methods are Utilized The methods used to identify structural rearrangements differ dependent on the size and type of rearrangements to be detected. For instance, standard cytogenetics and FISH are typically used for studying balanced structural rearrangements; whereas molecular cytogenetics, particularly Array-CGH and SNP arrays, are generally used for studying smaller unbalanced structural rearrangements. DNA sequencing methods are not commonly used to investigate structural rearrangements because they are too expensive and difficult to use routinely for this purpose. Overall there is a need to develop better methods for detecting structural rearrangements in cancers, particularly for DNA sequencing approaches, because it is common for fewer methods to be utilized to detect structural rearrangements in cancer research and routine diagnostics, thus sacrificing important detail in exchange for cheaper, quicker and simpler analysis. Further, currently structural rearrangements are mainly detected using standard cytogenetics. This is mostly done in liquid tumors (i.e., fCML, AML, and ALL, not so much for chronic lymphocytic leukemia, multiple myeloma, nonhodgkin lymphoma, and hodgkin lymphoma). In most solid tumors structural rearrangements are not routinely characterized with the exception of sarcomas and some prostate and lung cancers. Lastly, it should also be acknowledged that there is only a limited resolution achievable by standard and molecular cytogenetic techniques. So even in those cancers where structural rearrangements are often investigated, there are probably many more small-scale structural rearrangements, especially small inversions or CNAs, that are not identified by these methodologies and are thus not routinely diagnosed.

Medical Implications From Studying Structural Rearrangements in Cancer The primary objective of detecting and evaluating structural rearrangements in cancers is to use the results for clinical management, with the most important aim being the improvement of patient survival. Notably, there are many structural rearrangements that have produced beneficial medical implications, with the t(9;22) translocation and the resulting BCR/ABL fusion protein being a prime example. Firstly, structural rearrangements can function as disease biomarkers in cancer diagnosis, where the presence of BCR/ABL fusion alongside histopathological features can be used to diagnose CML. In addition, structural rearrangements can be used as biomarkers

Chromosome Rearrangements and Translocations

403

for other applications; for instance, the level of the BCR/ABL fusion transcript is used as a measurement to guide prognosis and treatment success, as well as influencing clinical decisions regarding disease management strategies. Furthermore, the BCR/ABL fusion protein resulting from Philadelphia chromosomes exemplifies how researching structural rearrangements can improve anticancer therapies by enabling personalized treatment approaches. This is because BCR/ABL results in constitutive tyrosine kinase signaling, and small molecule inhibitors of tyrosine kinase activity of BCR/ABL fusion protein (e.g., Imatinib, Dasatinib, and Nilotinib) have been developed. In fact, these drugs have been so efficacious that they have reduced the number of CML cells until the point at which a patient is declared to be in molecular remission with no detectable BCR/ABL fusion transcript, and they have greatly increased the survival time of many CML patients. Presently, studies are underway to investigate whether tyrosine kinase inhibitor treatment in CML patients can be stopped when certain criteria are met. First results indicate that about 50% of the CML patients can be considered cured and remain in permanent remission even after tyrosine kinase inhibitor treatment has ceased. Thus the study of chromosomal structural rearrangements which started in the early 1970s with Janet Rowley’s pioneering cytogenetic work and her discoveries of the t(9;22) and t(8;21) translocations have now resulted in the first permanent cures of a blood cancer without chemotherapy just using targeted therapies.

Conclusion and Prospective Vision Future research will aim to resolve many of the unanswered questions surrounding structural rearrangements in cancer; such as, which regions of the genome are most susceptible to structural rearrangements, what the exact mechanisms are behind common structural rearrangements, why certain rearrangements are seen in some cancers but not others, and how often the presence of structural rearrangements results in cancer development. amongst many more. Further, future research will aim to increase the capacity of genetic and cytogenetic methods to detect structural rearrangements, and it is anticipated that such technological advancements especially cheaper and more accurate NGS techniques will enable the identification of many new structural rearrangements in cancers and further increase the understanding on the genetics and biology surrounding structural rearrangements. Additionally, methodological improvements alongside increased routine assessment of structural rearrangements, will also likely further enhance the medical implications of structural rearrangements in oncology clinics, perhaps in helping with faster and more efficient screening of such rearrangements in patients and facilitating the improvement of anticancer therapies through individualized medicine.

See also: Chronic Myelogenous Leukemia: Pathology, Genetics, Diagnosis, and Treatment. Genetic Instability.

Further Reading Aparicio, T., Baer, R., Gautier, J., 2014. DNA double-strand break repair pathway choice and cancer. DNA Repair 19, 169–175. Beroukhim, R., Mermel, C.H., Porter, D., Wei, G., Raychaudhuri, S., Donovan, J., Mc Henry, K.T., 2010. The landscape of somatic copy-number alteration across human cancers. Nature 463 (7283), 899. Das, K., Tan, P., 2013. Molecular cytogenetics: Recent developments and applications in cancer. Clinical Genetics 84 (4), 315–325. Forment, J.V., Kaidi, A., Jackson, S.P., 2012. Chromothripsis and cancer: Causes and consequences of chromosome shattering. Nature Reviews. Cancer 12 (10), 663. Glover, T.W., Wilson, T.E., Arlt, M.F., 2017. Fragile sites in cancer: More than meets the eye. Nature Reviews. Cancer 17 (8), 489. Grade, M., Difilippantonio, M.J., Camps, J., 2015. Patterns of chromosomal aberrations in solid tumors. In: Chromosomal instability in cancer cells. Springer International Publishing, Cham, pp. 115–142. Hasty, P., Montagna, C., 2014. Chromosomal rearrangements in cancer: Detection and potential causal mechanisms. Molecular & Cellular Oncology 1 (1), e29904. Mitelman, F., 2000. Recurrent chromosome aberrations in cancer. Mutation Research 462 (2), 247–253. Nambiar, M., Kari, V., Raghavan, S.C., 2008. Chromosomal translocations in cancer. Biochimica et Biophysica Acta 1786 (2), 139–152. Nambiar, M., Raghavan, S.C., 2011. How does DNA break during chromosomal translocations? Nucleic Acids Research 39 (14), 5813–5825. Nickoloff, J.A., De Haro, L.P., Wray, J., Hromas, R., 2008. Mechanisms of leukemia translocations. Current Opinion in Hematology 15 (4), 338. Rowley, J.D., 1973. A new consistent chromosomal abnormality in chronic myelogenous leukemia identified by quinacrine fluorescence and Giemsa staining. Nature 243 (5405), 290–293. Wan, T.S., 2014. Cancer cytogenetics: Methodology revisited. Annals of Laboratory Medicine 34 (6), 413–425.

Relevant Websites Atlas Genetics Oncology. (2017). http://atlasgeneticsoncology.orgdAtlas Genetics Oncology. OMIC Tools. (2017). https://omictools.com/mitelman-database-of-chromosome-aberrations-and-gene-fusions-in-cancer-tooldMitelman Database of Chromosomal Aberrations and Gene Fusions in Cancer. Wikipedia. (2017). https://en.wikipedia.org/wiki/Chromosomal_inversiondChromosomal inversion. Wikipedia. (2017a). https://en.wikipedia.org/wiki/Chromosomal_rearrangementdChromosomal rearrangement. Wikipedia. (2017b). https://en.wikipedia.org/wiki/Chromosomal_translocationdChromosomal translocation. Wikipedia. (2017c). https://en.wikipedia.org/wiki/Copy-number_variation#Types_and_chromosomal_rearrangementsdCopy-number variation. Wikipedia. (2017d).https://en.wikipedia.org/wiki/Cytogenetics#Human_abnormalities_and_medical_applicationsd Cytogenetics. Wikipedia. (2017e). https://en.wikipedia.org/wiki/DNA_repairdDNA repair.

404

Chromosome Rearrangements and Translocations

Wikipedia. Wikipedia. Wikipedia. Wikipedia. Wikipedia.

(2017f). https://en.wikipedia.org/wiki/Homologous_recombinationdHomologous recombination. (2017g). https://en.wikipedia.org/wiki/Microhomology-mediated_end_joiningdMicrohomology mediated end joining. (2017h). https://en.wikipedia.org/wiki/Non-homologous end_joiningdNonhomologous end joining. (2017i). https://en.wikipedia.org/wiki/Replication_slippagedReplication slippage. (2017j). https://en.wikipedia.org/wiki/Structural_variationdStructural Variation.

Chronic Lymphocytic Leukemia: Pathology and Genetics Julio Delgado and Neus Villamor, University of Barcelona, Barcelona, Spain Ferran Nadeu, IDIBAPS, Barcelona, Spain Elı´as Campo, University of Barcelona, Barcelona, Spain © 2019 Elsevier Inc. All rights reserved.

Nomenclature CBA Chromosome banding analysis CNA Copy number alterations CLL Chronic lymphocytic leukemia CLPD Chronic lymphoproliferative disorders FISH Fluorescent in-situ hybridization GEP Gene-expression profiling IGHV Immunoglobulin heavy chain variable region MBL Monoclonal B-cell lymphocytosis mCLL CLL with mutated IGHV genes RT Richter’s transformation SLL Small lymphocytic lymphoma SNP Single nucleotide polymorphism uCLL CLL with unmutated IGHV genes WES whole exome sequencing WGS Whole genome sequencing WHO World Health Organization

Introduction Chronic lymphocytic leukemia (CLL) is a malignancy of CD5þ B-cells characterized by the accumulation of small, matureappearing neoplastic lymphocytes in the blood, bone marrow and lymphoid tissues, thereby resulting in lymphocytosis, marrow infiltration, lymphadenopathy and splenomegaly. The vast majority of CLL cases are thought to originate from monoclonal B-cell lymphocytosis (MBL), a precursor state that differs from CLL essentially by the concentration of circulating tumor cells. CLL is biologically and clinically very heterogeneous, with some patients never requiring any therapy and others displaying an aggressive course with limited response to therapy.

Definition The World Health Organization (WHO) requires the presence of a CLL-type monoclonal B-cell population of  5  109/L in the peripheral blood for the diagnosis of CLL. Small lymphocytic lymphoma (SLL) is the same disease but the term SLL refers to cases with blood involvement of < 5  109/L with documented nodal, splenic, or other extramedullary involvement. The WHO also defines monoclonal B-cell lymphocytosis (MBL) as the presence of the same type of B-cell population of < 5  109/L in the absence of lymphadenopathy, organomegaly, or other extramedullary disease.

Epidemiology CLL is the most prevalent form of leukemia. In the USA it is estimated that approximately 20,000 new cases will be diagnosed during 2017, representing the 1.2% of all new cancer cases in that country. The age-adjusted incidence rate is 4.7 cases per 100,000 men and women per year, with a clear difference between men (6.4 cases/year) and women (3.3 cases/year). In Europe, the incidence rate is similar (4.9 cases/year), with a similar sex distribution (5.9 for males and 4.0 for females). The median age at diagnosis is 70 years, while the median age at death is 80 years. One of the most notorious epidemiological characteristic of CLL is its racial diversity. Within the USA, the incidence rate is highest in Caucasian males (6.8 cases/year) and females (3.6 cases/year) compared with African Americans (5.1 and 2.5 cases/year for males and females, respectively), while Asian Americans and Pacific Islanders have the lowest incidence (1.6 and 0.7 cases/year for males and females, respectively). For unknown reasons, the incidence of CLL is very low in the Far East, including China and Japan,

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65256-X

405

406

Chronic Lymphocytic Leukemia: Pathology and Genetics

where it is estimated to occur at a frequency that is approximately 10% of that seen in Western countries. Genetic rather than environmental factors have been invoked for these differences, since Japanese who settled in Hawaii do not have a higher incidence than those living in Japan. The incidence of CLL is higher than expected among relatives of patients with the disease. Recent genome-wide association studies have identified common risk single nucleotide polymorphisms associated with sporadic CLL. The risk of CLL associated with each of these variants is however modest at best. Although families with CLL provide evidence for genetic susceptibility, no rare alleles of large effect have thus far been discovered.

Clinical Presentation Patients with CLL usually are asymptomatic at the time of diagnosis. A small proportion of patients consult a physician because of painless swelling of lymph nodes, which may wax and wane, but does not disappear. < 5% of patients present with the general symptoms including one or more of the following: (i) unintentional weight loss  10%of body weight within the previous 6 months; (ii) fevers of > 38  C for  2 weeks without evidence of infection; (iii) drenching night sweats without evidence of infection; and (iv) extreme fatigue. In some patients, the disease presents with features of acquired immunodeficiency (e.g. infections), autoimmune phenomena such as hemolytic anemia, thrombocytopenia or pure red cell aplasia, or exaggerated reactions to arthropod stings or bites. The most common finding on physical examination is lymphadenopathy, usually located in cervical, supraclavicular and axillary sites. The spleen may also be enlarged and, as is the case with enlarged lymph nodes, the splenomegaly is usually painless. Enlargement of the liver may also be rarely noted at diagnosis. In those cases, the liver is usually nontender and firm. Infiltration of the skin (leukemia cutis) is seen in < 5% of cases. In contrast to other lymphoid malignancies, gastrointestinal and meningeal involvement are rarely seen in CLL.

Laboratory Abnormalities The hallmark of the disease is an increased white blood cell count with a high percentage of small, mature-looking lymphocytes (i.e. lymphocytosis). The lymphocyte count is extremely variable, ranging from 10 to 200–300  109/L. In patients diagnosed with CLL on routine analysis, anemia is found in < 10% of the patients. Importantly, anemia is not always due to the infiltration of the bone marrow by the disease; other causes such as autoimmunity, iron, folic acid or vitamin B12 deficiency may account for it and should be investigated in all patients with anemia. Likewise, a marked thrombocytopenia (i.e. < 20  109/L) should raise the possibility of an immune origin, particularly in the absence of anemia. Hypogammaglobulinemia is frequent (30% of patients) and tends to worsen over the course of the disease. Serum immunofixation can demonstrate a monoclonal band, usually of the immunoglobulin M type, in around 10–15% of patients. A positive direct antiglobulin test is observed in around 5% of the patients at the time of diagnosis, with clinically apparent autoimmune hemolytic anemia being less frequent.

Morphology Peripheral Blood and Bone Marrow Before the advent of immunophenotyping, most chronic lymphoproliferative disorders (CLPD) with leukemic expression were diagnosed as CLL, but advances in the morphological evaluation of peripheral blood and lymphoid tissues, immunohistochemistry, flow cytometry, cytogenetics and molecular genetics have facilitated the differential diagnosis. Morphologically, leukemic CLL cells are characteristically small, mature-looking lymphocytes with scanty cytoplasm and a dense nucleus with clumped chromatin (Fig. 1). CLL cells are extremely fragile and can become partially broken during the preparation of blood smears; these cells are known as ‘basket’, ‘smudge’ of ‘Gumprecht cells’ which, although not specific, are typical of CLL (Fig. 1). Depending on the morphologic features of the tumor cells, two variants are recognized: typical and atypical CLL. In typical CLL > 90% of leukemic lymphocytes have the previously mentioned morphology. In atypical CLL, an increased number of atypical cells (prolymphocytes, centrocytes, centroblasts, large stimutated cells, etc.) is observed in the lymphocyte differential count. Atypical CLL can be further subdivided in CLL with increased prolymphocytes (CLL/PL), when the percentage of prolymphocytes ranges from 11% to 55%, and CLL mixed cellularity, when the atypical lymphocytes represent > 15% of all lymphocytes. CLL with atypical morphology is associated with trisomy 12 and it is often necessary to request additional tests to rule out other lymphoproliferative disorders.

Lymph Nodes and Bone Marrow Neither bone marrow or lymph node histological examination is required for CLL diagnosis, but it may be useful in difficult or atypical cases. Involved lymph nodes show a characteristic morphological picture with a diffuse effacement of the architecture by a population of small lymphocytes with scant cytoplasm, round nuclei, clumped chromatin and absent nucleoli. In some cases, lymph nodes may be only partially involved by the disease, with tumor cells infiltrating the areas between reactive follicles or surrounding these follicles with a perifollicular pattern. Most lymph nodes have a variable number of pale areas called proliferation centers (formerly named pseudofollicles) (Fig. 2). These areas have a continuum of small to larger cells with broader cytoplasm and

Chronic Lymphocytic Leukemia: Pathology and Genetics

407

Fig. 1 CLL in peripheral blood. Peripheral blood smear showing increased number of small-sized lymphocytes with partially aggregated chromatin. Notice also one ‘smudge’ or ‘Gumprecht’ cell (May-Grünwald-Giemsa, 1000).

Fig. 2 CLL morphology and phenotype in a lymph node. (A) The lymph node at low power shows pale areas corresponding to proliferation centers (H&E, 2); (B) The lymph node shows an effacement of the architecture by a diffuse proliferation of small cells (H&E, 10); (C) The tumor cells are small with round nuclei, clumped chromatin and scarce cytoplasms. Occasional cells have larger nuclei with evident central nucleoli corresponding to prolymphocytes (H&E, 40); (D) CD20 immunohistochemical staining shows the diffuse positivity of tumor cells, which is stronger in the cells of the proliferation centers; (E) CD3 is positive in infiltrating reactive T cells; (F) CD5, in contrast, is diffusely positive in tumor cells. Occasional cells with stronger staining correspond to infiltrating reactive T-cells that have a similar distribution as CD3 positive cells; (G) LEF1 is diffusely positive in CLL cells with a nuclear staining pattern; (H) CD23 is diffusely positive in CLL cells and it is stronger in the cells of the proliferation centers; I: The expression of the proliferation antigen KI67 is higher in tumor cells of the proliferation centers than in cells from diffuse areas.

408

Chronic Lymphocytic Leukemia: Pathology and Genetics

round nuclei with central nucleoli corresponding to prolymphocytes and paraimmunoblasts. These proliferation centers are characteristic of CLL and not present in any other lymphoid neoplasms. A few mitotic figures are usually found in proliferation centers. In some cases, proliferation centers may be large, becoming confluent and occupying an area broader than a 20 field. They may have also a higher number of proliferating cells with a Ki67 index > 40% and > 2.4 mitosis per proliferation center. These cases have been called “accelerated” or “histologically aggressive CLL” and are associated with unfavorable genetic alterations. These patients usually have clinical manifestations of progressive disease and a clinical outcome that is intermediate between typical CLL and Richter’s transformation. The bone marrow is virtually always infiltrated by the same population of cells with interstitial, nodular and/or diffuse patterns. Proliferation centers are less common than in lymph nodes but may be seen in some cases. In the spleen, tumor cells tend to expand white pulp nodules, but may also infiltrate the red pulp. Proliferation centers may be seen, but are usually less prominent and frequent than in lymph nodes.

Immunophenotype Flow Cytometry Immunophenotyping is essential for the diagnosis of CLL. Immunophenotyping usually distinguishes B- from T-cell disorders, establishes the monoclonal nature of the proliferation by showing immunoglobulin light chain restriction and defines the characteristic profile of CLL cells. CLL cells are phenotypically very different from normal mature and precursor B-cells. Typically, their levels of surface immunoglobulin, CD20, CD22, CD79b, CD81, and FMC7 are low or absent compared to normal B-cells. They also co-express CD5, CD43, and ROR1, and overexpress CD23 and CD200 (Fig. 3 and Table 1). The differential antigenic expression between CLL and normal B-cells has allowed the design of flow cytometry tests able to quantify minimal residual disease with a sensitivity  0.01%. Since there is no single marker exclusively expressed on CLL cells, a composite immunophenotype that integrates five markers into a scoring system helps distinguish CLL form other CLPD (Matutes score). In this score, one point is given for each of the following markers: (i) positive CD5 expression, (ii) positive CD23 expression, (iii) weak kappa/lambda chain expression, (iv) negative or weakly positive CD79b expression, and (v) negative FMC7 expression. Patients with CLL usually score of 4 or 5 points, while patients with other lymphoproliferative disorders usually score 0–2 points. Beyond its diagnostic and therapeutic value, immunophenotypic analysis has been shown to be useful for the estimation of prognosis. Expression of CD38, ZAP70 and CD49d are all predictors for an unfavorable outcome in terms of treatment-free interval and overall survival. Although there is an overlap on the expression of these markers and the IGHV mutational status (see below), some studies have demonstrated that they have independent prognostic value. High sensitive flow cytometry assays, able to detect at least a CLL cell in 10.000 normal cells, are used to assess minimal residual disease in patients receiving intensive therapy and its presence is associated with a worse outcome.

Immunohistochemistry

0 0

CD56 + lgk PE

1E2

1E3 0

1E3 1E2 0 0

CD23 PE

1E3 0

0

1E3 1E4 1E5

ROR1 APC

0 1E2 1E3 1E4 1E5

CD79b PE

0

1E3 1E4 1E5

CD43 APCH7

1E3 1E2 0 0

1E3 1E4 1E5

CD22 APC 1E5

1E4 1E3 1E2 0

0

1E4

1E3 1E4 1E5

1E5

1E4

1E5

CD43 APCH7

1E5

1E4

1E4

0 1E2 1E3 1E4 1E5

CD200 APC

CD43 APC

CD5 PE

1E2 1E3 1E4 1E5

1E5

1E3

1E2

CD5 PerCPCy5-5

1E5 1E4

1E3

0

0 1E2 1E3 1E4 1E5

CD79 PerCPCY5-5

0

1E3

1E5

1E4

CD81 FITC

1E2

1E5

1E4

CD19 PC7

1E3

1E5

CD79b FITC

1E4

FMC7 FITC

1E5

CD20 + CD4 PB

CD8 + lgL FITC

Immunophenotype may be studied in tissues using immunohistochemical methods (Fig. 2). Tumor cells are diffusely positive for the mature B-cell markers CD20 and CD79a with coexpression of CD5 and CD23. CD20 and CD23 expression is stronger in cells

1E4 1E3 1E2 0

0

1E3 1E4 1E5

CD81 APCH7

0

1E2 1E3 1E4 1E5

CD22 PerCPCy5-5

Fig. 3 Phenotypic profile of CLL cells in peripheral blood. Dot plots of B-cells from the peripheral blood and bone marrow of a patient with CLL stained with different combinations of antibodies. In each plot a countour showing the reactivity of the normal mature B-cells (cyan) or B-cell precursors (light violet) is depicted. CLL are painted in dark magenta, residual normal mature B cells in cyan and normal B-cell precursors in light violet.

Chronic Lymphocytic Leukemia: Pathology and Genetics Table 1

CD19 mIga CD20 CD22a CD79ba CD5a CD23a CD200 CD43 CD10 CD103 CD123 CD25 CD11c FMC7

409

Immunophenotypic profile of B-cell lymphoproliferative disorders with leukemic expression CLL

MCL

LPL

B-PLL

FL

HCL

MZL

þþ þ þ þ þ or þþ þþ þþþ þþþ þ    - or þ - or þ 

þþ þþ þþ þþ þþ þþ   - or þ      þ

þþ þ or þþ þ or þþ þ or þþ þ or þþ - or þ - or þ þ       þ

þþ þþ þþ þþ þþ - or þ - or þ - or þ - or þ    þ  þ

þ þþ þþ þþ þþ  þ þ  þ    - or þ þ

þþ þþ þþþ þþþ þþ   þþþ   þ þ þ þ þ

þþ þþ þþ þþ þþ  þ þ - or þ    - or þ - or þ þ

CLL, chronic lymphocytic leukemia; MCL, mantle-cell lymphoma; LPL, lymphoplasmacytic lymphoma; B-PLL, B-cell prolymphocytic leukemia; FL, follicular lymphoma; HCL, hairy-cell leukemia; MZL, marginal zone lymphoma. a Indicates antigens included in the immunophenotypic score for CLL.

located in the proliferations centers (Fig. 2). The CD5 staining of neoplastic B-cells is weaker than that of T-cells, and this double intensity is helpful in identifying the tumor subpopulation. Proliferation centers contain numerous T-cells, most of which are CD4þ, and in some cases a fine network of dendritic cells. Cells in proliferation centers have increased expression of the proliferation-associated marker Ki67, co-expression of survivin and MUM1/IRF4. These cells may also express MYC protein, NOTCH1, and cyclin D1 in up to about 30% of cases, but they are not associated with the associated genomic alterations. TP53 protein expression may be found in > 30% of the cells in CLL with TP53 mutations. This finding may suggest the presence of mutations, but should not substitute cytogenetic or molecular tests.

Monoclonal B-Cell Lymphocytosis In the absence of lymphadenopathy, organomegaly, cytopenias and clinical symptoms, the presence of up to 5  109/L monoclonal B-lymphocytes in blood is defined as monoclonal B-cell lymphocytosis (MBL). MBL represents a miscellaneous category ranging from patients with “low-count” clonal B-cell populations (< 0.5  109/L), detected through general population studies, to patients with absolute “high-count” ( 0.5  109/L) lymphocytosis that are referred for consultation. This is important because low count MBL has significant differences from CLL, an extremely limited, if any, chance of progression, and, until new evidence is provided, does not require routine follow-up outside of standard medical care. In contrast, high-count MBL requires routine follow-up, and has very similar biomarkers as early stage CLL. MBL is classified into three categories on the basis of phenotype: (1) CLL-type, when the cells have the typical CLL phenotype; (2) atypical CLL-type, when the clonal cells have strong CD20 and/or surface immunoglobulin, while CD23 may be negative; (3) non–CLL-type, which is considered when clonal cells are CD5 negative and usually have a stronger expression of CD20 and surface immunoglobulins. In atypical-CLL and non-CLL-type MBL, additional cytogenetic and molecular studies are recommended to rule out other leukemic lymphoid neoplasms, particularly mantle cell lymphoma and splenic marginal zone lymphoma. The prevalence of MBL is closely associated with the sensitivity of the flow cytometry test used. Initial studies using basic B-cell panels revealed that the MBL prevalence was 0.6%, but this figure reached 5–12% in adults over 60 years when a more complex and sensitive flow cytometry test was used. The highest reported prevalence of CLL-like MBL occurs in first-degree relatives of CLL patients. Studies in both the UK and USA showed a high prevalence of CLL-like MBL (13.5–18%) in individuals with a family history of CLL who had normal blood counts. Indeed, a fourfold increased prevalence of MBL was observed in families with a genetic predisposition to CLL compared to the general population. High-count MBL and early stage CLL seem to share the same immune system perturbation and in particular the increased risk of infections. Conversely, it has been reported that patients with MBL frequently have a personal history of pneumonia, meningitis and even infectious episodes among their siblings or children, suggesting that exposure to infectious agents may trigger immune events leading to MBL. High-count MBL does not always progress to CLL, a phenomenon that occurs in only 1–2% cases per year, but disease progression may occur over time and therefore lifelong monitoring is recommended for these cases, similar to what is current practice for patients with early stage CLL. Another important question is whether all CLL cases are preceded by MBL or if some CLL cases develop de novo. To address this question, a study evaluated 77,469 healthy adults who were enrolled in a US Cancer Screening Trial. Among 45 participants who

410

Chronic Lymphocytic Leukemia: Pathology and Genetics

were subsequently diagnosed with CLL during the study and had a prediagnostic peripheral blood sample available, MBL was present in 44 (98%) patients, suggesting that virtually all CLL cases are preceded by MBL. Although confirmatory studies are necessary, the concept of tissue-based MBL of CLL type is in discussion as there is a subset of cases with lymph node involvement by CLL cells that do not appear to have a significant rate of progression. In one retrospective study, lymph nodes with CLL/SLL in which proliferation centers were not observed and patients in whom lymphadenopathy was < 1.5 cm based on computed tomography scans were the best candidates for this new diagnostic category of tissue-based MBL.

Richter’s Transformation Richter’s transformation (RT) represents the development of an aggressive lymphoma, most commonly a diffuse large B-cell lymphoma in a patient previously (or simultaneously) diagnosed with CLL. The diagnosis needs to be substantiated through the biopsy of a lymph node or another lymphoid tissue. Rarely, RT presents in peripheral blood or bone marrow smears in the form of large immunoblasts with reticular chromatin, prominent nucleoli and basophilic cytoplasm admixed with small CLL lymphocytes. In this context, immunophenotyping is equally important as these cells tend to have a CLL phenotype with minor deviations. RT is reported to occur in 3–5% of patients with CLL, with a 10-year-cumulative incidence of around 10%. Clinical and laboratory clues that should raise the possibility of RT include sudden worsening of the general status, appearance of B symptoms, enlarging lymph nodes or spleen and dramatically increasing serum lactate dehydrogenase. Risk factors predicting for the development of RT are high CD38 expression, unmutated and stereotyped IGHV genes and presence of NOTCH1 mutations. In 80% of cases, the transformation occurs in the CLL clone, which usually acquires TP53, CDKN2A, and MYC abnormalities and/ or increased number of chromosomal alterations. The prognosis of these patients is poor. In the remaining 20% of patients, the large-cell lymphoma arises in a CLL-unrelated B-cell clone, and these patients have a similar prognosis compared with patients with de novo diffuse large B-cell lymphoma. There are infrequent cases (< 1%) of CLL transformation into Hodgkin’s lymphoma. The disease predominates in older men and the most common histological subtype is mixed cellularity. Most cases originate in CLL with mutated IGHV genes and are Epstein-Barr virus (EBV) positive. The diagnosis of Hodgkin’s lymphoma requires the presence of diagnostic cells associated with an appropriate microenvirontment of reactive T and inflammatory cells. Some CLL cases may have isolated Reed-Sternberg cells, frequently EBV positive, but without this microenvirotnment. These cases should not be diagnosis as Hodgkin’s lymphoma. The presence of EBV-positive Reed-Sternberg cells or Hodgkin’s lymphoma may be associated with treatments including fludarabine.

Differential Diagnosis with Other Lymphoproliferative Disorders Usually, the diagnosis of CLL is straightforward after the morphologic evaluation of peripheral smears and determination of the leukemic phenotype by flow cytometry. However, other chronic lymphoproliferative disorders (CLPD) with leukemic expression can have overlapping morphological and immunophenotypical features and in those cases genetic and molecular tests can be useful in the differential diagnosis.

Morphology CLL with atypical morphology is characterized by a variable percentage (> 10%) of atypical lymphocytes (e.g. prolymphocytes, lymphoplasmacytoid cells, centrocytes, centroblasts or activated lymphocytes) resembling other lymphoid neoplasms such as follicular lymphoma, mantle cell lymphoma, lymphoplasmacytic lymphoma and marginal zone lymphoma. The presence of > 55% prolymphocytes in the blood arbitrarily separates CLL from B-cell prolymphocytic leukemia, a completely separate disorder. In lymph nodes the presence of proliferation centers is very characteristic of CLL. However, in the few cases in which these structures are absent the differential diagnosis may need to be established with other small B-cell lymphomas. Atypical CLL cells infiltrating lymph nodes may have irregular nuclei that suggest mantle cell lymphoma. Some CLL cases may have residual reactive germinal centers without mantle zones similar to those seen in mantle cell lymphoma. The expression of LEF1 and lack of cyclin D1 would facilitate the CLL diagnosis.

Immunophenotyping Immunophenotyping of peripheral blood cells is necessary to make the differential diagnosis (Table 1). The characteristic phenotype of CLL is shown in Figs. 2 and 3. The dim expression of B-cell markers and the coexpression of CD5 are infrequent in other CLPD. The positivity of CD10 helps to distinguish CLL from follicular lymphoma as it is negative in the former and usually positive in the latter. The expression of CD43 and overexpression of CD23 and CD200 may provide further information in the differential diagnosis of CLL and other CLPD. For instance, these three antigens tend to be negative or dimly expressed in MCL, another CD5þ CLPD. CD200 is a glycoprotein belonging to the Ig superfamily and this antigen is being considered as a target for therapy in

Chronic Lymphocytic Leukemia: Pathology and Genetics

411

CD200þ tumors, including CLL, but also in rheumatoid arthritis and other autoimmune disorders. The differential diagnosis with lymphoplasmacytic lymphoma is sometimes difficult as it could express a CLL-like phenotype. The analysis of the plasma cell compartment and molecular tests may be useful. Finally, the receptor tyrosine-like orphan receptor one (ROR1) is an embryonic glycoprotein highly and specifically expressed on the membrane of malignant, but not normal B-cells. In particular, ROR1 is highly expressed in CLL and hairy-cell leukemia, intermediately expressed in mantle-cell lymphoma and diffuse large B-cell lymphoma and scarcely expressed in follicular lymphoma. Moreover, ROR1 expression in CLL patients is not influenced by therapy. In contrast, normal B-cells have no significant expression of ROR1. Several studies are currently testing the diagnostic and therapeutic use of monoclonal antibodies targeting this molecule.

Cytogenetics and Molecular Biology In contrast with what is observed in lymphoma, reciprocal balanced chromosomal translocations are rare in CLL. Indeed, fluorescent in-situ hybridization (FISH) tests investigating the presence of t(11;14)(q13;q32), leading to cyclin D1 rearrangement, are useful for distinguishing CLL from mantle-cell lymphoma. Since CLL does not display specific chromosomal aberrations, cytogenetics and FISH are generally performed for prognostic purposes. Chromosomal deletions and amplifications can be detected in up to 90% of patients. The most frequent genetic abnormalities in patients with CLL are del13q14 as isolated abnormality (14–60% of cases), del11q22-q23 (10–32%), trisomy 12 (11–18%), del 17p13 (3–27%) and del6q (2–9%), depending on the time-point at which the study is performed, and whether or not the disease is resistant to therapy. These structural abnormalities will be discussed in detail in the following section. Similarly, CLL is not characterized by specific gene mutations. MYD88 L265P mutation is detected in near all lymphoplasmacytic lymphoma and in only 3% of CLL. In the appropriate clinical, morphological and phenotypic context, the determination of MYD88 L265P mutation is useful in the differential diagnosis of CLL and this type of lymphoma.

Genetics of CLL Immunogenetics CLL can be divided into two main subsets with different clinical behavior based on the presence or absence of mutations in the variable region of the immunoglobulin genes, reflecting the stage of normal B-cell differentiation from which they originate. CLL cells expressing unmutated IGHV genes (uCLL) arise from B-cells that have not undergone differentiation in germinal centers. CLL cells with mutated IGHV (mCLL) originate from post-germinal center B-cells. Of note, these IGHV mutations occur in normal B-cells during an immune response to antigens and should not be considered pathological. Patients with uCLL typically have a more aggressive disease than those with mCLL in terms of time to first treatment, overall survival and also progression-free survival after therapy. Another interesting aspect of CLL immunogenetics is that the repertoire of immunoglobin molecules produced by CLL patients in general is considerably more limited than that of normal controls, reflecting the biased use of certain IGHV genes that have restricted somatic mutation and limited junctional and heavy–light chain combinatorial diversity. Moreover, in one third of patients the tumor expresses immunoglobulin ‘stereotypes’, which are stretches of the variable region that can also be identified in the immunoglobulins produced by the CLL cells of other patients. This limited immunoglobulin diversity provides compelling evidence that CLL cells are selected based on the binding activity of their surface immunoglobulin, which plays a crucial part in its pathogenesis. There are hundreds of stereotyped subsets, each defined by a unique VH CDR3 motif, and are classified as major (containing 20 or more sequences) or minor subsets. As of today, 19 different major subsets have been identified, representing 41% of al stereotyped cases and 12% of all CLL cases. Examples are subset #1, which is defined by the IGHD6–19 and IGHJ4 genes, subset #2 (IGHV3–21) and subset #4 (IGHV4–34). Interestingly, these subsets have different clinical characteristics. For instance, patients from subset #4 have a very indolent clinical course, even better than that of patients carrying del13q14, while patients from subsets #1 and #2 have a very aggressive disease with a prognosis similar to patients with TP53 aberrations.

Structural Abnormalities The importance of structural abnormalities in the outcome of patients with CLL is well established. Initial efforts were made using conventional chromosome banding analysis (CBA), but this technique is able to identify abnormalities in only 50% of patients, probably due to the low mitotic rate of CLL cells. These results were consistently improved with the advent of FISH so that, with a relatively low number of probes, genomic aberrations can be found in 80% of patients. In view of these results, FISH became the gold-standard method for cytogenetic assessment and has remained so to date even though it only evaluates specific genomic regions and not the entire genome. In parallel, the results of CBA could be improved when tumor cells were incubated with novel mitogens (e.g. DSP30 and IL2), which are capable of detect aberrations in 80% of patients, including 20–35% of those who have “normal” FISH results. Moreover, single-nucleotide polymorphism (SNP)-arrays also interrogate the entire genome and could be a suitable alternative to FISH for detecting deletions or amplifications. SNP-arrays do not detect balanced rearrangements (e.g. translocations), but these are rare in CLL. Irrespective of the technique used, the most common structural aberration in patients with CLL are:

412

•

•

• •

•

Chronic Lymphocytic Leukemia: Pathology and Genetics

Del13q14, which is present in 50–60% of cases. This deletion, affecting the microRNAs miR15-a and miR16–1, reduces the expression of these genes, which are implicated in the regulation of apoptosis and the cell cycle. In addition, translocations involving 13q14 with a variety of chromosomal partners are also relatively frequent and generally disrupt the miR15-a/miR16–1 locus as well. When del13q14 is the only cytogenetic aberration, it is usually associated with early, non-progressive disease and favorable prognosis. Trisomy 12 is generally considered the second most frequent aberration in patients with CLL, occurring in 15–20% of patients, although the gene involved remains unknown. It often appears as a unique cytogenetic alteration but it can also be associated with others. Importantly, both del13q14 and trisomy 12 are nowadays associated with a favorable/intermediate prognosis and considered early CLL-founding events, since they are usually present in most tumor cells. Trisomy 12 is associated with atypical morphology and immunophenotype (e.g. FMC7 positivity, CD23 and CD43 negativity, strong surface immunoglobulin, CD20 and CD22 expression). It is also associated with NOTCH1 mutations, particularly when trisomy 12 is not part of a complex karyotype. Deletions of chromosome 11q22-q23 may imply loss of the ATM and/or BIRC3 function. They are usually identified in younger, male patients with bulky lymphadenopathy and aggressive disease. The disease is usually less responsive to conventional monotherapy and tends to relapse early. It is statistically associated with the presence of SF3B1 mutations. Del17p13 alters the function of the TP53 gene. Most patients with 17p deletion also have TP53 mutations on the other allele, but 2–5% of patients present mutations of TP53 in absence of del17p13. Both del17p13 and TP53 mutations are associated with progressive disease resistant to conventional therapy, although there is also some degree of heterogeneity. In about a third of cases, ATM, TP53 and BIRC3 genes are inactivated by either point mutations or mutations plus copy number neutral loss of heterozygosity. Both del11q22-q23 and del17p13 are initially subclonal events that expand over time as a function of their proliferative advantage or environmental pressures (e.g. therapy). Novel candidate CLL driver genes recently identified in regions of copy number alterations (CNA) as detected by SNP arrays are ZNF292 (6q15 loss), SP140/SP110 (2q37 loss), SMARCC1/SETD2 (3p21 loss), NFKB2 (10q24 loss) and MGA (15q15.1 loss).

Since these structural abnormalities may coexist, a hierarchical model was devised in which patients are assigned to one of the following groups (from unfavorable to favorable prognosis): (i) del17p13; (ii) del11q22-q23 in the absence of del17p13; (iii) trisomy 12 in the absence of del17p13 or del11q22-q23; and (iv) isolated del13q14. Patients without any of the previous abnormalities have a prognosis that is intermediate between those with trisomy 12 and those with del13q14. However, this and other models combining chromosomal alterations and mutations need to be confirmed in larger cohorts of patients. Genomic complexity, either defined as complex karyotype by CBA or increased CNA by SNP arrays has been consistently associated with adverse clinical outcome. Genomic complexity frequently co-exists with TP53 disruption and modulates the poor prognosis of the latter. More recently, extreme genomic complex alterations called chromothripsis have been identified in 2–3% of patients. In chromothripsis focal regions of the genome undergo extensive chromosome fragmentation and reorganization causing structural genomic complexity which is also associated with TP53 disruption and poor outcome. It has also been associated with SETD2 inactivation, a gene that has been shown to cooperate with chromosomal aberrations in other types of leukemia. Translocations are rare in CLL, but cases of t(14;19)(q32;q13) involving the immunoglobulin heavy chain gene and BCL3 loci, or even t(14;18)(q32;q21), involving the immunoglobulin heavy chain locus and BCL2, have been observed. Of note, the use of novel mitogens has increased the detection of chromosomal abnormalities as detected by conventional cytogenetics.

Mutations in Coding Regions All cancers arise as a result of somatically acquired changes in the tumor cell genome. This does not mean, however, that all these abnormalities are involved in cancer development and, indeed, it is likely that some make no contribution at all. It is, therefore, of utmost importance to distinguish ‘driver’ from ‘passenger’ mutations. Driver mutations are causally implicated in oncogenesis, confer growth advantage and are positively selected by the microenvironment of the tissue in which the cancer arises. Passenger mutations are not selected, do not confer growth advantage and therefore do not contribute to cancer development. Passenger mutations are found within cancer genomes because somatic mutations without functional consequences often occur during cell division by different mechanisms. Thus, any given cell acquiring a driver mutation will already harbor innocuous passenger mutations within its genome, which will be carried along and therefore will be present in all tumor cells. In general, CLL is a very heterogeneous disease from a genomic point of view, in clear contrast to other diseases such as hairy cell leukemia and Waldenström’s macroglobulinemia, where BRAF and MYD88 mutations, respectively, are almost diagnostic. The advent of next generation sequencing (NGS) techniques together with potent bioinformatic pipelines has fuelled cost- and time-effective studies of cancer genomes. The first NGS-based study performed in patients with CLL was the whole genome sequencing (WGS) of four patients including two uCLL and two mCLL cases. This study identified NOTCH1, MYD88, XPO1 and KLHL6 as potential CLL drivers, a hypothesis that was validated by targeted sequencing of additional samples. The results from this study and those from an independent cohort confirmed the role of NOTCH1 activating mutations as a CLL driver. Moreover, NOTCH1 mutations are more common in uCLL compared to mCLL, and associated with resistance to rituximab therapy, shorter survival and high risk of transformation to aggressive lymphoma (RT). MYD88 mutations, on the other hand, had been previously observed in a proportion of cases of diffuse large B-cell lymphoma and virtually all cases of WM. In CLL, MYD88 mutations are associated with young age, mCLL and a normal life expectancy. Finally, the role of XPO1 mutations remains unclear due to the

Chronic Lymphocytic Leukemia: Pathology and Genetics

413

low frequency at which it appears in CLL patients (< 2%), even though selective protein inhibition appears promising for the treatment of CLL and other malignancies. This initial study has ben followed by whole genome and exome sequencing (WES) of larger cohorts comprising around 1000 patients in total. These studies confirmed the role as CLL drivers of previously described genes (e.g. TP53 and ATM) but also revealed other recurrently mutated genes not previously reported, including SF3B1, CHD2, POT1, FBXW7, DDX3X and BIRC3. In one cohort, thirty-six significantly mutated genes were identified together with 23 additional genes that were either significantly mutated in one subgroup (uCLL or mCLL), had truncating mutations, or were infrequent but had been described in other malignancies. The most recurrently mutated genes were NOTCH1 (12.6%), ATM (11%), SF3B1 (8.6%), BIRC3 (8.8%), CHD2 (6%), TP53 (5.3%) and MYD88 (4%). Novel CLL drivers were also identified, including ZNF292, ARID1A, ZMYM3 and PTPN11, and also less frequent mutations in well-known oncogenes (KRAS and NRAS), tumor suppressor genes (CDKN1B and CDKN2A) and transcription factors (IKZF3). In the other cohort, 44 recurrently mutated genes and 11 recurrently CNAs were identified. Some examples, not included in the previous list, are: MGA (3.2%) and FUBP1 (1.7%), both of which modulate MYC expression together with PTPN11; MAP2K1 (2%), which belongs to the MAPK-ERK pathway together with KRAS, NRAS and BRAF; and RPS15 (4.3%), a component of the S40 ribosomal unit. SF3B1 mutations cause aberrant splicing of specific genes and are more frequent in patients with aggressive disease. Other candidate genes such as POT1 and CHD2 have been further validated as bona fide CLL drivers by functional studies. Truncating mutations in DDX3X or BIRC3 have been associated with refractoriness to therapy and adverse outcome. Globally, these mutations could be grouped into eight main signaling pathways: B-cell receptor signaling (MYD88, IRAK1, TLR2, IRF4, CARD11), cell cycle regulation (PTPN11, KRAS, NRAS, BRAF, CDKN1B, CDKN2A), apoptosis (TP53, BAX), DNA damage response (ATM, TP53, POT1, CHEK2), chromatin remodeling (ARID1A, SETD1A, ZMYM3, IKZF3, ASXL1, CHD2), NF-kB signaling (TRAF3, BIRC3, NFKB2), NOTCH signaling (NOTCH1, FBXW7), and RNA metabolism (MGA, ZNF292, SF3B1, RPS15, DDX3X, XPO1, FUBP1).

Mutations in Noncoding Regions The role of non-coding mutations in cancer pathogenesis is starting to emerge after decades of controversy. A recent WGS study of CLL found recurrent mutations in two regions of interest. The first was located in the 30 -UTR of NOTCH1 in 2% of patients, suggesting that in about 20% of CLL tumors with NOTCH1 mutations, these are located in non-coding regions. The mutations create new splicing sites and result in the deletion of the protein’s PEST domain, thus increasing its stability in a similar way as previously described for the typical p.P2514fs*4 mutation. The second mutational region was found in 8% of patients and is located in a small intergenic region of chromosome 9p13. Several functional studies confirmed that this region contains an active enhancer for PAX5, a critical transcription factor for B-cells whose locus is located 330 kilobases away from this region. Remarkably, tumor cells with this mutation had a significantly reduced protein expression of PAX5 compared to wild-type CLL or even normal B-cells. Subsequent studies revealed that this mutation is also present in other lymphoid malignancies such as diffuse large B-cell lymphoma and follicular lymphoma, raising the possibility that PAX5 enhancer mutations might constitute driver events contributing to the development of these tumors.

Genomic Evolution As most tumors, CLL is composed of a variety of subclones with diverse genomic aberrations. These subclones may evolve differently over the years in response to intrinsic (microenvironment) or extrinsic (therapy) pressures in a process that has been termed “clonal evolution”. The first evidence of clonal evolution in CLL came from studies using FISH and SNP arrays. However, it was the digital quantification allowed by NGS that confirmed and extended this concept. According to these studies, the so-called “clonal” mutations are found in all or almost all tumor cells and probably constitute founder alterations (e.g. trisomy 12, del13q14, MYD88 mutations), whereas “subclonal” mutations are probably acquired over the course of the disease (e.g. mutations in TP53, ATM, SF3B1, NRAS) and their prevalence correlates with the number of treatments received by the patient. In line with this, direct pairwise relationships have identified that, for instance, del11q22-q23 or trisomy 12 usually precede the acquisition of ATM and NOTCH1 mutations, respectively. Of note, therapy may eradicate dominant clones and select subclones containing driver mutations, which would eventually propel disease progression and treatment refractoriness in most cases. Clonal evolution can be, however, very heterogeneous, including dominant clones that remain stable over time, but also subclones that switch dominance before and after treatment. Of special interest, several studies have shown that small subclones carrying SF3B1 mutations may evolve without treatment pressure and become the dominant tumor population at progression. In this regard, ultra-deep NGS analysis allows for the detection of very small subclones harboring these mutations (down to 0.3% of all tumor cells), which are invisible by traditional Sanger sequencing and non-detected in previous WGS/WES studies. In one study, patients with small TP53 mutated subclones only detectable by NGS had the same poor outcome compared to patients with clonal TP53 mutations detected by Sanger sequencing. The same approach was applied to other recurrently mutated genes (NOTCH1, SF3B1, BIRC3, POT1, XPO1, etc.), and the same conclusion was reached: subclonal mutations in these genes may have prognostic impact in CLL. In contrast, for some genes the proportion of tumor cells carrying the mutation is prognostically relevant. As such, a number of studies have revealed that highfrequency NOTCH1 mutations shorten patients’ survival, whereas low-frequency mutations do not. Moreover, ultra-deep targeted NGS using a panel of 28 genes plus SNP arrays, all molecular tests performed after one simple DNA extraction, were able to identify

414

Chronic Lymphocytic Leukemia: Pathology and Genetics

one driver abnormality in 86% of patients with the disease (almost as high as what you would achieve using WES or WGS). All together, these results have led to the progressive incorporation of NGS panels in the prognostic evaluation of patients with CLL. A great example of how different subclones may evolve depending on therapy comes from the expanding role of ibrutinib in CLL. Ibrutinib is a Bruton tyrosine kinase inhibitor that has been approved for patients with relapsed/refractory CLL and also for those with TP53 disruption. Shortly after this agent was introduced into the CLL therapeutic armamentarium, several studies revealed that relapse on this agent is predominantly due to acquired mutations in the BTK or in the PLCG2 gene, the enzyme immediately downstream of BTK. Of note, clinical resistance is preceded by a prolonged period of asymptomatic expansion of BTK- or PLCG2-mutated clones, suggesting that, perhaps, these patients could be switched to alternative agents before the patient develops refractory disease or even Richter’s transformation.

Gene Expression (Transcriptome) More than a decade ago, microarray gene expression profiling (GEP) studies were crucial for the characterization of the different gene signatures observed in patients with CLL. These seminal studies were responsible, among many other things, for the identification of ZAP70 expression as one of the best surrogate markers for IGHV mutational status; or ROR1 as one of the most CLLspecific surface antigens. Subsequent GEP studies were also of utmost importance in the delineation of the cellular origin of CLL and characterization of the different anatomic compartments of the disease. However, the transcriptional landscape of CLL at high resolution has remained elusive until the development of novel technologies for deep RNA sequencing such as RNAseq. Using this technology, global transcriptome signatures were identified, including transcriptional elements such as transposable elements, long non-coding RNAs and pseudogenes, all of which are by definition invisible to GEP microarrays.

Epigenomic Profile The advent of NGS and genome-wide methylation arrays has recently allowed investigators to explore the whole genome methylation pattern of CLL and normal B-cell subsets. CLL cells harbor a global hypomethylation in gene bodies, which correlates imperfectly with gene expression. Moreover, there is a striking correlation between the methylation pattern of uCLL and normal naïve B cells, and also between mCLL and normal memory B cells. These results supported the idea that naïve B-cells could potentially be the cell of origin of uCLL, whereas mCLL arises from memory B-cells, a conclusion also reached by an independent research group using gene expression profiling and similar genome wide methylation analysis. In addition, the authors could classify CLL patients according to their methylation profile into 3 groups with a distinct clinical outcome: naïve B-cell like, intermediate and memory Bcell like. These 3 groups had different IGHV mutational status, IGHV usage and other molecular characteristics such as NOTCH1 and SF3B1 mutations, and their diverse prognostic impact was validated in two independent cohorts.

Integrating Genomic, Epigenomic and Transcriptomic Data In the last decade, NGS technology has provided scientists with a wealth of information that requires careful interpretation. For instance, WGS revealed that each patient with CLL or MBL tends to bear from 240 to 5416 point base substitutions or small insertions/deletions, yielding an average of 0.87 mutations per megabase. This mutation rate is low compared to melanoma or lung carcinoma, but in agreement with previous estimates of less than one mutation per megabase for leukemias. Moreover, there are marked differences in the genomic and epigenomic pattern according to the tumor subtype (uCLL vs. mCLL), thus reflecting the molecular mechanisms implicated in their respective development. The number of mutations (excluding the immunoglobulin loci) is higher in mCLL than in uCLL even though the number of driver mutations is significantly lower in the former compared with the latter. In other words, mCLL accumulates a significantly higher proportion of passenger mutations, and not necessarily driver mutations, as they pass through the germinal centre. Indeed, the proportion of mCLL cases with at least one driver mutation identified is 83% compared to 96% for uCLL cases. Equally, CLL and MBL patients have a comparable number of mutations, but a significantly different number of driver mutations, which is consistent with the notion that progression from MBL to CLL is generally determined by the increasing accumulation of driver alterations. On the other hand, NGS and SNP arrays have identified a variety of CLL drivers associated with adverse prognosis. Interestingly, apart from having specific driver genes mutated, what confers a particularly dismal prognosis to patients is their accumulation, so that an increasing number of driver alterations correlates with a progressively shorter time to first treatment and overall survival. Moreover, epigenomics and genomics are clearly linked in CLL. The three epigenetic subgroups defined according to the methylome appear to have a distinct genomic profile as well. Patients from the naïve B-cell like subgroup generally had a higher tumor burden at diagnosis, as measured by beta2-microglobulin, and almost exclusively unmutated IGHV genes, with preferential use of the VH1–69 gene. Phenotypic markers linked with adverse outcome, such as high CD38, ZAP70 or CD49d expression were more frequent in this subgroup, and this also applies to some genomic aberrations such as trisomy 12, del11q22–23 and NOTCH1 and ATM mutations. In contrast, patients included in the memory B-cell like subgroup were portrayed as bearing mutated IGHV genes in almost all cases (frequently the VH4–34 gene) together with the absence of poor prognostic markers. Interestingly, though, patients categorized as ‘intermediate’, also have specific genomic aberrations. These patients, apart from having an intermediate outcome, bear unmutated IGHV genes in 22% of cases and, when they are mutated, the mutation rate is low (median 4%) compared to the memory B-cell like subgroup (median 7%). These tumors preferentially use the IGHV1–18, IGHV3–21 and IGHV3–23 genes and the incidence of

Chronic Lymphocytic Leukemia: Pathology and Genetics

415

SF3B1 and MYD88 mutations is remarkably high compared to the naïve B-cell like or memory B-cell like subgroups, thus confirming that this subset of patients deserves being separated from the other two subsets.

Applying this Knowledge into Current and Future Medical Practice Since a good proportion of patients with CLL have a normal life expectancy, and clinical trials have not shown to date any advantage for initiating treatment in asymptomatic patients, the gold-standard of care for the majority of patients at diagnosis is careful observation without therapy. Once the disease becomes symptomatic, patients are usually treated with DNA-damaging agents such as fludarabine, cyclophosphamide or chlorambucil in combination with monoclonal antibodies (i.e. chemoimmunotherapy). More recently, BCR inhibitors such as ibrutinib or idelalisib, or BCL2 antagonists such as venetoclax, have been added to the treatment armamentarium after promising results obtained in clinical trials. Despite these advances, CLL remains incurable with current therapeutic options and new agents are desperately needed for patients with high-risk disease. New genomic developments could improve patient outcome in two ways: (1) allowing for a more precise prognostic estimation and, (2) identifying novel therapeutic targets. The first aspect has already been put into practice in a variety of publications evaluating the prognostic value of these new driver genes in combination with classical markers, and also as predictors of response to therapy. Examples of these are:

• • • • •

Patients with TP53 abnormalities are unlikely to respond to conventional chemotherapeutic agents and clearly benefit from the use of ibrutinib or idelalisib. Indeed, both agents have been approved by the European Medicines Agency for patients with CLL and TP53 disruption (del17p13 or TP53 mutations). Patients with NOTCH1 mutations do not benefit from the addition of rituximab to the FC (fludarabine plus cyclophosphamide) backbone, which still remains the gold standard frontline therapy for younger and fitter patients. Patients harboring SF3B1 and NOTCH1 mutations have very short progression-free survival after chemoimmunotherapy (i.e. rituximab, fludarabine and cyclophosphamide), probably suggesting that they should be considered for inclusion into appropriate clinical trials whenever possible. Both complex karyotype and del17p13 are associated with disease progression on ibrutinib, suggesting that markers that predict poor prognosis with standard therapies also confer risk with novel therapies, albeit a delayed risk. Moreover, as previously mentioned, disease progression on ibrutinib is very frequently preceded by the emergence of BTK- or PLCG2-mutated clones. Patients with mCLL devoid of TP53 abnormalities or del11q22–23 obtain very good responses and a very prolonged progression-free survival with conventional chemoimmunotherapy (i.e. fludarabine, cyclophosphamide and rituximab).

The second important message from NGS studies is that the prognostic value of these aberrations is not only evident when they are detectable using conventional Sanger sequencing, but also when they are only present in very small subclones, which could change the way we approach molecular studies in our daily practice. The uncovering of intra-tumor heterogeneity and the description of the Darwinian process of clonal evolution leading to the expansion of resistant subclones upon therapy, initially proposed by Nowell in 1976, could change the way we approach CLL therapy. Indeed, half of all driver mutations are subclonal, and the clone composition remains stable over time for most untreated patients, probably owing to the slow growing capacity of CLL cells. In contrast, however, most patients requiring therapy experience a clonal evolution characterized by the expansion of highly fit clones and, perhaps, conventional therapy should be combined with agents targeting the subclonal driver(s) that could potentially expand upon disease relapse. In this regard, genomic studies have identified a number of pathways that could be potential targets for therapeutic intervention:

•

•

•

B-cell receptor (BCR) signaling: The importance of BCR signaling in CLL biology is well established since all B-cells require these signals for their survival and proliferation. As such, pharmacological inhibition of several proteins downstream the BCR (Syk, BTK, PI3K) is remarkably effective in this disease. Some of these genes (i.e. BTK and PLCG2) may acquire mutations upon BTK inhibition. Moreover, point mutations affecting key molecules belonging to this and other tightly related pathways (e.g. Toll-like receptors) have also been identified (MYD88, IRAK1, TLR2, IRF4, CARD11). As already stated, the BTK inhibitor ibrutinib and the PI3K inhibitor idelalisib have approved for therapy of patients with CLL. Cell cycle regulation: CLL cells overexpress a variety of cyclins and cyclin-dependent kinases, mostly through BCR activation, but also through point mutations in cyclin-dependent kinase inhibitors (CDKN1B, CDKN2A) and other similar molecules (PTPN11). Pharmacological inhibition of cyclin-dependent kinases has been tested in CLL with mixed results, mostly due to safety concerns (not lack of efficacy), but novel oral inhibitors (e.g. palbociclib) have been approved in solid tumors and hold promise for CLL. Furthermore, the MAPK/ERK pathway is disrupted in a proportion of patients with CLL, with mutations identified in KRAS, NRAS, or BRAF. Interestingly, specific inhibitors of BRAF or MEK have been already approved for a number of solid tumors and could be of interest. Also, multikinase inhibitors of the MAPK/ERK pathway, such as sorafenib, have showned a remarkable anti-CLL activity in vitro. Apoptosis: As in most malignancies, CLL cells try to evade apoptosis either by disrupting DNA damage response (mutations in ATM, TP53, POT1, or CHEK2), disrupting pro-apoptotic molecules (mutations in BAX) or overexpressing anti-apoptotic molecules (BCL2 or MCL1). Over the past years, several BCL2 inhibitors have been developed and one of them, venetoclax, has remarkable efficacy in CLL, including patients with TP53 and ATM disruption and has been already approved by the US Food and Drug Administration for the treatment of CLL.

416

•

• •

Chronic Lymphocytic Leukemia: Pathology and Genetics

NOTCH signaling: This pathway is involved in cell-fate decisions during development through its nuclear translocation and transcription of target genes including HES1 and MYC. In CLL, NOTCH1 is constitutively expressed thereby promoting apoptosis evasion and cell survival. In some patients, this is due to NOTCH1 mutations which stabilize the protein. Some agents, called gamma-secretase inhibitors, are able to induce apoptosis in vitro and inhibit the constitutive Notch activation that is characteristic of NOTCH1-mutated CLL cells. Moreover, MYC itself could be a potential target thanks to the novel Bromodomain and Extra-Terminal motif (BET) protein inhibitors, which are currently being investigated in many different lymphoid malignancies. RNA metabolism: The identification of mutations in the spliceosome-RNA processing machinery (SF3B1, RPS15, DDX3X) has led to the development of spliceosome modulators. These agents induce selective cytotoxicity of CLL cells in vitro when compared with normal lymphocytes or cells from other B-cell malignancies. NF-kB signaling: The NF-kB signaling pathway is an important therapeutic target since it is constitutively activated in CLL through the canonical and alternative (non-canonical) pathways. In some patients this is due to point mutations (TRAF3, BIRC3, NFKBIE) or structural abnormalities (NFKB2), and pharmacological blockade of this pathway (e.g. using proteasome inhibitors or direct IKK inhibitors) leads to anti-CLL activity in vitro. Moreover, BET inhibitors block downstream oncogenic NFkB-driven transcriptional programs in in vitro and in vivo lymphoma models. Of note, TRAF3 serves as a negative regulator of the non-canonical NF-kB signaling pathway and targets NIK for constant ubiquitination and degradation. As a result, TRAF3mutated tumors result in non-canonical NF-kB activation through NIK overexpression and, indeed, NIK has emerged as a very attractive target for therapy.

Conclusions CLL is the most common leukemia in adults in the developed world. The diagnosis of CLL requires the phenotypic analysis of tumor cells. Small lymphocytic lymphoma is the same disease with tissue involvement and low tumor cell counts in peripheral blood. Morphology and immunophenotype are the initial, and usually sufficient, diagnostic investigations. In cases with morphological atypical features, additional tests such as molecular genetics and/or histology are required to confirm the diagnosis. Some other markers, such as CD38 and ZAP70 expression, are also useful for prognostic purposes. A monoclonal population immunophenotypically identical to that found in CLL or other CLPD can be detected in up to 3–10% of healthy people. However, only those with ‘high-count’ MBL behave as early stage CLL and need to be clinically followed, while avoiding cumbersome and anxiety-triggering medical examinations in those with minute (‘low-count’) monoclonal populations. Regarding the genetics of the disease, several research groups have discovered significant prognostic associations coupled with new driver genes and signaling pathways whose role in cancer was previously unknown or poorly understood. Massive scale epigenomics and transcriptomics have supplied the foundations for the cellular origin of the disease, and recent studies have uncovered non-coding mutations as drivers for the disease. We are now in a position to compose a fully integrated model of biological factors that could be assimilated into daily practice, even though the list of genes that should be evaluated remains undefined. Moreover, some drivers could be targeted pharmacologically, even when present at subclonal level, thus avoiding clonal selection and disease refractoriness. This outburst of genomic knowledge is just starting to expand into the clinic, but in the coming years, as our understanding broadens and ongoing technological innovation propels new achievements, we will certainly learn how to apply it in our daily practice.

Further Reading Falchi, L., Keating, M.J., Marom, E.M., Truong, M.T., Schlette, E.J., Sargent, R.L., Trinh, L., Wang, X., Smith, S.C., Jain, N., Estrov, Z., O’Brien, S., Wierda, W.G., Lerner, S., Ferrajoli, A., 2014. Correlation between FDG/PET, histology, characteristics, and survival in 332 patients with chronic lymphoid leukemia. Blood 123, 2783–2790. Giné, E., Martinez, A., Villamor, N., López-Guillermo, A., Camos, M., Martinez, D., Esteve, J., Calvo, X., Muntañola, A., Abrisqueta, P., Rozman, M., Rozman, C., Bosch, F., Campo, E., Montserrat, E., 2010. Expanded and highly active proliferation centers identify a histological subtype of chronic lymphocytic leukemia (“accelerated” chronic lymphocytic leukemia) with aggressive clinical behavior. Haematologica 95, 1526–1533. Kipps, T.J., Stevenson, F.K., Wu, C.J., Croce, C.M., Packham, G., Wierda, W.G., O’Brien, S., Gribben, J., Rai, K., 2017. Chronic lymphocytic leukemia. Nature Reviews Disease Primers 3, 17008. Kulis, M., Heath, S., Bibikova, M., Queirós, A.C., Navarro, A., Clot, G., Martínez-Trillos, A., Castellano, G., Brun-Heath, I., Pinyol, M., Barberán-Soler, S., Papasaikas, P., Jares, P., Beà, S., Rico, D., Ecker, S., Rubio, M., Royo, R., Ho, V., Klotzle, B., Hernández, L., Conde, L., López-Guerra, M., Colomer, D., Villamor, N., Aymerich, M., Rozman, M., Bayes, M., Gut, M., Gelpí, J.L., Orozco, M., Fan, J.B., Quesada, V., Puente, X.S., Pisano, D.G., Valencia, A., López-Guillermo, A., Gut, I., López-Otín, C., Campo, E., MartínSubero, J.I., 2012. Epigenomic analysis detects widespread gene-body DNA hypomethylation in chronic lymphocytic leukemia. Nature Genetics 44, 1236–1242. Landau, D.A., Carter, S.L., Stojanov, P., McKenna, A., Stevenson, K., Lawrence, M.S., Sougnez, C., Stewart, C., Sivachenko, A., Wang, L., Wan, Y., Zhang, W., Shukla, S.A., Vartanov, A., Fernandes, S.M., Saksena, G., Cibulskis, K., Tesar, B., Gabriel, S., Hacohen, N., Meyerson, M., Lander, E.S., Neuberg, D., Brown, J.R., Getz, G., Wu, C.J., 2013. Evolution and impact of subclonal mutations in chronic lymphocytic leukemia. Cell 152, 714–726. Landau, D.A., Tausch, E., Taylor-Weiner, A.N., Stewart, C., Reiter, J.G., Bahlo, J., Kluth, S., Bozic, I., Lawrence, M., Böttcher, S., Carter, S.L., Cibulskis, K., Mertens, D., Sougnez, C.L., Rosenberg, M., Hess, J.M., Edelmann, J., Kless, S., Kneba, M., Ritgen, M., Fink, A., Fischer, K., Gabriel, S., Lander, E.S., Nowak, M.A., Döhner, H., Hallek, M., Neuberg, D., Getz, G., Stilgenbauer, S., Wu, C.J., 2015. Mutations driving CLL and their evolution in progression and relapse. Nature 526, 525–530. Landgren, O., Albitar, M., Ma, W., Abbasi, F., Hayes, R.B., Ghia, P., Marti, G.E., Caporaso, N.E., 2009. B-cell clones as early markers for chronic lymphocytic leukemia. New England Journal of Medicine 360, 659–667.

Chronic Lymphocytic Leukemia: Pathology and Genetics

417

Nadeu, F., Delgado, J., Royo, C., Baumann, T., Stankovic, T., Pinyol, M., Jares, P., Navarro, A., Martín-García, D., Beà, S., Salaverria, I., Oldreive, C., Aymerich, M., SuárezCisneros, H., Rozman, M., Villamor, N., Colomer, D., López-Guillermo, A., González, M., Alcoceba, M., Terol, M.J., Colado, E., Puente, X.S., López-Otín, C., Enjuanes, A., Campo, E., 2016. Clinical impact of clonal and subclonal TP53, SF3B1, BIRC3, NOTCH1, and ATM mutations in chronic lymphocytic leukemia. Blood 127, 2122–2130. Puente, X.S., Beà, S., Valdés-Mas, R., Villamor, N., Gutiérrez-Abril, J., Martín-Subero, J.I., Munar, M., Rubio-Pérez, C., Jares, P., Aymerich, M., Baumann, T., Beekman, R., Belver, L., Carrio, A., Castellano, G., Clot, G., Colado, E., Colomer, D., Costa, D., Delgado, J., Enjuanes, A., Estivill, X., Ferrando, A.A., Gelpí, J.L., González, B., González, S., González, M., Gut, M., Hernández-Rivas, J.M., López-Guerra, M., Martín-García, D., Navarro, A., Nicolás, P., Orozco, M., Payer, Á.R., Pinyol, M., Pisano, D.G., Puente, D.A., Queirós, A.C., Quesada, V., Romeo-Casabona, C.M., Royo, C., Royo, R., Rozman, M., Russiñol, N., Salaverría, I., Stamatopoulos, K., Stunnenberg, H.G., Tamborero, D., Terol, M.J., Valencia, A., López-Bigas, N., Torrents, D., Gut, I., López-Guillermo, A., López-Otín, C., Campo, E., 2015. Non-coding recurrent mutations in chronic lymphocytic leukemia. Nature 526, 519–524. Seifert, M., Sellmann, L., Bloehdorn, J., Wein, F., Stilgenbauer, S., Dürig, J., Küppers, R., 2012. Cellular origin and pathophysiology of chronic lymphocytic leukemia. Journal of Experimental Medicine 209, 2183–2198. Swerdlow, S.H., Campo, E., Pileri, S.A., Harris, N.L., Stein, H., Siebert, R., Advani, R., Ghielmini, M., Salles, G.A., Zelenetz, A.D., Jaffe, E.S., 2016. The 2016 revision of the World Health Organization classification of lymphoid neoplasms. Blood 127, 2375–2390.

Chronic Myelogenous Leukemia: Pathology, Genetics, Diagnosis, and Treatment ska, Science to the Point, Archamps Technopole, Archamps, France Katarzyna Szyman Sophie Park, Grenoble-Alpes University and Institute for Advanced Biosciences, Grenoble, France © 2019 Elsevier Inc. All rights reserved.

Abbreviations ALP Alkaline phosphatase AP Accelerated phase BP Blast phase (blast crisis) CBC Complete blood count CD Cluster of differentiation CML Chronic myelogenous leukemia CP Chronic phase DLI Donor lymphocyte infusion FDA US Food and Drug Administration FISH Fluorescence in situ hybridization HSCT Hematopoietic stem cell transplantation ICD-O International Classification of Diseases for Oncology LSCs Leukemic stem cells NCCN US National Comprehensive Cancer Network RT-PCR Reverse transcriptase polymerase chain reaction TKI Tyrosine kinase inhibitor WBC White blood cells WHO World Health Organization

Definition and Classification Chronic myelogenous leukemia (CML), BCR-ABL positive (ICD-O code: 9875/3) is a myeloproliferative neoplasm that originates from an abnormal pluripotent bone marrow stem cell and is consistently associated with the t(9;22)(q34.1;q11.2) reciprocal chromosomal translocation which generates the BCR-ABL1 fusion gene located on the so called Philadelphia (Ph) chromosome. The BCR-ABL1 is found in all myeloid lineages as well as in some lymphoid cells and endothelial cells. CML, BCR-ABL positive is also called CML, Philadelphia chromosome positive; CML, t(9;22)(q34;q11); chronic granulocytic leukemia, Philadelphia chromosome positive; chronic granulocytic leukemia, t(9;22)(q34;q11); chronic granulocytic leukemia, BCR/ABL; and chronic myeloid leukemia.

Epidemiology and Risk Factors Burden CML accounts for 15%–20% of adult leukemias. It affects 1–2 individuals per 100,000 annually, with a slight male predominance. The incidence increases with age: the disease is rare in children (< 0.1 case per 100,000 population) and is most frequent in the elderly (over 2.5 cases per 100,000). No significant geographical variations in the CML incidence have been reported, however patients in areas of lower socioeconomic status tend to be younger at disease onset. CML which has reached the blast phase used to be associated with a dismal outcome. However, with the introduction of targeted therapies and more accurate monitoring methods, the 10-year survival rates for CML patients have risen to approximately 80%– 90%.

Etiology and Risk Factors The etiology of CML is largely unknown. Except for acute radiation exposure (such as in atomic bomb survivors), no environmental or lifestyle factors have been identified. There seem to be no familial predisposition, either.

418

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65257-1

Chronic Myelogenous Leukemia: Pathology, Genetics, Diagnosis, and Treatment

419

Presentation and Diagnosis The natural clinical course of untreated CML consists of two or three phases: chronic phase (CP) and blast phase (BP; also called blast crisis) which are usuallydbut not alwaysdseparated by accelerated phase (AP; sometimes also called acute phase because of its resemblance to acute leukemia). Most patients are diagnosed in CP which usually has an insidious onset. Nearly 50% of newly diagnosed CP-CML cases are asymptomatic and are discovered incidentally with abnormally high white blood cell (WBC) counts at routine blood testing and/ or with splenomegaly at physical examination. Common findings at presentation are nonspecific fatigue, malaise, night sweats, and weight loss, and they may occur long after the disease onset. Loss of energy and reduced exercise tolerance may appear during CP after several months. Many patients have symptoms related to enlargement of the spleen, with about half of the patients presenting with palpable splenomegaly. The enlarged spleen may encroach on the stomach and cause early satiety and decreased food intake leading to weight loss. Spleen infarction may manifest as “gripping” pain in the left upper abdominal quadrant. Splenomegaly may also be associated with a hypermetabolic state manifesting as low-grade fevers, excessive sweating, and chronic fatigue. The spleen size usually correlates with the peripheral blood granulocyte counts (the largest spleen in patients with the highest WBC counts). Hepatomegaly, usually asymptomatic, may also occur but it is less common. It usually results from the extramedullary hematopoiesis occurring in the spleen. Visual changes, seizures, cerebral or myocardial infarctions, and priapism may occur in patients with extremely high WBC counts (over 300,000 cells/mL) as manifestations of leukostasis (abnormal WBC clumping) and hyperviscosity. The retina may show papilledema, venous obstruction, and hemorrhages on funduscopy. Untreated CP-CML inevitably progresses to blast crisis (via AP or without it) within 3–5 years. About 5% of patients are diagnosed in AP or BP (without a recognized CP). Bleeding, petechiae, and ecchymoses are frequently prominent symptoms of AP-CML. Fever is usually associated with infections. Bone pain and fever, very large spleen, and/or increased bone marrow fibrosis are common harbingers of the blast phase. Typical symptoms of the blast crisis are due to increasing anemia, thrombocytopenia, basophilia, a rapidly enlarging spleen, and failure of the usual medications to control leukocytosis and splenomegaly. The initial workup consists of a physical examination, including palpation of the spleen and the liver, and a complete blood count (CBC) with a differential and a chemistry profile. Patients should also be tested for infections with hepatitis viruses. CML can usually be diagnosed based on the peripheral blood findings combined with the detection of the Philadelphia chromosome and/or the BCR/ABL1 fusion protein and transcript using cytogenetic and molecular techniques (fluorescence in situ hybridization (FISH) with dual BCR/ABL1 probes and reverse transcriptase polymerase chain reaction (RT-PCR), respectively). Bone marrow aspiration is necessary to obtain sufficient material for karyotyping and morphological evaluation which allow to determine the CML phase. For patients diagnosed with advanced-phase disease (AP or BP), flow cytometry to determine the cell lineage and mutational analyses are recommended. In a majority of patients, bone marrow biopsy is not necessary for diagnosis. However, it remains an important alternative in case of atypical peripheral blood findings or when a cellular aspirate cannot be obtained. The major initial finding at the blood test is neutrophilic leukocytosis, with neutrophil left shift (increased proportion of immature neutrophils) and WBC counts commonly exceeding 30,000 cells/mL at diagnosis. Mild eosinophilia and basophilia are often present already in CP and increase with the disease progression to advanced phases. Increased platelet levels (thrombocytosis) are common at presentation, with the counts that may exceed 1,000,000 cells/mL. Thrombocytopenia is uncommon in CP. Other laboratory abnormalities include hyperuricemia which reflects high bone marrow cell turnover, as well as markedly elevated serum vitamin B-12–binding protein correlated with the degree of leukocytosis (this protein is synthesized by granulocytes). At cytogenetic and molecular analyses, the BCR-ABL1 fusion is found in all myeloid lineages as well as in some lymphoid cells and endothelial cells, in all CML phases. The presence of additional chromosomal abnormalities in Ph-positive cells (“clonal evolution”) is associated with the disease progression to AP and/or BP. Of note, very low levels of BCR-ABL1 transcripts are detected in approximately 30% of healthy individuals, with an incidence which increases with age. The risk of developing CML is extremely low in these individuals and they do not require treatment. Disease progression manifests with a declining performance status, constitutional signs such as fever and weight loss, recurrent bone pain, and symptoms related to increasing anemia, thrombocytopenia, basophilia, increasing WBC counts with increasing proportion of blasts, and enlarging spleen. The disease status should be reevaluated in this setting. The clinical and morphological boundaries between different CML phases are not very sharp, and the diagnostic criteria may vary between different centers. In particular, some investigators insist that a cut-off of 30% (and not 20% as defined by the World Health Organization, WHO) sets the boundary of the blast crisis, and this threshold has been used in many clinical trials, including those which led to the approval of tyrosine kinase inhibitor (TKI) therapy for treatment of CML patients. However, using either of the two cut-offs does not seem to have any prognostic significance for BP-CML patients. The diagnostic criteria for the three CML phases defined by WHO are summarized in Table 1. As the established hematological/cytogenetic criteria, in particular in AP-CML patients, are not sufficient for accurate prognostication, provisional criteria of response to therapy (mainly treatment with TKIs) have been proposed. They seem to improve the prognostic utility of the established hematological/cytogenetic criteria but this remains to be clinically validated.

420

Chronic Myelogenous Leukemia: Pathology, Genetics, Diagnosis, and Treatment

Table 1

Diagnostic criteria for different chronic myelogenous leukemia (CML) phases as defined by the World Health Organization (WHO)

Chronic phase (CP) CML

Accelerated phase (AP) CML

Blast phase (BP) CML

Established hematological/cytogenetic criteria The presence of the Philadelphia chromosome and/or the BCR/ABL1 fusion protein or transcript confirmed by cytogenetic and molecular techniques At least 20% of blasts in the bone marrow or peripheral 90 85 65 85 60 10

Overall survival data are approximations. T1 tumor limited to the submucosa; T2 tumor infiltrating into the muscularis propria; T3 tumor infiltrating into the subserosal fat; T4 tumor penetrating the serosal membrane and/or infiltrating an adjacent organ. N0 no lymph node metastasis; N1 metastasis in 1–3 regional lymph nodes; N2 metastases in four or more regional lymph nodes, M1 metastases in any distant organ.

advanced stages, metastases can be found in almost any organ. In a clinical context, stage grouping is practiced facilitating decisions on therapy. Stage grouping is explained in Table 1. Stage determines prognosis as well as the therapeutic approach, notably as regards the need for adjuvant therapy.

Molecular Pathology Colorectal cancer is a classic example of stepwise progression, initially developing as an aberrant crypt focus, passing on to a still benign adenoma, and ultimately progressing to an invasive adenocarcinoma with the capacity to metastatic spread. The cell of origin most likely is a crypt base stem cell, from which a clone of aberrant transformed cells arises through the accumulation of multiple genetic abnormalities. The key signaling pathway affected in colorectal cancer is the Wnt pathway through inactivating mutation of APC, activating mutation of KRAS and subsequent mutation of TP53 and genes in the TGF-b pathway (notably SMAD4) which confer invasive and metastatic capacity (Fig. 7). As discussed in the paragraph on pathology, many subtypes of colorectal cancer can be distinguished, differing in morphology, genetic background, molecular profile, clinical behavior and response to therapy. We will discuss the three main pathways involved, their relationship to familial colorectal cancer syndromes and how molecular pathology impacts on clinical decision making.

The Chromosomal Instability (CIN) Pathway The prototypical pattern of the development of colorectal cancer is along the chromosomal instability (CIN) pathway, which represents about 85% of sporadic colorectal cancers. Chromosomal instability is reflected in a high number of allelic losses and gains, including recently discovered chromosomal translocations involving the NTRK1 gene. CIN cancers have a relatively low gene point mutation rate, contrary to mismatch repair deficient microsatellite instable colorectal cancers. CIN colorectal cancers almost invariably carry a mutation in the Wnt pathway genes APC (70%) or CTNNB1 (30%). KRAS mutations, usually in codons 12 and 13, occur in 45% of cases and constitutively activate the MAP-kinase/ERK signaling pathway which renders them insensitive to

+ KRAS

Normal mucosa

- P16 - APC

Aberrant crypt focus

+ telomerase

Low grade adenoma

- P53

High grade adenoma

- SMAD4

Carcinoma

Metastases

Fig. 7 The classical adenoma-carcinoma sequence of the development of colorectal cancer (after Vogelstein), which follows a CIN pattern of carcinogenesis. The initial lesion is called aberrant crypt focus and is characterized by a slight disturbance of crypt architecture and cytonuclear atypia of crypt cells. These often carry an activating (þ) KRAS mutation. When an inactivating () APC mutation occurs epithelial dysplasia (as is shown in the second panel) ensues and an adenoma develops. Low grade adenomas are mostly small and tubular with little cytonuclear and architectural features of dysplasia. Upon telomerase activation (þ) high grade features emerge: larger size, villous architecture and high grade cytonuclear features (loss of polarity, nuclear pleomorphism, mitotic activity). Progression towards invasive carcinoma is often accompanied by a TP53 mutation and progression towards by further molecular events, such as inactivation of SMAD4 (). Adapted from Bosman, F.T. (2014). World Cancer report 2014, In: Stewart, B.W., Wild, C.P. (eds.) ISBN 978–92–832-0429-9. Page 392. IARC publications. With permission from the publisher.

436

Colorectal Cancer: Pathology and Genetics

anti-EGFR therapy. TP53 mutations are found in 70% of cases. In advanced colorectal cancer loss of SMAD4 (which functions downstream of the TGF-b2 receptor) occurs, which confers poor prognosis. Evidence emerging from The Cancer Genome Atlas (TCGA) studies has added to these classical molecular events > 20 genes that are frequently mutated, including ARID1A, SOX9 and FAM123B/WXT. In aberrant crypt foci, small patches of mucosa in which regular crypt architecture is disturbed, mutations in KRAS or APC gene are found. Aberrant crypt foci with an APC mutation are morphologically dysplastic, which signifies progression towards adenoma. Subsequent telomerase is activated, which confers unlimited lifespan and ultimately TP53 mutation is acquired which is associated with progression from low grade to high-grade adenoma. The TGF-b pathway, including the SMAD family, is involved in the progression from a non-invasive adenoma to an invasive carcinoma. This is illustrated in Fig. 7.

The Microsatellite Instability (MIN) Pathway Around 15% of colorectal carcinomas are characterized by microsatellite instability. This is a result of a deficient DNA mismatch repair system. This corrects mismatches (single base mismatches or short deletions or insertions) that have occurred during DNA replication. When this does not function, mutations accumulate and mismatch repair deficient cells effectively are hypermutated. This is reflected in the variable length of microsatellite repeats, hence the term microsatellite instability. Microsatellite instability can be visualized using a set of defined microsatellite sequences, which are almost invariably affected (the mononucleotides BAT25 and BAT26, and the dinucleotide repeats D2S123, D5S346, and D17S250). Mismatch repair deficiency is due to loss of function of one of the proteins that make up the effector complex of this system: MLH1, MSH2, MSH6, and PMS2. This can be due to a mutation of one of the responsible genes. Most cases of MSI colorectal carcinoma, however, are due to promoter methylation of MLH1 which silences expression of the gene. Patients with Lynch syndrome carry a germ-line mutation of one of the mismatch repair genes. Loss of function of these proteins can be visualized using immunohistochemical staining (Fig. 8). This is the most frequently used test approach for mismatch repair deficiency. Sporadic MSI cancers show a relatively high frequency of BRAF gene mutations, which are almost invariably of V600E type. Morphologically, MSI cancers also follow an adenoma-carcinoma sequence. Precursor adenomas may be of the conventional type but more often of the sessile serrated type with dysplasia. The ensuing carcinomas are more often located in the right colon, show mucinous or medullary histology and a host response characterized by a high number of TIL. It is assumed that the hypermutated state of these cancers induces numerous cancer-cell related neoepitopes, which evoke a strong host immune response. MSI

Fig. 8 Mismatch repair deficiency analysis using immunohistochemistry panels (A–C) and microsatellite instability. Immunohistochemical staining for MLH1 (A) and PMS2 (D) shows loss of expression (note nuclear staining of normal cells surrounding the cancer cells). MSH2 and MSH6 (panel B and C) show nuclear staining of cancer cells reflecting normal expression. Panels E and F show microsatellite analysis of a MS stable tissue sample (E) showing a single peak for the microsatellite marker (BAT26) whereas the existence of two peaks (F) indicates microsatellite instability. From Bosman, F. and Yan P. (2015). Molecular pathology of colorectal cancer. Polish Journal of Pathology 65, 257–266. https://doi.org/10.5114/pjp. 2014.48094.

Colorectal Cancer: Pathology and Genetics

437

cancers have a better prognosis than microsatellite stable (MSS) cancers but tend to be less responsive to 5-Fluorouracil, the standard chemotherapeutic drug used for colorectal cancer (adjuvant) treatment. Recent developments are quite exciting in that MSI cancers (not only colorectal cancers) appear to be highly sensitive to immune-checkpoint (PD1) blocking therapy, which potentially changes completely the treatment of advanced colorectal carcinoma.

The CpG Island Methylator Phenotype (CIMP) Pathway This pathway is also known as the “serrated pathway,” as it closely associated with the development and progression of sessile serrated adenomas (Fig. 9). It shows significant overlap with the microsatellite instability pathway. It is assumed that the earliest molecular event is promoter hypermethylation affecting a wide variety of genes (hence the term CpG island methylator phenotype) early on affecting MLH1 with microsatellite instability as a result, which then facilitates the accumulation of additional genetic abnormalities. In most cases, the BRAF gene is mutated (typically the V600E mutation). In more advanced stages of the pathway additionally affected genes correspond to those involved in the CIN pathway.

Colorectal Carcinoma in the Context of inflammatory Bowel Disease (IBD) Inflammatory bowel disease is the generic term used to cover two specific entities: Crohn’s disease and ulcerative colitis. Ulcerative colitis is a chronic inflammatory disease of the colon and/or rectum and in most cases restricted to the mucosa and submucosa. Crohn’s disease is also a chronic inflammatory disease but can affect any part of the gastrointestinal tract although most commonly the terminal ileum. The pattern of the inflammatory reaction can be quite similar to that of ulcerative colitis but in a significant proportion of Crohn’s disease cases, the inflammation is granulomatous. Inflammation extends through all layers of the bowel wall which may lead to stenosis through fibrosis but also perforation and formation of fistulae. Both conditions have a pattern of extra-intestinal manifestations. Important in this context is that in both conditions longstanding active disease is associated with increased risk of large bowel cancer or (in Crohn’s disease) small bowel cancer. In this setting, carcinoma develops through a “dysplasia-carcinoma” sequence. The transcription factor NF-kB provides the link between inflammation and carcinogenesis, as it induces expression of the pro-inflammatory mediators COX-2 and TNF-a, which both play a role in colorectal carcinogenesis. Genome abnormalities in colorectal carcinoma in the context of IBD are similar to those in sporadic CRC but TP53 mutations often occur early. APC mutations are less frequent.

Normal mucosa

BRAF mut Hyperplastic polyp CpG island hypermethylation Sessile serrated adenoma/polyp

SSA/P with low grade dysplasia

MLH1 methylation microsatellite instability

SSA/P with high grade dysplasia

Carcinoma

accumulation of additional oncogene and tumor suppressor gene mutations

Fig. 9 Molecular pathway involved in the development of sessile serrated adenomas, which largely corresponds to the CIMP pathway. From Bosman, F.T. (2013). Molecular pathology of colorectal cancer. In: Chung, L., Eble, L.N. (eds.) Molecular surgical pathology. New York: Springer, pp. 1– 16. ISBN 978–1–4614-4899-0.

438

Colorectal Cancer: Pathology and Genetics

Colorectal Cancer Genetics: Familial Colorectal Cancer Syndromes About 20% of all colorectal carcinomas occur in a familial context: more family members are affected than would be expected by chance. An individual in a family with one or more cases of colorectal cancer has a two to sixfold chance to be affected by this cancer, depending on the number of affected family members. However, in only half of such families can a familial cancer syndrome be identified. Genome-wide association studies have pinpointed a variety of chromosomal loci and single nucleotide polymorphisms, associated with increased colorectal cancer risk. These might be less (if at all) penetrant or represent variants of which the effect depends on gene-environment interactions. For a substantial proportion of familial colorectal cancer cases, therefore, the genetic background still needs to be resolved. Consequently, about 10% of all colorectal carcinomas arise in the context of a known familial cancer syndrome. Their number gradually increases. The first discovered syndrome is familial adenomatous polyposis coli. A later discovery was non-polyposis hereditary colon cancer or Lynch syndrome. More recent is the MutYH associated polyposis or MAP syndrome and a recently discovered syndrome is associated with DNA polymerase E/D1 deficiency. Other rare polyposis syndromes with increased colorectal cancer risk exist (e.g., Peutz-Jeghers syndrome, juvenile polyposis syndrome and Cowden syndrome).

Familial Adenomatous Polyposis About 1% of all colorectal cancers occur in the context of familial adenomatous polyposis, of which the Gardner syndrome is a variant. The syndrome is characterized by the development early in life of an increasing number (between hundreds and thousands) of adenomatous polyps in the colon characteristically in the second decade of life (Fig. 10) If untreated these will progress to colorectal cancer; mean age of colorectal cancer diagnosis in FAP patients is 39 years. Extracolonic manifestations include adenomas in the stomach and the proximal duodenum, which can progress to carcinoma, and fundic gland polyps in the stomach. FAP-associated benign lesions are desmoids, lipomas, fibromas, sebaceous and epidermoid cysts in the skin, skeletal osteomas, dental abnormalities and congenital hyperplasia of the retinal pigment epithelium (known as CHRPE). Less common extracolonic malignancies are malignant neoplasms of the thyroid, brain, adrenal, hepatoblastoma in the liver and pancreaticobiliary tumors. The syndrome is caused by an autosomal dominantly inherited mutation of the adenomatous polyposis (APC) gene on chromosome 5. The APC protein plays a crucial role in the Wnt signaling pathway: in a normal cell it binds b-catenin (which is also integrated into the E-cadherin-catenin cell-adhesion complex) which will be then be degraded in the proteasome. Loss of APC results in accumulation of b-catenin, which migrates to the nucleus with transcription factor activity, promoting transcription of genes of which the protein products stimulate growth, such as MYC and cyclin D1 (Fig. 11). In the context of FAP, > 400 different mutations have been reported in the APC gene. Different mutations are associated with different phenotypes and distinct genotype-phenotype associations have emerged, as listed in Table 2. Most important are the mutations reported in exons 3, 4, 6, and 15 in association with attenuated FAP. In attenuated FAP fewer adenomatous polyps are found and colorectal cancer tends to develop at a more advanced age. Upper gastrointestinal lesions are associated with mutations in exon 15 and those beyond codon 1400. Mutations beyond codon 1444 are associated with the presence of desmoid tumors and those in codons 767–1513 with osteomas. Thyroid cancers are mostly seen with mutations in exons 1–15 and CHRPE’s with mutations in exons 9–15 (Table 3). The Gardner syndrome was first described before the discovery of the FAP syndrome or the APC gene. It refers to a FAP subtype with numerous gastrointestinal polyps in combination with other neoplasms including osteomas, desmoid tumors, epidermoid cysts and also dental anomalies. APC mutations in Gardner syndrome occur in a region encoding for the b-catenin-binding site of the protein. The use of the term Gardner syndrome is discouraged, as it clearly constitutes one of the subtypes of FAP.

The MUTYH-Associated Polyposis (MAP) Syndrome Patients with the MUTYH-Associated Polyposis (MAP) syndrome have a FAP phenotype but lack a mutation in the APC gene. It is estimated that about 2% of colorectal cancers occur in the context of this syndrome. MAP is an autosomal recessive disorder caused

Fig. 10 Colon mucosa from a patient with familial adenomatous polyposis. The mucosa is diffusely seeded with polyps varying in size. Inset shows higher magnification.

Colorectal Cancer: Pathology and Genetics

439

Fig. 11 Schematic representation of the canonical Wnt pathway in the absence (A) or presence (B) of a signaling event. In the absence of signaling b-catenin forms a complex initially with axin and APC, downstream completed with GSK3b and CK1a, which results in ubiquitination of b-catenin and ultimately its degradation. A proportion of b-catenin participates in the e-cadherin/catenin cell-adhesion complex. In the presence of a signaling event (B) b-catenin is not degraded, translocates to the nucleus where it forms a transcription factor in complex with, among others, TCF/LEF, pygopus and Bcl9, activating expression of target genes. From https://www.google.nl/search? q¼wntþpathway&source¼lnms&tbm¼isch&sa¼X&ved¼0ahUKEwjo0pX1sv_XAhVsC8AKHW8hCggQ_AUICigB&biw¼1920&bih¼925#imgrc¼PUdZ3WdhmeWRM.

Table 2

Genotype-phenotype associations for extracolonic manifestations in familial adenomatous polyposis (FAP) and attenuated familial addenomatosis (aFAP)

Syndrome phenotype

Genotype

Characteristics

FAP FAP with Desmoid tumors Duodenal polyposis CHRPE Osteomas Thyroid cancer aFAP

Numerous different mutations reported (over 400)

Frequent extracolonic manifestations

Mutations beyond codon 1444 Mutations in exon 15, beyond codon 1400 Mutations in exons 9–5 Mutations in codons 767–1513 Mutations in exons 1–15 >35 mutations described (exons 3, 4, 6, 9, and 15; introns 3 and 9)

Other tumor types and desmoid tumors are less frequent Patterns of extracolonic manifestations not associated with specific mutations Often duodenal polyposis No CHRPE’s

440 Table 3

Colorectal Cancer: Pathology and Genetics Familial polyposis syndromes with polyp characteristics and responsible gene

Syndrome Autosomal dominant Lynch syndrome Familial adenomatous polyposis Polymerase proofreading-associated polyposis Peutz-Jeghers syndrome Juvenile polyposis syndrome Cowden syndrome Autosomal recessive MUTYH associated polyposis syndrome

Polyp characteristics

Gene responsible

Few adenomatous polyps Numerous adenomatous polyps Multiple mostly adenomatous or serrated polyps Hamartomatous polyps also in upper GI tract Multiple hamartomatous polyps Multiple adenomatous, hamartomatous or inflammatory polyps

MLH1, MSH2, MSH6, PMS2 APC POLE, POLD1 LKB1/STK11 SMAD4/BMPR1A pTEN

Multiple adenomatous or serrated polyps

MUTYH

by biallelic mutations in the Mut Y homolog (MUTYH) gene, located on chromosome 1p. The gene encodes a protein with an important function in the DNA base excision repair pathway. MAP syndrome resembles familial adenomatous polyposis in that multiple adenomatous polyps of the colorectum develop, with a very high lifetime risk of colorectal cancer. The number of adenomas ranges from very few to hundreds and therefore MAP resembles attenuated familial adenomatous polyposis. Extracolonic manifestations occur, including duodenal adenomatous polyps and adeno(carcino)mas, CHRPE’s and breast cancer.

Hereditary Non-Polyposis Colon Cancer or Lynch Syndrome About 3% of cases of colorectal cancers occur in the context of hereditary non-polyposis colon cancer or Lynch syndrome, the most common hereditary colorectal cancer predisposition syndrome. Lynch syndrome patients carry an 80% lifetime risk of developing colorectal cancer. Cancer occurs also in other organs, including (in decreasing order of frequency) in the endometrium, ovary, duodenum, urinary tract, stomach, pancreas, biliary tree, and brain. Lynch syndrome is autosomal dominant and is caused by mutations in one of the DNA mismatch repair genes mutL homolog 1 (MLH1) on chromosome 3p21, mutS homolog 2 (MSH2) on chromosome 2p16, mutS homolog 6 (MSH6) on chromosome 2p16 and postmeiotic segregation 2 (PMS2) on chromosome 7p22. The most frequent (about 80%) are MLH1 and MSH2 mutations. In the mismatch repair protein complex, which altogether is composed of seven DNA mismatch repair proteins (in addition to those previously mentioned MLH3, MSH3 and PMS1), these work sequentially to initiate repair of DNA mismatches, base mismatches or small insertions or deletions that occur during DNA replication. MLH1 deficiency is relatively frequent in cancer, not through a mutation of the gene but through epigenetic silencing by methylation of the MLH1 promoter. MSH2 deficiency can also occur through gene-silencing by transcriptional read-through of TACSTD1, a gene directly upstream of MSH2 and encoding the Ep-CAM protein. Because of a germline deletion of the last exons of TACSTD1, its transcription is extended into MSH2 with functional loss of MSH2 activity. Around 15% of colorectal cancers are MisMatch Repair deficient (dMMR) and these characteristically show microsatellite instability (MSI), increased variability in length of the nucleotide repeats that occur throughout the genome. The large majority of these dMMR cancers show epigenetic silencing of MLH1 through promoter methylation. Only patients with a germline mutation in one of the MMR genes have the Lynch syndrome phenotype. Not all replication errors due to MMR are pathogenic. Mutations are significant when they occur in key genes in processes such as cell proliferation, apoptosis or DNA repair, with functional significance in the context of cancer development. MSI is detected through analysis of a consensus set of gene markers frequently affected by dMMR (the mononucleotides BAT25 and BAT26, and the dinucleotide repeats D2S123, D5S346, and D17S250) which show variations in length in MSI. When three or more markers are unstable this is designated as MSI-H(igh). dMMR can also be detected through immunohistochemical staining of the four MMR genes mentioned above. Loss of staining for one of the proteins qualifies the case as MSI (Fig. 8). In case of MLH1 loss, the MLH1 gene might be mutated or, alternatively (and most frequently) silenced through promoter methylation. To diagnose such a case as sporadic or Lynch syndrome, the methylation status of the promoter needs to be assessed. When non-methylated the patient will be further screened for MLH1 mutation. BRAF mutation status can be considered a surrogate marker, as in Lynch syndrome-associated colorectal cancer BRAF mutation is exceptionally rare. dMMR cancers are hypermutated: they show a very high frequency of gene point mutations and significantly fewer chromosomal rearrangements and allelic imbalances. While of colorectal cancers about 15% is MSI, this is much more frequent in gastric and esophageal cancer, non-small cell lung cancer and squamous cell carcinoma of head and neck. This has gained clinical significance recently as MSI cancers have been found to respond to immune checkpoint blocking therapy (PD1/PD-L1 inhibitors): the PD1 inhibitor Pembrolizumab was approved for recurrent solid tumors, regardless of the histological tumor type. In Lynch syndrome, the dMMR colorectal cancers develop through an adenoma-carcinoma sequence with adenomatous polyps as precursor lesions. However, even though polyps may be multiple, they are never sufficiently numerous to call the condition polyposis. Common characteristics of (Lynch syndrome-associated and sporadic) dMMR colorectal cancers are typical localization in the right colon, mucinous or medullary histotype, and a striking lymphoid host reaction, partly as TIL’s and partly as peritumoral

Colorectal Cancer: Pathology and Genetics

441

secondary lymphoid follicles. The prognosis is better than that of mismatch repair competent cancers, but mismatch repair-deficient cancers respond less well to adjuvant chemotherapy. Therapeutic options have changed significantly with the discovery of their response to immune-checkpoint blocking drugs.

DNA-Polymerase Epsilon/D1 Deficiency Syndrome As with the three syndromes described above only a fraction of the familial cases of colorectal cancer could be explained, the search has been on to further identify inherited risk factors, through whole genome sequencing combined with linkage disequilibrium studies in families with multiple colorectal adenomas and early-onset colorectal cancer. Such studies identified germline mutation of the DNA-polymerase epsilon and D1 gene (POLE and POLD1) as a new autosomal dominant syndrome with high-penetrance predisposition to colorectal cancer. The exonuclease domain of the protein has proofreading activity and removes wrongly incorporated nucleotides during DNA replication. Mutations in the part of the gene encoding for this domain result in lack of fidelity of DNA replication. The ensuing hypermutated phenotype contributes to tumorigenesis. The associated syndrome is called polymerase proofreading-associated polyposis. In families with germline POLE mutations, a predisposition for multiple adenomas and early-onset is characteristic but neoplasms in endometrium, ovary, brain, pancreas, small intestine and skin (melanoma) have also been reported. POLD1 mutations also predispose carriers to multiple colorectal adenomas and carcinoma and in addition endometrial and breast cancer. The phenotype associated with polymerase proofreading-associated polyposis is not fully defined yet and will be further detailed as more families with this syndrome are identified.

Consensus Molecular Subtypes (CMS) of Colorectal Cancer, Tumor Microenvironment and the Immune Response In recent years a series of independent publications proposed sub-classifications of colorectal cancer based upon molecular criteria, varying from expression profiles to genome aberrations. These classifications shared certain characteristics in approach and resulting classification but were far from identical. To resolve this conundrum an international consortium composed of the groups responsible for the original publications worked out a gene expression-based classification system, which is now known as the “consensus molecular subtypes” (CMS) of colorectal cancer. This classification distinguishes four CMSs of colorectal cancer. Characteristics of CMS1 are MSI, mutations of BRAF, infiltration with TIL composed of Th1 cells and cytotoxic T-lymphocytes and activation of immune evasion pathways. CMS2 tumors are characterized by high chromosomal instability (CIN) and activation of Wnt and MYC pathways. In the CMS3 category, KRAS mutations are frequent, and metabolic pathways are disrupted. Finally, CMS4 tumors show a strong desmoplastic reaction along with high expression of mesenchymal genes and angiogenesis in which activation of the transforming growth factor beta (TGF-b) plays a central role. The biological significance of molecular subtyping initially was supported by a difference in prognosis between the subtypes, notably with poor prognosis for CMS4. Between the four CMS’s significant differences in the tumor microenvironment have been reported with high numbers of cytotoxic T-lymphocytes and macrophages in CMS1 and 4. In contrast, in CMS4 (myo)fibroblast proliferation and collagen deposition are higher than in the other CMS’s. These differences go along with different patterns of expression of chemokines, mediators of inflammation, genes involved in regulation of the immune response and of angiogenesis. With the stellar increase of interest in immunotherapy, a strong focus has developed on quantifying infiltrating immune cells which has become known as the immunoscore. However, understanding the role of the tumor microenvironment in biology and clinical behavior of colorectal cancer will require the comprehensive study of other cellular and molecular players.

Prognostic and Predictive Biomarkers Colorectal cancer patients are stratified according to prognostic parameters in view of eventual adjuvant chemotherapy. Central in this exercise remain the classical TNM stage parameters tumor extension (T) and lymph node metastasis (N). As these lack in precision (Table 1), more accurate parameters are proposed almost monthly. Most of these, however, do not hold up in subsequent validation studies and presently beyond TNM, (lymph)vascular invasion is reported as a rule and tumor budding as a sign of invasive activity might be added in the near future. In MSS cancers the small subgroup with the BRAF V600E mutation have a particularly poor prognosis. KRAS mutations status has no prognostic significance for overall survival of stage II/III patients, but does predict poor prognosis in advanced (stage IV) cases. MS status has gained in importance as MSI-H cancers have a better prognosis, even though they might respond less favorably to chemotherapy. Other prognostic factors have not made it into daily clinical practice. Recently published tests such as Oncotype DX (multiplex RT-PCR based) and Coloprint (array-based) might provide additional prognostic value, but this needs to be independently validated. As predictive biomarker, KRAS mutation status has become essential, notably for advanced colorectal cancer as KRAS mutated carcinomas do not respond to anti-EGFR therapy. The responsible mechanism is the position of KRAS downstream of EGFR in the MAPK pathway. Constitutive activation of KRAS renders approaches, blocking the upstream activity of EGFR, ineffective (Fig. 12). Of note, no KRAS mutated but also only about 40% of KRAS wild-type colorectal cancers respond to anti-EGFR treatment. This indicates that KRAS mutation status does identify a subset of non-responders but does not predict who will respond. More recently, the status of other genes involved in the pathway, including PIK3CA, BRAF, NRAS, and pTEN has been found to improve predictive

442

Colorectal Cancer: Pathology and Genetics

small molecular EGFR inhibitor EGFR ligand

EGFR-dimer

anti-EGFR antibody PP

P

PI3K RAS

EGFR

PTEN P

RAF

STAT

Akt

no dimerization, no phosphorylation, no activation

MEK mTOR MAPK

transcription activation

Fig. 12 Schematic representation of MAPK/mTOR signaling downstream of EGFR activation, and its role in insensitivity of KRAS mutated colorectal cancer for anti-EGFR treatment. Upon binding of an EGFR ligand to its receptor, EGFR dimerizes, phosphorylates and activates the signaling pathways. This can be inhibited by an anti-EGFR antibody or by a small molecular inhibitor. A mutation in KRAS, RAF or PI3K activates the pathway downstream of EGFR, thus circumventing EGFR inactivation. From Bosman, F. and Yan P. (2015). Molecular pathology of colorectal cancer. Polish Journal of Pathology 65, 257–266. https://doi.org/10.5114/pjp.2014.48094.

power, but as yet unidentified genes are likely to be involved. MSI status identifies a subgroup of colorectal cancer patients with a high chance of favorable response to immunotherapy, notably PD1/PD-L1 blocking.

Prospects For pathologists and for clinicians alike, heterogeneity of colorectal cancer remains a challenge. Which mechanisms are responsible for morphological heterogeneity? How can cancers that are histologically indistinguishable and in the same stage behave differently in terms of recurrence and response to chemotherapy? Some progress has been made with the introduction of molecular parameters and new molecular subtyping holds promise. However, this will have to be validated through molecular annotation of large series of colorectal cancers with detailed follow-up data. This will have to be repeated in connection with the emergence of new therapeutic targets and prognostic and predictive biomarkers. An exciting field is the host response to the growing tumor cells, which creates the tumor micro-environment. It has become clear that the characteristics of the immune response of the host, in terms of the type, location and density of tumor infiltrating lymphocytes in combination with their functional molecular orientation, which together determine the immune contexture, have a significant impact on tumor progression and response to therapy. It will be essential to collect longitudinally samples from tumors before and after treatment, as they progress in recurrent or metastatic sites, as circulating tumor cells or circulating tumor DNA, to improve our understanding of the dynamics of tumor evolution and progression under therapy. Molecular analysis of such samples will clarify which (sub)clones exist in the primary tumor, which of these manage to gain access to the circulation and/or develop into a metastasis and how they respond to the selection pressure exerted by systemic therapy. Detailed knowledge of such tumor dynamics will ultimately allow the development of what has become known as “precision medicine,” to improve outcome of patients with colorectal cancer.

See also: Colorectal Cancer: Diagnosis and Treatment.

Further Reading Bijlsma, M.F., Sadanandam, A., Tan, P., Vermeulen, L., 2017. Molecular subtypes in cancers of the gastrointestinal tract. Nature Reviews. Gastroenterology & Hepatology 14, 333–342.

Colorectal Cancer: Pathology and Genetics

443

Bohanes, P., LaBonte, M.J., Winder, T., Lenz, H.J., 2011. Predictive molecular classifiers in colorectal cancer. Seminars in Oncology 38, 576–587. Boland, C.R., Goel, A., 2010. Microsatellite instability in colorectal cancer. Gastroenterology 138, 2073–2087. Fearon, E.R., 2011. Molecular genetics of colorectal cancer. Annual Review of Pathology 6, 479–507. Fearon, E.R., Vogelstein, B., 1990. A genetic model for colorectal tumorigenesis. Cell 61, 759–767. Gala, M., Chung, D.C., 2011. Hereditary colon cancer syndromes. Seminars in Oncology 38, 490–499. Gibson, J.A., Odze, R.D., 2016. Pathology of premalignant colorectal neoplasia. Digestive Endoscopy 28, 312–323. Guinney, J., Dienstmann, R., Wang, X., de Reyniès, A., Schlicker, A., Soneson, C., Marisa, L., Roepman, P., Nyamundanda, G., Angelino, P., Bot, B.M., Morris, J.S., Simon, I.M., Gerster, S., Fessler, E., De Sousa E Melo, F., Missiaglia, E., Ramay, H., Barras, D., Homicsko, K., Maru, D., Manyam, G.C., Broom, B., Boige, V., Perez-Villamil, B., Laderas, T., Salazar, R., Gray, J.W., Hanahan, D., Tabernero, J., Bernards, R., Friend, S.H., Laurent-Puig, P., Medema, J.P., Sadanandam, A., Wessels, L., Delorenzi, M., Kopetz, S., Vermeulen, L., Tejpar, S., 2015. The consensus molecular subtypes of colorectal cancer. Nature Medicine 21, 1350–1356. Hamilton, S., Bosman, F., Boffetta, P., et al., 2010. Carcinoma of the colon and the rectum. In: Bosman, F.T., Carneiro, F., Hruban, R.H., Theise, N.D. (Eds.), WHO Classification of Tumors of the Digestive. System, 4th edn. IARC press, Lyon, pp. 134–146. Kinzler, K.W., Vogelstein, B., 1996. Lessons from hereditary colorectal cancer. Cell 87, 159–170. Morley-Bunker, A., Walker, L.C., Currie, M.J., Pearson, J., Eglinton, T., 2016. Translating colorectal cancer genetics into clinically useful biomarkers. Colorectal Disease 18, 749–762. Müller, M.F., Ibrahim, A.E., Arends, M.J., 2016. Molecular pathological classification of colorectal cancer. Virchows Archiv 469, 125–134. Snover, D.C., 2011. Update on the serrated pathway to colorectal carcinoma. Human Pathology 42, 1–10. The Cancer Genome Atlas Network, 2012. Comprehensive molecular characterization of human colon and rectal cancer. Nature 487, 330–337. Xie, J., Itzkowitz, S.H., 2008. Cancer in inflammatory bowel disease. World Journal of Gastroenterology 14, 378–389.

Relevant Websites https://www.cancer.gov/types/colorectal/hp - National Cancer Institute. Colorectal CancerdHealth Professional Version http://www.nlm.nih.gov/medlineplus/colorectalcancer.html - MedlinePlus. Colorectal Cancer. http://www.cancer.org/cancer/colonandrectumcancer/index - American Cancer Society. Colorectal Cancer. http://seer.cancer.gov/statfacts/html/colorect.html - National Cancer Institute. Surveillance, Epidemiology, and End Results (SEER) Program. Cancer Stat Facts: Colorectal Cancer. http://www.uptodate.com/contents/molecular-genetics-of-colorectal-cancer - UpToDate. Molecular genetics of colorectal cancer

Copy Number Variations in Tumors Tao Huang, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China © 2019 Elsevier Inc. All rights reserved.

Glossary Biomarker An easy-to-measure indicator of the disease status or subtype. Copy number variation The number of copies within a genome region is different from the normal number of copies, 2. Pan-cancer Diverse tumor types that share similar cancer features yet show different characteristics. Targeted sequencing Resequencing of an isolated region of genome. Whole genome sequencing The sequencing of the complete DNAs within an organism.

Human is a diploid organism with two copies of alleles for autosomal and pseudo autosomal genes. If there are no longer two copies, CNV (copy number variation) happens. When there are more than two copies, it is called gain or amplification. When there are less than two copies, it is called loss or deletion. CNVs are very common in diseases, especially in tumors when the biological systems dysfunction on multiple levels. CNV is usually the upstream change that triggers the cascade of signaling chaos and therefore can be the driver of disease. In tumors, CNV is found to be a very good biomarker for subtyping and survival prognosis. And it is helpful for understanding tumorgenesis and how mutations happen and in which order. In this review, we will introduce the latest computational methods for CNV detections, especially the challenges of tumor CNV detection, and how to use the CNV data to investigate the mechanisms of several major types of cancers.

The Computational Methods for NGS Based CNV Detection With the rapid development of next-generation sequencing (NGS) technology, CNV detection becomes easier and many computational methods have been proposed. The difficulties of NGS based CNV detection are filtering the noise, determining the region of CNV and estimating the copy numbers. The most widely used CNV software include PennCNV, CNVkit, and CNVtools, GISTIC and others. CNVkit is highly recommended for targeted sequencing. To summary the principles underlying these software, there are several approaches: read-depth based methods, paired-end mapping based methods, split-read based methods and sequence assembly based methods. The four methods for CNV detection are shown in Fig. 1. The first three methods need to map the reads onto the reference genome while the last methods need to assemble the sequence from the beginning. (A)

(B)

More mapped reads Amplification

Less mapped reads Deletion

Reference Genome

(C)

Mapped distance too close

Deletion

Mapped distance too far away

Insertion

(D)

Reads Two splits from one read were mapped to two distant locations

Deletion

To determine where to split needs a lot of computation One split can be mapped while the other split can't

Insertion

Assembled Genome

Deletion

Reads

Assembled Genome

Insertion

Reference Genome Fig. 1 The four methods for CNV detection. (A) Read-depth based method. (B) Paired-end mapping based method. (C) Split-read based method. (D) Sequence assembly based method.

444

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65047-X

Copy Number Variations in Tumors

445

The Read-Depth Based Methods The read-depth based methods are most intuitive and straightforward. These methods analyze the read depth at each position of the genome and find the differences across regions. It assumes that the read depth and copy number are positively correlated. This is generally correct, since the reads are randomly generated and should cover the genome without bias. Therefore, when a region is amplified or deleted, the reads within that region will change accordingly. But it is observed that the GC content can significantly affect the read depths. Usually, the region with extremely high or low GC content will have fewer reads. Another disadvantage of read depth methods is that the region with a lot of repeats tend to have more mapped reads since the mapping software could not assign the read onto the right region. This problem will become more serious when the length of reads is short. The simple but not perfect solution is to only consider the uniquely mapped reads or assign such reads onto a randomly chosen region. CNVnator is a representative CNV detection software that uses read-depth approaches.

The Paired-End Mapping Methods The paired-end mapping methods need to break the DNA sequence into fragments with fixed length and then sequencing from both ends. Since the reads are paired, the mapped position of two reads within a pair can provide important information of the genome structure. It can not only detect CNV but also structural variation. When the distance between the mapped positions of two reads from a read pair is significantly short or long, this read pair is discordantly mapped and indicates CNVs between them. By clustering such discordantly mapped read pairs, the CNV regions and copy numbers can be detected. BreakDancer is one of the software that use paired-end mapping approaches.

The Split-Read Based Methods The split-read based methods split the reads into smaller fragments and map the split fragments onto reference genome. To analyze the break point of the read with fragments that are mapped onto different regions, the CNV position can be accurately detected and the position can be as accurate as one base pair. If two fragments form one read are mapped onto two positions with a gap, it indicates a deletion between these two positions. If one fragment from a read can be mapped but the other fragment from this read can’t be mapped, it indicates an insertion. How to split the reads is difficult to decide. To try more split methods means more computational time. And if the length of the original read is short, it will be difficult to split them and expect the even smaller fragments can be mapped onto the genome correctly. One of the split-read based software is Pindel. Generally speaking, the split-read based methods are slower.

The Sequence Assembly Based Methods The sequence assembly based method does not require a reference genome and does not do mapping. It assembles the genome de novo and then detect the CNVs by analyzing the assembled genomes. Since the whole genome assembly is quite difficult, it usually does de novo assembly and local assembly to generate sequence clusters and compares with close reference genomes. Similar as read-depth based methods, the assembly based methods performs poorly at genome regions with many repeats. Velvet is a representative software that uses assembly based methods.

The Challenges of Tumor CNV Detection Although there are already many software for CNV detections, it is still difficult to perform CNV detection for tumor samples. The main challenge is that the tumor samples usually are not pure tumor cells and include normal cells. This will contaminate the tumor sample and make the CNV detection from NGS data difficult. Even with careful selection to make sure the samples are from tumor, the heterogeneity of tumor cells is still troublesome. Single cell sequencing may be a good solution for this problem. But the small amount of DNA in single cell sequencing makes the sequencing procedure and analysis challenging. We need to distinguish the duplication during PCR, low coverage due to the small amount of DNA and actual CNVs. Second challenge is that most methods assume that the genome is stable and most regions are diploid. But in tumor samples, it is quite common to have large structural variation and aneuploidy on a genome wide scale. Such systematic errors may cause great bias for CNV detection. It is better to detect structural variations before detecting CNVs for tumor samples. The contamination with normal cells, tumor heterogeneity, and tumor aneuploidy are very common in tumor samples and will make the CNV detection difficult.

The CNVs Contribute to Tumor Initiation and Progression With the accumulation of more and more CNV data, several CNV databases have been built, such as CNVD and SCAN. And large tumor CNV datasets, such as TCGA (The Cancer Genome Atlas) and METABRIC (Molecular Taxonomy of Breast Cancer International Consortium), have been released.

446

Copy Number Variations in Tumors

Various pathological CNVs have been identified to be associated with tumor initiation and progression. For example, CNVs of well-known breast cancer genes BRCA1 and BRCA2, are observed in breast cancer patients and suggest that the CNVs contribute to breast cancer. We will discuss the functions of CNVs in several major cancers.

CNVs in Breast Cancer We downloaded the GISTIC processed TCGA breast cancer (BRAC) CNV data and METABRIC breast cancer data. The CNV status were quantified as “ 2” for homozygous deletion, “ 1” for heterozygous deletion, “0” for diploid, “1” for low-level gain, and “2” for high-level amplification. In TCGA and METABRIC breast cancer dataset, the most common deep deletion genes were CSMD1, MYOM2, CDKN2A, DLGAP2, ERICH1, PTEN, RPL23AP53, ZNF596, ARHGEF10, CDKN2B, and KBTBD11; the most frequent high-level amplification genes were TRPS1, MYC, CASC8, LINC00536, PVT1, CCAT1, EIF3H, MIR1205, MIR1208, and TMEM75. In breast cancer, gene amplification is more common than deletion. The frequencies of amplification and deletion in TCGA and METABRIC breast cancer datasets were shown in Fig. 2. In Fig. 3, we plotted the top three deletion genes, CSMD1, MYOM2, CDKN2A, and the top three amplification genes, TRPS1, MYC, CASC8, in TCGA and METABRIC breast cancer dataset. It can be seen that the frequencies of the top amplification genes were much higher than the top deletion genes. The top amplification genes were consistently amplified in two datasets while the top deletion genes were deleted in most samples but amplified in a small proportion of other samples. To investigate the CNV consistency in breast cancer, we plotted the amplification and deletion frequencies in both TCGA and METABRIC breast cancer dataset in Fig. 4. It can be seen that although the deletion frequencies of CSMD1, MYOM2, and CDKN2A were high in both TCGA and METABRIC, there were considerable large proportions of breast cancer samples from both TCGA and METABRIC with amplification of these genes. For top amplification genes TRPS1, MYC and CASC8, there were no deletions in 3253 TCGA and METABRIC breast cancer patients. The CNV status of genes were reported to be associated with breast cancer stages, overall survival and disease-free survival. More specifically speaking, the breast cancer patients with deletions of C20orf46 and SCUBE2 had shorter overall survival time. And the

Fig. 2 The frequencies of amplification and deletion in TCGA and METABRIC breast cancer datasets. In breast cancer, the amplification is more common than deletion.

Copy Number Variations in Tumors

447

Study of origin CSMD1

6%

MYOM2

5%

CDKN2A

4%

TRPS1

25%

MYC

24%

CASC8

24%

Genetic Alteration

No alterations

Study of origin

Breast Invasive Carcinoma (TCGA, Provisional)

Amplification

Deep Deletion

Breast Cancer (METABRIC, Nature 2012 & Nat Commun 2016)

Fig. 3 The OncoPrint of the deletion genes, CSMD1, MYOM2, CDKN2A and the amplification genes, TRPS1, MYC, CASC8, in TCGA and METABRIC breast cancer dataset. CSMD1, MYOM2, CDKN2A were the top three deletion genes and TRPS1, MYC, CASC8 were the top three amplification genes. The patients without CNVs of these genes were not shown.

gain of ZFP14, GSTM2, and JAK1 tended to have bad effect on overall survival while the loss of PDGFRA tended to have good effect on extend the overall survival. The loss of PDGFRA can not only extend survival time, but also postpone breast cancer recurrence. The gain of ZFP14 and LCE3C will significantly accelerate breast cancer recurrence. CNVs are not only biomarkers for breast cancer subtyping or prognosis prediction, but also useful for understanding how breast cancer begins and in which temporal order the mutations occur. With the oncogenetic tree model, the chronological CNV mutation timeline can be inferred. The general mutation orders of key driver genes in breast cancer were as follows: the ERBB2 CNV mutation happened first, then the copy numbers of AKT2, KRAS, PTEN, and CCND1 became abnormal and the mutation changes of PIK3CA happened after the changes of KRAS. The CNV pattern of breast cancer varies a lot in different patients. Age is an important factor. For example, the prevalence of PIK3CA CNV in young breast cancer patients under 45 years old was much smaller than its prevalence in elderly patients over 70 years old. The prevalence of TP53 mutations was the highest in patients between 46 and 69 years old. The mutation of GATA3 only occurred in young patients under 45 years old but not in old patients while the mutations of CDH1 were only found in patients over 45 years old. Another factor is ethnicity. For example, the amplification of 3q26.1 was detected in Chinese breast cancer patients but not in other populations.

CNVs in Lung Cancer Non-small cell lung cancer (NSCLC) is the most common subtype of lung cancer and it includes adenocarcinoma and squamous cell carcinoma. In a CNV array dataset of over 200 lung adenocarcinoma patients and about 100 lung squamous cell carcinoma patients, 266 discriminative CNV probes were identified. They were able to classify the adenocarcinoma patients and squamous cell carcinoma patients nearly perfectly. They were from pathways, like hsa04310 Wnt signaling pathway, hsa04510 Focal adhesion, and hsa04512 ECM-receptor interaction. They located in the chromosome regions of 2q34, 10p15, 18q11, 3q26, 8p23, 3p21, 3q27, 22q12, Xq13, 2q36, 10p11, and 10p12. The CNV pattern of lung adenocarcinoma and squamous cell carcinoma are very significantly different. There are many published array-based CNV signatures. Will these signatures still be useful if the NGS is used? To investigate this, we selected 20 CNV genes that can classify the adenocarcinoma patients and squamous cell carcinoma patients and mapped these genes on TCGA lung adenocarcinoma and TCGA lung squamous cell carcinoma dataset. The OncoPrint of the 20 CNV signatures derived from array data in the NGS based TCGA lung adenocarcinoma and TCGA lung squamous cell carcinoma dataset was shown in Fig. 5. It can be seen that the array-based CNV signatures can still classify the NGS lung adenocarcinoma and lung squamous cell carcinoma samples correctly. The genes ESRRG and FZD8 were amplified more often in lung adenocarcinoma; CD69 was deleted more frequently in lung adenocarcinoma. In lung squamous cell carcinoma, TP63, NAALADL2, SOX2, BCL6, DGKG, DLG1, IL1RAP, LPP, LRRC15, MED12L, NLGN1, PAK2, PLSCR1, PLSCR2, TNIK, USP13, and ZIC4 were amplified more frequently. The array-based results and the NGS-based results were consistent. These results proved that the accumulated CNV array signatures can be adopted in the NGS-based methods. The CNV is not just an adenocarcinoma or squamous cell carcinoma biomarker, but also indicates drivers for dysfunctions of multiple levels in lung cancer. By integrating CNV with gene expression, microRNA expression, DNA methylation in TCGA squamous cell lung carcinoma dataset, a whole genome wide integrative regulatory network can be constructed using regression model. The CNV of FSCB, TCL6, MECOM, BCHE, ANO9, FBXO17, AGAP2, and C9orf53 were found to be the key drivers of the gene expression, microRNA expression, DNA methylation dysfunctions associated with lung squamous cell lung carcinoma. There have been many reports of how CNV affects gene expression through dosage effects. The regulatory roles of CNV in the integrative network of lung squamous cell lung carcinoma confirmed that CNV could regulate gene expression with dosage effect.

448

Copy Number Variations in Tumors

Fig. 4 The amplification and deletion frequencies of CSMD1, MYOM2, CDKN2A, TRPS1, MYC, and CASC8 in TCGA and METABRIC breast cancer dataset. The amplification and deletion frequencies of CSMD1 (A), MYOM2 (B), CDKN2A (C), TRPS1 (D), MYC (E), and CASC8 (F) in TCGA and METABRIC breast cancer dataset.

CNVs in Colorectal Cancer In TCGA colon and rectal cancer dataset, the CNVs were measured using Affymetrix SNP 6.0 arrays. Meanwhile the methylation, microRNA and mRNA were quantified with Illumina HumanMethylation27 BeadChip, Agilent miRNA-Seq and Illumina RNASeq, respectively. The regulatory effect of CNV, methylation and microRNA on mRNA levels can be modeled as a linear regression. Based on how well the CNV, methylation and microRNA data can predict the gene expression level, the CNV-regulated genes, the methylation-regulated genes and the microRNA-regulated genes can be identified. The functions of these genes were investigated. It was found that the CNV-regulated genes have very different functions compared with the methylation-regulated genes and the microRNA-regulated genes. This finding indicates that the CNV is a unique regulator of gene expression. Based on the TCGA colorectal cancer dataset, the CNVs prefer to regulate genes of cellular macromolecule metabolic process, gene expression, macromolecule modification, cellular protein modification process, protein modification process, transferase

Copy Number Variations in Tumors

449

Study of origin ESRRG

4%

FZD8

1.2%

CD69

3%

TP63

22%

NAALADL2

24%

SOX2

25%

BCL6

21%

DGKG

22%

DLG1

22%

IL1RAP

21%

LPP

22%

LRRC15

21%

MED12L

15%

NLGN1

23%

PAK2

22%

PLSCR1

12%

PLSCR2

12%

TNIK

24%

USP13

24%

ZIC4

13%

Genetic Alteration

No alterations

Study of origin

Lung Adenocarcinoma (TCGA, Provisional)

Amplification

Deep Deletion

Lung Squamous Cell Carcinoma (TCGA, Provisional)

Fig. 5 The OncoPrint of the 20 CNV signature derived from array data in the NGS based TCGA lung adenocarcinoma and TCGA lung squamous cell carcinoma dataset. The array based CNV signature can still classify the NGS lung adenocarcinoma and lung squamous cell carcinoma samples correctly.

activity, transferring phosphorus-containing groups, protein kinase activity, phosphotransferase activity, and alcohol group as acceptor. The CNV-regulated genes are not like the methylation-regulated genes which are mostly from anatomical structure morphogenesis, immune system process, regulation of multicellular organismal process, response to external stimulus and calcium ion binding, or the microRNA-regulated genes that are enriched onto single-organism process, single-organism cellular process, biological regulation, protein binding, receptor activity, signal transducer activity, molecular transducer activity, and signaling receptor activity. Through regulating key genes in colorectal cancer, CNV plays important roles for prognosis. For example, the CNV of UQCRB is a prognostic predictor for colorectal cancer. The receiver operative characteristic curve (AUC) of UQCRB CNV gain was 0.891. The CNVs of CHD8, CD47, and RERG, ARHGDIB are associated with colorectal cancer risk. The CNV of TNFRSF10C is associated with colorectal cancer metastasis. Similarly, there are also population specific CNVs in colorectal cancer. It was reported that the ZNF217 and CYP24A1 tend to be gained simultaneously in Chinese colorectal cancer patients.

CNVs in Pan-Cancer The TCGA dataset is a large multi-omics dataset with about 30 types of cancers. The CNVs of 19 genes (RPS15, IL17RC, CUL2, SMPD3, MIR4703, CDKN2A, RFFL, CTBP2, MMD2, SEMA6A, ZFPM1, CDC25A, ZMYND11, KBTBD6, CELF5, EGFR, PIGL, ZNF503-AS1, RBFOX1) were found to be able to classify six major cancer types in TCGA: BRCA (breast invasive carcinoma),

450

Copy Number Variations in Tumors

CDKN2A

8%

EGFR

3%

Genetic Alteration

No alterations

Amplification

Deep Deletion

Fig. 6 The OncoPrint of CDKN2A and EGFR in MSK-IMPACT. Within the MSK-IMPACT cohort which included over 10,000 patients and over 360 types of cancers, 8% of them had CDKN2A deep deletion and 3% of them had EGFR amplification. The patients without CNVs of these two genes were not shown.

COAD/READ (colon adenocarcinoma/rectum adenocarcinoma), GBM (glioblastoma multiforme), KIRC (kidney renal clear cell carcinoma), OV (ovarian serous cystadenocarcinoma) and UCEC (uterine corpus endometrioid carcinoma). The accuracy of the 19-gene CNV signature was over 75%. Another large pan-cancer CNV dataset is MSK-IMPACT in which over ten thousand cancer samples were included but only over four hundred genes were measured. The OncoPrint view of CNV will show how broadly CNV happens in cancer and how different their CNV pattern are. In Fig. 6, we plotted the OncoPrint of two genes from the 19 gene CNV signature discovered from TCGA dataset, CDKN2A and EGFR. Within the MSK-IMPACT cohort which included over 10,000 patients and over 360 types of cancers, 8% of them had CDKN2A deep deletion and 3% of them had EGFR amplification.

The Clinical Applications of CNVs in Tumors As we discussed above, the CNV patterns were different in tumor and normal samples, different tumor subtype samples, tumor patients with different drug response, tumor patients with different survival time. CNV is an important biomarker. Unlike gene expression which can be easily disturbed and shows randomness, CNV is more robust and stable. Since it is known that the normal copy number for human is 2, it doesn’t require complicated normalization baselines like gene expression based methods, such as RT-PCR, microarray or RNA-Seq. All these characteristics of CNV make it a perfect tumor signature. In clinical practice, the targeted NGS-based CNV panels are widely used and highly reproducible. In a study with one cohort of 391 samples and another cohort of 2375 samples, 37 unique CNV events were detected by nine targeted NGS panels with 100% sensitivity. This assessment of CNV applications in clinical laboratory testing shows how reliable CNV biomarkers are. There are many commercial CNV panels. For example, NanoString Technologies has an nCounterÒ v2 Cancer Copy Number Assay which includes 87 common tumor CNV genes, such as AKT, BRCA, ERBB2, MYC, PIK3CA, and PTEN. This panel can be used for formalin-fixed, paraffin-embedded (FFPE) samples. ArcherDX, Inc. has a comprehensive thyroid and lung (CTL) kit that can detect common oncogenic CNV drivers in lung cancers. NGeneBio Co., Ltd. also has a lung cancer kit called LUNGaccuTestÔ that can easily detect lung cancer CNVs. Unlike whole genome CNV analysis, the CNV panel analysis focus on a small set of genes but requires much higher accuracies since the results will directly affect the clinical practice, such as how to treat the cancer patients. Among the open source software, CNVkit is designed for targeted DNA sequencing and is a good choice for panel CNV data analysis. Of course, there are commercial software that are designed for specific platforms. For example, the NanoString platform has nCounterÒ Digital Analyzer. Overall, CNV can be a great biomarker for tumors. It can be applied for tumor subtyping, drug response prediction, survival time prediction. If we can choose the right genes based on genome-wide CNV investigation, the targeted CNV panel can be easily designed and applied. We believe that more and more CNV applications in tumor diagnostics will be developed and applied.

Prospective Vision The CNV analysis becomes more and more widely used. There are two big challenges. One is the CNV analysis in single-cell and ctDNA (circulating tumor DNA). When the amount of DNA is small or the genome is incomplete, the CNV measurements will become difficult and we must adjust these effects to distinguish actual copy number deletion and artificial low DNA fractions in a specific region. Another challenge is the integrative analysis of CNV with other multi-omics data. The CNV is only one level of data in the complex system of cancer which involves dysfunctions at almost all levels of regulation including SNP, CNV, methylation, mRNA, microRNA, lncRNA (Long non-coding RNA), circRNA (Circular RNA), protein, PTM (post-translational modification), and metabolites. We should not only study them on each level, but also integrate them and get a systems level picture of how tumors develop and progress.

See also: Chromosome Rearrangements and Translocations. Metastatic SignaturesdThe Tell-Tale Signs of Metastasis. Mutations: Driver Versus Passenger.

Copy Number Variations in Tumors

451

Further Reading de Araujo Lima, L., Wang, K., 2017. PennCNV in whole-genome sequencing data. BMC Bioinformatics 18 (Suppl 11), 383. Hehir-Kwa, J.Y., Pfundt, R., Veltman, J.A., 2015. Exome sequencing and whole genome sequencing for the detection of copy number variation. Expert Review of Molecular Diagnostics 15 (8), 1023–1032. Huang, T., Yang, J., Cai, Y.-D., 2015. Novel candidate key drivers in the integrative network of genes, microRNAs, methylations, and copy number variations in squamous cell lung carcinoma. BioMed Research International 2015, 358125. Huang, T., Li, B.-Q., Cai, Y.-D., 2016. The integrative network of gene expression, microRNA, methylation and copy number variation in colon and rectal cancer. Current Bioinformatics 11 (1), 59–65. Li, B.Q., You, J., Huang, T., Cai, Y.D., 2014. Classification of non-small cell lung cancer based on copy number alterations. PLoS One 9 (2), e88300. Li, X.C., Liu, C., Huang, T., Zhong, Y., 2016. The occurrence of genetic alterations during the progression of breast carcinoma. BioMed Research International 2016, 5237827. Mermel, C.H., Schumacher, S.E., Hill, B., Meyerson, M.L., Beroukhim, R., Getz, G., 2011. GISTIC2.0 facilitates sensitive and confident localization of the targets of focal somatic copy-number alteration in human cancers. Genome Biology 12 (4), R41. Pirooznia, M., Goes, F.S., Zandi, P.P., 2015. Whole-genome CNV analysis: Advances in computational approaches. Frontiers in Genetics 6, 138. Qiu, F., Xu, Y., Li, K., Li, Z., Liu, Y., et al., 2012. CNVD: Text mining-based copy number variation in disease database. Human Mutation 33 (11), E2375–81. Xu, L., Hou, Y., Bickhart, D.M., Song, J., Liu, G.E., 2013. Comparative analysis of CNV calling algorithms: Literature survey and a case study using bovine high-density SNP data. Microarrays (Basel, Switzerland) 2 (3), 171–185. Zhang, N., Wang, M., Zhang, P., Huang, T., 2016. Classification of cancers based on copy number variation landscapes. Biochimica et Biophysica Acta (BBA) - General Subjects 1860 (11, Part B), 2750–2755.

Relevant Websites CNVkit, Genome-wide copy number from high-throughput sequencingdhttp://cnvkit.readthedocs.io/en/stable/. PennCNVdhttp://penncnv.openbioinformatics.org/en/latest/. CNVtools, CNVtoolsdhttp://www.bioconductor.org/packages/release/bioc/html/CNVtools.html. BreakDancerdhttp://breakdancer.sourceforge.net/. SCAN, SNP and CNV Annotation Databasedhttp://www.scandb.org/newinterface/.

Defective 5-Methylcytosine Oxidation in Tumorigenesis Gerd P Pfeifer and Seung-Gi Jin, Center for Epigenetics, Van Andel Research Institute, Grand Rapids, MI, United States © 2019 Elsevier Inc. All rights reserved.

Glossary DNA methylation The addition of methyl groups from the methyl group donor S-adenosyl-L-methionine onto carbon five of cytosine residues in DNA resulting in the formation of 5-methylcytosine. This enzymatic reaction is carried out by DNA methyltransferases and occurs preferentially at CpG sequences in most tissues. DNA demethylation The loss of 5-methylcytosine from DNA resulting in replacement of 5-methylcytosine with cytosine. Demethylation can occur passively by continuing DNA replication in the absence of DNA methyltransferases, or actively, by TET protein-mediated oxidation of the methyl group of 5-methylcytosine. Epigenetics The study of mechanisms that cause changes in gene expression based on chromatin-linked events that do not involve direct changes in the DNA sequence. IDH Isocitrate dehydrogenase, an enzyme of the tricarboxylic acid (TCA) cycle that provides the TET enzyme cofactor alphaketoglutarate. TET proteins Ten-Eleven-Translocation (TET) proteins are 5-methylcytosine oxidases that oxidize the methyl group of 5methylcytosine leading to sequential formation of 5-hydroxymethylcytosine, 5-formylcytosine and 5-carboxylcytosine. TDG Thymine DNA glycosylase, an enzyme that was initially reported to remove thymine from T/G mispairs. TDG has strong activity to excise 5-formylcytosine and 5-carboxylcytosine from DNA. The resulting abasic site is processed by base excision repair. After TET-initiated 5-methylcytosine oxidation and base excision repair involving TDG, the outcome of these reactions is DNA demethylation.

Introduction 5-Methylcytosine (5mC) is an enzymatically produced modified cytosine base that has been known to exist in mammalian DNA for about 70 years. These decades of research have provided important clues as to its presumed functional role but there is no complete understanding yet. Numerous correlative studies have shown that the presence of 5mC at gene control elements such as promoters and enhancers is not compatible with gene expression. 5mC can directly interfere with binding of transcription factors. This modified base is often associated with inactive, repressed chromatin characterized by lack of histone acetylation. In some genomic regions, 5mC tends to co-exist with nucleosomes methylated at lysine 9 of histone H3 (H3K9me3), for example at repetitive DNA regions, or with nucleosomes methylated at lysine 36 of histone H3 (H3K36me3), when found in gene bodies. One major role of 5mC seems to be the suppression of unwanted gene activity, in particular its presence is thought to ensure the silenced state of retroviral and repetitive elements in the genome. This function could be described as aiming to reduce transcriptional noise genome-wide leaving unmethylated genomic regions as recognizable landmarks associated with active genes. A surprising discovery in the field was made in 2009 when two research groups reported that 5mC is not the only modified base in mammalian genomes. An oxidized form of 5mC, 5-hydroxymethylcytosine (5hmC) was detected initially in embryonic stem cells and in certain types of neurons in the brain. The levels of 5hmC in brain tissue can be quite substantial and may represent up to 1% of all cytosines, being equivalent to a quarter to a third of all 5mC in neuronal cell types. In most other tissues, the levels of 5hmC range from about 005% to 0.2% of all cytosines, which is about 20–80 times lower than the levels of 5mC. Predominantly, 5hmC occurs at CpG dinucleotides. Like 5mC, which is a product of DNA methyltransferase activity, 5hmC is produced enzymatically. Mammalian genomes encode three evolutionary conserved proteins that can catalyze the conversion of 5mC to 5hmC by enzymatic oxidation. They are named TET1, TET2, and TET3. This name comes from the fact that TET1 was initially described as a gene involved in a translocation in leukemia involving chromosomes 10 and 11 (Ten-Eleven-Translocation 1, or TET1). The enzymatic reaction involves the transfer of molecular oxygen to the methyl group and requires the cofactors alpha-ketoglutarate (aKG), Fe2 þ and ascorbate, the latter being important to preserve the 2þ state of iron (Fig. 1). Although 5hmC is a modified DNA base that can be readily detected and quantitated by antibody-based assays or by mass spectrometry-based approaches in tissue samples, in vitro studies with purified TET proteins showed that the reaction does not stop at the 5hmC base, but can proceed to create 5-formylcytosine (5fC), and 5-carboxylcytosine (5caC) (Fig. 2). However, in cultured cells or in tissues, the latter oxidation products are only barely detectable suggesting that they are either not effectively produced in vivo or that they are eliminated rapidly. One plausible pathway to process 5caC bases would be a decarboxylation reaction leading instantaneously back to unmodified cytosine (Fig. 2). Decarboxylation of 5caC does not occur spontaneously to a significant extent. So far, decarboxylase activity has not been demonstrated as an enzymatic function of TET proteins, and 5caC DNA decarboxylase enzymes have not been identified. Instead, it was shown that 5fC and 5caC can be removed from DNA by the base excision repair (BER) pathway initiated by thymine DNA glycosylase (TDG). This enzyme was previously known

452

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65059-6

Defective 5-Methylcytosine Oxidation in Tumorigenesis

453

Fig. 1 Formation of 5-hydroxymethylcytosine by TET protein catalytic activity. The TET enzymes oxidize 5-methylcytosines at CpG sequences in DNA. They require oxygen, alpha-ketoglutarate, Fe2 þ and ascorbic acid as cofactors.

Fig. 2 Sequential enzymatic oxidation of 5-methylcytosine by TET proteins results in DNA demethylation. The TET enzymes oxidize 5mC to 5hmC, 5fC and 5caC at CpG sequences in DNA. 5hmC is observed as a relatively stable DNA base in cells. When 5hmC undergoes further oxidation, the pathway proceeds towards DNA demethylation via base excision repair that removes 5fC and 5caC from DNA. A 5caC decarboxylase is hypothetical and has not yet been identified.

to remove thymine from T/G mispairs in DNA, which are formed by spontaneous hydrolytic deamination of 5mC. This TDGinitiated DNA repair step seems to be important for cellular survival and genome integrity inasmuch as 5fC and 5caC are capable of arresting the progression of RNA and DNA polymerases on templates containing these oxidized bases. An alternative pathway for removal of 5caC bases that uses the SRAP1 autopeptidase-endonuclease has recently been described. It is still not well understood under what physiological conditions the TET-initiated 5mC oxidation pathway proceeds from 5hmC to the higher oxidized forms of 5mC resulting in complete elimination of the modified base (Fig. 2). The biological role of 5hmC (and of 5fC and 5caC) as specific signaling marks rather than them simply being intermediates in an active DNA demethylation process has also been discussed. Several candidate proteins that may interact with oxidized 5mC bases have been identified using a mass spectrometry-based screening approach. However, the biological roles of these proteins when recognizing oxidized 5mC bases are not understood. The first indication that the 5mC oxidation pathway may play an important role in tumorigenesis came from the discovery of common mutations in the TET2 gene in hematological malignancies. We will discuss these mutations as well as other genetic and epigenetic changes in tumors that affect the balance between 5-methylcytosine formation and 5-methylcytosine removal (DNA demethylation).

454

Defective 5-Methylcytosine Oxidation in Tumorigenesis

TET2 Mutations Around the same time when 5-hydroxymethylcytosine and the TET enzymes that produce this base modification were first reported, an analysis of hematological malignancies uncovered frequent mutations in the coding sequence of the TET2 gene. This comprehensive analysis of TET2 in over 400 hematopoietic tumor samples revealed TET2 somatic mutations in 7.6% of myeloproliferative neoplasms, 42% of chronic myelomonocytic leukemia (CMML) and in 12% of acute myeloid leukemia (AML) cases analyzed. Many additional studies confirmed the prevalence of TET2 mutations in AML, myelodysplastic syndrome (MDS) and in CMML. However, TET2 mutations are uncommon in solid tumors as revealed by numerous cancer genome sequencing studies. The TET2 mutations in hematological tumors are generally point mutations or deletions that lead to a lack of a functional protein, for example by truncation mutations or by point mutations in the catalytic domain. In many cases, one wildtype TET2 allele remains intact. The TET2 mutations are considered early events in the malignant transformation process. They also occur at low frequency during normal aging leading to clonal hematopoiesis. Tet2-deficient mouse models also show an expansion of myeloid progenitor cells leading to the accumulation of pre-leukemic cell clones and the development of myeloid malignancies resembling CMML, MPD-like myeloid leukemia, and MDS, similar to the human disease. Somewhat surprisingly, the same human hematopoietic malignancy may harbor mutations in TET2 and in the DNA methyltransferase gene DNMT3A. Such a finding suggests that TET2 and DNMT3A may have different genomic targets that are affected differentially by functional inactivation of either epigenetic enzyme. Mutation of both Tet2 and Dnmt3a in mouse models leads to more severe phenotypes than single mutations of these genes. It is likely that TET2 loss of function, even when only partial, will lead to an imbalance of the methylation–demethylation cycle and one expectation is that CpG methylation at TET2-targeted loci should be increased. However, published studies have shown both a predominant loss and a predominant gain of methylation in tumors harboring TET2 mutations. One reason for the seemingly disparate finding may be that the tumors analyzed carry additional mutations in other epigenetic modifier genes (e.g., DNMT3A, EZH2) and it is difficult to judge the impact of a single mutation on genomic methylation patterns. Ascorbic acid is an important co-factor for TET activities. Many standard cell culture media lack ascorbic acid, which could be one explanation for the extremely low levels of 5hmC generally found in cells propagated in tissue culture. Treatment of TET2-deficient cells with vitamin C enhances 5-hydroxymethylcytosine formation and suppresses human leukemic cell colony formation and leukemia progression. This effect is likely mediated by activation of TET3. TET2 and TET3, although they may also have unique gene targets, are believed to function redundantly in many tissues.

Mutations in TET1 and TET3 TET1 and TET3 are rarely mutated in human tumors and no such mutations were identified in an initial screening of a variety of hematological malignancies. According to the COSMIC database, TET1 is found mutated in a few percent, generally less than 5% of the cases of endometrial, stomach and intestinal tumors. TET3 mutations are only found very rarely in solid or hematological tumors. Although TET2 is expressed at high levels in cells of the hematopoietic system partially explaining its frequent mutation in hematological diseases, it is unknown why TET1 and TET3 are rarely affected by mutations in other tumor tissues.

IDH1 and IDH2 Mutations Isocitrate dehydrogenases are NAD(þ)- or NADP(þ)-dependent enzymes that catalyze the conversion of isocitrate to alphaketoglutarate by oxidation and decarboxylation reactions. These enzymes are associated with the tricarboxylic acid (TCA) cycle. Mutations at specific amino acids of IDH1 were initially found in human brain cancers including astrocytomas, oligodendrogliomas and some glioblastomas. They are most common in lower grade gliomas (grade II or grade III). Mutations in IDH1 or IDH2 are also seen in acute myeloid leukemias, cholangiocarcinomas and in chondrosarcomas. The most common IDH1 mutations affect an amino acid within the active site of the protein and changes codon 132 from arginine to histidine. The change results in a gain of function producing a neomorphic enzyme that produces 2-hydroxyglutarate instead of alpha-ketoglutarate. Indeed, 2hydroxyglutarate accumulates to high levels in IDH1-mutant cells. The theory proposes that 2-hydroxyglutarate is a competitive inhibitor of a number of alpha-ketoglutarate-dependent dioxygenase enzymes. In fact, the 2-hydroxyglutarate metabolite was shown to inhibit TET enzymatic activities leading to a defect in 5mC oxidation and thus to an accumulation of 5mC at genomic location that would normally be targeted by TET activities. As a result, IDH1-mutant gliomas are characterized by an overabundance of hypermethylated CpG islands (also called CpG island methylator phenotype or “CIMP”). In some instances, these hypermethylation events, for example when occurring at promoters of genes, will lead to silencing of the linked gene. One other class of dioxygenase enzymes that are effectively inhibited by 2-hydroxyglutarate are the large group of alpha-ketoglutarate-dependent histone lysine demethylases of the JmjC-domain family. The effect of IDH1 mutation will be an increase of the methylated state of different lysines on histone tails, for example, resulting in a potential dysregulation of gene expression. Since many lysine demethylases may be affected simultaneously, the disruption of the epigenetic state at specific gene loci is expected to be complex. IDH1 or IDH2 mutations are relative early occurrences in glioma pathogenesis. They are a very common event in lower grade gliomas occurring in up to 80% of astrocytomas and oligodendrogliomas. Different molecular pathways seem to follow IDH1/2

Defective 5-Methylcytosine Oxidation in Tumorigenesis

455

mutations during tumor progression. These events include the loss of chromosomes 1p and 19q and the acquisition of TERT promoter mutations in oligodendrogliomas and mutations of TP53 and ATRX in astrocytomas. On the other hand, mutation or amplification of the epidermal growth factor receptor gene (EGFR) occur rarely in IDH1-mutant brain tumors but are common in IDH1-wildtype primary glioblastomas. Mutations in IDH1/2 and TET2 do not co-occur in AML suggesting that the same downstream targets may be impacted by these mutations and that inhibition of TET2 activity by 2-hydroxyglutarate is indeed a key mechanism in disease pathogenesis.

Succinate and Fumarate Two successively operating mitochondrial enzymes of the TCA cycle, succinate dehydrogenase (SDH) and fumarate hydratase (FH) show heterozygous mutations in their genes that cause predispositions to two types of inherited cancer syndromes. SDH mutations are linked to paraganglioma (PGL) and phaeochromocytoma and FH mutations predispose to leiomyomatosis and renal cell carcinoma. Mutation of these enzymes will lead to accumulation of their substrate metabolites, succinate and fumarate and dysregulation of metabolic pathways. Similar to 2-hydroxyglutarate, these metabolites are also competitive inhibitors of alpha-ketoglutaratedependent dioxygenases, that is, lysine demethylases, 5-methylcytosine oxidases and prolyl hydroxylases. These events will likely lead to perturbations of epigenetic states of DNA and histones. However, these two genes are rarely mutated in other types of cancers.

General Loss of 5-Hydroxymethylcytosine in Human Solid Tumors When viewed across the broad spectrum of multiple types of solid tumors, mutations in TET2, IDH1/2, FH, and SDH are still relatively rare and frequently occur in only a few tumor types such as gliomas and AML. The vast majority of human tumor specimens do not contain these aberrations. However, in 2011 it was discovered that the major product of 5mC oxidase activity, 5hmC, is strongly depleted in all solid tumors analyzed. These tumors included cancers of the lung, breast, colon, liver, pancreas, prostate, melanoma and several others. When quantitated with sensitive techniques such as liquid chromatography coupled with tandem mass spectrometry, the extent of 5hmC depletion in tumors can reach 80%–90%. This degree of loss of 5hmC is much greater than the loss of 5mC, which often is only in the range of 10%–30% in tumors when compared to corresponding normal tissue. Therefore, loss of 5hmC during tumorigenesis does not simply reflect the loss of its precursor, 5mC. The loss of 5hmC may occur as an early event in carcinogenesis. For example, the loss is seen as a consequence of exposure of rodent liver to a nongenotoxic carcinogen, phenobarbital. Currently, it is unknown if other carcinogenic substances may also impair 5hmC maintenance in normal cells. Loss of 5hmC, as monitored by immunocytochemistry or immunofluorescence staining, is characteristic of malignant lesions and may be useful as a biomarker for cancer. For example, when used for assessing the malignant potential of melanocytic lesions in the skin, benign nevi show strong staining for 5hmC but this staining is lost in primary or metastatic malignant melanomas. Interestingly, this modified base is also reduced in proliferating normal tissue including, for example intestinal crypt stem cell compartments. The mechanisms that lead to a general depletion of 5hmC in tumors are not well understood and multiple pathways are probably involved. It has been known from earlier in vitro studies that the presence of 5hmC, in place of 5mC, on a parental DNA strand is incompatible with DNA methyltransferase 1 (DNMT1) activity that normally copies symmetrical methylation patterns at CpG sites shortly after DNA replication. This lack of DNMT1 activity on 5hmC-containing substrates leads to a loss of both 5mC and 5hmC in replicated daughter strand DNA molecules. However, in malignant tissue the global loss of 5hmC is much greater than the loss of 5mC so that other or additional mechanisms need to be invoked. A general reduction of TET gene expression, TET protein levels or TET enzymatic activity in tumor versus normal cells is one likely explanation. However, gene expression data from The Cancer Genome Atlas (TCGA) does not confirm a loss of TET gene expression across multiple tumor entities. To the contrary, TET gene expression levels are often increased in tumors. TET protein levels have rarely been measured in primary matched tumor-normal tissues and neither have their activities. There is also the possibility that there is reduced availability of co-factors in tumor tissues. For example, levels of ascorbic acid may be reduced in tumor patients or more specifically, in their tumor tissues. Metabolic reprogramming is a widespread phenomenon in tumor cells (e.g., the Warburg effect). As one possibility, the availability of the cofactor alpha-ketoglutarate may be limited in tumors. Equally important could be the maintenance of a correct balance between the dioxygenase enzyme cofactor alpha-ketoglutarate and their inhibitors such as succinate and fumarate. All three of these compounds are part of the TCA cycle and even subtle disturbances in the ratios of these metabolites may affect TET enzymatic activities. These parameters would need to be assessed systematically in normal and in tumor tissue. One other unresolved issue is whether the depletion of 5hmC in cancer is perhaps only a consequence of tumor formation or enhanced cell proliferation or whether these changes have a causative role in tumorigenesis. There are not many studies that have addressed these issues other than those based on inactivation of Tet genes. Tet2 inactivation in mice leads to an expansion of clonal hematopoiesis and emergence of premalignant clones as discussed earlier. Reintroduction of active TET2 into a melanoma cell line with low 5hmC levels led to suppression of melanoma growth and increased tumor-free survival in mice. Interestingly, combined loss of Tet1 and Tet2 in mice promotes B cell tumors, but not myeloid malignancies. However, germline deletion of Tet2 in combination with acute loss of Tet3 leads to a rapid, progressive leukocytosis, which developed within a few weeks into aggressive myeloid leukemia in 100% of the gene-targeted mice. These combinations of Tet genotypes are not usually found in human tumors;

456

Defective 5-Methylcytosine Oxidation in Tumorigenesis

however, the data leave open the possibility that other mechanisms of TET protein inactivation may occur and may be critically involved in human cancers as well.

Perturbation of DNA Methylation Patterns as a Consequence of Defective 5mC Oxidation If one considers the state of methylation at a CpG-rich target to be the result of a balance between de novo methylation and DNA demethylation initiated by TET proteins, then the expectation is that loss of TET activity will shift that balance towards 5mC. These predictions have generally proven to be correct in experimental systems in which one or more Tet genes have been disabled. For example, genome-wide mapping of 5hmC has shown that enhancers are enriched for this modification. The results imply that certain TET proteins may be targeted to enhancers and that they are critically involved in regulating enhancer DNA methylation and thus the enhancer’s accessibility and function. Indeed, when Tet2 was deleted in embryonic stem cells, extensive loss of 5hmC occurred at enhancers and this loss was accompanied by DNA CpG hypermethylation at the same enhancer sequences. Combined loss of all three Tet genes in mouse embryonic stem cells inhibits their differentiation potential. These studies further revealed that promoters became hypermethylated upon loss of TET activities and this loss led to a deregulation of genes implicated in embryo development and differentiation. In mouse embryos, loss of all three Tet genes leads to a gastrulation defect also involving hypermethylation of normally demethylated critical gene control elements. In patient samples, the contribution of TET2 mutations to DNA hypermethylation patterns is more difficult to appreciate because of the simultaneous occurrence of a number of other epigenetic modifier mutations in the same malignant cell populations. Cases of myeloproliferative neoplasms with TET2 mutations tend to show not only decreased levels of 5hmC but also distinct sets of hypermethylated genes. Hypermethylation of enhancers has also been reported for human AML cases with TET2 mutations. Therefore, similar as in mouse ES cells, loss of TET2 function in human hematopoietic and other somatic stem cells may be associated with a phenotype of enhancer hypermethylation and dysfunction that may contribute to the formation of leukemias and other malignancies (Fig. 3). It is also conceivable that the global loss of 5hmC in solid tumors, which occurs by unknown mechanisms, will lead to DNA hypermethylation of regulatory gene sequences. At a global level, the extent of 5hmC loss does not seem to correlate with the number of DNA hypermethylation events observed in a tumor. Whether loss of 5hmC formation at specific sites (e.g., in regions flanking promoters or near enhancers) will lead to DNA hypermethylation of the same targets is more difficult to assess. It requires genome-wide high-resolution mapping of 5hmC in normal and tumor tissues, but single base mapping of 5hmC in mammalian genomes is still prohibitively expensive. Some of the initial studies using these approaches suggest that the presence of 5hmC is indeed negatively correlated with DNA methylation changes.

Prospective Vision Perturbation of DNA methylation patterns in tumors, a phenomenon known for several decades, includes broad, genome-scale DNA hypomethylation and localized DNA hypermethylation at CpG-rich sequences. The loss of 5-hydroxymethylcytosine in

Fig. 3 Model for changes of 5hmC and 5mC in cancer. The diagram shows a hypothetical gene. Unmethylated CpG-rich sequences at enhancers and promoters in normal cells are protected from aberrant de novo methylation by TET-protein-catalyzed removal of erroneously introduced 5mC. In this scenario, TET-mediated 5mC oxidation leads to conversion of 5mC back to cytosine. Open circles, unmethylated CpG sequences; black circles, methylated CpG sequences (5mC), red circles, hydroxymethylated CpG sequences (5hmC). Loss of 5hmC near promoters and enhancers may be accompanied by de novo methylation of gene control elements in tumors.

Defective 5-Methylcytosine Oxidation in Tumorigenesis

457

tumors greatly exceeds the extent of the loss of 5-methylcytosine. This loss is not limited to tumors carrying mutations in TET genes or in IDH genes but occurs across a broad spectrum of malignancies. There is a need to understand the mechanisms of 5hmC loss since this process may be mechanistically linked to tumorigenesis. It is also tempting to speculate that there is an intrinsic connection between 5hmC perturbation and the changes in 5mC that are observed in tumors. Targeted genetic studies in mice will be needed to dissect the tissue-selective roles of specific TET gene inactivation in cancers other than those of the hematopoietic system. Manipulation of co-factor pathways relevant for TET protein function will also be required to understand the role of altered metabolism in the imbalance of DNA methylation and DNA demethylation pathways during tumor formation. Diagnostic assays assessing the DNA methylation status of a few specific genes are already in use for clinical tests with the goal of early cancer detection and patient management. Since widespread changes in 5hmC are also a ubiquitous characteristic of tumor genomes, similar assays analyzing 5hmC patterns in tissue or blood DNA samples could be developed for aiding cancer detection and diagnosis.

Acknowledgements Work of the authors was supported by NIH grant CA160965.

See also: DNA Methylation Changes in Cancer: Cataloguing. DNA Methylation Changes in Cancer: Mechanisms. Environmental Exposures and Epigenetic Perturbations. Mutations in DNA Methyltransferases and Demethylases. TCA Cycle Aberrations and Cancer.

Further Reading Abdel-Wahab, O., Mullally, A., Hedvat, C., Garcia-Manero, G., Patel, J., Wadleigh, M., Malinge, S., Yao, J., Kilpivaara, O., Bhat, R., Huberman, K., Thomas, S., Dolgalev, I., Heguy, A., Paietta, E., Le Beau, M.M., Beran, M., Tallman, M.S., Ebert, B.L., Kantarjian, H.M., Stone, R.M., Gilliland, D.G., Crispino, J.D., Levine, R.L., 2009. Genetic characterization of TET1, TET2, and TET3 alterations in myeloid malignancies. Blood 114, 144–147. Bowman, R.L., Levine, R.L., 2017. TET2 in normal and malignant hematopoiesis. Cold Spring Harbor Perspectives in Medicine 7, a026518. Dai, H.Q., Wang, B.A., Yang, L., Chen, J.J., Zhu, G.C., Sun, M.L., Ge, H., Wang, R., Chapman, D.L., Tang, F., Sun, X., Xu, G.L., 2016. TET-mediated DNA demethylation controls gastrulation by regulating lefty-nodal signaling. Nature 538, 528–532. Dawlaty, M.M., Breiling, A., Le, T., Barrasa, M.I., Raddatz, G., Gao, Q., Powell, B.E., Cheng, A.W., Faull, K.F., Lyko, F., Jaenisch, R., 2014. Loss of Tet enzymes compromises proper differentiation of embryonic stem cells. Developmental Cell 29, 102–111. Hashimoto, H., Liu, Y., Upadhyay, A.K., Chang, Y., Howerton, S.B., Vertino, P.M., Zhang, X., Cheng, X., 2012. Recognition and potential mechanisms for replication and erasure of cytosine hydroxymethylation. Nucleic Acids Research 40, 4841–4849. He, Y.F., Li, B.Z., Li, Z., Liu, P., Wang, Y., Tang, Q., Ding, J., Jia, Y., Chen, Z., Li, L., Sun, Y., Li, X., Dai, Q., Song, C.X., Zhang, K., He, C., Xu, G.L., 2011. Tet-mediated formation of 5-carboxylcytosine and its excision by TDG in mammalian DNA. Science 333, 1303–1307. Hon, G.C., Song, C.X., Du, T., Jin, F., Selvaraj, S., Lee, A.Y., Yen, C.A., Ye, Z., Mao, S.Q., Wang, B.A., Kuan, S., Edsall, L.E., Zhao, B.S., Xu, G.L., He, C., Ren, B., 2014. 5mC oxidation by Tet2 modulates enhancer activity and timing of transcriptome reprogramming during differentiation. Molecular Cell 56, 286–297. Ito, S., Shen, L., Dai, Q., Wu, S.C., Collins, L.B., Swenberg, J.A., He, C., Zhang, Y., 2011. Tet proteins can convert 5-methylcytosine to 5-formylcytosine and 5-carboxylcytosine. Science 333, 1300–1303. Jin, S.G., Kadam, S., Pfeifer, G.P., 2010. Examination of the specificity of DNA methylation profiling techniques towards 5-methylcytosine and 5-hydroxymethylcytosine. Nucleic Acids Research 38, e125. Jin, S.G., Jiang, Y., Qiu, R., Rauch, T.A., Wang, Y., Schackert, G., Krex, D., Lu, Q., Pfeifer, G.P., 2011. 5-Hydroxymethylcytosine is strongly depleted in human cancers but its levels do not correlate with IDH1 mutations. Cancer Research 71, 7360–7365. Kohli, R.M., Zhang, Y., 2013. TET enzymes, TDG and the dynamics of DNA demethylation. Nature 502, 472–479. Kriaucionis, S., Heintz, N., 2009. The nuclear DNA base 5-hydroxymethylcytosine is present in Purkinje neurons and the brain. Science 324, 929–930. Lian, C.G., Xu, Y., Ceol, C., Wu, F., Larson, A., Dresser, K., Xu, W., Tan, L., Hu, Y., Zhan, Q., Lee, C.W., Hu, D., Lian, B.Q., Kleffel, S., Yang, Y., Neiswender, J., Khorasani, A.J., Fang, R., Lezcano, C., Duncan, L.M., Scolyer, R.A., Thompson, J.F., Kakavand, H., Houvras, Y., Zon, L.I., Mihm Jr., M.C., Kaiser, U.B., Schatton, T., Woda, B.A., Murphy, G.F., Shi, Y.G., 2012. Loss of 5-hydroxymethylcytosine is an epigenetic hallmark of melanoma. Cell 150, 1135–1146. Lorsbach, R.B., Moore, J., Mathew, S., Raimondi, S.C., Mukatira, S.T., Downing, J.R., 2003. TET1, a member of a novel protein family, is fused to MLL in acute myeloid leukemia containing the t(10;11)(q22;q23). Leukemia 17, 637–641. Moran-Crusio, K., Reavie, L., Shih, A., Abdel-Wahab, O., Ndiaye-Lobry, D., Lobry, C., Figueroa, M.E., Vasanthakumar, A., Patel, J., Zhao, X., Perna, F., Pandey, S., Madzo, J., Song, C., Dai, Q., He, C., Ibrahim, S., Beran, M., Zavadil, J., Nimer, S.D., Melnick, A., Godley, L.A., Aifantis, I., Levine, R.L., 2011. Tet2 loss leads to increased hematopoietic stem cell self-renewal and myeloid transformation. Cancer Cell 20, 11–24. Pfeifer, G.P., Kadam, S., Jin, S.G., 2013. 5-hydroxymethylcytosine and its potential roles in development and cancer. Epigenetics & Chromatin 6, 10. Rasmussen, K.D., Jia, G., Johansen, J.V., Pedersen, M.T., Rapin, N., Bagger, F.O., Porse, B.T., Bernard, O.A., Christensen, J., Helin, K., 2015. Loss of TET2 in hematopoietic cells leads to DNA hypermethylation of active enhancers and induction of leukemogenesis. Genes & Development 29, 910–922. Schuermann, D., Weber, A.R., Schar, P., 2016. Active DNA demethylation by DNA repair: Facts and uncertainties. DNA Repair (Amst) 44, 92–102. Scourzic, L., Mouly, E., Bernard, O.A., 2015. TET proteins and the control of cytosine demethylation in cancer. Genome Medicine 7, 9. Spruijt, C.G., Gnerlich, F., Smits, A.H., Pfaffeneder, T., Jansen, P.W., Bauer, C., Munzel, M., Wagner, M., Muller, M., Khan, F., Eberl, H.C., Mensinga, A., Brinkman, A.B., Lephikov, K., Muller, U., Walter, J., Boelens, R., van Ingen, H., Leonhardt, H., Carell, T., Vermeulen, M., 2013. Dynamic readers for 5-(hydroxy)methylcytosine and its oxidized derivatives. Cell 152, 1146–1159. Tahiliani, M., Koh, K.P., Shen, Y., Pastor, W.A., Bandukwala, H., Brudno, Y., Agarwal, S., Iyer, L.M., Liu, D.R., Aravind, L., Rao, A., 2009. Conversion of 5-methylcytosine to 5hydroxymethylcytosine in mammalian DNA by MLL partner TET1. Science 324, 930–935. Thomson, J.P., Lempiainen, H., Hackett, J.A., Nestor, C.E., Muller, A., Bolognani, F., Oakeley, E.J., Schubeler, D., Terranova, R., Reinhardt, D., Moggs, J.G., Meehan, R.R., 2012. Non-genotoxic carcinogen exposure induces defined changes in the 5-hydroxymethylome. Genome Biology 13, R93. Tsagaratou, A., Lio, C.J., Yue, X., Rao, A., 2017. TET methylcytosine oxidases in T cell and B cell development and function. Frontiers in Immunology 8, 220.

458

Defective 5-Methylcytosine Oxidation in Tumorigenesis

Valinluck, V., Sowers, L.C., 2007. Endogenous cytosine damage products alter the site selectivity of human DNA maintenance methyltransferase DNMT1. Cancer Research 67, 946–950. Wu, X., Zhang, Y., 2017. TET-mediated active DNA demethylation: Mechanism, function and beyond. Nature Reviews. Genetics 18, 517–534. Zhang, X., Su, J., Jeong, M., Ko, M., Huang, Y., Park, H.J., Guzman, A., Lei, Y., Huang, Y.H., Rao, A., Li, W., Goodell, M.A., 2016. DNMT3A and TET2 compete and cooperate to repress lineage-specific transcription factors in hematopoietic stem cells. Nature Genetics 48, 1014–1023.

Diabetes and Cancer Rachel Dankner, Sheba Medical Center, Tel Hashomer, Israel; Sackler Faculty of Medicine, School of Public Health, Tel Aviv University, Tel Aviv, Israel; and The Feinstein Institute for Medical Research, Manhasset, NY, United States © 2019 Elsevier Inc. All rights reserved.

Introduction Diabetes is a chronic progressive metabolic disease, defined by hyperglycemia. Diabetes is considered a worldwide epidemic. The world prevalence of diabetes for all ages is expected to double to 4.4% from the year 2000 to 2030. This means that by the year 2030 there will be 366 million people with diabetes living in the world. Older age is an important risk factor and the increase in the proportion of people > 65 years of age across the world is an important reason for future projections on diabetes prevalence. Obesity is another important factor and given the increasing prevalence of obesity, it is likely that these figures provide an underestimate of future diabetes prevalence. Diabetes is more prevalent in men, but there are more women with diabetes in the world due to women’s greater life-expectancy. In his landmark article, published in Science in 1956, Otto Warburg linked metabolism with cancer. In the decades since, numerous epidemiological studies have demonstrated links between diabetes and cancer. Several publications during the year 2009 suggested that treatment with the insulin analogue, glargine, may increase the risk of cancer incidence. These studies, together with criticisms of them, raised interest in the relationship between diabetes and cancer, which went well beyond the use of glargine. In December 2009, representatives from American and European diabetes and cancer organizations met and produced a consensus report. This document supported associations of diabetes with increased risk of certain cancers, and corroborated cancer as a complication of diabetes. Hyperglycemia, hyperinsulinemia, adiposity, subclinical inflammation, and diabetes treatment were all identified for their involvement in the relationship between the two diseases. More research was called for to elucidate the mechanistic relationships between the diseases, as well as the effects of diabetes medications on carcinogenesis. Since the issuance of the consensus report, the publication of observational and clinical studies supporting associations between diabetes and cancer has soared. In this article we will first present the epidemiological evidence for associations between diabetes and all-site as well as site-specific cancers. Next, we will present possible mechanisms that may underlie associations of diabetes with cancer in general, as well as with a number of specific sites. Glucose lowering medications (the term used in this article for anti-diabetes drugs) will be discussed as possible confounders in the association of diabetes with cancer, as well as causative factors for carcinogenesis. We have tried to take a critical approach and urge our readers to do the same.

The Epidemiological Evidence Epidemiological studies investigating associations of diabetes with cancer incidence or mortality are abundant. Diabetes is an exposure that cannot be imposed on study participants; thus, randomized controlled trials (RCTs) cannot be conducted to examine its association with cancer. Here, we present studies published since 2013, with an emphasis on systematic reviews and well-designed observational studies. Most of the meta-analyses conducted described considerable heterogeneity among included studies; and many lacked adjustments for potentially confounding factors. More recent studies more often adjusted for potentially confounding variables and some of them employed time-dependent approaches, which accounted for the date of diabetes diagnosis and assessed cancer incidence from only a certain period thereafter.

All-Site Cancer Estimated population attributable fractions (PAFs) for diabetes, as well as for high body mass index (BMI, > 25 kg/m2) to cancer incidence in the year 2012 were published for 12 cancers, for 175 countries, with stratification by age and sex. Overall, the contribution of diabetes was estimated as 2%. However, higher PAFs were calculated for cancers of the liver (14.5%), pancreas (12.8%), gallbladder (7.8%), and colorectum (2.9%) in men; and for the liver (15.8%), pancreas (12.6%), endometrium (10.8%), gallbladder (7.4%), colorectum (2.8%), and breast (2.2%) in women. The overall estimated PAF for high BMI was calculated as 3.9%, almost twice that of diabetes. During the 10–32-year lag (starting in 1980–2002 and spanning to 2012) between exposure to diabetes or high BMI, a disproportionate number of those with high BMI evidently converted to diabetes, and this may have further contributed to their risk for cancer incidence, thus indicting BMI as a double whammy. Substantial differences were observed across geographical regions, supporting possible effects of factors other than diabetes and overweight. Though not investigated in that study, these could include differences in genetic susceptibility, in fetal and childhood nutrition and growth, in the detection and treatment of diabetes, and in lifestyle factors including nutrition and physical activity. Most meta-analyses have reported positive associations of diabetes with cancer, yet some have only included a small number of cancer cases. A cohort of 2.3 million individuals followed for 11 years, using a time-dependent approach and a 2 year lag after diabetes diagnosis, reported hazard ratios (HRs) for all-site cancer of 1.42 (95% confidence interval (CI): 1.38, 1.46) in diabetic women and 1.33 (95%CI: 1.29, 1.36) in diabetic men, after adjusting for age, ethnic group, sociodemographic status (SES), and BMI. Several additional population-based studies have reported positive associations of diabetes with all-site cancer incidence. In studies

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65040-7

459

460

Diabetes and Cancer

that stratified by sex, associations were generally stronger in women, but this is apparently due in large part to the lower incidence of prostate cancer reported for diabetic men, as well as to the increased risks of female-specific cancers. For non-sex-specific cancers, risks were generally similar between men and women. Potentially confounding factors were considered to varying degrees. The vast majority of studies that distinguished the type of diabetes, investigated type 2 (T2D). Higher risks of cancer were generally found for T2D than type 1 diabetes (T1D).

Site-Specific Cancers In this section, we discuss meta-analyses of associations between diabetes and site-specific cancers, presented alphabetically, as well as a number of selected observational studies.

Bladder cancer Differences between the sexes have been reported, with some studies finding a risk of bladder cancer in diabetic women and not men, and others finding the risk only in diabetic men and not women. Smoking may attenuate the risk. A number of studies have found an increased risk of bladder cancer, only or predominantly, during the first year following diabetes diagnosis. Such findings support the possibility that increased medical surveillance may explain observed associations of diabetes with bladder cancer. Nonetheless, a large historical population study, with a 2-year lagged analysis since diabetes diagnosis and 11 years follow up found a 35% increased risk in men and 70% in women, after adjusting for age, SES, and smoking.

Brain cancer Brain cancer incidence was reported as 28% increased in diabetic women and 52% increased in diabetic men at 2–11 years after diabetes diagnosis.

Breast cancer Several, but not all studies on the matter have shown an association of diabetes with an increased risk of breast cancer. However, these findings may not be relevant to Asian, Hispanic and African American populations, and this may be due, at least in part, to differences in mean body weight and fat distribution between populations. A large population study showed a 21% increased risk of breast cancer in postmenopausal, but not premenopausal women at 2–11 years after diabetes diagnosis and after adjustments for BMI, SES, ethnicity, and parity (HR ¼ 1.21 95%CI 1.12, 1.30). A number of studies have shown that the association of diabetes with breast cancer may depend on the type of breast cancer. Triple-negative breast cancer tumors and estrogen receptor negative tumors have been associated with increased risk.

Cancer of unknown primary Cancers of unknown primary (CUP) are metastatic malignant tumors for which the primary tumor is not identified; the prognosis of these cancers is generally very poor. Tobacco smoking is the only established risk factor for CUP. An association of diabetes with increased risk of CUP has been reported, particularly when the metastasis was to the liver and respiratory systems.

Colorectal cancer Several studies have reported an association of diabetes with increased risk of colorectal cancer, in the range of 21%–26%. Some have found differences in risk between the sexes. A large population study with a time-dependent analysis reported 44% increased HR in women and 53% in men, for the 2–11 years after diabetes diagnosis, controlling for BMI. Another time-dependent analysis found an association of longer duration of obesity with increased risk of colorectal cancer, after adjusting for age, sex, smoking, alcohol consumption, and statin use in the previous 6 months. That study did not find an assocition with coloretal cancer of diabetes treatment stage; the latter was assessed as 90-day intervals of treatment with insulin or with one or more non-insulin glucose lowering medications alone or in combination. Some studies have shown differences in colon cancer risk, according to the location of the malignancy. However, the data are inconclusive, as both proximal and distal cancer have been found to be of greater risk.

Endometrial cancer Several studies have found an association of diabetes with increased risk of endometrial cancer, which did not always remain statistically significant after adjusting for BMI.

Esophageal cancer A number of studies have found a positive association of T2D with the risk of esophageal cancer; however, this association appears not to present in Asian populations. Male sex is a risk factor for esophageal cancer.

Gallbladder cancer Several studies have found an increased risk of gallbladder cancer, similarly in men and women with diabetes, and independent of BMI, smoking, and a history of gallstones. Though, the possibility arises of confounding by alcohol consumption, the association was shown to persist in a study conducted in Israel, where alcohol consumption is not considered an important confounding factor.

Diabetes and Cancer

461

Gastric/stomach cancer Most, but not all studies on the matter have found an association of diabetes with gastric cancer. This risk may differ between the sexes, and may be particularly strong in East Asian populations.

Head and neck cancers Associations of T2D with increased incidence of oral cancer, oropharyngeal cancer, and nasopharyngeal cancer have been reported, compared to non-diabetic individuals, matched by sex, age, obesity, and a number of comorbidities.

Liver cancer Considerable evidence has accumulated for an association of diabetes with liver cancer, which appears to be stronger in non-Asian than Asian populations. Tobacco smoking and increased BMI, but not alcohol, have been identified as confounding factors. The association remains strong, even after excluding the first 2 years after diabetes diagnosis.

Lung cancer Several studies have shown a robust association of diabetes with increased risk of lung cancer, particularly among women. However, the risk may be lower in Asian populations.

Ovarian cancer Data seems conclusive for an association of diabetes with ovarian cancer, though the strength of the association varies between studies. Higher risks were found in studies published later and conducted outside of the United States or Europe, in studies that did not adjust for BMI or smoking status, and for T1D than T2D. An elevated risk of ovarian cancer was found to persist in diabetic women after adjustment for parity and the exclusion of women who had hysterectomies.

Pancreatic cancer The pancreas, together with the liver, is directly responsible for glucose metabolism and insulin secretion. Thus, if diabetes is indeed associated with cancer via its metabolic pathways, the association would be expected to be the strongest in these organs. Indeed, the incidence of pancreatic cancer has consistently been shown to be elevated in the diabetic population. The population attributable fraction of diabetes to pancreatic cancer was found to be 13.5% and the BMI adjusted 2–11-year risk reportedly three times greater in the incident diabetic than in the non-diabetic population. The increased risk of pancreatic cancer is strong, even after excluding the first 2 years following diabetes diagnosis; and was shown to remain significant even after 20 years, and after adjusting for BMI and smoking.

Prostate cancer Several studies have reported a lower risk for prostate cancer among men with than without diabetes, in both T1 and T2 diabetes. The risk appears to be lower in Europeans and Americans, and higher in Asians.

Thyroid cancer The data are not conclusive regarding an association of diabetes with thyroid cancer, neither for men or women.

Site-specific cancersdsummary As the evidence presented here shows, considerable differences were found between studies for associations of diabetes with several types of cancer. While confounding factors and differences in study design likely account for some of the differences observed, real differences seem to exist between populations, and between the sexes.

Pathophysiology and Mechanisms Relevant to the Association of Diabetes With Cancer in General Four main explanations are possible for the associations observed between diabetes and cancer; the relevance of these explanations differ among cancer types. Causality is the relationship by which the pathology of diabetes disease increases the risk of cancer. Since diabetes is diagnosed by hyperglycemia, we will consider direct cause as an effect of hyperglycemia on cancer risk. A second explanation for the associations observed between diabetes and cancer relates to shared risk factors. Insulin resistance and hyperinsulinemia, obesity, chronic inflammation, oxidative stress and alterations in sex hormones evidently comprise a metabolic milieu that predisposes to diabetes and that is conducive to cancer (Fig. 1). Shared lifestyle factors: physical inactivity, poor nutrition, smoking and excessive alcohol intake, as well as genetic susceptibility, may impact such metabolic milieu. A third possible explanation for the observed relationship between diabetes and cancer is reverse causality, in which the cancer pathology precedes the diabetes, and may even be the cause for hyperglycemia, though cancer is diagnosed only after diabetes. A fourth possibility is a spurious association, perceived due to biases or confounding factors, rather than a causal association between the diseases. We will discuss each of the four possible explanations for the observed associations between diabetes and cancer, considering cancer as a single entity. We will then present specific mechanisms by which diabetes may increase the risk of cancer, as pertains to specific cancers.

462

Diabetes and Cancer

Glucose lowering medications

DIABETES (Hyperglycemia)

A METABOLIC MILIEU THAT PREDISPOSES TO DIABETES AND THAT IS CONDUCIVE TO CANCER Obesity Insulin resistance

Hyperinsulinemia Chronic Inflammation Altered levels of sex hormones

CANCER

SHARED GENETIC RISK FACTORS

SHARED LIFESTYLE RISK FACTORS Poor nutrition Physical inactivity Smoking Excessive alcohol

Oxidative stress

Fig. 1 The association of diabetes with cancer. The lines are dotted to indicate that the relationships are suspected and not verified, as they are based on observational studies. The thicker the line, the more evidence is currently available in support of the relationship.

Diabetes as a Risk Factor for Cancer Distinguishing the impact on cancer of hyperglycemia from influences of pathophysiological conditions that accompany diabetes may be complex. The Warburg effect describes how irreversible damage to cell respiration leads some normal cells to become more proficient in using glucose for anaerobic glycolysis; these cells convert to highly proliferating undifferentiated cancer cells. However, this supposition does not infer that an increased supply of glucose, that is, hyperglycemia, promotes the development of cancer. Nonetheless, fatless hyperglycemic mice were shown to develop more aggressive tumors than normoglycemic mice. While this model neutralized the effect of obesity, it did not exclude hyperinsulinemia. Among diabetic mice that were insulin deficient, increased tumor growth was not observed. Taken together, while glucose is evidently essential for tumor growth, hyperglycemia in itself does not necessarily increase such growth. Still, knowledge accumulated since Warburg’s publication has substantiated the perception of cancer as a metabolic disorder. A number of publications have described direct effects of hyperglycemia on tumorigenesis. In nonmalignant human breast cancer cells, increased glucose uptake activated known oncogenic signaling pathways. Further, elevated glucose was shown to enhance the signaling of WNT/b-catenin in cancer cells, and this regulates cell proliferation and survival. In a mouse model, the glucose transporter GLUT1 was demonstrated as necessary for mammary tumorigenesis. Hyperglycemia was shown to promote features associated with cancer stem cells in premalignant and malignant pancreatic ductal epithelial cells, by activating signaling of the cytokine TGF-b. Hyperglycemia promotes low-grade inflammation; and inflammatory processes are part of the etiology of cancer. Chronic hyperglycemia causes irreversible glycation and oxidation of proteins and lipids in the extracellular matrix; this results in the formation of advanced glycation end-products (AGE). Interactions of AGE and their receptors (RAGE) increase the generation of reactive oxygen species (ROS) in adipocytes, thus inhibiting glucose uptake and stimulating insulin resistance. Extracellular matrix glycation and RAGE activation may promote tumor growth by damaging proteins and DNA; this provides a link for the association between diabetes and cancer. Further evidence of an effect of elevated glucose on tumorigenesis by means of inflammation is the demonstration of increased phosphorylation of p38 MAPK, an essential kinase for inflammation, in pancreatic tumors from diabetic animals and in pancreatic cancer cells treated with high glucose. The balance between oxidative and redox reactions in tumorigenesis is apparently complex. Along this line, James Watson, the Nobel prize winner, proposed that deficiencies rather than excesses in reactive oxygen species (ROS) may be the basis of diabetes and cancer, as well as other chronic diseases. He referred specifically to an insufficiency of the endoplasmic reticulum to generate the oxidative redox potential necessary to form the disulphide bonds that stabilize active proteins. Taken together, disruptions in oxidative or redox systems may link between diabetes and cancer. Poorly controlled diabetes may perpetuate a pro-inflammatory state that may precede hyperglycemia, as will be detailed further below. The demonstration of higher levels of ROS among diabetic individuals with poorer glycemic control supports a role of hyperglycemia in pro-inflammation and oxidative stress. Further, poor glycemic control was found to be associated with serum gamma-glutamyltransferase and high sensitivity C-reactive protein levels, which are markers of inflammation. Nonetheless, individuals with poor glycemic control may tend to have more acute hyperinsulinemia, which is known to be associated with

Diabetes and Cancer

463

pro-inflammatory effects and oxidative stress, as discussed below. This exemplifies the difficulty in distinguishing the effect of hyperglycemic from other pathophysiological components of diabetes. The epidemiological evidence is inconclusive regarding a possible association of glucose control and the risk of cancer. While some studies have shown positive associations with any or site-specific cancers, others found no association. However, these studies generally did not consider factors that are relevant to the clinical manifestations and treatment of diabetes. Studies on metabolic status in populations without known diabetes are not confounded by glucose lowering medications. Increased overall cancer risk has been reported in individuals without known diabetes yet with elevated HbA1c levels, as well as in diabetic patients. However, the great differences between cancer sites that have been reported for associations of HbA1c and cancer risk, highlight the importance of investigating cancer sites separately, in regard to associations with glycemia.

Shared Risk Factors Between Diabetes and Cancer In individuals with diabetes, a number of factors interact with each other, generally before the diagnosis of diabetes; and separately or in combination, they may promote malignancy. While these factors are related to the development of diabetes, due to their precedence, we will consider them as entities shared by diabetes and cancer, rather than as the basis for a direct causal relationship between diabetes and cancer.

Obesity Obesity is a major risk factor for both diabetes and cancer. Links between obesity and diabetes are complex and not fully understood; while most persons with T2D are obese, most obese persons do not have T2D. Certain cancers have long been associated with obesity. Already in 1980, endometrial cancer was found to be associated with excess body weight in postmenopausal women; and with increased levels of estradiol (E2) and decreased levels of sex hormone-binding globulin (SHBG). Among individuals with T2D, associations have been demonstrated of excess body weight with all cancer and with a number of site-specific cancers. Adipocyte (fat) cells are central to energy storage, and particularly responsive to insulin. In addition to their metabolic role, adipocytes serve as endocrine cells that secrete a wide variety of hormones, among them cytokines, including tumor necrosis factor-alpha (TNF-a); steroid hormones, including estrogen; and adipokines, including leptin and adiponectin. Abnormal secretion of these hormones provides a link of obesity with diabetes and cancer. Obesity is associated with increased leptin and decreased adiponectin levels; these changes in endocrine function appear to have a role in cancer development. Insulin, estrogens, and TNF-a stimulate the release of leptin. Adiponectin regulates glucose metabolism, increases insulin sensitivity, and reduces the production of inflammatory cytokines.

Insulin resistance and hyperinsulinemia Obesity, particularly abdominal obesity, leads to insulin resistance; insulin resistance can also contribute to obesity, and is a precursor of T2D. Insulin resistance is followed by compensatory hyperinsulinemia; this leads to beta cell compensation, which increases insulin production, resulting in hyperinsulinemia. The association between hyperinsulinemia and cancer was first hypothesized in relation to colon cancer; and was based on the recognition that similar factors are involved in hyperinsulinemia and colon cancer risk: namely, obesity, and abdominal obesity in particular; physical inactivity; and a low fiber, high sugar diet; the latter raises insulin levels by increasing the rate of blood glucose elevation and the blood glucose load. A number of epidemiological studies have substantiated the relationship between hyperinsulinemia and cancer. Moreover, factors related to hyperinsulinemia such as Cpeptide, a biomarker of pancreatic insulin secretion and serum insulin growth factor (IGF)-1 were found to be associated with increased cancer risk. A number of mechanisms have been proposed for the contribution of insulin resistance and hyperinsulinemia to cancer development. In addition to the metabolic effects of insulin and IGF, these hormones, together with their receptors and signal transduction networks, are recognized to have mitogenic effects. The binding of insulin receptor isoform A (IR-A) to IGF-II was described to result in mainly mitogenic effects; whereas, the binding of the same isoform to insulin results in mainly metabolic effects. Increased expression of IR-A has been detected in thyroid, colon, and breast tumors. Hyperinsulinemia impairs downregulation of insulin receptors in cancer cells; and affects signaling pathways of insulin and IGF-1, as well as plasma-free amino acid profiles; such activities may promote the development of both cancer and diabetes. Insulin and IGF-1, separately and in combination, increased cell proliferation and decreased apoptosis in colon cancer cells; and activated extracellular-signal regulated kinase 1/2 and c-Jun N-terminal kinase. All these activities are considered to contribute to the development of colon cancer. Data of cancer incidence among individuals with T1D can help discriminate the roles of hyperglycemia and hyperinsulinemia, since only hyperglycemia and not hyperinsulinemia presents in T1D. A study of five nationwide diabetes registrars in European reported higher incidence of cancers of the stomach, liver, pancreas, endometrium and kidney in persons with T1D, compared to the general population; yet the rates were lower than for persons with T2D. These data support a role for hyperglycemia in cancer incidence, yet infer that other factors are evidently involved. Effects of hyperglycemia and hyperinsulinemia may be additive or even synergistic. For example, the pro-inflammatory effect of hyperinsulinemia and hyperglycemia positively interact. Further, both hyperglycemia and hyperinsulinemia promote DNA

464

Diabetes and Cancer

damage; and this has been proposed as a link in the relationship between diabetes and cancer: the mutations generated from DNA damage can lead to carcinogenesis. Disruptions in circadian rhythms that regulate metabolic processes has also been attributed a role in carcinogenesis.

Altered concentrations of sex hormones Hyperinsulinemia affects concentrations of endogenous sexual hormones, and this may contribute to the development of both diabetes and cancer. High levels of insulin in the blood decrease the hepatic synthesis of SHBG, which leads to increased circulation of E2. Studies in both sexes have reported an inverse association of T2D with SHBG; and a positive association with E2. In women, elevated E2 has been shown to be associated with breast, endometrial, ovarian, and cervical cancers.

Oxidative stress and inflammation While hyperglycemia induces oxidative stress, which causes insulin resistance that results in hyperinsulinemia, as described above, hyperinsulinemia also contributes to oxidative stress and low-grade inflammation. Associations are well established of proinflammatory agents with diabetes, as well as with several types of cancer. Apparently, in reciprocal relationships of chronic inflammation with both diabetes and cancer, inflammation may precede or be induced by diabetes and its associated conditions, as well as precede or be induced by oncogenic changes. A number of mediators between insulin resistance and inflammation have received attention for their potential as links between diabetes and cancer. Elevated lactate promotes insulin resistance and inflammation, and has thus been coined “an interaction hub between diabetes and cancer.” Similarly, the adipokine, chemerin, has been identified as a link between diabetes and cancer due to its involvement in inflammation and insulin resistance. Specifically, chemerin activates an inflammatory response, which increases oxidative stress in adipose tissue and thus contributes to insulin resistance. Further evidence for inflammation as a mediator in the link between diabetes and cancer is the reported association of increased concentration of growth differentiation factor 15 (GDF-15) with cancer, among individuals with type 2 diabetes. The stress responsive cytokine GDF-15 was found to be increased in individuals before type 2 diabetes diagnosis, yet not to be an independent predictor of diabetes. In addition, considerable evidence has accumulated for a role of the nuclear receptor transcription factor peroxisome proliferator activated receptor-g (PPARg) at the crossroad of obesity, diabetes, and several cancers. PPARg is involved in the regulation of insulin and adipokine production and secretion; and of inflammatory pathways and differentiation. Thiazolidinediones, including rosiglitazone and pioglitazone, are synthetic ligands of PPARg, which are used for the treatment of diabetes and whose pathways have been suggested to have potential benefit in the development of cancer therapies.

Genetically predisposing factors Genetic factors may have both direct and indirect effects on the relationship between diabetes and cancer. Single nucleotide polymorphisms (SNPs) that were identified from genome-wide association studies as related to T2D, were associated with increased risk of a number of cancer types in diverse populations. Genetic factors may also affect the relationship between diabetes and cancer indirectly, by affecting shared predisposing factors. Along this line, the evidence does not support an independent relationship of the fat mass and obesity-associated (FTO) gene variant with cancer risk. Nonetheless, this variant may affect diabetes and cancer risk by its effect on obesity.

Lifestyle factors: diet, physical activity, alcohol consumption, smoking A number of lifestyle factors have been identified as risk factors for both diabetes and cancer; and, as such, may contribute to obesity, as well as to other aspects of the metabolic milieu conducive to cancer, described above and depicted in Fig. 1. Poor nutrition can contribute to imbalances in oxidation/redox systems, which can contribute to pro-inflammatory processes; and these, as discussed above, may increase risks of chronic illnesses including diabetes and some types of cancer. The role of diet in cancer development is particularly important to colorectal cancer, due to changes in the gut microbiota. Among the dietary factors that may be involved in the association between gut microbiota and the development of colorectal cancer are obesity, dietary patterns, and the consumption of red meat, sulfur, and fiber. Evidence is robust for the benefit of physical activity to both primary and secondary prevention of chronic diseases, including diabetes and cancer. Among individuals with diabetes, cancer incidence was increased among those reporting low levels of physical activity (< 2 h weekly), as well as among those who were overweight or obese, but not among those reporting high levels of physical activity ( 2 h weekly). According to James Watson’s hypothesis, the benefit of physical activity to diabetes and cancer, as well as to other chronic diseases is in creating redox potential; specifically, the generation of ROS that oxidize free sulfhydryl groups of cysteine into the disulphide bonds that stabilize physiologically active proteins. Elevated hazard ratios for diabetes and cancer were reported among women with a 20% increase in energy consumption; and decreased risks among women with a 20% increase in activity-related energy expenditure. Further, decreased risks of 25%, 14%, and 21% for diabetes, breast cancer, and colon cancer, respectively, were reported for individuals who were highly active (total weekly activity  8000 metabolic equivalents (METs)) compared with inactive individuals (< 600 METs). Physical activity may affect breast cancer risk through biological pathways that involve adiposity, sex hormones, insulin resistance, adipokines, and chronic inflammation. Several studies have reported an association of moderate alcohol consumption with a reduced risk for diabetes. However, some data have shown such association only for women. An association of occasional heavy drinking with an increased risk of diabetes

Diabetes and Cancer

465

was also reported in women. Associations have been found of moderate alcohol consumption with increased risk of breast cancer and male colorectal cancer; and of light alcohol consumption with a reduced risk of lung cancer in both men and women. Several studies have reported an association of cigarette smoking, including passive smoking with T2D. The cancer population attributable risk of smoking has been estimated as 20%, that is, the elimination of smoking could reduce one-fifth of all cancers. Up to one-third of cancer deaths has been estimated as preventable from smoking cessation; this includes the effect of passive smoking.

Reverse Causality Reverse causality is a possible explanation for associations between diabetes and certain types of cancer. In a study of new users of diabetes medications, patients with upper gastrointestinal cancers (esophageal, stomach, pancreatic, liver cancers) were more likely than patients with other cancer diagnoses to have initiated insulin use during the 6 months prior to the cancer diagnosis. This supports the possibility of reverse causation, that is, that upper gastrointestinal cancers promoted diabetes, rather than the converse. To account for reverse causality, studies of the relationship between the two diseases have often excluded from analysis the early period, ranging from 3 months to 2 years, following the diabetes diagnosis.

Biases and Confounding Factors Relevant to the Exploration of the Relationship Between Diabetes and Cancer A main difficulty in assessing the relationship between diabetes and cancer is the difficulty in identifying and mitigating the biases that may arise. In addition to biological mechanisms and shared risk factors, factors that are often not considered may influence the relationship between the diseases; these include: the duration and latency periods of both diseases, competing risks such as cardiovascular disease; glucose lowering medications, confounding by indication and cancer screening. Since both diabetes and cancer may have long latency periods, their dates of diagnosis are somewhat arbitrary. Changes in glucose levels, and in insulin sensitivity and insulin secretion have been documented 3–6 years prior to the diagnosis of diabetes. For cancer subtypes, latency periods vary greatly, and should be considered in the determination of associations with diabetes. While not always a reflection of true biological disease onset, the dates of diagnosis of diabetes and cancer are important disease parameters, which generally reflect the initiation of treatment; the latter affects disease progression and possibly risks of other diseases. Immortal time bias occurs from an inadequate definition of the follow up time of the exposed group, which ignores the period prior to diabetes diagnosis, thus implying that the study outcome, that is, cancer incidence, cannot occur during that period. The period before diabetes was diagnosed should not be ignored or attributed to the exposed arm, but rather to the unexposed group. This bias dilutes the strength of diabetes association with cancer risk towards the null hypothesis of no association. Detection bias or ascertainment bias arises due to differences between persons with and without diabetes in cancer screening and surveillance; such differences may vary according to population. For example, United States and Israeli studies showed higher rates of screening among diabetic than non-diabetic men for colorectal cancer and for prostate cancer; however, in Korea, a lower rate of cancer screening was reported among people with than without diabetes; and a lower rate of screening for breast and cervical cancer was reported among Spanish women with than without diabetes. Several studies have shown particularly elevated cancer diagnoses during the first months, and up to 2 years following diabetes diagnosis, which could be due to increased cancer surveillance, as well as to reverse causality. Detection/surveillance bias may be especially relevant to cancers of the thyroid and prostate, which have relatively long latency periods. For these cancers, particularly, symptoms of the diagnosed cancer may not present during a patient’s lifetime; thus, the detection may be considered as an overdiagnosis, and the ensuing treatment may actually have negative consequences on physical and mental health, despite the cancer being asymptomatic. Detection bias can be mitigated in research by excluding from analysis an initial period, usually of up to 2 years following diabetes diagnosis, and adjusting for screening procedures.

Mechanisms Relevant to the Association of Diabetes With Certain Site-Specific Cancers In addition to the mechanisms described above, which may increase cancer risk in any organ, several site-specific mechanisms affecting particular organs have been described; some of them are described below.

Colorectal Cancer Disturbances in luminal factors may contribute to the increased risk of colorectal cancer in diabetic individuals; such factors include altered bile acid metabolism, delayed colonic transit, compositional change in gut microbiota, and reduction in the intestinal mucus barrier. Gut microbial dysbiosis has been reported in persons with T2D. NLRP6 inflammasomes have been reported to disrupt metabolism by altering gut microbiota and also to contribute to tumorigenesis. Various energy sensing pathways have been suggested to specifically link T2D with colorectal cancer. In addition, both diabetes and poor nutrition were shown to be independent and additive risk factors for colorectal cancer.

466

Diabetes and Cancer

Endometrial Cancer Estrogen and insulin have been shown to exhibit a synergistic effect in the development of type 1 endometrial carcinoma.

Liver Cancer T2D is associated with increased incidences of a number of liver diseases that are associated with hepatocellular cancer (HCC); among them, hepatitis B virus infection, which accounts for more than half of HCC cases; hepatitis C virus infection, which contributes to 25% of HCC cases worldwide; and both alcohol-induced liver disease and nonalcoholic fatty liver disease, and their progression to HCC. In addition, excessive alcohol consumption is a risk factor for both T2D and HCC, and a synergistic interaction of alcohol, diabetes and hepatitis on HCC has been observed.

Prostate Cancer While several studies have reported a lower risk for prostate cancer among diabetic men, a number of investigations have suggested that screening strategies may affect the association. In an Israeli study, the screening rate for the prostate-specific antigen (PSA) was 10% higher among diabetic than non-diabetic men, PSA positivity (> 4 mg/L) was 20% lower, and only low-moderate grade, not high-grade tumors were decreased in diabetic men. These findings suggest that lower reported risks of prostate cancer in diabetic men may be due to lower PSA levels compared to non-diabetic men, rather than to a lower actual risk. In concurrence, a large Australian study identified a history of diabetes, as well as a number of socioeconomic and health system factors, such as frequency of visits to primary care physicians, as factors that affect rates of PSA testing. The possibility that a lower rate of prostate cancer screening among Asian than European men may explain, at least in part, the increased risk of prostate cancer in Asian diabetic men and decreased risk in European diabetic men needs investigation.

Thyroid Cancer T2D as well as other metabolic diseases may increase the risk of thyroid cancer by affecting the level of thyroid stimulating hormone, iodine deficiency, chronic autoimmune thyroiditis and estrogen-dependent signaling.

Glucose Lowering Medications and Cancer Risk Although diabetes is an exposure that cannot be imposed on study participants; glucose lowering medications are such an exposure and can be studied in RCTs with cancer as an outcome. Since mainly diabetic patients use these medications, separating effects of the disease from effects of the treatment is challenging. Phase-3 RCTs, as well as observational studies, provide evidence to this paradigm, each with its inherent methodological limitations. The pharmacologic treatment of T2D changes with increasing hyperglycemia, the indication of disease severity. Recommendations for the treatment of T2D consist of first-line treatment with metformin, followed by the successive addition of other oral glucose-lowering medications, or insulin, as needed to maintain glucose control, which is measured by the HbA1c level. T1D is treated by insulin, sometimes with the addition of oral glucoselowering medications. Diabetes progression, together with its complications, the increased cancer risk over time, and the sequential nature of diabetes treatment and its combinations, makes it difficult to disentangle associations of diabetes pharmacotherapy and cancer risk. The biases discussed above regarding all investigations of diabetes and cancer are also relevant to observational studies on glucose lowering medications. In addition, a number of parameters are particularly relevant to investigations of medications, among them: precise classification of drug exposure: dosages, duration and continuity. Factors such as increased hyperglycemia and the time of initiation of treatments are especially challenging to the investigation of associations of glucose lowering medications with cancer risk. Thus, disease severity may be a confounder between the glucose lowering medication investigated and cancer risk, a phenomenon called “confounding by indication.” This bias is particularly relevant to non-randomized studies of diabetes therapies; whereby, for example, patients on metformin and insulin are compared, without considering that generally, metformin is a first-line treatment, and insulin is administered at a later stage when oral agents do not sufficiently reduce hyperglycemia. Another example arises in observational investigations of dose response relationships, when patients are stratified according to cumulative time under treatment or cumulative dose. However, such an approach may be confounded by disease severity, disease duration, and survival bias. Proper adjustments and using a time dependent statistical approach can attenuated these methodological limitations. Adjustment for disease duration and the use of statistical methods like propensity scoring or instrumental variables can overcome, to some extent, lack of randomization and fundamental differences in the compared treatment groups, but they do not account for unmeasured confounders. In addition, while time varying variables: factors like HbA1c level, BMI and disease duration affect the selection of glucose lowering medications over the course of the disease, treatments also affect HbA1c levels and BMI. Associations of BMI and HbA1c with cancer risk may therefore confound the true association between glucose lowering medications and cancer. Not accounting for competing risks between diseases can result in a selection bias, as for example happens when

Diabetes and Cancer

467

a diabetic patient dies due to another cause before treatment initiation. Excluding this patent from the analysis will result in immortal time bias. Time-dependent analysis, which accounts for the timing of changes in medications, can mitigate biases that arise in this context. In addition, such approach can overcome a differential misclassification bias, such as when patients are classified as using a medication of interest throughout the study period even though they may have used it only for a portion of the period. Referring unexposed person-years to the exposed group dilutes the magnitude of the association towards the null, and the investigated medication may even appear to show a protective effect with cancer, as often happens with metformin. Glucose lowering medications are divided into classes according to their mechanism of action on various organs and tissues (Fig. 2). The diversity of these classes of medications reflects the complexity of diabetes as a multi-organ metabolic disease. The classes of medications can be categorized into: insulins (human and analogs); insulin secretagogues (including sulphonylureas, meglitinides, GLP-1 agonists, and DPP-4 inhibitors); insulin sensitizers (including biguanides and thiazolidinediones); and other medications, which at present include glucose absorption disrupters acting in the intestine (alpha-glucosidase inhibitors) and glucose reabsorption disrupters acting in the kidneys (SGLT2 inhibitors). Deciphering the association between glucose lowering medications and cancer risk can help elucidate the mechanisms underlying the diabetes cancer association. For example, findings that sulfonylureas and exogenous insulin are positively associated with cancer risk, and medications that decrease insulin resistance such as metformin and perhaps thiazolidinediones negatively associated with cancer risk, are congruent with a role of hyperinsulinemia in cancer risk.

Insulin Insulin remains the most potent glucose-lowering agent. However, weight gain and risks of hypoglycemia may accompany improved glycemic control. In addition, all insulins to some extent act as a growth stimulating factor, raising concerns regarding their potential contribution to tumorogenicity. Since 2009, several studies have investigated cancer risks of insulin analogues (Aspart, Determir, Glargine, Glulisine, and Lispro), compared to human insulin. The insulin analogue glargine was associated with an increased risk of breast cancer risk in a number of observational studies that were later criticized for methodological flaws. RCTs have generally not found insulin treatment of any type to be associated with increased cancer risk, though some case-control and cohort studies showed increased risk. Further, the CARING five-country cohort study did not find consistently higher risk for 10 cancers following 5 or more years use of insulin glargine or insulin detemir compared to human insulin. A dose-response increased breast cancer risk was shown in women 40 years and older who used glargine compared to Hagedorn (NPH); this risk was evident only after 6 or more years of use. However, concerns that arise from the study’s methodology are confounding by indication and differential misclassification of the outcome due to the screening policy for breast cancer in England. While there is no compelling

Fig. 2 Site of action of glucose-lowering medications on target organs and tissues contribution to hyperglycemia. Site of action of glucose-lowering medications on target organs are shown in the dotted circles. Contribution of different tissues to hyperglycemia is indicated by the direction of the small arrows indicating the direction of change in function of the organs in diabetes. HGP, hepatic glucose production; GLP-1Ra, GLP-1 receptor agonists; TZDs, thiazolidinediones; DPP4i, DPP4 inhibitors; SGLT2i, SGLT2 inhibitors; SU, sulfonylureas; AGi, a-glucosidase inhibitors. Adapted from Impact of glucose-lowering drugs on cardiovascular disease in type 2 diabetes. European Heart Journal 2015, 36(34), 2288–2296. https://doi.org/10. 1093/eurheartj/ehv239.

468

Diabetes and Cancer

evidence that any clinically available insulin analogue or human insulin increases breast cancer risk, insulin might induce breast tumor progression by upregulating mitogenic signaling pathways.

Sulfonylureas Sulfonylureas have been used to treat T2DM patients since the 1960s, although their use has declined with the emergence of new glucose lowering medications. Their main mechanism of action is to enhance insulin secretion by b-cells; the resultant hyperinsulinemia overcomes, in part, insulin resistance in liver and muscle, and leads to decreases in blood glucose and HBA1c. In vitro studies demonstrated that sulfonylureas accelerate b-cell failure, a process that was shown in an in vivo study to be reversible. Their main side effect is increasing the risk of hypoglycemia. Several observational studies of sulfonylureas have found an increased cancer risk, but such association is not evident in RCTs. Differences in cancer risk between types of sulfonylureas have been reported and warrant further research.

Metformin Metformin, a biguanide, has been in use for over 50 years and its safety profile is well known. Metformin’s glucose lowering effect is most probably confined to the liver and gut: reduction in hepatic glucose production and modest weight loss due to an anorectic effect and gastrointestinal side effects (diarrhea, abdominal discomfort, and flatulence). While not completely clear, the glucose lowering mechanism of metformin apparently involves inhibition of mitochondrial complex I, which increases the AMP/ATP ratio and leads to activation of the energy sensor Adenosine Monophosphate (AMP)-Activated Protein Kinase (AMPK). A direct anticancer mechanism of metformin may involve AMPK-independent and AMPK-dependent effects; while reductions in blood glucose, hyperinsulinemia, and IGF-1 level are possible indirect effects. Already in 2005, the first study to report an association between glucose lowering treatment and cancer suggested a protective effect of metformin. Numerous observational studies have reported a protective effect of metformin on all-site or specific cancer risk; however, those that accounted for biases were less likely to report such. Moreover, RCTs have generally not shown an association between metformin and cancer incidence. The metformin protective effect reported in observational studies has been claimed to result from time-related biases, as discussed above, and from overlooking aspects of the natural history of T2D. Several studies that accounted for time-related biases reported metformin not to be associated with a decreased risk of prostate, lung, colorectal, breast, or pancreas cancers. A number of clinical trials are in the process of investigating a protective effect of metformin on breast cancer risk in overweight or obese women. Metformin is thought to have antineoplastic activity directly via the modulation of cancer stem cell energy pathways, and indirectly by reducing hyperinsulinemia and glycemic levels. Clinical trials in prostate cancer patients are in process, with metformin as an additive to treatment arms, to examine a possible benefit in cancer survival.

Thiazolidinediones The thiazolidinediones (TZDs, pioglitazone and rosiglitazone) are insulin sensitizers that activate peroxisome proliferator-activated receptors-g (PPAR-g). TZDs induce insulin secretion, preserve b-cell function, and improve insulin sensitivity. RCTs have generally shown no association of TZDs with all-site cancer; though some case-control and cohort studies have shown a protective effect. In December 2016, the U.S. Food and Drug Administration (FDA) re-issued a drug safety communication to the labelling information, having concluded that use of pioglitazone may be linked to an increased risk of bladder cancer. In France the drug has been removed from the market altogether. An analysis of close to 3 million individuals from several cohort and RCT studies showed higher risk of pioglitazone for bladder cancer for European populations, larger cumulative drug dose and longer use. Observational studies have generally not shown a significant association of bladder cancer with ever pioglitazone users compared with diabetic patients who had never used it, although an increased risk for bladder cancer may be possible after longer than 1-year exposure to pioglitazone.

Meglitinides Meglitinides (repaglinide, mitiglinide and nataglinide) are insulin secretagogues that can cause weight gain and hypoglycemia. The evidence shows no association between Meglitinide use and all-site cancer risk.

Alpha-Glucosidase Inhibitors (AGIs) AGIs (acarbose, miglitol, and voglibose) impede the breakdown of complex carbohydrates in the gastrointestinal tract, thus slowing carbohydrate absorption and reducing postprandial hyperglycemia. AGIs also increase plasma GLP-1 levels. Treatment with AGIs is usually associated with weight loss. RCTs have not shown an association between AGIs and cancers.

Dipeptidyl Peptidase-4 Inhibitors Dipeptidyl peptidase-4 inhibitors (DPP4i) block the degradation of GLP-1, glucose-dependent insulinotropic polypeptide GIP, and other peptides, including brain natriuretic peptide. This class of drugs has a modest effect in reducing HbA1c and no effect on body

Diabetes and Cancer

469

weight. Early concerns about an increased risk of acute pancreatitis and pancreatic cancer associated with DPP4i use have, for the most part, been refuted in more-recent studies. However, laboratory findings have raised concern that DPP4i can accelerate tumor metastasis. Numerous RCTs have shown no association between DPP-4is and cancer risk.

Incretins The incretins, (GLP-1 receptor agonistsdGLP-1RAs) mimic the action of endogenous glucagon-like peptide-1 (GLP-1). They enhance glucose homeostasis through: stimulation of insulin secretion; inhibition of glucagon secretion; direct and indirect suppression of endogenous glucose production; and delayed gastric emptying. The result is suppression of appetite, decreased postprandial hyperglycemia, and enhanced insulin sensitivity secondary to weight loss. Laboratory studies suggest GLP-1RAs signaling promotes intestinal growth, and may promote colonic tumorigenesis. Several RCTs have reported no association of GLP-1RAs with cancer risk; a number of them specifically investigated the risk of pancreatic cancer. The possibility arises for a decreased risk of pancreatic cancer following long-term use of incretin therapies. GLP-1 analogues were demonstrated not to increase breast cancer incidence compared to DPP-inhibitors. The incretin therapies GLP-1 and Exendin-4 were shown to inhibit migration and promote apoptosis of human ovarian cancer cells, thus supporting the potential for a protective effect of these therapies on cancer risk.

Sodium-Glucose Co-Transporter-2 Inhibitors Sodium-glucose co-transporter-2 inhibitors (SGLT2i) (dapagliflozin) represent the newest class of oral glucose lowering agents approved for the treatment of T2D in the United States and Europe. Inhibition of the SGLT2 transporter leads to increased urinary glucose excretion, resulting in a decline in plasma glucose concentrations. Due to their unique mechanism of action, the SGLT2i can be combined with all glucose-lowering medications, including insulin. The evidence shows no association of dapagliflozin with cancer risk. Long term RCTs are needed to rule out an association with bladder or urethral malignancies.

Glucose Lowering MedicationsdSummary of the Evidence Meta-analyses from systematic reviews of observation studies (cohort and case control studies) showed increased risk for insulin and sulfonylureas, and a protective effect for metformin and TZDs. However, meta-analyses of RCTs showed no increased cancer risk conferred by any glucose lowering medication. We note that RCTs published to date on the association between glucose lowering medications and cancer risk were not designed for this purpose but were conducted to demonstrate cardiovascular safety and HbA1c reduction efficacy. These trials included diabetic patients with a specific risk profile to address their designated purposes; thus, some lacked statistical power to investigate cancer risk due to a short length of follow up or a small sample size. Further, since metformin, which is often used as the comparator or as background therapy, is the first-line oral therapy for type 2 diabetes, its comparison with other therapies that could be of lower clinical safety and effectiveness would be unethical. While metaanalyses overcome problems of small sample size, they are still subject to the limitations inherent in each of the included studies. Moreover, high heterogeneity between studies can reduce the validity of the findings. To date, no study was published that accounts for the full complexity of diabetes disease, which exposes patients to sequentially changing treatments. Glucose lowering medications potentially act both as confounders and as causative factors in the association between diabetes and cancer. Thus, a research implication is that glucose lowering medications may confound associations of diabetes with cancer risk less than previously thought. A temporal association is consistently demonstrated between diabetes and cancer, which is not explained by glucose control or by glucose lowering medications. A clinical implication is that cancer risk need not be considered in the selection of diabetes treatments, in most circumstances. Still, the possibility that metformin, and perhaps also TZDs, may reduce cancer risk seems worthy of further investigation.

Concluding Remarks Associations of diabetes with cancer risk are well established. Explanations are abundant and vary according to cancer type. Diabetes is evidently temporally associated with cancer risk directly, by means of hyperglycemia; as well as indirectly by means of shared risk and predisposing factors. Accounting for the numerous factors that can confound the association between the diseases and the potential biases is challenging and must be addressed. The evidence does not support causality of glucose lowering medications on cancer risk.

See also: Hormones and Cancer. Metformin.

Further Reading Dang, C.V., 2012. Links between metabolism and cancer. Genes and Development 26, 877–890. Dankner, R., Boffetta, P., Balicer, R.D., et al., 2016a. Time-dependent risk of cancer after a diabetes diagnosis in a cohort of 2.3 million adults. American Journal of Epidemiology 18, 1098–1106.

470

Diabetes and Cancer

Dankner, R., Boffetta, P., Keinan-Boker, L., et al., 2016b. Diabetes, prostate cancer screening and risk of low- and high-grade prostate cancer: An 11 year historical population follow-up study of > 1 million men. Diabetologia 59, 1683–1691. Farmer, R.E., Ford, D., Forbes, H.J., Chaturvedi, N., Kaplan, R., Smeeth, L., Bhaskaran, K., 2017. Metformin and cancer in type 2 diabetes: A systematic review and comprehensive bias evaluation. International Journal of Epidemiology 46, 728–744. Johnson, J.A., Bowker, S.L., 2011. Intensive glycaemic control and cancer risk in type 2 diabetes: A meta-analysis of major trials. Diabetologia 54, 25–31. Johnson, J.A., Carstensen, B., Witte, D., et al., 2012. Diabetes and cancer (1): Evaluating the temporal relationship between type 2 diabetes and cancer incidence. Diabetologia 55, 1607–1618. Kong, A.P.S., Xu, G., Brown, N., et al., 2013. Diabetes and its comorbiditiesdWhere east meets west. Nature Reviews Endocrinology 9, 537–547. Kyu, H.H., Bachman, V.F., Alexander, L.T., et al., 2016. Physical activity and risk of breast cancer, colon cancer, diabetes, ischemic heart disease, and ischemic stroke events: Systematic review and dose-response meta-analysis for the global burden of disease study 2013. BMJ 354, i3857. Pearson-Stuttard, J., Zhou, B., Kontis, V., et al., 2017. Worldwide burden of cancer attributable to diabetes and high body-mass index: A comparative risk assessment. The Lancet Diabetes and Endocrinology 6, 95–104. Shanik, M.H., Xu, Y., Skrha, J., et al., 2008. Insulin resistance and hyperinsulinemia: Is hyperinsulinemia the cart or the horse? Diabetes Care 31 (Suppl 2), S262–S268. Suissa, S., Azoulay, L., 2012. Metformin and the risk of cancer: Time-related biases in observational studies. Diabetes Care 35, 2665–2673. Tsilidis, K.K., Kasimis, J.C., Lopez, D.S., et al., 2015. Type 2 diabetes and cancer: Umbrella review of meta-analyses of observational studies. BMJ 350, 1–11. Wu, L., Zhu, J., Prokop, L.J., Murad, M.H., 2015. Pharmacologic therapy of diabetes and overall cancer risk and mortality: A meta-analysis of 265 studies. Scientific Reports 5, 10147. Zheng, C., Beresford, S.A., Van Horn, L., et al., 2014. Simultaneous association of total energy consumption and activity-related energy expenditure with risks of cardiovascular disease, cancer, and diabetes among postmenopausal women. American Journal of Epidemiology 180, 526–535.

Diet and Cancer Livia SA Augustin, National Cancer Institute IRCCS G. Pascale Foundation, Naples, Italy; and Clinical Nutrition and Risk Factor Modification Centre, St Michael’s Hospital, Toronto, ON, Canada Concetta Montagnese, National Cancer Institute IRCCS G. Pascale Foundation, Naples, Italy Ilaria Calabrese, University of Naples Federico II, Naples, Italy Giuseppe Porciello, Elvira Palumbo, and Sara Vitale, National Cancer Institute IRCCS G. Pascale Foundation, Naples, Italy Stephanie Nishi, National Cancer Institute IRCCS G. Pascale Foundation, Naples, Italy; and University of Toronto, Toronto, ON, Canada © 2019 Elsevier Inc. All rights reserved.

Introduction Diet plays a key role in the prevention of chronic diseases including cancer. Cancer is one of the leading cause of death accounting for 8.2 million deaths globally (International Agency for Research on Cancer, 2014). Cancer rates vary across geographical regions and population migration studies observed a 20-fold variation in rates (Rastogi et al., 2008; Ziegler et al., 1993; King et al., 1985) suggesting that environmental factors including diet play may be important contributors. Historically there is evidence supporting a role of diet in carcinogenesis from observations by Peyton Rous who, in Rous (1914) found that food restriction in mice reduced tumor metastasis (Rous, 1914), while in the 1920s obesity was found to be associated with greater cancer mortality and in the 1930s high plant food consumption was suggested to be protective in cancer development (Stocks et al., 2009). Doll and Peto (1981) estimated that approximately 35% of cancer deaths could be avoided by dietary modifications (Doll and Peto, 1981) and this seems to be confirmed by more recent epidemiological studies (McCullough and Giovannucci, 2004; Willett, 1995). Generally, high calorie diets as well as high consumption of red and processed meat are linked to increased cancer risk whereas high vegetable and fruit intake, particularly citrus fruit and vegetables from the brassica family, are consistently found to be protective against cancer development (Kushi et al., 2012; Pan et al., 2012; Santarelli et al., 2008; Willett, 1995; Bosetti et al., 2012; Turati et al., 2015b). There is a paucity of data from clinical trials of diet in cancer patients owing to the challenges and high costs of such trials. Below sections targeting nutrients, selected food groups and dietary patterns are discussed.

Dietary Carbohydrates and Glycemic Index Amongst dietary compounds, carbohydrates have been implicated in the etiology of cancer at various sites (Franceschi et al., 1997; Giovannucci, 1995; Augustin et al., 2002) while sources of carbohydrates such as whole grains have been found to reduce cancer mortality in many studies and have been included in the guidelines for breast cancer recurrence of the American Cancer Society (Runowicz et al., 2016). In a recent meta-analysis published in 2016 of ten large international epidemiological studies showed that higher consumption of whole grains compared to low or no consumption reduced cancer mortality by 12% (Zong et al., 2016). This suggest that not all dietary carbohydrates are the same and some may give protection while others increase risk. Specifically it has been proposed that the extent of the blood glucose rise produced by carbohydrates and captured by the glycemic index (GI), may play a differential role in cancer development (Augustin et al., 2002). The GI is a ranking of carbohydrate foods based on the ability to raise blood glucose concentration by the food ingested (Jenkins et al., 1981). In equicarbohydrate amounts, low GI foods increase blood glucose levels mildly while high GI foods result in larger rises in blood glucose and insulin (Jenkins et al., 1981). Hyperglycemia and hyperinsulinemia have been suggested to increase cancer risk. McKeown-Eyssen (1994) and Giovannucci (1995) independently suggested that blood glucose and insulin may be important factors promoting malignant transformation and tumor growth. Indeed hyperglycemia, both fasting and postprandial, and even at concentrations below the diabetes threshold have been associated with a higher risk of cancer incidence and death independent of body weight (Jee et al., 2005; Stattin et al., 2007; Stocks et al., 2009) and cancer site (Crawley et al., 2014). A meta-analysis showed that blood glucose concentration above 110 mg/dL (6.11 mmol/L), increased risk of all cancers by 32% compared to glycemic concentrations below 110 mg/dL (6.11 mmol/L) (Crawley et al., 2014). Fasting blood glucose
140 mg/dL (
7.8 mmol/L) was associated with significantly higher (23%–29%) death rates from major cancer sites including colorectum, liver and pancreas, compared with blood glucose concentration < 90 mg/dL (< 5.0 mmol/L) (Jee et al., 2005). Furthermore, hyperglycemia can enhance cancer progression and promotion (Li et al., 2015), and in a dose–response meta-analysis it was shown that every 10 mg/dL (0.56 mmol/L) increase in blood glucose concentration corresponded to a 15% increased risk of pancreatic cancer (Liao et al., 2015). Reducing hyperglycemia pharmacologically by use of antihyperglycemic medications (e.g., metformin, acarbose) (Noto et al., 2012; Tseng et al., 2015) and consuming low GI diets instead of high GI diets has shown benefits in reducing the risk of major cancers especially colorectum, breast and endometrium (Augustin et al., 2001; Barclay et al., 2008; Turati et al., 2015a; Choi et al., 2012). Some evidence may also suggest reduced colorectal cancer recurrence (Meyerhardt et al., 2012). The mechanisms involved in increasing risk of cancer with high GI foods may include at least two pathways, hormonal and oxidative stress.

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65132-2

471

472

Diet and Cancer

Potential Mechanisms Regarding whole grains these carbohydrate foods are rich in dietary fiber, vitamins, minerals and phytochemicals which are protective against chronic diseases (Slavin, 2003). Dietary fiber down-regulates inflammation, probably through the production of short chain fatty acids, particularly butyrate and propionate, when fiber is fermented in the colon by gut microbiotia. There are however other mechanisms of action beyond fiber which have been suggested and include (Fardet, 2010). Furthermore, dietary fiber may reduce insulin resistance by decreasing the rate of carbohydrate absorption and increasing insulin-like growth factor binding protein 3 concentrations. These mechanisms are explained further below in relation to the dietary GI.

Insulin-like growth factord1 (IGF-1) High GI foods induce higher glycemic and insulinemic responses and insulin is known to have growth promoting and proliferating effects (Giovannucci, 2003; Kaaks and Lukanova, 2001) and to affect recurrence in breast cancer patients (Goodwin et al., 2002). Insulin is able to bind to the insulin-like growth factor (IGF)-1 receptor and produce the cascade of reactions typically seen with IGF1 activation. IGF-1 is a peptide, similar in molecular structure to insulin, produced mainly by the liver in response to growth hormone, but can be synthesized by almost any tissue in the body (Olivecrona et al., 1999). IGF-1 is the primary circulating growth factor in the IGF system, comprising of other IGFs and their binding proteins, regulating cellular proliferation, differentiation and apoptosis. Meta-analyses of recent studies confirm previous reports regarding elevated serum levels of IGF-I to be associated with an increased risk of colorectal cancer (Chi et al., 2013; Rinaldi et al., 2010), breast cancer (Endogenous et al., 2010) and prostate cancer (Rowlands et al., 2009). Studies in healthy people of low versus high GI diets and changes in circulating IGF-1 and its binding proteins suggest that low GI diets beneficially affect the IGF-1 axis and therefore may lead to an environment that is less conducive to tumor growth compared to high GI foods (Brand-Miller et al., 2005; Runchey et al., 2012).

Oxidative stress Damage to DNA by reactive metabolites has been widely accepted as a major cause of cancer (Ames and Gold, 1991; Beckman and Ames, 1997; Halliwell, 2007). Oxidative stress can activate a variety of transcription factors leading to the expression of over 500 different genes, including those for growth factors, inflammatory cytokines, chemokines, cell cycle regulatory molecules, and antiinflammatory molecules (Reuter et al., 2010). Activation of these pathways via oxidative damage has been implicated in the initiation, promotion, and progression of a normal cell into a tumor cell (Reuter et al., 2010; Valko et al., 2007). Consumption of high glycemic index foods have been suggested to increase oxidative stress through formation of free radicals that are capable of damaging biological molecules and hence initiating abnormal cell growth through gene mutation. Hyperglycemia contributes to increased oxidative stress through an incremental generation of reactive oxygen species (ROS) during the mitochondrial oxidative metabolism of carbohydrates which depletes antioxidant defenses (Brownlee, 2005; Ceriello et al., 1998; Title et al., 2000). Reducing the GI of a meal using acarbose has been shown to significantly reduce markers of oxidative stress in individuals with impaired glucose tolerance (Inoue et al., 2006; Monnier et al., 2006; Quagliaro et al., 2005). Compared to a high GI diet, consumption of a low GI diet was associated with an increased plasma total antioxidant capacity in overweight and obese men (Botero et al., 2009) and with a lower marker of oxidative stress in overweight and obese women (Arikawa et al., 2015).

Conclusions and Future Directions Despite population differences in dietary habits and carbohydrate intakes low GI diets seem to protect from cancer risk compared to high GI diets particularly for specific cancer sites: colorectum (WHS study) (Higginbotham et al., 2004), breast (ORDET and EPIC studies) (Sieri et al., 2007; Romieu et al., 2012) and endometrium (NBSS and NHS studies) (Silvera et al., 2005; Cui et al., 2011). Populations who could benefit the most from consuming low GI diets may be those characterized by a high prevalence of overweight and/or obesity since the strongest associations have been found in people with higher body weight (Coleman et al., 2014; Franceschi et al., 2001) and this is possibly due to the underling insulin resistance in people who are overweight or obese which would result in higher insulin responses for the same dietary GI rank. In an environment characterized by sedentary behavior, overeating and obesity, higher circulating blood glucose as a result of high GI food consumption, could represent a greater burden for an already stressed metabolism caused by obesity and lack of physical activity. Lowering the GI of foods may therefore be particularly relevant in affluent societies and in countries with high consumption of dietary carbohydrates as in the Mediterranean region (Favero et al., 1997; Slimani et al., 2002; Wirfalt et al., 2002). Regarding prevention of cancer recurrence (secondary prevention), the guidelines for cancer survivors by the American Cancer Society include foremost maintenance of a healthy body weight, engaging in physical activity and consuming a healthy diet which are consistent with guidelines for cancer prevention and heart disease prevention (Kushi et al., 2012). There is no mention of carbohydrate quality beyond whole grains, with regards to longer survival. At present, there is a lack of clinical trial investigating low GI diets in secondary cancer prevention. Nevertheless, when summarizing the mechanistic evidence linking glycemia and insulinemia to carcinogenesis and the evidence from primary prevention studies, there is a general support for cancer protecting effects of lower fluctuations of blood glucose and insulin (Rock et al., 2012) and hence potentially for low GI diets. Considering the longer survival of people diagnosed with cancer and therefore their

Diet and Cancer

473

potential risk of developing or worsening other chronic conditions such as diabetes, the inclusion in cancer guidelines for a preference of low GI whole grains and low GI carbohydrates in general would be a sensible decision.

Plant Based Proteins and Cancer Plant based dietary patterns have been associated with reduced risk of developing several chronic diseases, including cancer. Consistent with these findings, guidelines support dietary patterns lower in animal-based foods and higher in plant-based foods, emphasizing beans/legumes including soy, whole grains, nuts/seeds, and vegetables, namely sources of plant protein (USDA, 2010; USDA, 2015; Kushi et al., 2012; Schüz et al., 2015). From a global standpoint, production of plant-based proteins is considered to have lower environmental burden, land and water requirements coupled with fewer greenhouse gas emissions, as well as having a lower financial cost compared with animal-based proteins (Pimentel and Pimentel, 2003; Reijnders and Soret, 2003). These reasons, along with concerns for animal welfare, likely explain the increasing interest in the potential of using plant-based protein sources. The following focuses on cancer morbidity and mortality related to plant compared with animal protein consumption, summarizing available evidence and considering implications for future research.

Plant Based Protein Sources Plant-based proteins can be found in a variety of food sources, including soy beans and other legumes, nuts/seeds, whole grains containing gluten (e.g., seitan) as well as other vegetables although the amounts are usually small (USDA Food Composition Database, 2018). In Europe and the United States, cereals are the principal contributors to plant protein intake providing 40%–70% of plant protein. However, animal products are still the main source of dietary protein providing between 55% and 71%, depending on the country, with red meats contributing 16%–35% to animal protein intake (Halkjaer et al., 2009; Phillips et al., 2015; Camilleri et al., 2013).

Evidence for Plant Based Protein Compared to Animal Protein and Cancer Risk Numerous studies have assessed the impact of animal protein consumption on various types of cancer, however, few have focused specifically on cancer and consumption of plant protein. Evidence regarding carcinogenic or anticarcinogenic effects of plant compared with animal protein is mixed. Yet, protein is not consumed in isolation, but as part of a food matrix. Thus, in epidemiological studies, it can be difficult to attribute any observed benefit solely to protein as the finding may be due to other nutrients found in the food source. As well, findings from observational studies may be confounded since individuals consuming a more plant-based diet may also tend to participate in “healthier” lifestyle behaviors, such as being a nonsmokers, consuming less alcohol, having a lower body weight, being more physically active, and/or consuming a generally healthier diet, all of which may contribute to decreased cancer risk (Tharrey et al., 2018; Davey et al., 2003). Due to these limitations the following will focus on research where protein was explicitly the nutrient of interest in the comparison of animal and plant based dietary exposures. Much of the available evidence for plant protein compared to animal protein in relation to specific cancers pertains to colorectal and breast cancers. In a meta-analysis of observational studies on colorectal cancer risk, comprising 21 studies and 8187 cases, showed no evidence of significance between dietary animal protein or vegetable protein intake (Lai et al., 2017). The association between dietary protein sources and breast cancer risk was assessed in a meta-analysis of 46 prospective cohort studies (60,615 cases and 2,749,307 participants, with a mean follow-up of 3.9–65 years). Findings indicated higher total red meat, fresh red meat, and processed meat intake were risk factors for breast cancer, whereas evaluation of plant protein sources showed soy protein exhibited a protective association, and nuts (peanuts and tree nuts) had no adverse association. In terms of cancer mortality, animal and plant protein intake has been prospectively examined in US health professionals (121,700 women and 51,529 men) followed for up to 32 years with repeated measures of dietary intake. Animal and plant protein consumption comprised 14% (9%–22%) and 4% (2%–6%) of total energy intake, respectively. Plant protein was inversely associated with cancer mortality when age was taken into account, however, when the model was further adjusted for dietary and lifestyle factors, such as smoking status, body mass index and physical activity, this relationship disappeared (Song et al., 2016).

Potential Mechanisms of Action: Plant vs. Animal Proteins The mechanisms involved require further investigation. Proposed potential mechanisms relate to: (1) the reduction and/or displacement of animal protein intake with plant protein and (2) the food matrix within which plant protein tends to be contained. In reducing animal protein intake it is suggested there would be a subsequent reduction of the carcinogenic byproducts from consumption of red meats, such as heterocyclic amines and polycyclic aromatic hydrocarbons formed during high temperature cooking (Steck et al., 2007; Zheng and Lee, 2009). Animal protein is also a source of fat and heme iron, which have been implicated

474

Diet and Cancer

in the production of oxidative stress and inflammation, where experimental data suggest oxidative stress and inflammation have a pro-carcinogenic role in cancer development (Farvid et al., 2014; Fonseca-Nunes et al., 2014; Hedlund et al., 2008; NegreSalvayre et al., 2010; Samraj et al., 2015). Plant protein is found in food matrices that tend to include dietary fiber and antioxidants, thus possibly playing a role in cancer protection by lessening the negative effects of inflammation, insulin resistance and hyperinsulinemia by reducing IGF-1 concentrations, as well as oxidative damage. Additionally, plant foods are known to contain antioxidants, including vitamin C, vitamin E, and carotenoids, yet many plant proteins and peptides have also been identified as novel antioxidant agents, and are thought to contribute to carcinogenic protection by interfering with oxidative damage to DNA, lipids, and proteins (Lu et al., 2010; Zhang et al., 2010; Xue et al., 2009).

Conclusions and Recommendations and Future Directions It is important for overall population health to continue to support dietary patterns with lower animal-based foods and higher plant-based foods (USDA, 2010; USDA, 2015; Kushi et al., 2012; Schüz et al., 2015). Cancer specific recommendations still require further research, yet, at this time it appears that consumption of plant-based proteins do not have adverse results thus supporting their intake. Higher quality research is recommended, differentiating cancer types and investigating cancer mortality rates for high versus low vegetable protein consumption.

Soy Foods and Cancer Over the years epidemiological studies have consistently shown health benefits of soy foods related to the prevention of chronic diseases, including cancer (Messina, 2010). Of particular interest for human health are soy isoflavones, bioactive compounds in soybeans, that have shown anticancer properties due to their estrogen-like activity, as well as their ability to modulate cell cycle, apoptosis, differentiation, proliferation, and cell signaling (Rizzo and Baroni, 2018). Breast cancer incidence rates are lower in soy food consuming countries such as Asian populations (Pisani et al., 1999). Asian epidemiologic studies show that higher soy consumption is associated with an approximate one-third reduction in breast cancer risk (Chen et al., 2014). Opposite, in last decades, as Westernization of Asian cultures has occurred, breast cancer rates among Asian populations have steadily increased. In 1995 the “early intake” hypothesis emerged, in animal models, that exposure to soy isoflavones early in life reduce breast cancer risk (Lamartiniere et al., 1995a, b, 2000). Data from epidemiological studies indicate that higher soy intake early in life, such as during childhood and/or adolescence, is associated with 25%–60% reductions in risk (Wu et al., 2002; Shu et al., 2001; Lee et al., 2009). The “early-intake” hypothesis could also explain discrepancies in epidemiologic data from different countries. The estrogen-like activity of isoflavones by soybeans is at the base of data reporting soy isoflavones adversely affect the prognosis of breast cancer patients. Results from animal studies indicate that isoflavones activity stimulates the growth of existing mammary tumors (Ju et al., 2001; Allred et al., 2001; Helferich et al., 2008). However, results from in vitro and animal studies are controversial and not in line with human studies. The European Prospective Investigation into Cancer and Nutrition Study cohort (EPIC) reported no increase in cancer risk with higher soy isoflavone intake (Zamora-Ros et al., 2013). Regarding the association between soy food intake in breast cancer patients, clinical and prospective epidemiological data clearly show that soy consumption after breast cancer diagnosis was associated with reductions in breast cancer recurrence and beneficial effects were similar in Asian and non-Asian populations (Guha et al., 2009; Caan et al., 2011; Shu et al., 2009; Kang et al., 2010; Zhang et al., 2012). A meta-analysis of five cohort studies showed that soy isoflavones intake might be associated with lower recurrence in ER negative, ER þ/PR þ breast cancer (Chi et al., 2013). Furthermore, it is reported that exposure to soy isoflavones was associated with reduced mortality and reduced cancer recurrence also in women with ER(þ) and ER() breast cancer (Messina, 2016a, b; Shu et al., 2009). Results from in vivo studies shown that isolated isoflavones possibly interfere with the efficacy of tamoxifene, an antiestrogen prescribed to women with ER þ tumors (Ju et al., 2008; Jones et al., 2002), whereas other studies in humans report benefits of combined soy food intake and tamoxifen therapy use on the inhibition of breast cancer growth improving pharmacological efficacy (Kang et al., 2010; Nechuta et al., 2012). As for breast cancer, prostate cancer incidence varies throughout the world and it appear higher in Western countries than in the Asian population. Specifically, the higher soy food intake in Asian population is associated with as much as a 50% reduction in prostate cancer risk (Zhang et al., 2016) even if prostate cancer has a lower incidence rate in Asian population and this phenomenon has been linked to difference in intestinal bacteria responsible for conversion of daidzein to equol in the gut (Akaza, 2012). Clinical studies indicate that isoflavone exposure has beneficial effects in prostate cancer patients (Messina et al., 2006; Kwan et al., 2010; Ide et al., 2010). A recent meta-analysis supports the existing evidence which indicates that the total soy food intake is associated with reduced prostate cancer risk (Applegate et al., 2018). Nevertheless, some studies reported controversial results on the effects of soy on prostate cancer (Bosland et al., 2010; Fleshner et al., 2011). Differences in study designs limit the implications of results. The European Food Safety Agency (EFSA) commented on the possible association between the intake of isoflavones and adverse effect on sex-hormone responsive tissues (such as breast and endometrium), after reviewing 25 clinical studies and concluded that

Diet and Cancer

475

35–150 mg per day of isoflavones from supplements do not adversely affect the endometrium (EFSA Panel on Food Additives and Nutrient Sources added to Food ANS, 2015). No effect of oral isoflavone supplementation on endometrial thickness overall was found in a meta-analysis of 23 randomized controlled trials in peri- and postmenopausal women (Liu et al., 2016). Moreover, data from a meta-analysis of 10 observational studies found that dietary soy isoflavones from soy beans and soy foods were inversely associated with endometrial cancer risk, and the protective effects were found for both Asian and non-Asian populations. Evidence reported from a meta-analysis of 17 epidemiologic studies (including case-control and prospective cohort studies) revealed that soy isoflavone consumption, particularly with soy foods/products, was significantly associated also with reduced risk of colorectal cancer in Asian populations (Yu et al., 2016). Another meta-analysis of prospective studies showed higher isoflavone intakes significantly decreased the risk of stomach and lung cancer while nearly significantly decreased the risk of breast and colorectal cancer (Grosso et al., 2017).

Mechanisms of Action Many biological explanations have been proposed for the cancer protective effects of soy foods (Grosso et al., 2017). Soy isoflavones play a competitive role with endogenous estrogens for binding of estrogen receptors (ERs) and stimulate cell proliferation (Trock et al., 2006; Taylor et al., 2009; Guha et al., 2009). Isoflavones act as weak ER agonists or antagonists, depending on the cell type and estrogen environment (Larkin et al., 2008). ER agonist/antagonist properties of isoflavones have been widely studied and because of their tissue-selective effects isoflavones are classified as selective ER modulators. Soy may act as an ER antagonist (Dorjgochoo et al., 2011) and soy isoflavones intake was associated with protective effects in both ER and ER þ breast cancer patients (Chi et al., 2013). The mechanisms by which soy intake may improve the prognosis of breast cancer patients is not obvious, taking into account also data showing a lack of effect of isoflavones on breast cancer markers. To date, neither soy protein nor isoflavones supplements seem to affect markers of breast cancer including mammographic density and breast cell proliferation (Hooper et al., 2010; Wu et al., 2015; Sartippour et al., 2004; Palomares et al., 2004; Cheng et al., 2007; Khan et al., 2012). Different mechanisms have been proposed to explain the role that isoflavones may play in prostate cancer prevention and modulation. Clinical and animal studies report that isoflavones are able to inhibit metastasis (Xu et al., 2009; Jin et al., 2013). Also, cell-based studies show that phytoestrogens exert a chemopreventive effect through binding ERb, expressed in prostate epithelial cells, regulating genes that control cell cycle and apoptosis (Applegate et al., 2018; Mahmoud et al., 2014). Several mechanisms for the protective role of early isoflavone exposure have been proposed (Messina and Hilakivi-Clarke, 2009). The protective effect of early isoflavone exposure may be related to isoflavones ability to impact mammary gland development, changing cells in ways that make them permanently less likely to be transformed into cancer cells. Early life exposure to phytoestrogens has been demonstrated to impact mammary gland development through accelerated terminal end bud differentiation. Furthermore, many similarities between the protective effect of early isoflavone exposure and early pregnancy have been observed (De Assis et al., 2011; Rahal and Simmen, 2011; Mishra et al., 2011). Soy isoflavones also act independently of ER activity exhibiting antiproliferative, antioxidant and antiinflammatory properties in vitro and in vivo (Guha et al., 2009; Rizzo and Baroni, 2018; Zaheer and Humayoun Akhtar, 2017). Isoflavones act as antioxidants and reduce the long-term cancer risk by preventing DNA damage. Genistein in particular is a potent antioxidants and it seems to increase the production of SOD which removes free radicals from cells (Zaheer and Humayoun Akhtar, 2017). Isoflavones interact with epigenetic modifications, such as hypermethylation of tumor suppressor genes. In vitro studies have demonstrated that phytoestrogens act on chromatin and modify transcription through the demetylation and acetylation of histones in breast cancer cell lines (Dagdemir et al., 2013; Zaheer and Humayoun Akhtar, 2017; Grosso et al., 2017).

Conclusions and Recommendations Evidence indicates that soy foods, mainly by virtue of their isoflavones content, as having a number of health benefits. Despite the many benefits, soy foods have not been embraced by the medical community. Mainly the estrogen-like properties of isoflavones has led to concerns that soy may exert untoward effects in some subpopulation (e.g., postmenopausal women), increasing breast cancer risk and getting worse the prognosis of breast cancer patients. As a result, clinicians treating breast cancer patients frequently caution them to either avoid soy foods or use them in moderation. However, evidence published over the past decades indicates the safety of isoflavones exposure with respect to breast and endometrial cancer. The position of AICR and WFRC are that soy foods are safe and even beneficial if consumed by women with a breast cancer diagnosis (American Institute for Cancer Research (AICR); World Cancer Research Fund International (WCRF). In 2015, after a literature evaluation, EFSA concluded that isoflavones do not adversely affect the breast, uterus and thyroid (the three organs that were investigated) in peri and postmenopausal women (EFSA Panel on Food Additives and Nutrient Sources added to Food ANS, 2015). Clinical studies did not support the hypothesis of an increased risk of breast cancer and supplementation with 150 mg/day of soy isoflavones (up to 30 months) of had no adversely effect on endometrial thickness and in the uterus (EFSA Panel on Food Additives and Nutrient Sources added to Food ANS, 2015).

476

Diet and Cancer

As suggested by the early hypothesis, it could be reasonable to encourage young women to consume one serving per day of soy foods because epidemiological data suggest it may be protective. However, the consumption of a variety of soy foods, such as soy flour and tofu, should be encouraged rather than purified forms of phytoestrogens. A moderate consumption of soy foods, such as one to two standard servings daily of soy foods (one serving averages approximately 25 mg isoflavones) is not links to increased breast cancer risk.

Future Directions Although soy and isoflavones offer health benefits to humans, controversial results from different studies have raised doubts about the validity of such health claims. More studies are therefore needed to clarify the controversial results. Nutraceutical values of soybean, including isoflavone content of soy foods, can vary considerably due to the soybean type, cultivar and food processing. Labeling of soy foods with isoflavones concentrations could be a helpful information for consumers and scientists. Dietary guidelines for soy foods and soy isoflavones consumption may improve cancer prevention and management.

Dietary Fats Dietary fats include oils, butter and lard. They are composed of a 3-carbon backbone to which fatty acids are attached. Fatty acids are formed by long, short or medium hydrocarbon or aliphatic chain linked to a carboxylic acid end. The aliphatic chain of fatty acids can present double or single bonds between their carbon molecules, indicating unsaturated fats or saturated fats (SFA), respectively. Fatty acids with one double bound are called monounsaturated fats (MUFA), while polyunsaturated fats (PUFA) are fatty acids with more than one double bound. A type of unsaturated fat that occurs in small amounts in nature are transunsaturated fatty acids, with one or more of their double bonds in the “trans” rather than the common “cis” configuration. Cholesterol is not a fat but an organic molecule of the sterol family (similar to wax) and it is not used for calorie production hence it is not oxidized. Fats and oils are most abundant lipids in nature. They represent the major component of storage fat or adipose tissue in the human body and the major contributor to energy intake in diet. The fatty acids profile in serum and tissues reflects the composition of dietary fats. Fat-rich foods are formed by SFA (with no double-bond), MUFA (e.g., oleic acid) and PUFA (n  3, n  6 fatty acids, e.g., linolenic acid, omega 3 fatty acids). SFAs are found in dairy products such as butter, cream, cheese and milk, in meat and certain plants (e.g., palm oil, coconut oil). Dietary sources of MUFAs are olives, rapeseed, peanuts, sesame and safflower oil. Specific types of fish contain abundant levels of n  3 PUFAs, such as eicosapentaenoic acid (EPA) and docosahexanoic acid (DHA). Linoleic acid (LA, n  6 PUFAs) is present in oil of safflower, grape seed, hemp, corn, wheat germ, cotton-seed and soybean, while a-linoleic acid (ALA, n  3 PUFAs) is in flaxseed, walnuts, some vegetables and canola oils (Burdge, 2004). Trans-fatty acids are found especially in dairy products and meat, but in high amounts in baked foods (biscuits, crackers, cookies, etc.) and margarines. Cholesterol dietary sources are mainly egg yolk, organ meat, shrimp, squid and shell.

Dietary Fats and Cancer Prevention Many studies have suggested that a high consumption of fat through diet is related to an increased incidence of breast, colon, pancreatic and prostate cancer (Othman, 2007). Independently from type of fat, excess dietary fat intake has been associated with cancer recurrence (Dal Maso et al., 2008; McEligot et al. 2006; Vernieri et al. 2018). However, when intervention studies with low fat diets were conducted in the USA no effect on primary (Prentice et al. 2006) or secondary prevention of breast cancer were observed (Pierce et al. 2007) unless the dietary change involved weight loss (Chlebowski et al. 2006).

Saturated Fatty Acids Direct associations between SFAs intake and risk of breast cancer in postmenopausal breast cancer have been found but not in premenopausal women (Cao et al., 2016; Xia et al., 2015). Higher intakes of total fat and SFAs compared to lower intakes were associated with higher risk of breast cancer subtypes, in particular ER and PR positive breast cancer (Sieri et al., 2014). When investigating exposure to SFA consumption and mortality incidence, a recent meta-analysis of four studies found 50% higher risk of breast cancer mortality in the highest compared to the lowest SFA intakes. In colorectal cancer total fat and SFAs did not show any significant associations in an Asian population (Zhong et al., 2013) but in a large European population SFA was directly and significantly associated with colorectal cancer risk (May-Wilson et al. 2017). A meta-analysis of endometrial cancer studies and dietary fat a 45% increased risk was found in case-control studies only (Zhao et al., 2016a, b). Evidence from observational studies (both case-control and prospective) indicate that there is no apparent relationship

Diet and Cancer

477

between SFAs and pancreatic cancer risk. Case-controls study on prostate cancer and dietary fats shows that high consumption of SFAs was associated with a higher aggressiveness of prostate cancer (PC) which seemed to be related mainly to the SFA effects on serum cholesterol levels. Other studies confirmed a role of dietary cholesterol on PC aggressiveness (Allott et al., 2017; Mensink et al. 2003; Platz et al., 2006; Platz et al., 2009).

Monounsaturated Fat (MUFA) Most recent epidemiologic evidence has shown that foods rich in MUFAs could reduce breast cancer risk (Pelucchi et al., 2011) but others do not support these results. High versus low consumption of MUFAs to SFAs ratio was associated with a significantly reduced risk of colorectal cancer in three case–control studies (Rosato et al., 2016). Inverse associations were observed between MUFAs and endometrial cancer risk in cohort studies, while nonsignificant associations were observed in case-control studies (Zhao et al., 2016b). No significant relationship between MUFAs and pancreatic cancer risk were observed in cohort studies (Yao and Saturated, 2015) and conflicting results were seen for prostate cancer. MUFAs are found both in vegetable oils especially olive oil and canola oil but also in meat. This could explain the heterogeneity of results in international studies particularly between North American populations with high meat to olive oil intakes and Mediterranean populations who tend to consume larger amounts of olive oil and less meat. Indeed, when investigating vegetable oil and olive oil a protection is seen in breast cancer risk of 12% and 26%, respectively, in a meta-analysis of cohort and case-control studies (Xin et al. 2015). This protection was supported by results of the PREDIMED clinical trial conducted in Spain where the arm consuming higher extra virgin olive oil showed a 68% reduction in breast cancer risk after 4 years of intervention compared to the control group (Toledo et al. 2015).

Polyunsaturated Fat PUFA Numerous studies (epidemiological, clinical, animal model and in vitro) confirm a preventive role of n  3 PUFAs in different types of cancer sites, including breast, colon and prostate (Berquin et al., 2008; Huerta-Yépez et al., 2016) while high dietary intake of n  6 PUFAs could increase risk of cancer development (Lawrence, 2013). In vitro and in vivo recent studies indicate positive association between n  3 PUFAs and cancer risk (Sawada et al., 2012), moreover data from epidemiologic studies confirm a protective role of this type of fats on cancer risk (Huerta-Yépez et al., 2016). Colorectal cancer is most widely developed in populations who consume a Western diet which is characterized by high fat and are especially rich in n  6 PUFAs particularly arachidonic acid (AA) which is associated with cancer promotion (Huerta-Yépez et al., 2016). Overall, n  3 PUFAs fats and fish may have positive effects on carcinogenesis and lower risk of colorectal cancer (Hall et al., 2008; Ronco et al., 2010). Results from meta-analysis indicate that high intakes of PUFAs was associated to reduced risk of pancreatic cancer, especially in case-control studies but not clearly in prospective studies while a high n  6/n  3 PUFA ratio was correlated with higher risk of high-grade prostate cancer (Williams et al., 2011). In recent meta-analysis, consumption of foods rich in n  3 PUFAs such as marine fish, EPA and DHA, was associated with lower risk of breast cancer (Yang et al., 2014). Meta-analysis on endometrial cancer risk indicate a relationship between intake of fish and lower risk of cancer in European studies (Hou et al., 2017). Results from prospective cohort studies confirm the benefit of n  3 PUFAs on breast cancer incidence (Zheng et al., 2013) which may be due to improvements in immune system for EPA/DHA in supplementation studies (Paixão et al., 2017).

Trans-Fatty Acid Intakes of commercially produced trans-fatty acid are associated with several diseases, including several types of cancers (Thompson et al., 2008; Mozaffarian et al., 2006; Stender and Dyerberg, 2004). Data from correlational studies on trans-fatty acid intake and breast cancer indicate a positive correlation between intake and risk in premenopausal women (Hu et al., 2011) but studies on ruminant-derived trans-fatty acids do not confirm this association (Kolahdouz Mohammadi et al., 2017). However trans fatty acids in ruminants are different than those in industrial products, they are 10-fold lower in concentration and the position of the double bond is located on different carbons. Only few studies observed direct associations between intakes of trans-fatty acids and colon cancer or prostate cancer (Hu et al., 2010).

Cholesterol Dietary cholesterol is found only in animal foods. There is an increased evidence of an important role of dietary cholesterol in increasing breast cancer risk (Li et al., 2016) and this may be due partly to the cholesterol metabolite, 27-hydroxycholesterol, which has estrogenic activity increasing proliferation of ER þ breast cancer cells (Nelson et al., 2013). High dietary cholesterol may also increase risk of pancreatic (Chen et al., 2015) and colorectal cancers in some studies (Järvinen et al., 2001; Lee et al., 2009), but not all (Nagata et al., 2001) and of endometrial and prostate cancers (Allott et al., 2017).

478

Diet and Cancer

Mechanisms of Action Potential mechanisms of dietary fats in the etiology of cancer include pathways linked to excess adiposity, reactive oxygen species (ROS), hyperinsulinemia, adipokine secretion and other key mechanism of inflammation which could induce an environment favorable to tumor growth (Austin et al., 2011; Sung et al., 2011). Dietary fats affect eicosanoids synthesis which are involved in the inflammatory cascade which in turn may stimulate cancer growth (Larsson et al., 2004; Escrich et al., 2011). ROS production is linked to higher consumption of dietary fat and induce oxidative stress that may promote carcinogenesis (Kang, 2002, Baracos et al., 2004). Fatty acids regulate transcription and expression of genes influencing cancer (Jump, 2004; Mandal et al., 2010). Different fatty acids can differently induce cell growth, proliferation, differentiation and motility of cancer cells (Zadra et al., 2013; Navarro-Tito et al., 2008). N  3 PUFAs for example reduce inflammation and seem to be protective against tumor growth (Schley et al., 2005).

Guidelines and Recommendations The recommendations are to limit the intake of red meat and processed meats as much as possible since they are rich sources of SFA, AA, cholesterol and increase inflammation. It is important to avoid excess dietary fat which would lead of positive energy balance and therefore to overweight and obesity. The preferred sources of fats linked to better health and lower cancer risk are MUFA and PUFAs (ACS, American Cancer Society; WCRF, World Cancer Research Fund International).

Cruciferous Vegetables and Related Sulphoraphane Cruciferous vegetables are part of the Brassica genus of plants and include arugula, broccoli, Brussels sprouts, cabbage, cauliflower, kale, turnip greens and mustard, among others. Cruciferous vegetables have been associated with protection against a range of cancers (AICR’S Foods That Fight CancerdBroccoli and Cruciferous Vegetables; NIHdCruciferous Vegetables and Cancer Prevention). They are a rich source of carotenoids (beta-carotene, lutein, zeaxanthin), vitamins C, dietary fiber and a major source of glucosinolate-derived bioactive compounds which have been shown to have anticarcinogenic properties. Glucosinolates are present in almost every member of Cruciferae family and these sulfur-containing compounds are responsible for the pungent aroma and bitter flavor that distinguish cruciferous vegetables from other vegetables (Verhoeven et al., 1996; Higdon et al., 2007). Chemically, glucosinolates are composed of thiohydroximate-O-sulfonate group linked to glucose, and an alkyl, aralkyl, or indolyl side chain (R) (Agerbirk and Olsen, 2012). Glucosinolates (Glucoraphanin) undergo hydrolysis upon contact with the enzyme myrosinase, which is present within plant tissues, and they form biologically active compounds such as indoles (indole-3-carbinol) and isothiocyanates (sulforaphane) that have well known anticarcinogenic properties in in vitro and animal studies (Bosetti et al., 2012; Zhang et al., 1992). Isothiocyanates (ITCs) in plants have antibacterial and antifungal activity and provide important protection from insect and herbivore attack (Rosa et al., 1997). Glucoraphanin glucosinolate is found in the vacuoles of intact plant cells and it is spatially separated from myrosinase, which is in the cytoplasm cells. After cell rupture, induced by chewing during human consumption or by tissue damage during freezing, thawing, and chopping of edible plants, glucoraphanin comes into contact with myrosinase and is converted into sulforaphane and sulforaphane nitrile.

Raw vs Cooked Storage, processing and cooking can change ITCs formation from glucosinolate and affect the cruciferous vegetables anticancer activity (Barba et al., 2016; Jones et al., 2010). Intake of raw cruciferous vegetables provides two to nine times the amount of ITCs in humans compared with similar intakes of their cooked counterparts (Rouzaud et al., 2004; Getahun and Chung, 1999; Conaway et al., 2000; Shapiro et al., 1998). Studies demonstrated that inverse association between cruciferous vegetable intake and cancer risk appear to be stronger with raw vegetable consumption than with their cooked counterparts. It is notable that ITCs may be largely reduced by cooking procedures due to heat-inactivating myrosinase, reducing the formation of sulforaphane (Rouzaud et al., 2004; Getahun and Chung, 1999; Conaway et al., 2000). Human intestinal microflora also possess myrosinase activity and is able to partially hydrolyze ingested glucosinolates, but studies have shown that isothiocyanate exposure after the consumption of cooked cruciferous vegetables is 60%–90% less than that after the ingestion of raw cruciferous vegetables (Rouzaud et al., 2004; Getahun and Chung, 1999; Conaway et al., 2000; Shapiro et al., 1998). Among thermal processing, steaming is preferred over boiling because glucosinolates can leach into the boiling water and because the action of myrosinase at internal temperatures exceeding 70 C is inhibited (Jones et al., 2010). Freezing at  20 C is one way to preserve the content of nutrients and improve health benefits of cruciferous vegetables. Freezing at  20 C resulted in the formation of more sulforaphane of broccoli sprouts which was probably due to the enhancement of

Diet and Cancer

479

myrosinase activity and to the damage caused by ice crystals that brings glucoraphanin into contact with myrosinase (Guo et al., 2015). However, freezing at  20 C and  40 C had a negative impact on the ascorbic acid content in broccoli leading to the enhancement of myrosinase activity (Guo et al., 2015).

Evidence That Cruciferous Vegetables Can Reduce Cancer Risk There is substantial evidence from human studies that cruciferous vegetable consumption may reduce the risk of several common cancer (IARC 2004, WCRF/AICR 2007; Verhoeven et al., 1997; Turati et al., 2015).

Prostate Cancer A meta-analysis of 13 studies found that high intakes of cruciferous vegetables were associated with the decreased risk of prostate cancer (Liu et al., 2012). Cohort studies have examined a wide range of daily cruciferous vegetable intakes and found little or no association with prostate cancer risk (Schuurman et al., 1998; Giovannucci et al., 2003; Key et al., 2004). However, some casecontrol studies have found that people who ate greater amounts of cruciferous vegetables had a lower risk of prostate cancer (Kolonel et al., 2000; Jain et al., 1999).

Colorectal Cancer Results from cohort studies have generally found no association between cruciferous vegetable intake and colorectal cancer risk (McCullough et al., 2003; Flood et al., 2002; Michels et al., 2000). Two meta-analysis reported consistent findings on the protective effects of cruciferous vegetables for colorectal cancer suggesting that intake of cruciferous vegetables may reduce the risk of colorectal cancer in humans (Wu et al., 2013; Tse and Eslick, 2014). In an updated report from the World Cancer Research Fund and the American Institute for Cancer Research (WCRF/AICR), dietary fiber intake, including cruciferous vegetables, has been upgraded as a convincing protective dietary factor for CRC (WCRF Colorectal cancer).

Breast Cancer Data from a meta-analysis of 13 studies suggest that cruciferous vegetables consumption may reduce the risk of breast cancer with a significantly reduced risk in postmenopausal but not in premenopausal women (Liu and Lv, 2013). Despite strong evidence of a protective role of cruciferous vegetables in breast cancer in animal and in in-vitro studies, data in human studies are often inconsistent. Some epidemiological studies reported no association (Smith-Warner et al., 2001) or a weak association with breast cancer risk (Zhang et al., 1999). No evidence of association between cruciferous vegetables consumption and breast cancer prognosis has been observed in breast cancer patients (Nechuta et al., 2013; Peng et al., 2017).

Overall Significant reductions of colorectal and breast cancer risk in subjects consuming cruciferous vegetables at least once a week (compared to non-consumers) were reported from an analysis of a large network of case-control studies while no significant associations were seen in prostate cancer risk (Bosetti et al., 2012). Similar findings have been reported from a systematic review of Mediterranean countries (Turati et al., 2015).

Mechanisms Similar to other plants, cruciferous vegetables contain various protective compounds including dietary fiber, several antioxidants and vitamins (i.e., carotenoids, polyphenols, vitamin C, and folate) (IARC 2004; Verhoeven et al., 1997; Higdon et al., 2007). In addition, cruciferae are rich unique sources of glucosinolates whose major breakdown products (indoles and ITCs) have shown anticarcinogenic properties. Indoles and ITCs may protect against cancer by modulating Phase I and Phase II enzymes, inhibiting carcinogenesis, stimulating cell cycle arrest and apoptosis, and by reducing inflammation and angiogenesis (Navarro et al., 2011; Lenzi et al., 2014).

480

Diet and Cancer

Dietary Fiber Cruciferous vegetables are good sources of dietary fiber which may play a protective role in colorectal cancer (Aune et al., 2011).

Mechanisms Modulating Estrogen Receptor-a Expression There is strong evidence from cell and animal experiments of a preventive role of cruciferous vegetables in breast carcinogenesis by regulating the expression of estrogen receptor, altering the metabolism of estrogen, or suppressing cyclooxygenase 2 (COX-2) (Lin et al., 2017; Fowke et al., 2000; Kang et al., 2009). The chemopreventive effect of indole-3-carbinol on breast cancer may be linked to its ability of decreasing estrogen receptor-a expression (Bosetti et al., 2002; Fowke et al., 2000). Indeed indole-3-carbinol are able to bind to the estrogen receptor (ER), repressing ER signaling, downregulating the expression of estrogen-responsive genes (Meng et al., 2000) and thus preventing the development of estrogen-dependent cancers including breast, endometrial and cervical cancers. Besides indole-3-carbinol also ITCs can modulate expression of sex hormones and their receptors. ITCs function as ERa disruptors to abrogate mitogenic estrogen signaling in ER-positive breast cancer cells, which provides a molecular explanation for the growth inhibitory function of ITCs in breast cancer development, and a rational for further exploration of ITCs as chemopreventive agents for human mammary carcinogenesis. (Kang et al., 2009).

GST Polymorphism and Colorectal Cancer Several studies have investigated the different capacity of individuals to metabolize and eliminate glucosinolate byproducts of hydrolysis and this theory may help explain the differential health benefits of cruciferous vegetables (Lampe and Peterson, 2002; Lin et al., 1998). Approximately half of the population lacks GSTM1 (a variant of glucosinolate enzymes) due to gene deletion (Lin et al., 1998) and the homozygous deletion of GSTM1 gives rise to the null phenotype in which there is no expression of GSTM1 protein. The hypothesis is that individuals with the null genotype of GSTM1 or GSTT1 polymorphisms would have less conjugation activity, and thus permitting ITCs to remain biologically active potentially providing greater protection against cancer (Cotton et al., 2000). Findings from a meta-analysis suggest a positive correlation between glucosinolate polymorphism and the protective effects of cruciferous vegetables consumption against colorectal cancer risk in individuals with a single null GSTT1 genotype (Tse and Eslick, 2014).

Epigenetic Modulation Cruciferous vegetables may have epigenetic properties modulating cancer prevention. Most cancers are characterized by the overexpression of histone deacetylases and DNA methyltransferases, and the mis-expression of micro RNAs. Sulforaphane and indole-3-carbinol from cruciferous vegetables may regulate micro RNAs and inhibit histone deacetylases and DNA methyltransferases (Royston and Tollefsbol, 2015).

Recommendations Dietary Guidelines generally recommend consuming a variety of vegetables each day because of their different content of bioactive compounds (USDA Choose my Plate). Usually cruciferous vegetables fall into the “dark-green vegetables” category. It is suggested to eat cruciferous vegetables as part of a healthy diet and to prefer eating them raw or after steaming and that sulforaphane appears to be an effective and safe chemopreventive molecule and a promising tool to fight cancer (Lenzi et al., 2014).

Pro-Inflammatory Vegetables: Is There Any Evidence? High plasma levels of C reactive protein (CRP) are expression of chronic inflammatory status and are associated with poor prognosis of tumors. The inflammatory cytokines are released into the blood by the tumor and by the tumor-associated macrophages, their presence in the blood and the presence of high levels of CRP may indicate that the tumor is more aggressive. It is a common belief that the Solanaceae plants (tomatoes, eggplants, peppers, and potatoes) may increase inflammation however there seems to be a lack of evidence (Berrino, 2016). In order to keep inflammation low, it is preferable to consume a diet rich in plant foods such as vegetables, fruit and whole grains with some omega-3 fats (flaxseeds, soybeans, wild herbs). Evidence of a possible association between the intake of tomatoes and tomato juice and inflammation has given mixed results: reduction in CRP levels in some studies (Jacob et al., 2008) or no significant change in CRP and other inflammatory mediators in others (Blum et al., 2007). Furthermore, isolated lycopene supplementation has not shown antiinflammatory effect (Markovitis

Diet and Cancer

481

et al., 2009). Tomatoes, eggplants, peppers, and potatoes are rich the glycoalkaloids a- solanine and adchaconine which are plants’ natural defenses. These compounds however may stimulate the formation of polyamines, metabolites that induce cell proliferation, reduce immune defenses against tumors and promote metastatic migration of tumor cells and angiogenesis (Soda, 2011).

Supplements: Soy Isoflavones Soy isoflavones are phytochemical compounds that are found in soybeans. These bioactive compounds are also known as phytoestrogen because of their chemical structure similar to animal endogenous estrogens which allows them to bind to the estrogen receptor (Setchell, 1998; Cabot, 2003). Mean isoflavone intake among adults range from 30 to 50 mg/day in Asian countries consuming traditional soy foods whereas is less than 3 mg/day in United States, Canada and Europe (Goodman-Gruen and Kritz-Silverstein, 2001; Bai et al., 2014; Van Erp-Baart et al., 2003; Van der Schouw et al., 2005). Soy isoflavones occur mainly as glycosides form, bound to a sugar molecule (Murphy et al., 2002). After food fermentation during processing or gut digestion the sugar moiety is hydrolyzed, thereby allowing absorption to occur (Rowland et al., 2003). Breaking of this glycosidic bond leaves the isoflavones in their simple aglycone form (Murphy et al., 1999, 2002; Yamabe et al., 2007; Otieno et al., 2007). It is still not clear whether the effects are due only to the aglycone or its glucosidic portion or both. Soy isoflavones occur mostly in the form of genistein, daidzein and glycitein which account for approximately 50%, 40%, and 10%, respectively, of the total soy isoflavones content (Murphy et al., 2002). Some human studies reported a wide inter-individual variability in isoflavones metabolism (Jackson et al., 2011) which depends primarily on intestinal flora polymorphisms (Moon et al., 2006; Nechuta et al., 2012a,b). Some individuals host intestinal bacteria that convert the isoflavone daidzein into the isoflavonoid equol (Jackson et al., 2011; Setchell, 1998; Setchell and Clerici, 2010) and this may explain some of the discrepancies in results seen for the health effects of soy foods in human studies (Bolca et al., 2007). It has been suggested that individuals with equol-producing intestinal bacteria may benefit from soy food consumption more than those without (Setchell et al., 2002; Setchell and Clerici, 2010; Setchell, 1998).

Sources Plants synthetize isoflavones through environmental stresses, such as infections or paucity of nutrients (Lozovaya et al., 2005). Also, different climatic conditions and different cultivation practices lead to variable bean dimension as well as isoflavone content (Howitz and Sinclair, 2008). Isoflavones content can vary depending on temperature and soil moisture. During plant growth, the highest isoflavone concentrations occurring at low temperatures and high soil moisture (Rizzo and Baroni, 2018). Isoflavones are contained in different legumes, such as soy, kidney beans, navy beans, red clover and Japanese arrowroot called kudzu, but only soy beans and soy foods provide the main dietary sources of isoflavones in the human diet (Messina, 1999; Mazur et al., 1998). Soy foods include traditional Asian foods such as tofu, soy milk, tempeh and edamame. Each gram of soy protein in soybeans and traditional soy foods is associated with approximately 3.5 mg of isoflavones (Messina et al., 2006a,b,c). One serving of a traditional soy food, such as 100 g of tofu or 250 mL soymilk, typically provides about 25 mg isoflavones. In 1999 USDA in conjunction with Iowa State University created a database providing data on the isoflavone (including daidzein, genistein, glycitein and total isoflavones) content of selected food items (USDA).

Health Effects Beneficial effects have been associated with soy consumption, some of which may depend on phytoestrogen content (Rizzo and Baroni, 2018; Messina et al., 2006a,b,c). Epidemiological studies in Asian countries showed that a traditional diet rich in phytoestrogens was associated with a lower risk of chronic diseases and report on possible benefits in hormone-dependent prostate cancer (Applegate et al., 2018; Zuniga et al., 2013), colon cancer (Yu et al., 2016; Grosso et al., 2017a,b), breast and ovarian cancers (Loibl et al., 2011; Magee et al., 2004; Patisaul and Jefferson, 2010; Peeters et al., 2002; Grosso et al., 2017a), endometrial cancer (Zhong et al., 2018) and stomach cancer (Grosso et al., 2017b). Some common potential mechanisms of action through which isoflavones could exert protective effects in carcinogenesis have been reported. Isoflavones are structurally similar to 17b-estradiol, the primary endogenous estrogen (Tham et al., 1998; Guha et al., 2009). These molecules play a competitive role with endogenous estrogens for binding to ERa and ERb although they have lower estrogenic potency compared to estradiol (Breinholt and Larsen, 1998). Isoflavones possessed mixed ER agonist and antagonist properties, depending on the cell type and concentration of estrogen present in the surroundings (Larkin et al., 2008). It has been suggested that soy isoflavones may act as selective tissue estrogenic activity regulators and selective estrogen receptor modulator also with different mechanisms than direct receptor interactions (Nilsson and Gustafsson, 2002; Meegan and Lloyd, 2003; Riggs and Hartmann, 2003; Smith and O’Malley, 2004). ERa and ERb display distinct expression patterns in men and women, thus phytoestrogens do not exert their activity as classical estrogen agonists (Matthews and Gustafsson, 2003; Hooper et al.,

482

Diet and Cancer

2009). For example, the ability of genistein to bind to ERb may be a key factor in the inhibition of prostatic carcinogenesis but other mechanisms include cellular proliferation, apoptosis and differentiation (Mahmoud et al., 2014). Indeed, in addition to ERdepending activity, anticancer effects of soy have been largely ascribed to isoflavones which can modulate cell cycle, apoptosis, differentiation, proliferation and cell signaling (Messina et al., 2006a,b,c; Zhou et al., 1999; Vitale et al., 2016). Soy isoflavones, as a polyphenol subclass, also act exhibiting antioxidant, anti-proliferative and anti-inflammatory properties in vitro and in vivo (Tham et al., 1998; Guha et al., 2009; Patel et al., 2001). Newly discovered anticarcinogenic mechanisms of action of phytoestrogens include epigenetic modifications. In breast cancer cell lines phytoestrogens modulated chromatin transcription through epigenetic modifications such as methylation and acetylation of histones (Dagdemir et al., 2013; Grosso et al., 2017a,b). A protective action of soy isoflavones during the cancer promotion phase may explain the reduced breast cancer risk and reduced recurrence in Asian women exposed to soy since infancy. In 1995 the “early intake” hypothesis emerged showing that early isoflavone exposure may be protective against breast cancer in adulthood in animal studies. This effect may be related to isoflavones ability to impact on mammary gland development, changing cells in ways that make them permanently less likely to be transformed into cancer cells. Animal studies and epidemiological evidence support this hypothesis (Russo et al., 2005; Brown et al., 2010; De Assis et al., 2011; Rahal and Simmen, 2011; Mishra et al., 2011).

Concerns The estrogen-like properties of isoflavones have been a concern suggesting it may increase estrogenic activities and thus cancer cell proliferation particularly in those who already had breast cancer. This concern stems mainly from animal studies where pharmacological doses of isoflavones were able to stimulate the growth of existing tumors in athymic ovariectomized mice implanted with estrogen-sensitive human breast cancer (Messina, 2016). However clinical and epidemiological studies in humans indicate that isoflavone exposure is safe for all women and it improves the prognosis of breast cancer patients (Grosso et al., 2017a,b; Rizzo and Baroni, 2018). This may be due partly to the selective nature of isoflavones in binding to estrogen receptors and to their other anticancer properties mentioned above. In 2015, after a comprehensive and multiyear evaluation of the literature, the European Food Safety Authority (EFSA) concluded that in postmenopausal women, intakes of 35–150 mg per day of isoflavones from food supplements do not adversely affect sex hormones-responsive tissues such as breast and utero after 2.5 years of treatment (EFSA Panel on Food Additives and Nutrient Sources added to Food ANS, 2015). Data from this systematic review does not support the hypothesis of an increased risk of breast cancer nor of an effect on mammographic density or proliferation marker Ki-67 expression (Wu et al., 2015). Moreover, higher isoflavone intakes in breast cancer patients were associated with reduced mortality and cancer recurrence in women with ER (þ) and ER(-) breast cancer (Shu et al., 2009) and no contro-indications have been observed for women under treatment with tamoxifeen or anastrozole, two antihormonal drugs commonly used to treat breast cancer (Guha et al., 2009; Nechuta et al., 2012a,b; Kang et al., 2010).

Recommendations Up to three servings/day (up to 100 mg/day of isoflavones) as consumed by Asian populations for thousands of years do not seem to be associated to increased breast cancer risk (American Institute for Cancer Research, AICR). Considering one serving averages about 7 g of protein and 25 mg isoflavones, a moderate soy isoflavone consumption is achieved with 1–2 standard servings daily of soybeans or soy foods.

Green Tea and Green Tea Extracts (GTEs) Next to water, tea is the beverage most widely consumed worldwide, consisting of an infusion or decoction made from the leaves of a woody plant, Camellia sinensis (L.) Kuntze, belonging to the family Theaceae. The varieties of tea, derived from the leaves of the same plant, are created through different treatments and have different degrees of enzymatic oxidation of polyphenols by polyphenol oxidase (fermentation). Green tea is a nonoxidized nonfermented product which has similar phenolic compounds to unprocessed tea leaves. To preserve the polyphenols present in green tea, the leaves are arranged on bamboo surfaces and are exposed to the sun for a few hours, then they are steamed (Japanese Style Green Tea) or heated dried (Chinese Style Green Tea) to inactivate enzymes involved in the modification of polyphenols. Drying phases are alternated with folding or rolling allowing water to evaporate. When the leaves are well dried they are ready to be refined (dust and debris are eliminated) and packaged (Kosinska and Andlauer, 2014). The most common polyphenols present in green tea are catechins: (-)-epigallocatechin-3-gallate (EGCG) y 59% of total catechins, (-)-epigallocatechin (EGC) (y 19%), (-)-epicatechin-3-gallate (ECG) (y 13.6%), (-)-epicatechin (EC), (þ)-catechin (C), (-)-gallocatechin gallate (GCG) (y6.4%) and glycosylated flavonols (Lin et al., 2008). They contribute to a higher antioxidant capacity of green tea and its sensory properties. Green tea products are also available as highly concentrated extracts (GTEs) for reconstituted tea drinks (in liquid and powdered forms) or supplements.

Diet and Cancer

483

Recently, the EFSA ANS Panel provided a scientific opinion on the safety of green tea catechins from dietary sources including food supplements and infusions as a result of concerns about the possible hepatotoxicity, concluding that the catechins contained in green tea infusion and reconstituted drinks are risk-free. However, there is evidence from interventional clinical trials that intake of doses above 800 mg EGCG/day as a food supplement, may be a health concern due to increased serum transaminases (EFSA, Opinion 2018). Previously peculiar to Asia, green tea is now gaining popularity in Western countries because its consumption seems to be linked to health benefits, especially for its potential anticancer effects in primary and secondary prevention (Cabrera et al., 2006) and in cancer therapy (Lecumberri et al. 2013). Several preclinical evidences have shown that EGCG is a multipotent chemopreventive and anticancer agent in vitro and in various animal models (Fujita et al., 1989; Fujiki and Suganuma, 2002; Gupta et al., 2001; Yamane et al., 1995; Yang et al., 2009; Yiannakopoulou, 2014). The mechanisms involved include inhibition of NF-kB signaling, protection from intracellular oxidation and DNA damage and apoptosis in tumor cells (Roy et al., 2010; Sadava et al., 2007). However, in contrast to the consistent results in in vitro and animal models, results from human studies are mixed (Johnson et al., 2012; Yuan, 2013). Two high quality systematic reviews assessed the link between green tea consumption and the risk of incidence, mortality and recurrence of all cancer types (Bohem et al., 2009; Sturgeon et al., 2009). The authors concluded that despite the associations between green tea consumption and decreased risk for some cancers, the overall evidence is limitated in quantity and quality. In intervention studies an inhibitory role of oral supplementation of green tea on a precancerous lesion of the oral cavity was shown (Li et al., 1999). A limited moderate to strong evidence has been investigated for lung cancer and seem inversely associated with green tea intake among never smokers but not among smokers, probably due to the potential confounding effect of smoking that cancel out the potential protective effect of green tea (Zhong et al., 2001). In prostate cancer, observational studies do not support a role of tea intake in reducing risk but some phase II clinical trials suggested that green tea catechins may have chemopreventive properties against prostate carcinogenesis (Bettuzzi et al., 2006). The role of green tea infusion on cancer of the liver, colorectal and breast is not yet clear. However, two case control studies investigated the role of genotypes and suggested that differing angiotensin converting and folate metabolism activity genotypes in women probably play a role in the risk reduction of breast cancer with green tea (Inoue et al., 2008; Yoan et al., 2007). A field to be investigate is the interaction between green tea/GTEs supplementation and cancer drug treatments. Studies showed an interference between green tea supplementation and the chemotherapy drug Bortezomib (Golden et al., 2009) and Sunitinib (Ge et al., 2011). Until now, very little research has been done on the use of green tea or green tea extracts in treating cancer. Phase I and II trials of daily oral somministration of a green tea extract to patients with a chronic lymphocytic leukemia, who did not take any other treatment, showed that the number of leukemic cells and the size of their lymphones were reduced in a third of the participants (Shanafelt et al., 2009, 2013). More evidence is needed to confirm these preliminary results. At present there is no conclusive evidence that green tea helps to prevent or treat cancer in people because the number of studies especially randomized controlled trials are lacking and studies conducted thus far have utilized different amounts and types of green tea in different settings which makes it difficult to compare results.

Mediterranean Diet The traditional Mediterranean diet represents the dietary pattern of the Mediterranean countries in 1960s (Kromhout et al., 1989; Sofi et al., 2008; Sofi et al., 2013). The traditional Mediterranean diet is characterized by a high intake of vegetables, legumes, fruits, nuts, unprocessed cereals and extra virgin olive oil as the main source of fat, it is also characterized by a low intake of saturated fats, meat and poultry, moderately high intake of fish, moderately low intake of dairy products and ethanol, mainly wine and generally during meals (Willett et al., 1995; Sofi et al., 2014). The Mediterranean diet contributes to the intake of important vitamins and phytochemicals including b-carotene, vitamins B, C, and E, folic acid, polyphenols and it contains on average: 55%–60% carbohydrates (80% complex carbohydrates and 20% simple sugars mainly from fruit), 10%–15% proteins (40% plant protein), and 25%– 30% fats (mainly monousaturated fat from olive oil). The traditional Mediterranean diet received a great attention in the 1970s by the American physiologist Ancel Keys who lived for many years in Southern Italy and conducted the Seven Countries Study, the first major epidemiological study to investigate the relationship between diet and lifestyle in cardiovascular disease across countries and cultures (Menotti et al., 1989; Serra-Majem et al., 2006). This study suggested that the Mediterranean diet is protective against cardiovascular disease and cancer (Ferro-Luzzi and Branca, 1995). Numerous epidemiological studies showed that the Mediterranean diet is associated with a risk reduction of overall mortality, cardiovascular disease incidence and mortality, total cancer incidence and mortality, 25% reduction in colorectal cancer, 15% in breast cancer, and 10% in prostate, pancreas and endometrial cancers (Bloomfield et al., 2016).

Mediterranean Diet in Breast Cancer High adherence to Med Diet has been shown to reduce incidence and recurrence of breast cancer (Bloomfield et al., 2016). A large European observational study supports a protective role of the Med Diet and its components in breast cancer risk (Buckland et al.,

484

Diet and Cancer

2013). In international studies outside the Mediterranean basin, reductions in breast cancer risk have been observed with higher consumption of components of the Med Diet such as dietary fiber, low glycemic index carbohydrates, whole grains, fruit and vegetables and dietary marine u – 3 (Farvid et al., 2016; Kushi et al., 2012; Nicodemus et al., 2001; Zheng et al., 2013). The intervention study PREDIMED (Prevención con Dieta Mediterránea), showed that a Spanish Med Diet supplemented with at least 20% of calories as extra virgin olive oil reduced the incidence of breast cancer by 68% after 4.8 years (Toledo et al., 2015). Possible mechanisms seem to be linked to reduced inflammation, higher antioxidant, antiproliferative and apoptosisinducing properties of specific compounds present in vegetables, fruit, nuts, extra virgin olive oil and fish, flavonoids such as polyphenols present in olive oil and red wine, dietary fiber and vitamin C present in vegetables and fruit, lignans in whole grains and seeds, monounsaturated fat and squalene in olive oil, and omega-3 fatty acids in fish (Dussaillant et al., 2016; Jing et al., 2013) improving insulin sensitivity and decreasing insulin-like growth factors (Barnard et al., 2006; Breneman and Tucker, 2013). In addition the antioxidant content in extra virgin olive oil (Carruba et al., 2006) may decrease endogenous estrogens, increase sex-hormone binding globulin levels (Wu et al., 2009), neutralize free radicals, prevent DNA damage and reduce oxidative stress (Mitjavila et al., 2013; Visioli et al., 2004). These results are confirmed also by the DIANA-5 study (Diet and Androgens) (Villarini et al., 2012).

Mediterranean Diet in Colorectal Cancer The Med diet seems to be implicated also in a decreased risk of colorectal cancer although some studies are discordant. It seems that the positive effects of the Med Diet are more evident for the distal colon rather than the proximal colon (Donovan et al., 2017). Possible mechanisms are the antinflammatory effects as reported by preclinical studies. In vitro studies attributed this effect to the phenolic fraction and u – 3 content of olive oil and partly also to oleic acid, polyphenols from grapes, red wine, fruits and vegetables (Llor et al., 2003). Other studies assess that butyrate, produced from fiber by microorganisms in the gut, contribute to the oncoprotective effect. Future studies will focus on the role of the Med diet on epigenetic regulation and on changes of the microbiome as means of reducing colorectal cancer risk.

Mediterranean Diet in Prostate Cancer Approximately 10% of the incidence of prostate cancer can be prevented following a traditional Mediterranean dietary pattern (Capurso and Vendemiale, 2017), reducing risk by 4% (Bloomfield et al., 2016). Possible protective compounds may be flavonoids in fruit and vegetables which exert antiinflammatory, antioxidant, antimutagenic and antiproliferative activities (Vance et al., 2013). Important is the antioxidation ativity of lycopene from tomatoes which modulate inflammatory mechanisms of carcinogenesis (Gann et al., 1999; Graff et al., 2016). Numerous studies are also investigating the antiflammatory action of u – 3 derived from fish in prostate cancer prevention and progression (Thompson et al., 2014; Aucoin et al., 2017).

Mediterranean Diet in Endometrial Cancer Evidence associating endometrial cancer risk and specific dietary components are limitated. Howewer a 10% protection was found with the highest adherence to the Med Diet (Schwingshackl et al., 2017). Bioactive compounds and fiber of vegetables and fruit, as well as monounsatured fatty acid of extra virign olive oil, seem to modulate the estrogen levels and the inflammatory response. (Bravi et al., 2009; Filomeno et al., 2015; Lampe, 1999; Bandera et al., 2007; Fortner et al., 2017; Wang et al., 2011; Rose, 1990).

Mediterranean Diet in Pancreatic Cancer Studies on the association between the Med Diet and pancreatic cancer are scanty and discordant. Howewer some studies suggest a protection of 10% for an high adherence to the Med Diet (Trichopoulou et al., 2000) possibly due to the high consumption of vegetables and fruit because of vitamin C, D, E, and K, folate and phenolic compounds (Larsson et al., 2006; Nothlings et al., 2007; Rossi et al., 2012; Trichopoulou et al., 2000; Buckland et al., 2013; Giacosa et al., 2013; Davis-Yadley and Malafa, 2015).

Guidelines and Future Perspectives Scientific evidence supports healthy benefits of Mediterranean dietary pattern in primary and secondary prevention for the most common types of cancer. The World Cancer Research Fund (WCRF) and the American Institute for Cancer Research (AICR), based on scientific evidence, including the European code against cancer, provided lifestyle recommendations aimed to reduce the incidence and recurrence of the most common cancers worldwide. The recommendations are to avoid sugared beverages and processed

Diet and Cancer

485

meat and to limit calorie-dense and salty food, alcoholic beverages and red meat. The central recommendation is to “Eat mostly food of plant origin, with a variety of nonstarchy vegetables and of fruit every day and with unprocessed cereals and/or pulses within every meal”. these recommendations are at the base of the Med diet characteristics.(WCRF/AICR, 2007, www.iarc.it).

Fasting and Intermittent Fasting Diet In the last decades a range of dietary interventions have gained considerable popularity for the prevention and treatment of agerelated diseases, including cancer. Calorie restriction (CR) as well as fasting regimens are both dietary interventions of reduced caloric intake without malnutrition and have demonstrated a wide range of beneficial effects potentially able to prevent cancer and increase the efficacy of cancer therapies (Varady and Hellerstein, 2007; Mattson et al., 2017; Buono and Longo, 2018). Calorie restriction (CR) is a term commonly used to refer to a 20%–40% reduction in caloric intake without malnutrition (e.g., with an adequate micronutrient intake) (Mirzaei et al., 2016; Varady and Hellerstein 2007). As CR is not well tolerated by cancer patients and contraindicated for those at risk of weight loss, cachexia and sarcopenia, other caloric restricted approaches have been investigated, such as intermittent fasting (IF) which may be more suitable (Varady and Hellerstein, 2007; Rizza et al., 2014) Fasting refers to a complete absence of food intake. The fasting period for the organism is a time for standby and repair in order to harvest energy when food becomes available (Longo and Panda, 2016). Alternate cycle of feeding and fasting are at the base of fasting physiology. Fasting can be administered at various time frames and can be repeated many times. Both reducing daily energy intake and restricting the time of food intake to a few hours may trigger the fasting physiology (Longo and Panda, 2016). IF comprises different forms of fasting including alternate day fasting (ADF), periodic fasting (PF), time-restricted feeding (TRF) and Ramadan fasting (Longo and Mattson, 2014; Varady, 2016). ADF regimens generally involve a 24-h fast followed by a 24-h non-fasting period (Horne et al., 2015). PF only requires participants to fast two or more consecutive days per week, every 2 or more weeks (Hutchison and Heilbronn, 2016). TRF requires individuals to confine the period of food intake to a set period of time during the day, for example an 8-h feeding window where all food is consumed between 10:00 am and 6:00 pm (Rothschild et al., 2014). The Islamic ritual of Ramadan fasting is a form of time-restricted feeding, where food consumption is only permitted during the night time (Mazidi et al., 2015).

Evidence in Cancer Data from epidemiological and experimental studies suggest that calorie intake, the timing of food intake (e.g., fasting cycles), and dietary composition (e.g., protein, aminoacid, fat, mineral, vitamin and phytochemical intakes) are implicated in the pathogenesis of chronic diseases, including common types of cancer (Fontana et al., 2010; Mattson, 2005; Eyre et al., 2004). Chronic CR provides both beneficial and detrimental effects as well as major compliance challenges, whereas different fasting regimens, without malnutrition, are emerging as interventions with the potential to be widely used to prevent and treat cancer (Patterson and Sears, 2017; Buono and Longo, 2018; Lee et al., 2012b; O’Flanagan et al., 2017). First experiments in animal models demonstrated that fasting exerts a protective effect against DNA damage and promotes longevity (Longo and Fontana, 2010; Varady and Hellerstein, 2007). IF reduces the incidence of spontaneous tumors (Berrigan et al., 2002; Descamps et al., 2005), while PF can delay cancer progression as effectively as chemotherapy in animal models (Lee et al., 2012b). No data are available on the effects of IF on cancer rates in humans. Surrogate evidence that IF may reduce cancer risk can be derived from its effects on cancer risk factors. For example, most studies of ADF show benefits in weight reduction (Tinsley and La Bounty, 2015) and weight control is associated with a reduced risk of cancer (Renehan et al., 2008). IF has a beneficial effect on insulin resistance and diabetes which are cancer prognostic factors (Goodwin et al., 2015). Some studies have investigated the effect of fasting on the effectiveness of cancer therapy. Preliminary studies in 10 cancer patients fasted voluntarily during their chemotherapy and showed a decrease in the range of self-reported common side effects caused by chemotherapy compared to when they were on a standard diet (Safdie et al., 2009; Raffaghello et al., 2010b).

Mechanisms CR and fasting have been shown to promote stress resistance as well as longevity in a wide variety of animal models. These dietary regimens affect cellular metabolism and growth by down regulating conserved nutrient-signaling proteins and by activating stress resistance transcription factors (Fontana et al., 2010). Metabolic and molecular antitumorigenic mechanisms of CR are well established. Accumulating data have showed that CR tumor suppressive effects are mediated by enhanced apoptosis, modulation of systemic signals such as IGF-1, insulin, metabolic and inflammatory pathways, and reduced angiogenesis (O’Flanagan et al., 2017; Fontana et al., 2010). Indirect evidence suggests that CR and fasting regimens may share common mechanisms. To date no data from human studies on the effect of IF or PF in cancer prevention are available. However their effect on reducing IGF-1, insulin and glucose levels, and increasing IGFBP1 and ketone body levels could generate a protective environment that reduces DNA damage and carcinogenesis, while at the same time creating hostile conditions for cancer and pre-cancerous cells (Longo and Mattson 2014). Elevated circulating IGF-1 is associated with an increased risk of developing several cancers (Chan et al.,

486

Diet and Cancer

2000; Giovannucci et al., 2000) and individuals with severe IGF-1 deficiency caused by growth hormone receptor deficiency, rarely develop cancer (Guevara-Aguirre et al., 2011; Shevah and Laron, 2007). Findings in IGF-1 deficient subjects suggest that fasting may protect from cancer by reducing cellular and DNA damage but also by enhancing the programmed cell death of precancerous cells (Guevara-Aguirre et al., 2011). However, the effect of IF on IGF-1 levels in human studies has been variable. Some studies reported no changes in IGF-1 with weight loss due to intermittent fasting (Harvie et al., 2011), and others showed an increase in both IGFBP1 and IGFBP2 (Harvie et al., 2011; Fontana et al., 2010; Thissen et al., 1994). It is well established that a positive energy balance, as well as an increasing adiposity contributes to an increased risk of developing several types of cancer, including cancer of the colon, breast, prostate, endometrium, pancreas, liver and kidney, whereas weight loss lowers risk (Longo and Fontana, 2010; Calle and Kaaks, 2004). Evidence demonstrates the important role of dysregulation in signal transduction and metabolic pathway in tumor initiation and progression. Cancer cells are characterized by an abnormally high level of glucose requirements necessary to produce the glycolytic ATP, as well as by accumulation of a variety of mutations (e.g., the IGF-1 receptor) (Buono and Longo, 2018) making them potentially sensitive to the changes in growth factors and circulating nutrients. Furthermore, a differential stress resistance (e.g., cell mechanism used during fasting to protect normal cells but not cancer cells from chemotherapy) and differential stress sensitization (e.g., cell mechanism used during fasting to preferentially kill cancer cells and other damaged cells in combination with chemotherapy or other cytotoxic agents) of fasting conditions in normal and cancer cells have been showed in in vitro and animal models (Buono and Longo, 2018; Raffaghello et al., 2008). Fasting conditions can be protective in normal cells because they are capable of adapting to starvation (Raffaghello et al., 2008) whereas cancer cells are characterized by inability to adapt to multi-stressed environments (Cheng et al., 2014; Lee et al., 2012b).

Future Directions Several epidemiological studies clearly demonstrated that diet plays an important role in the initiation, promotion and progression of many common cancers (Kushi et al., 2006; Renehan et al., 2010). To date, there are no specific guidelines on the type of nutrition for each cancer type or cancer stage. According to the American Cancer Society, cancer patients receiving chemotherapy should increase calorie and protein intake (Doyle et al., 2006). Findings from in vitro and animal studies have reported beneficial effects of combined fasting regimens with the standard-ofcare pharmacotherapy. Dietary interventions mimicking CR can also be viewed as adjuvant anticancer strategies to be combined to standard cancer therapy (chemotherapeutic agents, ionizing radiation, and drugs with specific molecular targets). Clinical studies on the potential use of dietary treatments alongside conventional treatments are ongoing however the preliminary evidence suggests that these approaches may be a valid integrated therapy in cancer patients.

See also: Cancer Risk Reduction Through Lifestyle Changes. Dietary Factors and Cancer.

References Agerbirk, N., Olsen, C.E., 2012. Glucosinolate structures in evolution. Phytochemistry 77, 16–45. Akaza, H., 2012. Prostate cancer chemoprevention by soy isoflavones: Role of intestinal bacteria as the “second human genome.” Cancer Science 103, 969–975. Allott, E.H., Arab, L., Su, L.J., Farnan, L., Fontham, E.T., Mohler, J.L., Bensen, J.T., Steck, S.E., 2017. Saturated fat intake and prostate cancer aggressiveness: Results from the population-based North Carolina-Louisiana prostate Cancer project. Prostate Cancer and Prostatic Diseases 20 (1), 48–54. https://doi.org/10.1038/pcan.2016.39. Epub 2016 Sep 6. Allred, C.D., Ju, Y.H., Allred, K.F., Chang, J., Helferich, W.G., 2001. Dietary genistin stimulates growth of estrogen-dependent breast cancer tumors similar to that observed with genistein. Carcinogenesis 22, 1667–1673. Ames, B.N., Gold, L.S., 1991. Endogenous mutagens and the causes of aging and cancer. Mutation Research 250 (1–2), 3–16. Applegate, C.C., Rowles, J.L., Ranard, K.M., Jeon, S., Erdman, J.W., 2018. Soy consumption and the risk of prostate Cancer: An updated systematic review and meta-analysis. Nutrients 10 (1). Applegate, C.C., Rowles, J.L., Ranard, K.M., Jeon, S., Erdman, J.W., 2018. Soy consumption and the risk of prostate cancer: An updated systematic review and meta-analysis. Nutrients 10 (1). Arikawa, A.Y., Jakits, H.E., Flood, A., Thomas, W., Gross, M., Schmitz, K.H., Kurzer, M.S., 2015. Consumption of a high glycemic load but not a high glycemic index diet is marginally associated with oxidative stress in young women. Nutrition Research 35 (1), 7–13. https://doi.org/10.1016/j.nutres.2014.10.005. Aucoin, M., Cooley, K., Knee, C., Fritz, H., Balneaves, L.G., Breau, R., Fergusson, D., Skidmore, B., Wong, R., Seely, D., 2017. Fish-derived Omega-3 fatty acids and prostate Cancer: A systematic review. Integrative Cancer Therapies 16 (1), 32–62. Augustin, L.S., Dal Maso, L., La Vecchia, C., Parpinel, M., Negri, E., Vaccarella, S., Kendall, C.W., Jenkins, D.J., Francesch, S., 2001. Dietary glycemic index and glycemic load, and breast cancer risk: A case-control study. Annals of Oncology 12 (11), 1533–1538. Augustin, L.S., Franceschi, S., Jenkins, D.J., Kendall, C.W., La Vecchia, C., 2002. Glycemic index in chronic disease: A review. European Journal of Clinical Nutrition 56 (11), 1049–1071. https://doi.org/10.1038/sj.ejcn.1601454. Aune, D., Chan, D.S., Lau, R., et al., 2011. Dietary fibre, whole grains, and risk of colorectal cancer: Systematic review and dose-response meta-analysis of prospective studies. BMJ 343, d6617. Austin, G.L., Ogden, L.G., Hill, J.O., 2011. Trends in carbohydrate, fat, and protein intakes and association with energy intake in normal-weight, overweight, and obese individuals: 1971–2006. The American Journal of Clinical Nutrition 93 (4), 836–843.

Diet and Cancer

487

Bai, W., Wang, C., Ren, C., 2014. Intakes of total and individual flavonoids by US adults. International Journal of Food Science 65, 9–14. Bandera, E.V., Kushi, L.H., Moore, D.F., Gifkins, D.M., McCullough, M.L., 2007. Fruits and vegetables and endometrial cancer risk: A systematic literature review and meta-analysis. Nutrition and Cancer 58 (1), 6–21. Baracos, V.E., Mazurak, V.C., Ma, D.W., 2004. N-3 polyunsaturated fatty acids throughout the cancer trajectory: Influence on disease incidence, progression, response to therapy and cancer-associated cachexia. Nutrition Research Reviews 17, 177–192. Barba, F.J., Nikmaram, N., Roohinejad, S., Khelfa, A., Zhu, Z., Koubaa, M., 2016. Bioavailability of Glucosinolates and their breakdown products: Impact of processing. Frontiers in Nutrition 3, 24. Barclay, A.W., Petocz, P., McMillan-Price, J., Flood, V.M., Prvan, T., Mitchell, P., Brand-Miller, J.C., 2008. Glycemic index, glycemic load, and chronic disease riskdA metaanalysis of observational studies. The American Journal of Clinical Nutrition 87 (3), 627–637. Barnard, R.J., Gonzalez, J.H., Liva, M.E., Ngo, T.H., 2006. Effects of a low-fat, high-fiber diet and exercise program on breast cancer risk factors in vivo and tumor cell growth and apoptosis in vitro. Nutrition and Cancer 55 (1), 28–34. Beckman, K.B., Ames, B.N., 1997. Oxidative decay of DNA. The Journal of Biological Chemistry 272 (32), 19633–19636. Berquin, I.M., Edwards, I.J., Chen, Y.Q., 2008. Multi-targeted therapy of cancer by omega-3 fatty acids. Cancer Letters 269 (2), 363–377. https://doi.org/10.1016/j.canlet.2008.03.044. Epub 2008 May 13. Berrigan, D., Perkins, S.N., Haines, D.C., Hursting, S.D., 2002. Adult-onset calorie restriction and fasting delay spontaneous tumorigenesis in p53-deficient mice. Carcinogenesis 23, 817–822. Berrino, F., 2016. I quattro pilastri alimentari della dieta adiuvante le terapie oncologiche. In: Berrino, F. (Ed.), Il cibo dell’uomo. La via della salute tra conoscenza scientifica e antiche saggezze, pp. 121–135. FrancoAngeli/Self-help. Bettuzzi, S., Brausi, M., Rizzi, F., et al., 2006. Chemoprevention of human prostate cancer by oral administration of green tea catechins in volunteers with high-grade prostate intraepithelial neoplasia: A preliminary report from a one-year proof-of-principle study. Cancer Research 66, 1234–1240. Bloomfield, H.E., Koeller, E., Greer, N., MacDonald, R., Kane, R., Wilt, T.J., 2016. Effects on health outcomes of a Mediterranean diet with no restriction on fat intake: A systematic review and meta-analysis. Annals of Internal Medicine 165 (7), 491–500. Blum, A., Monir, M., Khazim, K., Peleg, A., Blum, A., 2007. Tomato-rich (Mediterranean) diet does not modify inflammatory markers. Clinical and Investigative Medicine 30, E70–E74. Bohem, K., Borrelli, F., Ernst, E., et al., 2009. Green tea (Camellia sinensis) for the prevention of cancer. Cochrane Database of Systematic Reviews 3, CD005004. Bolca, S., Possemiers, S., Herregat, A., Huybrechts, I., Heyerick, A., De Vriese, S., Verbruggen, M., Depypere, H., De Keukeleire, D., Bracke, M., et al., 2007. Microbial and dietary factors are associated with the equol producer phenotype in healthy postmenopausal women. The Journal of Nutrition 137, 2242–2246. Bosetti, C., Filomeno, M., Riso, P., Polesel, J., Levi, F., Talamini, R., Montella, M., Negri, E., Franceschi, S., La Vecchia, C., 2012. Cruciferous vegetables and cancer risk in a network of case-control studies. Annals of Oncology 23 (8), 2198–2203. https://doi.org/10.1093/annonc/mdr604. Bosetti, C., Filomeno, M., Riso, P., Polesel, J., Levi, F., Talamini, R., Montella, M., Negri, E., Franceschi, S., La Vecchia, C., 2012. Cruciferous vegetables and cancer risk in a network of case-control studies. Annals of Oncology 23, 2198–2203. Bosetti, C., Negri, E., Kolonel, L., et al., 2002. A pooled analysis of case-control studies of thyroid cancer. VII. Cruciferous and other vegetables (International). Cancer Causes & Control 13, 765–775. Bosland, M.C., Kato, I., Zeleniuch-Jacquotte, A., Schmoll, J., Enk Rueter, E., Melamed, J., Kong, M.X., Brown, N.M., Belles, C.A., Lindley, S.L., Zimmer-Nechemias, L.D., Zhao, X., Witte, D.P., Kim, M.O., Setchell, K.D., 2010. The chemopreventive action of equol enantiomers in a chemically induced animal model of breast cancer. Carcinogenesis 31, 886–893. Botero, D., Ebbeling, C.B., Blumberg, J.B., Ribaya-Mercado, J.D., Creager, M.A., Swain, J.F., Feldman, H.A., Ludwig, D.S., 2009. Acute effects of dietary glycemic index on antioxidant capacity in a nutrient-controlled feeding study. Obesity (Silver Spring) 17 (9), 1664–1670. https://doi.org/10.1038/oby.2009.203. Brand-Miller, J.C., Liu, V., Petocz, P., Baxter, R.C., 2005. The glycemic index of foods influences postprandial insulin-like growth factor-binding protein responses in lean young subjects. The American Journal of Clinical Nutrition 82 (2), 350–354. Bravi, F., Scotti, L., Bosetti, C., Zucchetto, A., Talamini, R., Montella, M., Greggi, S., Pelucchi, C., Negri, E., Franceschi, S., La Vecchia, C., 2009. Food groups and endometrial cancer risk: A case-control study from Italy. American Journal of Obstetrics and Gynecology 200 (3), 293. e1–7. Breinholt, V., Larsen, J.C., 1998. Detection of weak estrogenic flavonoids using a recombinant yeast strain and a modified MCF7 cell proliferation assay. Chemical Research in Toxicology 11, 622–629. Breneman, C.B., Tucker, L., 2013. Dietary fibre consumption and insulin resistance - the role of body fat and physical activity. The British Journal of Nutrition 110 (2), 375–383. Brown, N.M., Belles, C.A., Lindley, S.L., Zimmer-Nechemias, L.D., Zhao, X., Witte, D.P., Kim, M.O., Setchell, K.D., 2010. The chemopreventive action of equol enantiomers in a chemically induced animal model of breast cancer. Carcinogenesis 31, 886–893. Brownlee, M., 2005. The pathobiology of diabetic complications: A unifying mechanism. Diabetes 54 (6), 1615–1625. Buckland, G., Travier, N., Cottet, V., et al., 2013. Adherence to the mediterranean diet and risk of breast cancer in the European prospective investigation into cancer and nutrition cohort study. International Journal of Cancer 132 (12), 2918–2927. Buono, R., Longo, V.D., 2018. Starvation, stress resistance, and Cancer. Trends in Endocrinology and Metabolism 29, 271–280. Burdge, G., 2004. Alpha-linolenic acid metabolism in men and women: Nutritional and biological implications. Current Opinion in Clinical Nutrition and Metabolic Care 7 (2), 137–144. Caan, B.J., Natarajan, L., Parker, B., Gold, E.B., Thomson, C., Newman, V., Rock, C.L., Pu, M., Al-Delaimy, W., Pierce, J.P., 2011. Soy food consumption and breast cancer prognosis. Cancer Epidemiology, Biomarkers & Prevention 20, 854–858. Cabot, W., 2003. Phytoestrogens. Journal of the American Academy of Orthopaedic Surgeons 11, 153–156. Cabrera, C., Artacho, R., Giménez, R., 2006. Beneficial effects of green tea: A review. Journal of the American College of Nutrition 25, 79–99. Calle, E.E., Kaaks, R., 2004. Overweight, obesity and cancer: Epidemiological evidenceand proposed mechanisms. Nature Reviews. Cancer 4, 579–591. Camilleri, G.M., Verger, E.O., Huneau, J.F., Carpentier, F., Dubuisson, C., Mariotti, F., 2013. Plant and animal protein intakes are differently associated with nutrient adequacy of the diet of french adults. The Journal of Nutrition 143, 1466–1473. Cao, Y., Hou, L., Wang, W., 2016. Dietary total fat and fatty acids intake, serum fatty acids and risk of breast cancer: A meta-analysis of prospective cohort studies. International Journal of Cancer 138 (8), 1894–1904. https://doi.org/10.1002/ijc.29938. Epub 2015 Dec 9. Capurso, C., Vendemiale, G., 2017. The Mediterranean diet reduces the risk and mortality of the prostate Cancer: A narrative review. Front Nutr 4, 38. Carruba, G., Granata, O.M., Pala, V., Campisi, I., Agostara, B., Cusimano, R., Ravazzolo, B., Traina, A., 2006. A traditional Mediterranean diet decreases endogenous estrogens in healthy postmenopausal women. Nutrition and Cancer 56 (2), 253–259. Ceriello, A., Bortolotti, N., Crescentini, A., Motz, E., Lizzio, S., Russo, A., Ezsol, Z., Tonutti, L., Taboga, C., 1998. Antioxidant defences are reduced during the oral glucose tolerance test in normal and non-insulin-dependent diabetic subjects. European Journal of Clinical Investigation 28 (4), 329–333. Chan, J.M., Stampfer, M.J., Giovannucci, E., Ma, J., Pollak, M., 2000. Insulin-like growth factor I (IGF-I), IGFbinding protein-3 and prostate cancer risk: Epidemiological studies. Growth Hormone & IGF Research 10 (Suppl A), S32–S33. Chen, H., Qin, S., Wang, M., Zhang, T., Zhang, S., 2015. Association between cholesterol intake and pancreatic cancer risk: Evidence from a meta-analysis. Scientific Reports 5, 8243. https://doi.org/10.1038/srep08243.

488

Diet and Cancer

Chen, M., Rao, Y., Zheng, Y., Wei, S., Li, Y., Guo, T., Yin, P., 2014. Association between soy isoflavone intake and breast cancer risk for pre- and post-menopausal women: A metaanalysis of epidemiological studies. PLoS One 9, e89288. Cheng, C.W., et al., 2014. Prolonged fasting reduces IGF-1/PKA to promote hematopoietic-stem-cell-based regeneration and reverse immunosuppression. Cell Stem Cell 14, 810–823. Cheng, G., Wilczek, B., Warner, M., Gustafsson, J.A., Landgren, B.M., 2007. Isoflavone treatment for acute menopausal symptoms. Menopause 14, 468–473. Chi, F., Wu, R., Zeng, Y.C., Xing, R., Liu, Y., 2013. Circulation insulin-like growth factor peptides and colorectal cancer risk: An updated systematic review and meta-analysis. Molecular Biology Reports 40 (5), 3583–3590. https://doi.org/10.1007/s11033-012-2432-z. Chi, F., Wu, R., Zeng, Y.C., Xing, R., Liu, Y., Xu, Z.G., 2013. Post-diagnosis soy food intake and breast cancer survival: A meta-analysis of cohort studies. Asian Pacific Journal of Cancer Prevention 14, 2407–2412. Chlebowski, R.T., Blackburn, G.L., Thomson, C.A., Nixon, D.W., Shapiro, A., Hoy, M.K., Goodman, M.T., Giuliano, A.E., Karanja, N., McAndrew, P., Hudis, C., Butler, J., Merkel, D., Kristal, A., Caan, B., Michaelson, R., Vinciguerra, V., Del Prete, S., Winkler, M., Hall, R., Simon, M., Winters, B.L., Elashoff, R.M., 2006. Dietary fat reduction and breast cancer outcome: Interim efficacy results from the Women’s intervention nutrition study. Journal of the National Cancer Institute 98 (24), 1767–1776. Choi, Y., Giovannucci, E., Lee, J.E., 2012. Glycaemic index and glycaemic load in relation to risk of diabetes-related cancers: A meta-analysis. The British Journal of Nutrition 108 (11), 1934–1947. https://doi.org/10.1017/S0007114512003984. Coleman, H.G., Kitahara, C.M., Murray, L.J., Dodd, K.W., Black, A., Stolzenberg-Solomon, R.Z., Cantwell, M.M., 2014. Dietary carbohydrate intake, glycemic index, and glycemic load and endometrial cancer risk: A prospective cohort study. American Journal of Epidemiology 179 (1), 75–84. https://doi.org/10.1093/aje/kwt222. Conaway, C.C., Getahun, S.M., Liebes, L.L., et al., 2000. Disposition of glucosinolates and sulforaphane in humans after ingestion of steamed and fresh broccoli. Nutrition and Cancer 38, 168–178, 26. Cotton, S.C., Sharp, L., Little, J., Brockton, N., 2000. Glutathione S-transferase polymorphisms and colorectal cancer: A HuGE review. American Journal of Epidemiology 151, 7–32. Crawley, D.J., Holmberg, L., Melvin, J.C., Loda, M., Chowdhury, S., Rudman, S.M., Van Hemelrijck, M., 2014. Serum glucose and risk of cancer: A meta-analysis. BMC Cancer 14, 985. https://doi.org/10.1186/1471-2407-14-985. Cui, X., Rosner, B., Willett, W.C., Hankinson, S.E., 2011. Dietary fat, fiber, and carbohydrate intake in relation to risk of endometrial cancer. Cancer Epidemiology, Biomarkers & Prevention 20 (5), 978–989. https://doi.org/10.1158/1055-9965.EPI-10-1089. Dagdemir, A., Durif, J., Ngollo, M., Bignon, Y.J., Bernard-Gallon, D., 2013. Histone lysine trimethylation or acetylation can be modulated by phytoestrogen, estrogen or anti-HDAC in breast cancer cell lines. Epigenomics 5, 51–63. Dagdemir, A., Durif, J., Ngollo, M., Bignon, Y.J., Bernard-Gallon, D., 2013. Histone lysine trimethylation or acetylation can be modulated by phytoestrogen, estrogen or anti-HDAC in breast cancer cell lines. Epigenomics 5, 51–63. Dal Maso, L., Zucchetto, A., Talamini, R., Serraino, D., Stocco, C.F., Vercelli, M., Falcini, F., Franceschi, S., 2008. Prospective analysis of case-control studies on environmental factors and health (PACE) study group. Effect of obesity and other lifestyle factors on mortality in women with breast cancer. International Journal of Cancer 123 (9), 2188–2194. https://doi.org/10.1002/ijc.23747. Davey, G.K., Spencer, E.A., Appleby, P.N., Allen, N.E., Knox, K.H., Key, T.J., 2003. EPIC-Oxford: Lifestyle characteristics and nutrient intakes in a cohort of 33 883 meat-eaters and 31 546 non meat-eaters in the UK. Public Health Nutrition 6 (3), 259–269. Davis-Yadley, A.H., Malafa, M.P., 2015. Vitamins in pancreatic cancer: A review of underlying mechanisms and future applications. Advances in Nutrition 6 (6), 774–802. De Assis, S., Warri, A., Benitez, C., Helferich, W., Hilakivi-Clarke, L., 2011. Protective effects of prepubertal genistein exposure on mammary tumorigenesis are dependent on BRCA1 expression. Cancer Prevention Research 4, 1436–1448. De Assis, S., Warri, A., Benitez, C., Helferich, W., Hilakivi-Clarke, L., 2011. Protective effects of prepubertal genistein exposure on mammary tumorigenesis are dependent on BRCA1 expression. Cancer Prevention Research 4, 1436–1448. Descamps, O., Riondel, J., Ducros, V., Roussel, A.M., 2005. Mitochondrial production of reactive oxygen species and incidence of age-associated lymphoma in OF1 mice: Effect of alternate-day fasting. Mechanisms of Ageing and Development 126, 1185–1191. Doll, R., Peto, R., 1981. The causes of cancer: Quantitative estimates of avoidable risks of cancer in the United States today. Journal of the National Cancer Institute 66 (6), 1191–1308. Donovan, M.G., Selmin, O.I., Doetschman, T.C., Romagnolo, D.F., 2017. Mediterranean diet: Prevention of colorectal Cancer. Frontiers in Nutrition 4, 59. Dorjgochoo, T., Gu, K., Zheng, Y., Kallianpur, A., Chen, Z., et al., 2011. Soy intake in association with menopausal symptoms during the first 6 and 36 months after breast cancer diagnosis. Breast Cancer Research and Treatment 130, 879–889. Doyle, C., Kushi, L.H., Byers, T., Courneya, K.S., Demark-Wahnefried, W., Grant, B., et al., 2006. Nutrition and physical activity during and after cancer treatment: An American Cancer Society guide for informed choices. CA: A Cancer Journal for Clinicians 56, 323–353. Dussaillant, C., Echeverria, G., Urquiaga, I., Velasco, N., Rigotti, A., 2016. Revista Médica de Chile 144 (8), 1044–1052. Current evidence on health benefits of the mediterranean diet. EFSA Panel on Food Additives and Nutrient Sources added to Food (ANS), 2015. Risk assessment for peri- and post-menopausal women taking food supplements containing isolated isoflavones: Safety of isoflavones from food supplements in menopausal women. EFSA Journal 13, 4246. EFSA Panel on Food Additives and Nutrient Sources added to Food (ANS), 2015. Risk assessment for peri- and post-menopausal women taking food supplements containing isolated isoflavones: Safety of isoflavones from food supplements in menopausal women. EFSA Journal 13, 4246. Endogenous, Hormones, Group Breast Cancer Collaborative, Key, T.J., Appleby, P.N., Reeves, G.K., Roddam, A.W., 2010. Insulin-like growth factor 1 (IGF1), IGF binding protein 3 (IGFBP3), and breast cancer risk: Pooled individual data analysis of 17 prospective studies. The Lancet Oncology 11 (6), 530–542. https://doi.org/10.1016/S1470-2045(10) 70095-4. Escrich, E., Solanas, M., Moral, R., Escrich, R., 2011. Modulatory effects and molecular mechanisms of olive oil and other dietary lipids in breast cancer. Current Pharmaceutical Design 17 (8), 813–830. European Food Safety Authority (EFSA) Panel on Food Additives and Nutrient Sources added to Food (ANS), 2018. Scientific opinion on the safety of green tea catechins. EFSA Journal 16, 5239. Eyre, H., et al., 2004. Preventing cancer, cardiovascular disease, and diabetes: A common agenda for the American Cancer Society the American Diabetes Association, and the American Heart Association. CA: A Cancer Journal for Clinicians 54, 190–207. Fardet, A., 2010. New hypotheses for the health-protective mechanisms of whole-grain cereals: What is beyond fibre? Nutrition Research Reviews 23 (1), 65–134. https://doi.org/ 10.1017/S0954422410000041. Farvid, M.S., Cho, E., Chen, W.Y., Eliassen, A.H., Willett, W.C., 2014. Premenopausal dietary fat in relation to pre and post-menopausal breast cancer. Breast Cancer Research and Treatment 145, 255–265. Farvid, M.S., Cho, E., Eliassen, A.H., Chen, W.Y., Willett, W.C., 2016. Lifetime grain consumption and breast cancer risk. Breast Cancer Research and Treatment 159 (2), 335–345. Favero, A., Salvini, S., Russo, A., Parpinel, M., Negri, E., Decarli, A., La Vecchia, C., Giacosa, A., Franceschi, S., 1997. Sources of macro- and micronutrients in Italian women: Results from a food frequency questionnaire for cancer studies. European Journal of Cancer Prevention 6 (3), 277–287. Ferro-Luzzi, A., Branca, F., 1995. Mediterranean diet, Italian-style: Prototype of a healthy diet. The American Journal of Clinical Nutrition 61 (6 Suppl), 1338S–1345S. Filomeno, M., Bosetti, C., Bidoli, E., Levi, F., Serraino, D., Montella, M., La Vecchia, C., Tavani, A., 2015. Mediterranean diet and risk of endometrial cancer: A pooled analysis of three Italian case-control studies. British Journal of Cancer 112 (11), 1816–1821.

Diet and Cancer

489

Fleshner, N.E., Kapusta, L., Donnelly, B., Tanguay, S., Chin, J., Hersey, K., Farley, A., Jansz, K., Siemens, D.R., Trpkov, K., et al., 2011. Progression from high-grade prostatic intraepithelial neoplasia to cancer: A randomized trial of combination vitamin-E, soy, and selenium. Journal of Clinical Oncology 29, 2386–2390. Flood, A., Velie, E.M., Chaterjee, N., et al., 2002. Fruit and vegetable intakes and the risk of colorectal cancer in the breast Cancer detection demonstration project follow-up cohort. The American Journal of Clinical Nutrition 75, 936–943. Fonseca-Nunes, A., Jakszyn, P., Agudo, A., 2014. Iron and cancer riskdA systematic review and meta-analysis of the epidemiological evidence. Cancer Epidemiology, Biomarkers & Prevention 23, 12–31. Fontana, L., Partridge, L., Longo, V.D., 2010. Extending healthy life span–from yeast to humans. Science 328, 321–326. Fortner, R.T., Husing, A., Kuhn, T., et al., 2017. Endometrial cancer risk prediction including serum-based biomarkers: Results from the EPIC cohort. International Journal of Cancer 140 (6), 1317–1323. Fowke, J.H., Longcope, C., Hebert, J.R., 2000. Brassica vegetable consumption shifts estrogen metabolism in healthy postmenopausal women. Cancer Epidemiology, Biomarkers & Prevention 9, 773–779. Franceschi, S., Dal Maso, L., Augustin, L., Negri, E., Parpinel, M., Boyle, P., Jenkins, D.J., La Vecchia, C., 2001. Dietary glycemic load and colorectal cancer risk. Annals of Oncology 12 (2), 173–178. Franceschi, S., Favero, A., La Vecchia, C., Negri, E., Conti, E., Montella, M., Giacosa, A., Nanni, O., Decarli, A., 1997. Food groups and risk of colorectal cancer in Italy. International Journal of Cancer 72 (1), 56–61. Fujiki, H., Suganuma, M., 2002. Green tea and cancer prevention. Proceedings of the Japan Academy 78, 263–270. Fujita, Y., Yamane, T., Tanaka, M., et al., 1989. Inhibitory effect of (-)-epigallocatechin gallate on carcinogenesis with N-ethyl-N’-nitro-N-nitrosoguanidine in mouse duodenum. Japanese Journal of Cancer Research 80, 503–505. Gann, P.H., Ma, J., Giovannucci, E., Willett, W., Sacks, F.M., Hennekens, C.H., Stampfer, M.J., 1999. Lower prostate cancer risk in men with elevated plasma lycopene levels: Results of a prospective analysis. Cancer Research 59 (6), 1225–1230. Ge, J., Tan, B.X., Chen, Y., et al., 2011. Interaction of green tea polyphenol epigallocatechin-3-gallate with sunitinib: Potential risk of dimin- ished sunitinib bioavailability. Journal of Molecular Medicine (Berl) 89 (6), 595e602. Getahun, S.M., Chung, F.L., 1999. Conversion of glucosinolates to isothiocyanates in humans after ingestion of cooked watercress. Cancer Epidemiology, Biomarkers & Prevention 8, 447–451. Giacosa, A., Barale, R., Bavaresco, L., Gatenby, P., Gerbi, V., Janssens, J., Johnston, B., Kas, K., La Vecchia, C., Mainguet, P., Morazzoni, P., Negri, E., Pelucchi, C., Pezzotti, M., Rondanelli, M., 2013. Cancer prevention in Europe: The Mediterranean diet as a protective choice. European Journal of Cancer Prevention 22 (1), 90–95. Giovannucci, E., 1995. Insulin and colon cancer. Cancer Causes & Control 6 (2), 164–179. Giovannucci, E., 2003. Nutrition, insulin, insulin-like growth factors and cancer. Hormone and Metabolic Research 35 (11 12), 694–704. https://doi.org/10.1055/s-2004814147. Giovannucci, E., Pollak, M., Platz, E.A., Willett, W.C., Stampfer, M.J., Majeed, N., Colditz, G.A., Speizer, F.E., Hankinson, S.E., 2000. Insulin-like growth factor I (IGF-I), IGF-binding protein-3 and the risk of colorectal adenoma and cancer in the nurses’ health study. Growth Hormone & IGF Research 10 (Suppl A), S30–S31. Giovannucci, E., Rimm, E.B., Liu, Y., Stampfer, M.J., 2003. Willett WC. A prospective study of cruciferous vegetables and prostate cancer. Cancer epidemiology. Biomarkers & Prevention 12, 1403–1409. Golden, E.B., Lam, P.Y., Kardosh, A., et al., 2009. Green tea polyphenols block the anticancer effects of bortezomib and other boronic acid-based proteasome inhibitors. Blood 113 (23), 5927–5937. Goodman-Gruen, D., Kritz-Silverstein, D., 2001. Usual dietary isoflavone intake is associated with cardiovascular disease risk factors in postmenopausal women. The Journal of Nutrition 131, 1202–1206. Goodwin, P.J., Ambrosone, C.B., Hong, C.C., 2015. Modifiable lifestyle factors andbreast cancer outcomes: Current controversies and researchrecommendations. Adv. Exp. Med. Biol. 862, 177–192. Goodwin, P.J., Ennis, M., Pritchard, K.I., Trudeau, M.E., Koo, J., Madarnas, Y., Hartwick, W., Hoffman, B., Hood, N., 2002. Fasting insulin and outcome in early-stage breast cancer: Results of a prospective cohort study. Journal of Clinical Oncology 20 (1), 42–51. Graff, R.E., Pettersson, A., Lis, R.T., Ahearn, T.U., Markt, S.C., Wilson, K.M., Rider, J.R., Fiorentino, M., Finn, S., Kenfield, S.A., Loda, M., Giovannucci, E.L., Rosner, B., Mucci, L.A., 2016. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. The American Journal of Clinical Nutrition 103 (3), 851–860. Grosso, G., Bella, F., Godos, J., Sciacca, S., Del Rio, D., Ray, S., Galvano, F., Giovannucci, E.L., 2017a. Possible role of diet in cancer: Systematic review and multiple metaanalyses of dietary patterns, lifestyle factors, and cancer risk. Nutrition Reviews 75, 405–419. Grosso, G., Godos, J., Lamuela-Raventos, R., Ray, S., Micek, A., Pajak, A., Sciacca, S., D’Orazio, N., Del Rio, D., Galvano, F., 2017b. A comprehensive meta-analysis on dietary flavonoid and lignan intake and cancer risk: Level of evidence and limitations. Molecular Nutrition & Food Research 61. Grosso, G., Micek, A., Godos, J., Pajak, A., Sciacca, S., Galvano, F., Giovannucci, E.L., 2017. Dietary flavonoid and Lignan intake and mortality in prospective cohort studies: Systematic review and dose-response meta-analysis. American Journal of Epidemiology 1–13. Guevara-Aguirre, J., Balasubramanian, P., Guevara-Aguirre, M., Wei, M., Madia, F., Cheng, C.W., Hwang, D., Martin-Montalvo, A., Saavedra, J., Ingles, S., de Cabo, R., Cohen, P., Longo, V.D., 2011. Growth hormone receptor deficiency is associated with a major reduction in pro-aging signaling, cancer, and diabetes in humans. Science Translational Medicine 3, 7–13. Guha, N., Kwan, M.L., Quesenberry Jr., C.P., Weltzien, E.K., Castillo, A.L., Caan, B.J., 2009. Soy isoflavones and risk of cancer recurrence in a cohort of breast cancer survivors: The life after Cancer epidemiology study. Breast Cancer Research and Treatment 118, 395–405. Guha, N., Kwan, M.L., Quesenberry, C.P., Weltzien, E.K., Castillo, A.L., Caan, B.J., 2009. Soy isoflavones and risk of cancer recurrence in a cohort of breast cancer survivors: The life after cancer epidemiology study. Breast Cancer Research and Treatment 118, 395–405. Guo, L., Yang, R., Wanga, Z., Zhenxin, G., 2015. Effect of freezing methods on sulforaphane formation in broccoli sprouts. RSC Advances 5, 32290. Gupta, S., Hastak, K., Ahmad, N., Lewin, J.S., Mukhtar, H., 2001. Inhibition of prostate carcinogenesis in TRAMP mice by oral infusion of green tea polyphenols. Proceedings of the National Academy of Sciences of the United States of America 98, 10350–10355. Halkjaer, J., Olsen, A., Bjerregaard, L., Deharveng, G., Tjønneland, A., Welch, A., Crowe, F., Wirfält, E., Hellstrom, V., Niravong, M., 2009. Intake of total, animal and plant proteins, and their food sources in 10 countries in the European prospective investigation into cancer and nutrition. European Journal of Clinical Nutrition 63, S16–S36. Hall, M.N., Chavarro, J.E., Lee, I.M., Willett, W.C., Ma, J., 2008. A 22-year prospective study of fish, n-3 fatty acid intake, and colorectal cancer risk in men. Cancer Epidemiology, Biomarkers & Prevention 17, 1136–1143. Halliwell, B., 2007. Oxidative stress and cancer: Have we moved forward? The Biochemical Journal 401 (1), 1–11. https://doi.org/10.1042/BJ20061131. Harvie, M.N., Pegington, M., Mattson, M.P., Frystyk, J., Dillon, B., Evans, G., et al., 2011. The effects of intermittent or continuous energy restriction on weight loss and metabolic disease risk markers: A randomized trial in young overweight women. International Journal of Obesity 35, 714–727. Hedlund, M., Padler-Karvani, V., Varki, N.M., Varkia, A., 2008. Evidence for a human-specific mechanism for diet and antibody-mediated inflammation in carcinoma progression. Proceedings of the National Academy of Sciences of the United States of America 105, 18936–18941. Helferich, W.G., Andrade, J.E., Hoagland, M.S., 2008. Phytoestrogens and breast cancer: A complex story. Inflammopharmacology 16, 219–226. Higdon, J.V., Delage, B., Williams, D.E., Dashwood, R.H., 2007. Cruciferous vegetables and human cancer risk: Epidemiologic evidence and mechanistic basis. Pharmacological Research 55, 224–236.

490

Diet and Cancer

Higginbotham, S., Zhang, Z.F., Lee, I.M., Cook, N.R., Giovannucci, E., Buring, J.E., Liu, S., Women’s Health, S., 2004. Dietary glycemic load and risk of colorectal cancer in the Women’s health study. Journal of the National Cancer Institute 96 (3), 229–233. Hooper, L., Madhavan, G., Tice, J.A., Leinster, S.J., Cassidy, A., 2010. Effects of isoflavones on breast density in pre- and post-menopausal women: A systematic review and metaanalysis of randomized controlled trials. Human Reproduction Update 16, 745–760. Hooper, L., Ryder, J.J., Kurzer, M.S., Lampe, J.W., Messina, M.J., Phipps, W.R., Cassidy, A., 2009. Effects of oy protein and isoflavones on circulating hormone concentrations in pre- and post-menopausal women: A systematic review and meta-analysis. Human Reproduction Update 15, 423–440. Horne, B.D., Muhlestein, J.B., Anderson, J.L., 2015. Health effects of intermittent fasting: Hormesis or harm? A systematic review. The American Journal of Clinical Nutrition 102, 464–470. Hou, R., Yao, S.S., Liu, J., Wang, L.L., Wu, L., Jiang, L., 2017. Dietary n-3 polyunsaturated fatty acids, fish consumption, and endometrial cancer risk: A meta-analysis of epidemiological studies. Oncotarget 8 (53), 91684–91693. https://doi.org/10.18632/oncotarget.18295 eCollection 2017 Oct 31. Howitz, K.T., Sinclair, D.A., 2008. Xenohormesis: Sensing the chemical cues of other species. Cell 133, 387–391. Hu, J., La Vecchia, C., de Groh, M., Negri, E., Morrison, H., Mery, L., 2011. Canadian Cancer registries epidemiology research group. Dietary transfatty acids and cancer risk. European Journal of Cancer Prevention 20 (6), 530–538. https://doi.org/10.1097/CEJ.0b013e328348fbfb. Hu, J., La Vecchia, C., Negri, E., Mery, L., 2010. Nutrients and risk of colon cancer. Cancers (Basel) 2 (1), 51–67. https://doi.org/10.3390/cancers2010051. Huerta-Yépez, S., Tirado-Rodriguez, A.B., Hankinson, O., 2016. Role of diets rich in omega-3 and omega-6 in the development of cancer. Boletín Médico del Hospital Infantil de México 73 (6), 446–456. https://doi.org/10.1016/j.bmhimx.2016.11.001. Epub 2016 Nov 30. Hutchison, A.T., Heilbronn, L.K., 2016. Metabolic impacts of altering meal frequency and timing–does when we eat matter? Biochimie 124, 187–197. IARC. Cruciferous vegetables, isothiocyanates and indoles. IARC Handbooks of Cancer Prevention Vol. 9. Lyon: IARC Press, 2004 (accessed on 14 April 2018). Ide, H., Tokiwa, S., Sakamaki, K., Nishio, K., Isotani, S., Muto, S., Hama, T., Masuda, H., Horie, S., 2010. Combined inhibitory effects of soy isoflavones and curcumin on the production of prostate-specific antigen. Prostate 70, 1127–1133. Inoue, M., Robien, K., Wang, R., et al., 2008. Green tea intake, MTHFR/TYMS genotype and breast cancer risk: The Singapore Chinese health study. Carcinogenesis 29, 1967–1972. Inoue, I., Shinoda, Y., Nakano, T., Sassa, M., Goto, S., Awata, T., Komoda, T., Katayama, S., 2006. Acarbose ameliorates atherogenecity of low-density lipoprotein in patients with impaired glucose tolerance. Metabolism 55 (7), 946–952. https://doi.org/10.1016/j.metabol.2006.03.002. International Agency for Research on Cancer, 2014. In: Stewart, B.W., Wild, C.P. (Eds.), World Cancer Report 2014. Lyon, France. Jackson, R.L., Greiwe, J.S., Schwen, R.J., 2011. Emerging evidence of the health benefits of S-equol, an estrogen receptor beta agonist. Nutrition Reviews 69, 432–448. Jacob, K., Periago, M.J., Bohm, V., Berruezo, G.R., 2008. Influence of lycopene and vitamin C from tomato juice on biomarkers of oxidative stress and inflammation. The British Journal of Nutrition 99, 137–146. Jain, M.G., Hislop, G.T., Howe, G.R., Ghadirian, P., 1999. Plant foods, antioxidants, and prostate cancer risk: Findings from case-control studies in Canada. Nutrition and Cancer 34, 173–184. Järvinen, R., Knekt, P., Hakulinen, T., Rissanen, H., Heliövaara, M., 2001. Dietary fat, cholesterol and colorectal cancer in a prospective study. British Journal of Cancer 85 (3), 357–361. Jee, S.H., Ohrr, H., Sull, J.W., Yun, J.E., Ji, M., Samet, J.M., 2005. Fasting serum glucose level and cancer risk in Korean men and women. JAMA 293 (2), 194–202. https:// doi.org/10.1001/jama.293.2.194. Jenkins, D.J., Wolever, T.M., Taylor, R.H., Barker, H., Fielden, H., Baldwin, J.M., Bowling, A.C., Newman, H.C., Jenkins, A.L., Goff, D.V., 1981. Glycemic index of foods: A physiological basis for carbohydrate exchange. The American Journal of Clinical Nutrition 34 (3), 362–366. Jin, Y., Qu, S., Tesikova, M., Wang, L., Kristian, A., Maelandsmo, G.M., Kong, H., Zhang, T., Jeronimo, C., Teixeira, M.R., et al., 2013. Molecular circuit involving KLK4 integrates androgen and mTOR signaling in prostate cancer. Proceedings of the National Academy of Sciences of the United States of America 110, E2572–E2581. Jing, K., Wu, T., Lim, K., 2013. Omega-3 polyunsaturated fatty acids and cancer. Anti-Cancer Agents in Medicinal Chemistry 13 (8), 1162–1177. Johnson, R., Bryant, S., Huntley, L.A., 2012. Green tea and gree tea catechin extracts: An overview of the clinical evidence. Maturitas 73, 280–287. Jones, J.L., Daley, B.J., Enderson, B.L., et al., 2002. Genistein inhibits tamoxifen effects on cell proliferation and cell cycle arrest in T47D breast cancer cells. The American Surgeon 68, 575–577. Jones, R.B., Frisina, C.L., Winkler, S., Imsic, M., Tomkins, R.B., 2010. Cooking method significantly effects glucosinolate content and sulforaphane production in broccoli florets. Food Chemistry (2), 237–242. Ju, Y.H., Allred, C.D., Allred, K.F., Karko, K.L., Doerge, D.R., Helferich, W.G., 2001. Physiological concentrations of dietary genistein dose-dependently stimulate growth of estrogendependent human breast cancer (MCF-7) tumors implanted in athymic nude mice. The Journal of Nutrition 131, 2957–2962. Ju, Y.H., Doerge, D.R., Woodling, K.A., et al., 2008. Dietary genistein negates the inhibitory effect of letrozole on the growth of aromatase-expressing estrogen-dependent human breast cancer cells (MCF-7Ca) in vivo. Carcinogenesis 29, 2162–2168. Jump, D.B., 2004. Fatty acid regulation of gene transcription. Critical Reviews in Clinical Laboratory Sciences 41 (1), 41–78. Kaaks, R., Lukanova, A., 2001. Energy balance and cancer: The role of insulin and insulin-like growth factor-I. The Proceedings of the Nutrition Society 60 (1), 91–106. Kang, D.H., 2002. Oxidative stress, DNA damage, and breast cancer. AACN Advanced Critical Care 13 (4), 540–549. Kang, L., Ding, L., Wang, Z.Y., 2009. Isothiocyanates repress estrogen receptor alpha expression in breast cancer cells. Oncology Reports 21, 185–192. Kang, X., Zhang, Q., Wang, S., Huang, X., Jin, S., 2010. Effect of soy isoflavones on breast cancer recurrence and death for patients receiving adjuvant endocrine therapy. CMAJ 182, 1857–1862. Kang, X., Zhang, Q., Wang, S., Huang, X., Jin, S., 2010. Effect of soy isoflavones on breast cancer recurrence and death for patients receiving adjuvant endocrine therapy. CMAJ 182, 1857–1862. Key, T.J., Allen, N., Appleby, P., et al., 2004. Fruits and vegetables and prostate cancer: No association among 1104 cases in a prospective study of 130544 men in the European prospective investigation into Cancer and nutrition (EPIC). International Journal of Cancer 109, 119–124. Khan, S.A., Chatterton, R.T., Michel, N., Bryk, M., Lee, O., Ivancic, D., Heinz, R., Zalles, C.M., Helenowski, I.B., Jovanovic, B.D., et al., 2012. Soy isoflavone supplementation for breast cancer risk reduction: A randomized phase II trial. Cancer Prevention Research 5, 309–319. King, H., Li, J.Y., Locke, F.B., Pollack, E.S., Tu, J.T., 1985. Patterns of site-specific displacement in cancer mortality among migrants: The Chinese in the United States. American Journal of Public Health 75 (3), 237–242. Kolahdouz Mohammadi, R., Bagheri, M., Kolahdouz Mohammadi, M., Shidfar, F., 2017. Ruminant trans-fatty acids and risk of breast cancer: A systematic review and meta-analysis of observational studies. Minerva Endocrinologica 42 (4), 385–396. https://doi.org/10.23736/S0391-1977.16.02514-1. Epub 2016 Sep 14. Kolonel, L.N., Hankin, J.H., Whittemore, A.S., et al., 2000. Vegetables, fruits, legumes and prostate cancer: A multiethnic case-control study. Cancer epidemiology. Biomarkers & Prevention 9, 795–804. Kosinska, A., Andlauer, W., 2014. Antioxidant capacity of tea: Effect of processing and storage. In: Preedy, V. (Ed.), Processing and impact on antioxidants in beverages. Elsevier, pp. 109–120. Kromhout, D., Keys, A., Aravanis, C., Buzina, R., Fidanza, F., Giampaoli, S., Jansen, A., Menotti, A., Nedeljkovic, S., Pekkarinen, M., et al., 1989. Food consumption patterns in the 1960s in seven countries. The American Journal of Clinical Nutrition 49 (5), 889–894.

Diet and Cancer

491

Kushi, L.H., et al., 2006. American Cancer Society guidelines on nutrition and physical activity for cancer prevention: Reducing the risk of cancer with healthy food choices and physical activity. CA: A Cancer Journal for Clinicians 56, 254–281 quiz 313–314. Kushi, L.H., Doyle, C., McCullough, M., Rock, C.L., Demark-Wahnefried, W., Bandera, E.V., Gapstur, S., Patel, A.V., Andrews, K., Gansler, T., 2012. The American Cancer Society 2010 nutrition and physical activity guidelines advisory committee. American Cancer Society Guidelines on Nutrition and Physical Activity for Cancer Prevention. CA: A Cancer Journal for Clinicians 62, 30–67. Kushi, L.H., Doyle, C., McCullough, M., Rock, C.L., Demark-Wahnefried, W., Bandera, E.V., Gapstur, S., Patel, A.V., Andrews, K., Gansler, T., American Cancer Society N, Physical Activity Guidelines Advisory C, 2012. American Cancer Society guidelines on nutrition and physical activity for cancer prevention: Reducing the risk of cancer with healthy food choices and physical activity. Ca-A Cancer Journal for Clinicians, 62 (1), 30–67. Kushi, L.H., Doyle, C., McCullough, M., Rock, C.L., Demark-Wahnefried, W., Bandera, E.V., Gapstur, S., Patel, A.V., Andrews, K., Gansler, T., Nutrition American Cancer Society, and Committee Physical Activity Guidelines Advisory, 2012. American Cancer Society guidelines on nutrition and physical activity for cancer prevention: Reducing the risk of cancer with healthy food choices and physical activity. CA: A Cancer Journal for Clinicians 62 (1), 30–67. https://doi.org/10.3322/caac.20140. Kwan, W., Duncan, G., Van Patten, C., Liu, M., Lim, J., 2010. A phase II trial of a soy beverage for subjects without clinical disease with rising prostate-specific antigen after radical radiation for prostate cancer. Nutrition and Cancer 62, 198–207. Lai, R., Bian, Z., Lin, H., Ren, J., Zhou, H., Guo, H., 2017. The association between dietary protein intake and colorectal cancer risk: A meta-analysis. World Journal of Surgical Oncology 15, 169. Lamartiniere, C.A., Moore, J.B., Brown, N.M., Thompson, R., Hardin, M.J., Barnes, S., 1995b. Genistein suppresses mammary cancer in rats. Carcinogenesis 16, 2833–2840. Lamartiniere, C.A., Moore, J., Holland, M., Barnes, S., 1995a. Neonatal genistein chemoprevents mammary cancer. Proceedings of the Society for Experimental Biology and Medicine 208, 120–123. Lamartiniere, C.A., Zhao, Y.X., Fritz, W.A.G., 2000. Mammary cancer chemoprevention, in vivo mechanisms of action, potential for toxicity and bioavailability in rats. J. Women’s. Cancer 2, 11–19. Lampe JWand Peterson, S., 2002. Brassica, biotransformation and cancer risk: Genetic polymorphisms alter the preventive effects of cruciferous vegetables. The Journal of Nutrition 132, 2991–2994. Lampe, J.W., 1999. Health effects of vegetables and fruit: Assessing mechanisms of action in human experimental studies. The American Journal of Clinical Nutrition 70 (3 Suppl), 475S–490S. Larkin, T., Price, W.E., Astheimer, L., 2008. The key importance of soy isoflavone bioavailability to understanding health benefits. Critical Reviews in Food Science and Nutrition 48, 538–552. Larkin, T., Price, W.E., Astheimer, L., 2008. The key importance of soy isoflavone bioavailability to understanding health benefits. Critical Reviews in Food Science and Nutrition 48, 538–552. Larsson, S.C., Giovannucci, E., Wolk, A., 2006. Folate intake, MTHFR polymorphisms, and risk of esophageal, gastric, and pancreatic cancer: A meta-analysis. Gastroenterology 131 (4), 1271–1283. Larsson, S.C., Kumlin, M., Ingelman-Sundberg, M., Wolk, A., 2004. Dietary long-chain n-3 fatty acids for the prevention of cancer: A review of potential mechanisms. The American Journal of Clinical Nutrition 79 (6), 935–945. Lawrence, G.D., 2013. Dietary fats and health: Dietary recommendations in the context of scientific evidence. Advances in Nutrition 4 (3), 294–302. https://doi.org/10.3945/ an.113.003657. Lecumberri, E., Dupertuis, Y.M., Miralbell, R., Pichard, C., 2013. Green tea polyphenl epigallocatechin-3-gallate (EGCG) as adjuvant in cancer therapy. Clinical Nutrition 32, 894–903. Lee, C., et al., 2012b. Fasting cycles retard growth of tumors and sensitize a range of cancer cell types to chemotherapy. Science Translational Medicine 4, 124–127. Lee, S.A., Shu, X.O., Li, H., Yang, G., Cai, H., Wen, W., Ji, B.T., Gao, J., Gao, Y.T., Zheng, W., 2009. Adolescent and adult soy food intake and breast cancer risk: Results from the ShanghaiWomen’s health study. The American Journal of Clinical Nutrition 89, 1920–1926. Lee, S.A., Shu, X.O., Yang, G., Li, H., Gao, Y.T., Zheng, W., 2009. Animal origin foods and colorectal cancer risk: A report from the shanghai Women’s health study. Nutrition and Cancer 61 (2), 194–205. https://doi.org/10.1080/01635580802419780. Lenzi, M., Fimognari, C., Hrelia, P., 2014. Sulforaphane as a promising molecule for fighting cancer. Cancer Treatment and Research 159, 207–223. Li, J., Ma, J., Han, L., Xu, Q., Lei, J., Duan, W., Li, W., Wang, F., Wu, E., Ma, Q., Huo, X., 2015. Hyperglycemic tumor microenvironment induces perineural invasion in pancreatic cancer. Cancer Biology & Therapy 16 (6), 912–921. https://doi.org/10.1080/15384047.2015.1040952. Li, N., Sun, Z., Han, C., Chen, J., 1999. The chemopreventive effects of tea on human oral precancerous mucosa lesions. Proceedings of the Society for Experimental Biology and Medicine 220, 218–224. Li, C., Yang, L., Zhang, D., Jiang, W., 2016. Systematic review and meta-analysis suggest that dietary cholesterol intake increases risk of breast cancer. Nutrition Research 36 (7), 627–635. https://doi.org/10.1016/j.nutres.2016.04.009. Epub 2016 Apr 26. Liao, W.C., Tu, Y.K., Wu, M.S., Lin, J.T., Wang, H.P., Chien, K.L., 2015. Blood glucose concentration and risk of pancreatic cancer: Systematic review and dose-response metaanalysis. BMJ 349, g7371. https://doi.org/10.1136/bmj.g7371. Lin, L.Z., Chen, P., Harnly, J.M., 2008. New phenolic components and chromatographic profiles of green and fermented teas. Journal of Agricultural and Food Chemistry 56, 8130–8140. Lin, H.J., Probst-Hensch, N.M., Louie, A., Kau, I.H., Witte, J.S., et al., 1998. Glutathione transferase null genotype, broccoli, and lower prevalence of colorectal adenomas. Cancer Epidem Biomar 7, 647–652. Lin, T., Zirpoli, G.R., McCann, S.E., Moysich, K.B., Christine, B., Tang, A.L., 2017. Trends in cruciferous vegetable consumption and associations with breast Cancer risk: A casecontrol study. Current Developments in Nutrition 1, e000448. Liu, X., Lv, K., 2013. Cruciferous vegetables intake is inversely associated with risk of breast cancer: A meta-analysis. Breast 22, 309–313. Liu, B., Mao, Q., Cao, M., Xie, L., 2012. Cruciferous vegetables intake and risk of prostate cancer: A meta-analysis. International Journal of Urology 19, 134–141. Liu, J., Yuan, F., Gao, J., Shan, B., Ren, Y., Wang, H., Gao, Y., 2016. Oral isoflavone supplementation on endometrial thickness: A ma-analysis of randomized placebo-controlled trials. Oncotarget 7, 17369–17379. Llor, X., Pons, E., Roca, A., Alvarez, M., Mane, J., Fernandez-Banares, F., Gassull, M.A., 2003. The effects of fish oil, olive oil, oleic acid and linoleic acid on colorectal neoplastic processes. Clinical Nutrition 22 (1), 71–79. Loibl, S., Lintermans, A., Dieudonné, A.S., Neven, P., 2011. Management of menopausal symptoms in breast cancer patients. Maturitas 68, 148–154. Longo, V.D., Fontana, L., 2010. Calorie restriction and cancer prevention: Metabolic and molecular mechanisms. Trends in Pharmacological Sciences 31, 89–98. Longo, V.D., Mattson, M.P., 2014. Fasting: Molecular mechanisms and clinical applications. Cell Metabolism 19 (2), 181–192. Longo, V.D., Panda, S., 2016. Fasting, circadian rhythms, and time-restricted feeding in healthy lifespan. Cell Metabolism 14 (23), 1048–1059. Lozovaya, V.V., Lygin, A.V., Ulanov, A.V., Nelson, R.L., Daydé, J., Widholm, J.M., 2005. Effect of temperature and soil moisture status during seed development on soybean seed Isoflavone concentration and composition. Crop Science 45, 1934–1940. Lu, R.R., Qian, P., Sun, A., et al., 2010. Hempseed protein derived antioxidative peptides: Purification, identification and protection from hydrogen peroxide-induced apoptosis in PC12 cells. Food Chemistry 123 (4), 1210–1218.

492

Diet and Cancer

Magee, P.J., McGlynn, H., Rowland, I.R., 2004. Differential effects of isoflavones and lignans on invasiveness of MDA-MB-231 breast cancer cells in vitro. Cancer Letters 208, 35–41. Mahmoud, A.M., Yang, W., Bosland, M.C., 2014. Soy isoflavones and prostate cancer: A review of molecular mechanisms. The Journal of Steroid Biochemistry and Molecular Biology 140, 116–132. Mahmoud, A.M., Yang, W., Bosland, M.C., 2014. Soy isoflavones and prostate cancer: A review of molecular mechanisms. The Journal of Steroid Biochemistry and Molecular Biology 140, 116–132. Mandal, C.C., Ghosh-Choudhury, T., Yoneda, T., Choudhury, G.G., Ghosh-Choudhury, N., 2010. Fish oil prevents breast cancer cell metastasis to bone. Biochemical and Biophysical Research Communications 402 (4), 602–607. Markovitis, N., Ben Amotz, A., Levy, Y., 2009. The effect of tomato-derived lycopene on low carotenoids and enhanced systemic inflammation and oxidation in severe obesity. The Israel Medical Association Journal 11, 598–601. Matthews, J., Gustafsson, J.-A., 2003. Estrogen signaling: A subtle balance between ER alpha and ER beta. Molecular Interventions 3, 281–292. Mattson, M.P., 2005. Energy intake, meal frequency, and health: A neurobiologicalperspective. Annual Review of Nutrition 25, 237–260. Mattson, M.P., Longo, V.D., Harvie, M., 2017 Oct. Impact of intermittent fasting on health and disease processes. Ageing Research Reviews 39, 46–58. May-Wilson, S., Sud, A., Law, P.J., et al., 2017. Pro-inflammatory fatty acid profile and colorectal cancer risk: A Mendelian randomisation analysis. European Journal of Cancer 84, 228–238. Mazidi, M., Rezaie, P., Chaudhri, O., et al., 2015. The effect of Ramadan fasting on cardiometabolic risk factors and anthropometrics parameters: A systematic review. Pakistan Journal of Medical Sciences 31, 1250–1255. Mazur, W.M., Duke, J.A., Wahala, K., Rasku, S., Adlercreutz, H., 1998. Isoflavonoids and lignans in legumes: Nutritional and health aspects in humans. The Journal of Nutritional Biochemistry 9, 193–200. McCullough, M.L., Giovannucci, E.L., 2004. Diet and cancer prevention. Oncogene 23 (38), 6349–6364. https://doi.org/10.1038/sj.onc.1207716. McCullough, M.L., Robertson, A.S., Chao, A., et al., 2003. A prospective study of whole grains, fruits, vegetables and colon cancer risk. Cancer Causes & Control 14, 959–970. McEligot, A.J., Largent, J., Ziogas, A., Peel, D., Anton-Culver, H., 2006. Dietary fat, fiber, vegetable, and micronutrients are associated with overall survival in postmenopausal women diagnosed with breast cancer. Nutrition and Cancer 55 (2), 132–140. McKeown-Eyssen, G., 1994. Epidemiology of colorectal cancer revisited: Are serum triglycerides and/or plasma glucose associated with risk? Cancer Epidemiology, Biomarkers & Prevention 3 (8), 687–695. Meegan, M.J., Lloyd, D.G., 2003. Advances in the science of estrogen receptor modulation. Current Medicinal Chemistry 10, 181–210. Meng, Q., Yuan, F., Goldberg, I.D., Rosen, E.M., Auborn, K., Fan, S., 2000. Indole-3-carbinol is a negative regulator of estrogen receptor-alpha signaling in human tumor cells. The Journal of Nutrition 130, 2927–2931. Menotti, A., Keys, A., Aravanis, C., Blackburn, H., Dontas, A., Fidanza, F., Karvonen, M.J., Kromhout, D., Nedeljkovic, S., Nissinen, A., et al., 1989. Seven countries study. First 20year mortality data in 12 cohorts of six countries. Annals of Medicine 21 (3), 175–179. Mensink, R.P., Zock, P.L., Kester, A.D., Katan, M.B., 2003. Effects of dietary fatty acids and carbohydrates on the ratio of serum total to HDL cholesterol and on serum lipids and apolipoproteins: A meta-analysis of 60 controlled trials. The American Journal of Clinical Nutrition 77, 1146–1155. Messina, M.J., 1999. Legumes and soybeans: Overview of their nutritional profiles and health effects. The American Journal of Clinical Nutrition 70, 439S–450S. Review. Messina, M., 2010. A brief historical overview of the past two decades of soy and isoflavone research. The Journal of Nutrition 140 (7), 1350S–4S. Review. Messina, M., 2016a. Impact of soy foods on the development of breast Cancer and the prognosis of breast Cancer patients. Forschende Komplementärmedizin 23 (2), 75–80. Review. Messina, M., 2016b. Soy and health update: Evaluation of the clinical and epidemiologic literature. Nutrients 8, 754. Messina, M., 2016. Soy and health update: Evaluation of the clinical and epidemiologic literature. Nutrients 8, 754. Messina, M., Hilakivi-Clarke, L., 2009. Early intake appears to be the key to the proposed protective effects of soy intake against breast cancer. Nutrition and Cancer 61 (6), 792–798. Messina, M., Kucuk, O., Lampe, J.W., 2006. An overview of the health effects of isoflavones with an emphasis on prostate cancer risk and prostate-specific antigen levels. Journal of AOAC International 89, 1121–1134. Messina, M., Kucuk, O., Lampe, J.W., 2006b. An overview of the health effects of isoflavones with an emphasis on prostate cancer risk and prostate-specific antigen levels. Journal of AOAC International 89, 1121–1134. Messina, M., McCaskill-Stevens, W., Lampe, J.W., 2006a. Addressing the soy and breast cancer relationship: Review, commentary, and workshop proceedings. Journal of the National Cancer Institute 98 (18), 1275–1284. Review. Messina, M., Nagata, C., Wu, A.H., 2006c. Estimated Asian adult soy protein and isoflavone intakes. Nutrition and Cancer 55, 1–12. Meyerhardt, J.A., Sato, K., Niedzwiecki, D., Ye, C., Saltz, L.B., Mayer, R.J., Mowat, R.B., Whittom, R., Hantel, A., Benson, A., Wigler, D.S., Venook, A., Fuchs, C.S., 2012. Dietary glycemic load and cancer recurrence and survival in patients with stage III colon cancer: Findings from CALGB 89803. Journal of the National Cancer Institute 104 (22), 1702– 1711. https://doi.org/10.1093/jnci/djs399. Michels, K.B., Giovannucci, E., Joshipura, K.J., et al., 2000. Prospective study of fruit and vegetable consumption and incidence of colon and rectal cancers. Journal of the National Cancer Institute 92 (21), 1740–1752. Mirzaei, H., Di Biase, S., Longo, V.D., 2016. Dietary interventions, cardiovascular aging, and disease: Animal models and human studies. Circulation Research 118 (10), 1612–1625. Mishra, P., Kar, A., Kale, R.K., 2011. Prepubertal daidzein exposure enhances mammary gland differentiation and regulates the expression of estrogen receptor-alpha and apoptotic proteins. ISRN Oncology 2011, 896826. Mishra, P., Kar, A., Kale, R.K., 2011. Prepubertal daidzein exposure enhances mammary gland differentiation and regulates the expression of estrogen receptor-alpha and apoptotic proteins. ISRN Oncology 2011, 896826. Mitjavila, M.T., Fandos, M., Salas-Salvado, J., Covas, M.I., Borrego, S., Estruch, R., Lamuela-Raventos, R., Corella, D., Martinez-Gonzalez, M.A., Sanchez, J.M., Bullo, M., Fito, M., Tormos, C., Cerda, C., Casillas, R., Moreno, J.J., Iradi, A., Zaragoza, C., Chaves, J., Saez, G.T., 2013. The Mediterranean diet improves the systemic lipid and DNA oxidative damage in metabolic syndrome individuals. A randomized, controlled, trial. Clinical Nutrition 32 (2), 172–178. Monnier, L., Mas, E., Ginet, C., Michel, F., Villon, L., Cristol, J.P., Colette, C., 2006. Activation of oxidative stress by acute glucose fluctuations compared with sustained chronic hyperglycemia in patients with type 2 diabetes. JAMA 295 (14), 1681–1687. https://doi.org/10.1001/jama.295.14.1681. Moon, Y.J., Wang, X., Morris, M.E., 2006. Dietary flavonoids: Effects on xenobiotic and carcinogen metabolism. Toxicology In Vitro 20, 187–210. Mozaffarian, D., Katan, M.B., Ascherio, A., Stampfer, M.J., Willett, W.C., 2006. Trans fatty acids and cardiovascular disease. The New England Journal of Medicine 354, 1601–1613. Murphy, P.A., Barua, K., Hauck, C.C., 2002. Solvent extraction selection in the determination of isoflavones in soy foods. Journal of Chromatography. B, Analytical Technologies in the Biomedical and Life Sciences 777, 129–138. Murphy, P.A., Song, T., Buseman, G., Barua, K., Beecher, G.R., Trainer, D., Holden, J., 1999. Isoflavones in retail and institutional soy foods. Journal of Agricultural and Food Chemistry 47, 2697–2704. Nagata, C., Shimizu, H., Kametani, M., Takeyama, N., Ohnuma, T., Matsushita, S., 2001. Diet and colorectal adenoma in Japanese males and females. Diseases of the Colon and Rectum 44 (1), 105–111.

Diet and Cancer

493

Navarro, S.L., Li, F., Lampe, J.W., 2011. Mechanisms of action of isothiocyanates in cancer chemoprevention: An update. Food & Function 2, 579–587. Navarro-Tito, N., Robledo, T., Salazar, E.P., 2008. Arachidonic acid promotes FAK activation and migration in MDA-MB-231 breast cancer cells. Experimental Cell Research 314 (18), 3340–3355. https://doi.org/10.1016/j.yexcr.2008.08.018. Epub 2008 Sep 9. Nechuta, S.J., Caan, B.J., Chen, W.Y., et al., 2012a. Soy food intake after diagnosis of breast cancer and survival: An in-depth analysis of combined evidence from cohort studies of US and Chinese women. The American Journal of Clinical Nutrition 96, 123–132. Nechuta, S., Caan, B.J., Chen, W.Y., Kwan, M.L., Lu, W., Cai, H., Poole, E.M., Flatt, S.W., Zheng, W., Pierce, J.P., Shu, X.O., 2013. Postdiagnosis cruciferous vegetable consumption and breast cancer outcomes: A report from the after breast Cancer pooling project. Cancer Epidemiology, Biomarkers & Prevention 22, 1451–1456. Nechuta, S.J., Caan, B.J., Chen, W.Y., Lu, W., Chen, Z., Kwan, M.L., Flatt, S.W., Zheng, Y., Zheng, W., Pierce, J.P., et al., 2012. Soy food intake after diagnosis of breast cancer and survival: An in-depth analysis of combined evidence from cohort studies of US and Chinese women. The American Journal of Clinical Nutrition 96, 123–132. Nechuta, S.J., Caan, B.J., Chen, W.Y., Lu, W., Chen, Z., Kwan, M.L., Flatt, S.W., Zheng, Y., Zheng, W., Pierce, J.P., et al., 2012b. Soy food intake after diagnosis of breast cancer and survival: An in-depth analysis of combined evidence from cohort studies of US and Chinese women. The American Journal of Clinical Nutrition 96, 123–132. Negre-Salvayre, A., Auge, N., Ayala, V., Basaga, H., Boada, J., Brenke, R., Chapple, S., Cohen, G., Feher, J., Grune, T., Lengyel, G., Mann, G.E., Pamplona, R., Poli, G., PorteroOtin, M., Riahi, Y., Salvayre, R., Sasson, S., Serrano, J., Shamni, O., Siems, W., Siow, R.C.M., Wiswedel, I., Zarkovic, K., Zarkovic, N., 2010. Pathological aspects of lipid peroxidation. Free Radical Research 44, 1125–1171. Nelson, E.R., Wardell, S.E., Jasper, J.S., Park, S., Suchindran, S., Howe, M.K., Carver, N.J., Pillai, R.V., Sullivan, P.M., Sondhi, V., Umetani, M., Geradts, J., McDonnell, D.P., 2013. 27-Hydroxycholesterol links hypercholesterolemia and breast cancer pathophysiology. Science 342 (6162), 1094–1098. https://doi.org/10.1126/science.1241908. Nicodemus, K.K., Jacobs Jr., D.R., Folsom, A.R., 2001. Whole and refined grain intake and risk of incident postmenopausal breast cancer (United States). Cancer Causes & Control 12 (10), 917–925. Nilsson, S., Gustafsson, J.-A., 2002. Biological role of estrogen and estrogen receptors. Critical Reviews in Biochemistry and Molecular Biology 37, 1–28. Nothlings, U., Murphy, S.P., Wilkens, L.R., Henderson, B.E., Kolonel, L.N., 2007. Flavonols and pancreatic cancer risk: The multiethnic cohort study. American Journal of Epidemiology 166 (8), 924–931. Noto, H., Goto, A., Tsujimoto, T., Noda, M., 2012. Cancer risk in diabetic patients treated with metformin: A systematic review and meta-analysis. PLoS One 7 (3), e33411. https:// doi.org/10.1371/journal.pone.0033411. Olivecrona, H., Hilding, A., Ekstrom, C., Barle, H., Nyberg, B., Moller, C., Delhanty, P.J., Baxter, R.C., Angelin, B., Ekstrom, T.J., Tally, M., 1999. Acute and short-term effects of growth hormone on insulin-like growth factors and their binding proteins: Serum levels and hepatic messenger ribonucleic acid responses in humans. The Journal of Clinical Endocrinology and Metabolism 84 (2), 553–560. https://doi.org/10.1210/jcem.84.2.5466. O’Flanagan, C.H., Smith, L.A., McDonell, S.B., Hursting, S.D., 2017. When less may be more: Calorie restriction and response to cancer therapy. BMC Medicine 15 (1), 106. Othman, R.A., 2007. Dietary lipids and Cancer. The Libyan Journal of Medicine 2 (4), 180–184. Published online 2007 Dec 1. https://doi.org/10.4176/070831. Otieno, D.O., Ashton, J.F., Shah, N.P., 2007. Isoflavone phytoestrogen degradation in fermented soymilk with selected beta-glucosidase producing L. acidophilus strains during storage at different temperatures. International Journal of Food Microbiology 115, 79–88. Paixão, E.M.D.S., Oliveira, A.C.M., Pizato, N., Muniz-Junqueira, M.I., Magalhães, K.G., Nakano, E.Y., Ito, M.K., 2017. The effects of EPA and DHA enriched fish oil on nutritional and immunological markers of treatment naïve breast cancer patients: A randomized double-blind controlled trial. Nutrition Journal 16 (1), 71. https://doi.org/10.1186/s12937-0170295-9. Palomares, M.R., Hopper, L., Goldstein, L., Lehman, C.D., Storer, B.E., Gralow, J.R., 2004. Effect of soy isoflavones on breast proliferation in postmenopausal breast cancer survivors. Breast Cancer Research and Treatment 88, 4002. Pan, A., Sun, Q., Bernstein, A.M., Schulze, M.B., Manson, J.E., Stampfer, M.J., Willett, W.C., Hu, F.B., 2012. Red meat consumption and mortality: Results from 2 prospective cohort studies. Archives of Internal Medicine 172 (7), 555–563. https://doi.org/10.1001/archinternmed.2011.2287. Patel, R.P., Boersma, B.J., Crawford, J.H., Hogg, N., Kirk, M., Kalyanaraman, B., Parks, D.A., Barnes, S., Darley-Usmar, V., 2001. Antioxidant mechanisms of isoflavones in lipid systems: Paradoxical effects of peroxyl radical scavenging. Free Radical Biology & Medicine 31, 1570–1581. Patisaul, H.B., Jefferson, W., 2010. The pros and cons of phytoestrogens. Frontiers in Neuroendocrinology 31, 400–419. Patterson, R.E., Sears, D.D., 2017. Metabolic effects of intermittent fasting. Annual Review of Nutrition 37, 371–393. Peeters, P.H.M., Keinan-Boker, L., van der Schouw, Y.T., Grobbee, D.E., 2002. Phytoestrogens and breast cancer risk: Review of the epidemiological evidence. IARC Scientific Publications 156, 331–336. Pelucchi, C., Bosetti, C., Negri, E., Lipworth, L., La Vecchia, C., 2011. Olive oil and cancer risk: An update of epidemiological findings through 2010. Current Pharmaceutical Design 17 (8), 805–812. Peng, C., WP1, L., Zhang, C.X., 2017. Fruit and vegetable intake and breast cancer prognosis: A meta-analysis of prospective cohort studies. The British Journal of Nutrition 117 (5), 737–749. Phillips, S.M., Fulgoni, V.L., Heaney, R.P., Nicklas, T.A., Slavin, J.L., Weaver, C.M., 2015. Commonly consumed protein foods contribute to nutrient intake, diet quality, and nutrient adequacy. The American Journal of Clinical Nutrition 101, 1346S–1352S. Pierce, J.P., Natarajan, L., Caan, B.J., et al., 2007. Influence of a diet very high in vegetables, fruit, and fiber and low in fat on prognosis following treatment for breast cancer: The Women’s healthy eating and living (WHEL) randomized trial. Journal of the American Medical Association 298 (3), 289–298. Pimentel, D., Pimentel, M., 2003. Sustainability of meat-based and plant-based diets and the environment. The American Journal of Clinical Nutrition 78, 660S–3S. Pisani, P., Parkin, D.M., Bray, F., Ferlay, J., 1999. Estimates of the worldwide mortality from 25 cancers in 1990. International Journal of Cancer 83, 18–29. Platz, E.A., Leitzmann, M.F., Visvanathan, K., Rimm, E.B., Stampfer, M.J., Willett, W.C., et al., 2006. Statin drugs and risk of advanced prostate cancer. Journal of the National Cancer Institute 98, 1819–1825. Platz, E.A., Till, C., Goodman, P.J., Parnes, H.L., Figg, W.D., Albanes, D., et al., 2009. Men with low serum cholesterol have a lower risk of high-grade prostate cancer in the placebo arm of the prostate cancer prevention trial. Cancer Epidemiology, Biomarkers & Prevention 18, 2807–2813. Prentice, R.L., Caan, B., Chlebowski, R.T., et al., 2006. Low-fat dietary pattern and risk of invasive breast cancer: The Women’s Health Initiative randomized controlled dietary modification trial. Journal of the American Medical Association 295 (6), 629–642. Quagliaro, L., Piconi, L., Assaloni, R., Da Ros, R., Maier, A., Zuodar, G., Ceriello, A., 2005. Intermittent high glucose enhances ICAM-1, VCAM-1 and E-selectin expression in human umbilical vein endothelial cells in culture: The distinct role of protein kinase C and mitochondrial superoxide production. Atherosclerosis 183 (2), 259–267. https://doi.org/ 10.1016/j.atherosclerosis.2005.03.015. Raffaghello, L., et al., 2008. Starvation-dependent differentialstress resistance protects normal but not cancer cells against high-dose chemotherapy. Proceedings of the National Academy of Sciences of the United States of America 105, 8215–8220. Raffaghello, L., et al., 2010b. Fasting and differential chemotherapy protection in patients. Cell Cycle 9, 4474–4476. Rahal, O.M., Simmen, R.C., 2011. Paracrine-acting adiponectin promotes mammary epithelial differentiation and synergizes with genistein to enhance transcriptional response to estrogen receptor beta signaling. Endocrinology 152, 3409–3421. Rahal, O.M., Simmen, R.C., 2011. Paracrine-acting adiponectin promotes mammary epithelial differentiation and synergizes with genistein to enhance transcriptional response to estrogen receptor beta signaling. Endocrinology 152, 3409–3421. Rastogi, T., Devesa, S., Mangtani, P., Mathew, A., Cooper, N., Kao, R., Sinha, R., 2008. Cancer incidence rates among south Asians in four geographic regions: India, Singapore, UK and US. International Journal of Epidemiology 37 (1), 147–160. https://doi.org/10.1093/ije/dym219. Reijnders, L., Soret, S., 2003. Quantification of the environmental impact of different dietary protein choices. The American Journal of Clinical Nutrition 78, 664S–8S.

494

Diet and Cancer

Renehan, A.G., et al., 2010. Incident cancer burden attributable to excess body mass index in 30 European countries. International Journal of Cancer 126, 692–702. Renehan, A.G., Tyson, M., Egger, M., Heller, R.F., Zwahlen, M., 2008. Body-massindex and incidence of cancer: A systematic review and meta-analysis ofprospective observational studies. Lancet 371, 569–578. Reuter, S., Gupta, S.C., Chaturvedi, M.M., Aggarwal, B.B., 2010. Oxidative stress, inflammation, and cancer: How are they linked? Free Radical Biology & Medicine 49 (11), 1603– 1616. https://doi.org/10.1016/j.freeradbiomed.2010.09.006. Riggs, B.L., Hartmann, L.C., 2003. Selective estrogen-receptor modulatorsdMechanisms of action and application to clinical practice. The New England Journal of Medicine 348, 618–629. Rinaldi, S., Cleveland, R., Norat, T., Biessy, C., Rohrmann, S., Linseisen, J., Boeing, H., Pischon, T., Panico, S., Agnoli, C., Palli, D., Tumino, R., Vineis, P., Peeters, P.H., van Gils, C.H., Bueno-de-Mesquita, B.H., Vrieling, A., Allen, N.E., Roddam, A., Bingham, S., Khaw, K.T., Manjer, J., Borgquist, S., Dumeaux, V., Torhild Gram, I., Lund, E., Trichopoulou, A., Makrygiannis, G., Benetou, V., Molina, E., Donate Suarez, I., Barricarte Gurrea, A., Gonzalez, C.A., Tormo, M.J., Altzibar, J.M., Olsen, A., Tjonneland, A., Gronbaek, H., Overvad, K., Clavel-Chapelon, F., Boutron-Ruault, M.C., Morois, S., Slimani, N., Boffetta, P., Jenab, M., Riboli, E., Kaaks, R., 2010. Serum levels of IGF-I, IGFBP-3 and colorectal cancer risk: Results from the EPIC cohort, plus a meta-analysis of prospective studies. International Journal of Cancer 126 (7), 1702–1715. https://doi.org/ 10.1002/ijc.24927. Rizza, W., Veronese, N., Fontana, L., 2014 Jan. What are the roles of calorie restriction and diet quality in promoting healthy longevity? Ageing Research Reviews 13, 38–45. Rizzo, G., Baroni, L., 2018. Soy, soy foods and their role in vegetarian diets. Nutrients 5, 10. Review. Rizzo, G., Baroni, L., 2018. Soy, soy foods and their role in vegetarian diets. Nutrients 5, 10. Review. Rock, C.L., Doyle, C., Demark-Wahnefried, W., Meyerhardt, J., Courneya, K.S., Schwartz, A.L., Bandera, E.V., Hamilton, K.K., Grant, B., McCullough, M., Byers, T., Gansler, T., 2012. Nutrition and physical activity guidelines for cancer survivors. CA: A Cancer Journal for Clinicians 62 (4), 243–274. https://doi.org/10.3322/caac.21142. Romieu, I., Ferrari, P., Rinaldi, S., Slimani, N., Jenab, M., Olsen, A., Tjonneland, A., Overvad, K., Boutron-Ruault, M.C., Lajous, M., Kaaks, R., Teucher, B., Boeing, H., Trichopoulou, A., Naska, A., Vasilopoulo, E., Sacerdote, C., Tumino, R., Masala, G., Sieri, S., Panico, S., Bueno-de-Mesquita, H.B., Van-der, A.D., van Gils, C.H., Peeters, P.H., Lund, E., Skeie, G., Asli, L.A., Rodriguez, L., Navarro, C., Amiano, P., Sanchez, M.J., Barricarte, A., Buckland, G., Sonestedt, E., Wirfalt, E., Hallmans, G., Johansson, I., Key, T.J., Allen, N.E., Khaw, K.T., Wareham, N.J., Norat, T., Riboli, E., Clavel-Chapelon, F., 2012. Dietary glycemic index and glycemic load and breast cancer risk in the European prospective investigation into Cancer and nutrition (EPIC). The American Journal of Clinical Nutrition 96 (2), 345–355. https://doi.org/10.3945/ajcn.111.026724. Ronco, A.L., De Stéfani, E., Stoll, M., 2010. Hormonal and metabolic modulation through nutrition: Towards a primary prevention of breast cancer. Breast 19 (5), 322–332. https:// doi.org/10.1016/j.breast.2010.05.005. Epub 2010 Jun 9. Rosa, E.A.S., Heaney, R.K., Fenwick, G.R., Portas, C.A.M., 1997. Glucosinolates in crop plants. Horticultural Reviews 19, 99–215. Rosato, V., Guercio, V., Bosetti, C., Negri, E., Serraino, D., Giacosa, A., Montella, M., La Vecchia, C., Tavani, A., 2016. Mediterranean diet and colorectal cancer risk: A pooled analysis of three Italian case-control studies. British Journal of Cancer 115 (7), 862–865. https://doi.org/10.1038/bjc.2016.245. Epub 2016 Aug 18. Rose, D.P., 1990. Dietary fiber and breast cancer. Nutrition and Cancer 13 (1–2), 1–8. Rossi, M., Lugo, A., Lagiou, P., Zucchetto, A., Polesel, J., Serraino, D., Negri, E., Trichopoulos, D., La Vecchia, C., 2012. Proanthocyanidins and other flavonoids in relation to pancreatic cancer: A case-control study in Italy. Annals of Oncology 23 (6), 1488–1493. Rothschild, J., Hoddy, K.K., Jambazian, P., Varady, K.A., 2014. Time-restricted feeding and risk of metabolic disease: A review of human and animal studies. Nutrition Reviews 72, 308–318. Rous, P., 1914. The influence of diet on transplanted and spontaneous mouse tumors. The Journal of Experimental Medicine 20 (5), 433–451. Rouzaud, G., Young, S.A., Duncan, A.J., 2004. Hydrolysis of glucosinolates to isothiocyanates after ingestion of raw or microwaved cabbage by human volunteers. Cancer Epidemiology, Biomarkers & Prevention 13, 125–131. Rowland, I., Faughnan, M., Hoey, L., Wahala, K., Williamson, G., Cassidy, A., 2003. Bioavailability of phyto-oestrogens. The British Journal of Nutrition 89, S45–S58. Rowlands, M.A., Gunnell, D., Harris, R., Vatten, L.J., Holly, J.M., Martin, R.M., 2009. Circulating insulin-like growth factor peptides and prostate cancer risk: A systematic review and meta-analysis. International Journal of Cancer 124 (10), 2416–2429. https://doi.org/10.1002/ijc.24202. Roy, P., Nigam, N., Singh, M., et al., 2010. Tea polyphenols inhibit cyclooxygenase-2 expression and block activation of nuclear factor-kappa B and Akt in diethylnitrosoamine induced lung tumors in Swiss mice. Investigational New Drugs 28 (4), 466–471. Royston, K.J., Tollefsbol, T.O., 2015. The epigenetic impact of cruciferous vegetables on Cancer prevention. Current Pharmacology Reports 1, 46–51. Runchey, S.S., Pollak, M.N., Valsta, L.M., Coronado, G.D., Schwarz, Y., Breymeyer, K.L., Wang, C., Wang, C.Y., Lampe, J.W., Neuhouser, M.L., 2012. Glycemic load effect on fasting and post-prandial serum glucose, insulin, IGF-1 and IGFBP-3 in a randomized, controlled feeding study. European Journal of Clinical Nutrition 66 (10), 1146–1152. https://doi.org/10.1038/ejcn.2012.107. Runowicz, C.D., Leach, C.R., Henry, N.L., Henry, K.S., Mackey, H.T., Cowens-Alvarado, R.L., Cannady, R.S., Pratt-Chapman, M.L., Edge, S.B., Jacobs, L.A., Hurria, A., Marks, L.B., LaMonte, S.J., Warner, E., Lyman, G.H., Ganz, P.A., 2016. American Cancer Society/American Society of Clinical Oncology breast Cancer survivorship care guideline. CA: a Cancer Journal for Clinicians 66 (1), 43–73. https://doi.org/10.3322/caac.21319. Russo, J., Mailo, D., Hu, Y.F., Balogh, G., Sheriff, F., Russo, I.H., 2005. Breast differentiation and its implication in cancer prevention. Clinical Cancer Research 11, 931s–936s. Sadava, D., Whitlock, E., Kane, S.E., 2007. The green tea polyphenol, epigallocatechin 3- gallate inhibits telomerase and induces apoptosis in drug-resistant lung cancer cells. Biochemical and Biophysical Research Communications 360 (1), 233–237. Safdie, F.M., et al., 2009. Fasting and cancer treatment in humans: A case series report. Aging (Albany NY) 1, 988–1007. Samraj, A.N., Pearce, O.M., Laubli, H., Crittenden, A.N., Bergfeld, A.K., Banda, K., Gregg, C.J., Bingman, A.E., Secrest, P., Diaz, S.L., et al., 2015. A red meat-derived glycan promotes inflammation and cancer progression. Proceedings of the National Academy of Sciences of the United States of America 112, 542–547. Santarelli, R.L., Pierre, F., Corpet, D.E., 2008. Processed meat and colorectal cancer: A review of epidemiologic and experimental evidence. Nutrition and Cancer 60 (2), 131–144. https://doi.org/10.1080/01635580701684872. Sartippour, M.R., Rao, J.Y., Apple, S., Wu, D., Henning, S., Wang, H., Elashoff, R., Rubio, R., Heber, D., Brooks, M.N., 2004. A pilot clinical study of short-term isoflavone supplements in breast cancer patients. Nutrition and Cancer 49, 59–65. Sawada, N., Inoue, M., Iwasaki, M., Sasazuki, S., Shimazu, T., Yamaji, T., Takachi, R., Tanaka, Y., Mizokami, M., Tsugane, S., Japan Public Health Center-Based Prospective Study Group, 2012. Consumption of n-3 fatty acids and fish reduces risk of hepatocellular carcinoma. Gastroenterology 142 (7), 1468–1475. https://doi.org/10.1053/j.gastro.2012.02.018. Epub 2012 Feb 16. Schley, P.D., Jijon, H.B., Robinson, L.E., Field, C.J., 2005. Mechanisms of omega-3 fatty acid-induced growth inhibition in MDA-MB-231 human breast cancer cells. Breast Cancer Research and Treatment 92 (2), 187–195. Schuurman, A.G., Goldbohm, R.A., Dorant, E., van den Brandt, P.A., 1998. Vegetable and fruit consumption and prostate cancer risk: A cohort study in the Netherlands. Cancer epidemiology. Biomarkers & Prevention 7 (8), 673–680. Schüz, J., Espina, C., Villain, P., Herrero, R., Leon, M.E., Minozzi, S., Romieu, I., Segnan, N., Wardle, J., Wiseman, M., Belardelli, F., Bettcher, D., Cavalli, F., Galea, G., Lenoir, G., Martin-Moreno, J.M., Nicula, F.A., Olsen, J.H., Patnick, J., Primic-Zakelj, M., Puska, P., van Leeuwen, F.E., Wiestler, O., Zatonski, W., 2015. Working groups of scientific experts. European code against cancer 4th edition: 12 ways to reduce your cancer risk. Cancer Epidemiology 39 (Suppl 1), S1–10. Schwingshackl, L., Schwedhelm, C., Galbete, C., Hoffmann, G., 2017. Adherence to Mediterranean diet and risk of Cancer: An updated systematic review and meta-analysis. Nutrients 9 (10). Serra-Majem, L., Roman, B., Estruch, R., 2006. Scientific evidence of interventions using the Mediterranean diet: A systematic review. Nutrition Reviews 64 (2 Pt 2), S27–S47.

Diet and Cancer

495

Setchell, K.D., 1998. Phytoestrogens: The biochemistry, physiology, and implications for human health of soy isoflavones. The American Journal of Clinical Nutrition 68, 1333S– 1346S. Setchell, K.D., Brown, N.M., Lydeking-Olsen, E., 2002. The clinical importance of the metabolite equol-a clue to the effectiveness of soy and its isoflavones. The Journal of Nutrition 132, 3577–3584. Setchell, K.D., Clerici, C., 2010. Equol: Pharmacokinetics and biological actions. The Journal of Nutrition 140, 1363S–1368S. Shanafelt, T.D., Call, T.G., Zent, C.S., et al., 2009. Phase I trial of daily, oral polyphenol E in patients with asymptomatic rai stage 0 to II chronic lymphocytic leukemia. Journal of Clinical Oncology 27, 3808–3814. Shanafelt, T.D., Call, T.G., Zent, C.S., et al., 2013. Phase I trial of daily, oral polyphenol E in patients with asymptomatic rai stage 0 to II chronic lymphocytic leukemia. Cancer 119, 363–370. Shapiro, T.A., Fahey, J.W., Wade, K.L., Stephenson, K.K., Talalay, P., 1998. Human metabolism and excretion of cancer chemoprotective glucosinolates and isothiocyanates of cruciferous vegetables. Cancer Epidemiology, Biomarkers & Prevention 7, 1091–1100. Shevah, O., Laron, Z., 2007. Patients with congenital deficiency of IGF-I seem protected from the development of malignancies: A preliminary report. Growth Hormone & IGF Research 17, 54–57. Shu, X.O., Jin, F., Dai, Q., Wen, W., Potter, J.D., Kushi, L.H., Ruan, Z., Gao, Y.T., Zheng, W., 2001. Soyfood intake during adolescence and subsequent risk of breast cancer among Chinese women. Cancer Epidemiology, Biomarkers & Prevention 10, 483–488. Shu, X.O., Zheng, Y., Cai, H., Gu, K., Chen, Z., Zheng, W., Lu, W., 2009. Soy food intake and breast cancer survival. JAMA 302, 2437–2443. Shu, X.O., Zheng, Y., Cai, H., Gu, K., Chen, Z., Zheng, W., Lu, W., 2009. Soy food intake and breast cancer survival. JAMA 302, 2437–2443. Sieri, S., et al., 2014. Dietary fat intake and development of specific breast cancer subtypes. Journal of the National Cancer Institute 106 (5) pii: dju068. https://doi.org/10.1093/ jnci/dju068. Sieri, S., Pala, V., Brighenti, F., Pellegrini, N., Muti, P., Micheli, A., Evangelista, A., Grioni, S., Contiero, P., Berrino, F., Krogh, V., 2007. Dietary glycemic index, glycemic load, and the risk of breast cancer in an Italian prospective cohort study. The American Journal of Clinical Nutrition 86 (4), 1160–1166. Silvera, S.A., Rohan, T.E., Jain, M., Terry, P.D., Howe, G.R., Miller, A.B., 2005. Glycaemic index, glycaemic load and risk of endometrial cancer: A prospective cohort study. Public Health Nutrition 8 (7), 912–919. Slavin, J., 2003. Why whole grains are protective: Biological mechanisms. The Proceedings of the Nutrition Society 62 (1), 129–134. https://doi.org/10.1079/PNS2002221. Slimani, N., Fahey, M., Welch, A.A., Wirfalt, E., Stripp, C., Bergstrom, E., Linseisen, J., Schulze, M.B., Bamia, C., Chloptsios, Y., Veglia, F., Panico, S., Bueno-de-Mesquita, H.B., Ocke, M.C., Brustad, M., Lund, E., Gonzalez, C.A., Barcos, A., Berglund, G., Winkvist, A., Mulligan, A., Appleby, P., Overvad, K., Tjonneland, A., Clavel-Chapelon, F., Kesse, E., Ferrari, P., Van Staveren, W.A., Riboli, E., 2002. Diversity of dietary patterns observed in the European prospective investigation into Cancer and nutrition (EPIC) project. Public Health Nutrition 5 (6B), 1311–1328. https://doi.org/10.1079/PHN2002407. Smith, C.L., O’Malley, B.W., 2004. Coregulator function: A key to understanding tissue specificity of selective receptor modulators. Endocrine Reviews 25, 45–71. Smith-Warner, S.A., Spiegelman, D., Yaun, S.S., et al., 2001. Intake of fruits and vegetables and risk of breast cancer: A pooled analysis of cohort studies. JAMA 285 (6), 769–776. Soda, K., 2011. The mechanisms by which polyamines accelerate tumor spread. Journal of Experimental & Clinical Cancer Research 30, 95. Sofi, F., Cesari, F., Abbate, R., Gensini, G.F., Casini, A., 2008. Adherence to Mediterranean diet and health status: Meta-analysis. BMJ 337, a1344. Sofi, F., Macchi, C., Abbate, R., Gensini, G.F., Casini, A., 2013. Mediterranean diet and health. BioFactors 39 (4), 335–342. Sofi, F., Macchi, C., Abbate, R., Gensini, G.F., Casini, A., 2014. Mediterranean diet and health status: An updated meta-analysis and a proposal for a literature-based adherence score. Public Health Nutrition 17 (12), 2769–2782. Song, M., Fung, T.T., Hu, F.B., Willett, W.C., Longo, V., Chan, A.T., Giovannucci, E.L., 2016. Animal and plant protein intake and all-cause and cause-specific mortality: Results from two prospective US cohort studies. JAMA Internal Medicine 176 (10), 1453–1463. Stattin, P., Bjor, O., Ferrari, P., Lukanova, A., Lenner, P., Lindahl, B., Hallmans, G., Kaaks, R., 2007. Prospective study of hyperglycemia and cancer risk. Diabetes Care 30 (3), 561–567. https://doi.org/10.2337/dc06-0922. Steck, S.E., Gaudet, M.M., Eng, S.M., Britton, J.A., Teitelbaum, S.L., Neugut, A.I., Santella, R.M., Gammon, M.D., 2007. Cooked meat and risk of breast cancerdLifetime versus recent dietary intake. Epidemiology 18, 373–382. Stender, S., Dyerberg, J., 2004. Influence of trans fatty acids on health. Annals of Nutrition & Metabolism 48, 61–66. Stocks, T., Rapp, K., Bjorge, T., Manjer, J., Ulmer, H., Selmer, R., Lukanova, A., Johansen, D., Concin, H., Tretli, S., Hallmans, G., Jonsson, H., Stattin, P., 2009. Blood glucose and risk of incident and fatal cancer in the metabolic syndrome and cancer project (me-can): Analysis of six prospective cohorts. PLoS Medicine 6 (12), e1000201. https://doi.org/ 10.1371/journal.pmed.1000201. Sturgeon, J.L., Williams, M., van Servellen, G., 2009. Efficacy of green tea in the prevention of cancers. Nursing and Health Sciences 11, 436–446. Sung, M.K., Yeon, J.Y., Park, S.Y., Park, J.H.Y., Choi, M.S., 2011. Obesity-induced metabolic stresses in breast and colon cancer. Annals of the New York Academy of Sciences 1229 (1), 61–68. Taylor, C.K., Levy, R.M., Elliott, J.C., et al., 2009. The effect of genistein aglycone on cancer and cancer risk: A review of in vitro, preclinical, and clinical studies. Nutrition Reviews 67, 398–415. Tham, D.M., Gardner, C.D., Haskell, W.L., 1998. Potential health benefits of dietary phytoestrogens: A review of the clinical, epidemiological, and mechanistic evidence. The Journal of Clinical Endocrinology and Metabolism 83, 2223–2235. Tharrey, M., Mariotti, F., Mashchak, A., Barbillon, P., Delattre, M., Fraser, G.E., 2018. Patterns of plant and animal protein intake are strongly associated with cardiovascular mortality: The Adventist health Study-2 cohort. International Journal of Epidemiology 1–10. Thissen, J.P., Ketelslegers, J.M., Underwood, L.E., 1994. Nutritional regulation of the insulin-like growth factors. Endocrine Reviews 15, 80–101. Thompson Jr., I.M., Cabang, A.B., Wargovich, M.J., 2014. Future directions in the prevention of prostate cancer. Nature Reviews. Clinical Oncology 11 (1), 49–60. Thompson, A.K., Shaw, D.I., Minihane, A.M., Williams, C.M., 2008. Trans-fatty acids and cancer: The evidence reviewed. Nutrition Research Reviews 21, 174–188. Tinsley, G.M., La Bounty, P.M., 2015. Effects of intermittent fasting on body composition and clinical health markers in humans. Nutrition Reviews 73, 661–674. Title, L.M., Cummings, P.M., Giddens, K., Nassar, B.A., 2000. Oral glucose loading acutely attenuates endothelium-dependent vasodilation in healthy adults without diabetes: An effect prevented by vitamins C and E. Journal of the American College of Cardiology 36 (7), 2185–2191. Toledo, E., Salas-Salvado, J., Donat-Vargas, et al., 2015. Mediterranean diet and invasive breast cancer risk among women at high cardiovascular risk in the PREDIMED trial: A randomized clinical trial. JAMA Internal Medicine 175, 1752–1760. Toledo, E., Salas-Salvado, J., Donat-Vargas, C., Buil-Cosiales, P., Estruch, R., Ros, E., Corella, D., Fito, M., Hu, F.B., Aros, F., Gomez-Gracia, E., Romaguera, D., Ortega-Calvo, M., Serra-Majem, L., Pinto, X., Schroder, H., Basora, J., Sorli, J.V., Bullo, M., Serra-Mir, M., Martinez-Gonzalez, M.A., 2015. Mediterranean diet and invasive breast Cancer risk among women at high cardiovascular risk in the PREDIMED trial: A randomized clinical trial. JAMA Internal Medicine 175 (11), 1752–1760. Trichopoulou, A., Lagiou, P., Kuper, H., Trichopoulos, D., 2000. Cancer and Mediterranean dietary traditions. Cancer Epidemiology, Biomarkers & Prevention 9 (9), 869–873. Trock, B.J., Hilakivi-Clarke, L., Clarke, R., 2006. Meta-analysis of soy intake and breast cancer risk. Journal of the National Cancer Institute 98, 459–471. Tse, G., Eslick, G.D., 2014. Cruciferous vegetables and risk of colorectal neoplasms: A systematic review and meta-analysis. Nutrition and Cancer 66, 128–139. Tseng, Y.H., Tsan, Y.T., Chan, W.C., Sheu, W.H., Chen, P.C., 2015. Use of an alpha-glucosidase inhibitor and the risk of colorectal Cancer in patients with diabetes: A Nationwide, population-based cohort study. Diabetes Care. https://doi.org/10.2337/dc15-0563.

496

Diet and Cancer

Turati, F., Galeone, C., Gandini, S., Augustin, L.S., Jenkins, D.J., Pelucchi, C., La Vecchia, C., 2015a. High glycemic index and glycemic load are associated with moderately increased cancer risk. Molecular Nutrition & Food Research 59 (7), 1384–1394. https://doi.org/10.1002/mnfr.201400594. Turati, F., Rossi, M., Pelucchi, C., Levi, F., La Vecchia, C., 2015. Fruit and vegetables and cancer risk: A review of southern European studies. The British Journal of Nutrition 113, S102–S110. Turati, F., Rossi, M., Pelucchi, C., Levi, F., La Vecchia, C., 2015b. Fruit and vegetables and cancer risk: A review of southern European studies. The British Journal of Nutrition 113 (Suppl 2), S102–S110. https://doi.org/10.1017/S0007114515000148. U.S. Department of Agriculture and Agriculture Research Service, 2018. USDA food composition databases. https://ndb.nal.usda.gov/ndb/nutrients/report/nutrientsfrm? max¼25&offset¼0&totCount¼0&nutrient1¼203&fg¼20&fg¼9&fg¼16&fg¼12&fg¼11&subset¼0&sort¼c&measureby¼g. U.S. Department of Agriculture and U.S. Department of Health and Human Services, 2010. Dietary guidelines for Americans, 7th edn. U.S. Government Printing Office., Washington, DC. December 2010. U.S. Department of Health and Human Services and U.S. Department of Agriculture, 2015. 2015–2020 Dietary Guidelines for Americans, 8th edn. December. Available at: http:// health.gov/dietaryguidelines/2015/guidelines/. Valko, M., Leibfritz, D., Moncol, J., Cronin, M.T., Mazur, M., Telser, J., 2007. Free radicals and antioxidants in normal physiological functions and human disease. The International Journal of Biochemistry & Cell Biology 39 (1), 44–84. https://doi.org/10.1016/j.biocel.2006.07.001. Van der Schouw, Y.T., Kreijkamp-Kaspers, S., Peeters, P.H., Keinan-Boker, L., Rimm, E.B., Grobbee, D.E., 2005. Prospective study on usual dietary phytoestrogen intake and cardiovascular disease risk in western women. Circulation 111, 465–471. Van Erp-Baart, M.A., Brants, H.A., Kiely, M., Mulligan, A., Turrini, A., Sermoneta, C., Kilkkinen, A., Valsta, L.M., 2003. Isoflavone intake in four different European countries: The VENUS approach. The British Journal of Nutrition 89, S25–S30. Vance, T.M., Su, J., Fontham, E.T., Koo, S.I., Chun, O.K., 2013. Dietary antioxidants and prostate cancer: A review. Nutrition and Cancer 65 (6), 793–801. Varady, K.A., 2016. Impact of intermittent fasting on glucose homeostasis. Current Opinion in Clinical Nutrition and Metabolic Care 19 (4), 300–302. Varady, K.A., Hellerstein, M.K., 2007 Jul. Alternate-day fasting and chronic disease prevention: A review of human and animal trials. The American Journal of Clinical Nutrition 86 (1), 7–13. Verhoeven, D.T.H., Goldbohm, R.A., vanPoppel, G., Verhagen, H., vanden-Brandt, P.A., 1996. Epidemiological studies on brassica vegetables and cancer risk. Cancer Epidemiology, Biomarkers 5, 733–748. Verhoeven, D.T., Verhagen, H., Goldbohm, R.A., van den Brandt, P.A., van Poppel, G., 1997. A review of mechanisms underlying anticarcinogenicity by brassica vegetables. Chemico-Biological Interactions 103, 79–129. Vernieri, C., Nichetti, R.A., Pusceddu, S., Platania, M., Berrino, F., de Braud, F., 2018. Diet and supplements in cancer prevention and treatment: Clinical evidences and future perspectives. Critical Reviews in Oncology/Hematology 123, 57–73. https://doi.org/10.1016/j.critrevonc.2018.01.002. Epub 2018 Feb 2. Villarini, A., Pasanisi, P., Traina, A., Mano, M.P., Bonanni, B., Panico, S., Scipioni, C., Galasso, R., Paduos, A., Simeoni, M., Bellotti, E., Barbero, M., Macellari, G., Venturelli, E., Raimondi, M., Bruno, E., Gargano, G., Fornaciari, G., Morelli, D., Seregni, E., Krogh, V., Berrino, F., 2012. Lifestyle and breast cancer recurrences: The DIANA-5 trial. Tumori 98 (1), 1–18. Visioli, F., Grande, S., Bogani, P., Galli, C., 2004. The role of antioxidants in the mediterranean diets: Focus on cancer. European Journal of Cancer Prevention 13 (4), 337–343. Vitale, S.G., Laganà, A.S., Nigro, A., La Rosa, V.L., Rossetti, P., Rapisarda, A.M.C., La Vignera, S., Condorelli, R.A., Corrado, F., Buscema, M., et al., 2016. Peroxisome proliferatoractivated receptor modulation during metabolic diseases and cancers: Master and minions. PPAR Research 2016, 6517313. Wang, T., Rohan, T.E., Gunter, M.J., Xue, X., Wactawski-Wende, J., Rajpathak, S.N., Cushman, M., Strickler, H.D., Kaplan, R.C., Wassertheil-Smoller, S., Scherer, P.E., Ho, G.Y., 2011. A prospective study of inflammation markers and endometrial cancer risk in postmenopausal hormone nonusers. Cancer Epidemiology, Biomarkers & Prevention 20 (5), 971–977. WCRF/AICR, 2007. Food, Nutrition, physical activity and the prevention of Cancer: A global perspective. Washington. WCRF/AICR, 2007. Food, nutrition, physical activity and the prevention of Cancer: A global perspective. Willett, W.C., 1995. Diet, nutrition, and avoidable cancer. Environmental Health Perspectives 103 (Suppl 8), 165–170. Willett, W.C., Sacks, F., Trichopoulou, A., Drescher, G., Ferro-Luzzi, A., Helsing, E., Trichopoulos, D., 1995. Mediterranean diet pyramid: A cultural model for healthy eating. The American Journal of Clinical Nutrition 61 (6 Suppl), 1402S–1406S. Williams, C.D., Whitley, B.M., Hoyo, C., Grant, D.J., Iraggi, J.D., Newman, K.A., Gerber, L., Taylor, L.A., McKeever, M.G., Freedland, S.J., 2011. A high ratio of dietary n-6/n-3 polyunsaturated fatty acids is associated with increased risk of prostate cancer. Nutrition Research 31 (1), 1–8. https://doi.org/10.1016/j.nutres.2011.01.002. Wirfalt, E., McTaggart, A., Pala, V., Gullberg, B., Frasca, G., Panico, S., Bueno-de-Mesquita, H.B., Peeters, P.H., Engeset, D., Skeie, G., Chirlaque, M.D., Amiano, P., Lundin, E., Mulligan, A., Spencer, E.A., Overvad, K., Tjonneland, A., Clavel-Chapelon, F., Linseisen, J., Nothlings, U., Polychronopoulos, E., Georga, K., Charrondiere, U.R., Slimani, N., 2002. Food sources of carbohydrates in a European cohort of adults. Public Health Nutrition 5 (6B), 1197–1215. https://doi.org/10.1079/PHN2002399. Wu, A.H., Spicer, D., Garcia, A., Tseng, C.C., Hovanessian-Larsen, L., Sheth, P., Martin, S.E., Hawes, D., Russell, C., MacDonald, H., et al., 2015. Double-blind randomized 12month soy intervention had no effects on breast MRI fibroglandular tissue density or mammographic density. Cancer Prevention Research 8, 942–951. Wu, A.H., Spicer, D., Garcia, A., Tseng, C.C., Hovanessian-Larsen, L., Sheth, P., Martin, S.E., Hawes, D., Russell, C., MacDonald, H., et al., 2015. Double-blind randomized 12month soy intervention had no effects on breast MRI Fibroglandular tissue density or mammographic density. Cancer Prevention Research 8, 942–951. Wu, A.H., Wan, P., Hankin, J., Tseng, C.C., Yu, M.C., Pike, M.C., 2002. Adolescent and adult soy intake and risk of breast cancer in Asian-Americans. Carcinogenesis 23, 1491–1496. Wu, Q.J., Yang, Y., Vogtmann, E., Wang, J., Han, L.H., Li, H.L., Xiang, Y.B., 2013. Cruciferous vegetables intake and the risk of colorectal cancer: A meta-analysis of observational studies. Annals of Oncology 24, 1079–1087. Wu, A.H., Yu, M.C., Tseng, C.C., Stanczyk, F.Z., Pike, M.C., 2009. Dietary patterns and breast cancer risk in Asian American women. The American Journal of Clinical Nutrition 89 (4), 1145–1154. Xia, H., Ma, S., Wang, S., Sun, G., 2015. Meta-analysis of saturated fatty acid intake and breast Cancer risk. Medicine (Baltimore) 94 (52), e2391. https://doi.org/10.1097/ MD.0000000000002391. Xin, Y., Li, X.Y., Sun, S.R., Wang, L.X., Huang, T., 2015. Vegetable oil intake and breast Cancer risk: A meta-analysis. Asian Pacific Journal of Cancer Prevention 16 (12), 5125–5135. Xu, L., Ding, Y., Catalona, W.J., Yang, X.J., Anderson, W.F., Jovanovic, B., Wellman, K., Killmer, J., Huang, X., Scheidt, K.A., et al., 2009. MEK4 function, genistein treatment, and invasion of human prostate cancer cells. Journal of the National Cancer Institute 101, 1141–1155. Xue, Z., Yu, W., Liu, Z., Wu, M., Kou, X., Wang, J., 2009. Preparation and antioxidative properties of a rapeseed (Brassica napus) protein hydrolysate and three peptide fractions. Journal of Agricultural and Food Chemistry 57 (12), 5287–5293. Yamabe, S., Kobayashi-Hattori, K., Kaneko, K., Endo, H., Takita, T., 2007. Effect of soybean varieties on the content and composition of isoflavone in rice-koji miso. Food Chemistry 100, 369–374. Yamane, T., Takahashi, T., Kuwata, K., et al., 1995. Inhibition of N-methyl-N’-nitro-N-nitrosoguanidine-induced carcinogenesis by (-)-epigallocatechin gallate in the rat glandular stomach. Cancer Research 55, 2081–2084. Yang, B., Ren, X.L., Fu, Y.Q., Gao, J.L., Li, D., 2014. Ratio of n-3/n-6 PUFAs and risk of breast cancer: A meta-analysis of 274135 adult females from 11 independent prospective studies. BMC Cancer 14, 105. https://doi.org/10.1186/1471-2407-14-105.

Diet and Cancer

497

Yang, C.S., Wang, X., Lu, G., Picinich, S.C., 2009. Cancer prevention by tea: Animal studies, molecular mechanisms and human relevance. Nature Reviews. Cancer 9, 429–439. Yao X, Tian Z. Saturated, Monounsaturated and polyunsaturated fatty acids intake and risk of pancreatic Cancer: Evidence from observational studies. PLoS One 2015; 10(6):e0130870. doi: https://doi.org/10.1371/journal.pone.0130870. eCollection 2015. Yiannakopoulou, E.C., 2014. Effect of green tea catechins on breast carcinogenesis: A systematic review of in-vitro and in-vivo experimental studies. European Journal of Cancer Prevention 23, 84–89. Yoan, J.M., Koh, W.P., Sun, C.L., Lee, H.P., Yu, M.C., 2007. Green tea intake, ACE gene polymorphism and breast cancer risk among Chinese women in Singapore. Carcinogenesis 28, 1074–1078. Yu, Y., Jing, X., Li, H., Zhao, X., Wang, D., 2016. Soy isoflavone consumption and colorectal cancer risk: A systematic review and meta-analysis. Scientific Reports 6, 25939. Yu, Y., Jing, X., Li, H., Zhao, X., Wang, D., 2016. Soy isoflavone consumption and colorectal cancer risk: A systematic review and meta-analysis. Scientific Reports 6, 25939. Yuan, J.M., 2013. Cancer prevention by green tea: Evidence from epidemiologic studies. The American Journal of Clinical Nutrition 98, 1676S–1681S. Zadra, G., Photopoulos, C., Loda, M., 2013 Oct. The fat side of prostate cancer. Biochimica et Biophysica Acta 1831 (10), 1518–1532. https://doi.org/10.1016/j.bbalip.2013.03.010. Epub 2013 Apr 2. Zaheer, K., Humayoun Akhtar, M., 2017. An updated review of dietary isoflavones: Nutrition, processing, bioavailability and impacts on human health. Critical Reviews in Food Science and Nutrition 57, 1280–1293. Zamora-Ros, R., Ferrari, P., González, C.A., Tjønneland, A., Olsen, A., Bredsdorff, L., Overvad, K., Touillaud, M., Perquier, F., Fagherazzi, G., et al., 2013. Dietary flavonoid and lignan intake and breast cancer risk according to menopause and hormone receptor status in the European prospective investigation into Cancer and nutrition (EPIC) study. Breast Cancer Research and Treatment 139, 163–176. Zhang, Q., Feng, H., Qluwakemi, B., Wang, J., Yao, S., Cheng, G., Xu, H., Qiu, H., Zhu, L., Yuan, M., 2016. Phytoestrogens and risk of prostate cancer: An updated meta-analysis of epidemiologic studies. International Journal of Food Sciences and Nutrition 1–15. Zhang, S., Hunter, D.J., Forman, M.R., et al., 1999. Dietary carotenoids and vitamins A, C, and E and risk of breast cancer. Journal of the National Cancer Institute 91 (6), 547–556. Zhang, Y.F., Kang, H.B., Li, B.L., et al., 2012. Positive effects of soy isoflavone food on survival of breast cancer patients in China. Asian Pacific Journal of Cancer Prevention 13, 479–482. Zhang, Y., Talalay, P., Cho, C.-G., Posner, G., 1992. A major inducer of anticarcinogenic protective enzymes from broccoli: Isolation and elucidation of structure. Proceedings of the National Academy of Science USA 89, 2399–2403. Zhang, J., Zhang, H., Wang, L., Guo, X., Wang, X., Yao, H., 2010. Isolation and identification of antioxidative peptides from rice endosperm protein enzymatic hydrolysate by consecutive chromatography and MALDI-TOF/TOF MS/MS. Food Chemistry 119 (1), 226–234. Zhao, J., Lyu, C., Gao, J., Du, L., Shan, B., Zhang, H., Wang, H.Y., Gao, Y., 2016b Jul. Dietary fat intake and endometrial cancer risk: A dose response meta-analysis. Medicine (Baltimore) 95 (27), e4121. https://doi.org/10.1097/MD.0000000000004121. Zhao, H., Zhou, L., Shangguan, A.J., Bulun, S.E., 2016a. Aromatase expression and regulation in breast and endometrial cancer. Journal of Molecular Endocrinology 57 (1), R19– R33. https://doi.org/10.1530/JME-15-0310. Epub 2016 Apr 11. Zheng, J.S., Hu, X.J., Zhao, Y.M., Yang, J., Li, D., 2013. Intake of fish and marine n-3 polyunsaturated fatty acids and risk of breast cancer: Meta-analysis of data from 21 independent prospective cohort studies. BMJ 346, f3706. https://doi.org/10.1136/bmj.f3706. Zheng, J.S., Hu, X.J., Zhao, Y.M., Yang, J., Li, D., 2013. Intake of fish and marine n-3 polyunsaturated fatty acids and risk of breast cancer: Meta-analysis of data from 21 independent prospective cohort studies. BMJ 346, 346–f3706. f3706. Zheng, W., Lee, S.A., 2009. Well-done meat intake, heterocyclic amine exposure, and cancer risk. Nutrition and Cancer 61, 437–446. Zhong, X., Fang, Y.J., Pan, Z.Z., Li, B., Wang, L., Zheng, M.C., Chen, Y.M., Zhang, C.X., 2013. Dietary fat, fatty acid intakes and colorectal cancer risk in Chinese adults: a casecontrol study. European Journal of Cancer Prevention 22 (5), 438–447. https://doi.org/10.1097/CEJ.0b013e32835e88c4. Zhong, X.S., Ge, J., Chen, S.W., Xiong, Y.Q., Ma, S.J., Chen, Q., 2018. Association between dietary Isoflavones in soy and legumes and endometrial cancer: A systematic review and meta-analysis. Journal of the Academy of Nutrition and Dietetics 118 (4), 637–651. Zhong, L., Goldberg, M.S., Gao, Y.T., et al., 2001. Population-based case-control study of lung cancer and green tea consumption among women living in Shanghai, China. Epidemiology 12, 695–700. Zhou, J.R., Gugger, E.T., Tanaka, T., Guo, Y., Blackburn, G.L., Clinton, S.K., 1999. Soybean phytochemicals inhibit the growth of transplantable human prostate carcinoma and tumor angiogenesis in mice. The Journal of Nutrition 129, 1628–1635. Ziegler, R.G., Hoover, R.N., Pike, M.C., Hildesheim, A., Nomura, A.M., West, D.W., Wu-Williams, A.H., Kolonel, L.N., Horn-Ross, P.L., Rosenthal, J.F., Hyer, M.B., 1993. Migration patterns and breast cancer risk in Asian-American women. Journal of the National Cancer Institute 85 (22), 1819–1827. Zong, G., Gao, A., Hu, F.B., Sun, Q., 2016. Whole grain intake and mortality from all causes, cardiovascular disease, and Cancer: A meta-analysis of prospective cohort studies. Circulation 133 (24), 2370–2380. https://doi.org/10.1161/CIRCULATIONAHA.115.021101. Zuniga, K.E., Clinton, S.K., Erdman, J.W., 2013. The interactions of dietary tomato powder and soy germ on prostate carcinogenesis in the TRAMP model. Cancer Prevention Research 6, 548–557.

Further Reading Adlercreutz, H., Markkanen, H., Watanabe, S., 1993. Plasma concentrations of phyto-oestrogens in Japanese men. Lancet 342, 1209–1210. Agnoli, C., Grioni, S., Sieri, S., Palli, D., Masala, G., Sacerdote, C., Vineis, P., Tumino, R., Giurdanella, M.C., Pala, V., Berrino, F., Mattiello, A., Panico, S., Krogh, V., 2013. Italian Mediterranean index and risk of colorectal cancer in the Italian section of the EPIC cohort. International Journal of Cancer 132 (6), 1404–1411. AICR’S Foods That Fight CancerdBroccoli & Cruciferous Vegetables. 2018. http://www.aicr.org/foods-that-fight-cancer/broccoli-cruciferous.html (accessed on 14 April 2018). Alberti-Fidanza, A., Fidanza, F., Chiuchiu, M.P., Verducci, G., Fruttini, D., 1999. Dietary studies on two rural italian population groups of the seven countries study. 3. Trend of food and nutrient intake from 1960 to 1991. European Journal of Clinical Nutrition 53 (11), 854–860. American Institute for Cancer Research (AICR). (2018) Soy is Safe for Breast Cancer Survivors. Available online: http://www.aicr.org/press/press-releases/soy-safe-breast-cancersurvivors.html (accessed on 14 April 2018). American Institute for Cancer Research (AICR). 2018 Soybeans, soy nuts and edamame contain fiber. After a systematic review of the global scientific literature, AICR/WCRF weighed the strength of the evidence linking these factors to lower risk for several cancers. Available online: http://www.aicr.org/foods-that-fight-cancer/tab-content/soyresearch-1.html. (accessed on 14 April 2018). Aucoin, M., Cooley, K., Knee, C., Fritz, H., Balneaves, L.G., Breau, R., Fergusson, D., Skidmore, B., Wong, R., Seely, D., 2017. Fish-derived Omega-3 fatty acids and prostate Cancer: A systematic review. Integrative Cancer Therapies 16 (1), 32–62. https://doi.org/10.1177/1534735416656052. Bach, A., Serra-Majem, L., Carrasco, J.L., Roman, B., Ngo, J., Bertomeu, I., Obrador, B., 2006. The use of indexes evaluating the adherence to the Mediterranean diet in epidemiological studies: A review. Public Health Nutrition 9 (1A), 132–146.

498

Diet and Cancer

Balanza, R., Garcia-Lorda, P., Perez-Rodrigo, C., Aranceta, J., Bonet, M.B., Salas-Salvado, J., 2007. Trends in food availability determined by the food and agriculture Organization’s food balance sheets in Mediterranean Europe in comparison with other European areas. Public Health Nutrition 10 (2), 168–176. Bamia, C., Lagiou, P., Buckland, G., et al., 2013. Mediterranean diet and colorectal cancer risk: Results from a European cohort. European Journal of Epidemiology 28 (4), 317– 328. https://doi.org/10.1007/s10654-013-9795-x. Epub 2013 Apr 12. Bandera, E.V., Kushi, L.H., Moore, D.F., Gifkins, D.M., McCullough, M.L., 2007. Fruits and vegetables and endometrial cancer risk: A systematic literature review and metaanalysis. Nutrition and Cancer 58 (1), 6–21. https://doi.org/10.1080/01635580701307929. Berrino, F., 2016. Mediterranean diet and its association with reduced invasive breast cancer risk. JAMA Oncology 2 (4), 535–536. Bloomfield, H.E., Koeller, E., Greer, N., MacDonald, R., Kane, R., Wilt, T.J., 2016. Effects on health outcomes of a Mediterranean diet with no restriction on fat intake: A systematic review and meta-analysis. Annals of Internal Medicine 165 (7), 491–500. https://doi.org/10.7326/M16-0361. Boker, L.K., Van der Schouw, Y.T., De Kleijn, M.J., Jacques, P.F., Grobbee, D.E., Peeters, P.H., 2002. Intake of dietary phytoestrogens by Dutch women. The Journal of Nutrition 132, 1319–1328. Bravi, F., Scotti, L., Bosetti, C., Zucchetto, A., Talamini, R., Montella, M., Greggi, S., Pelucchi, C., Negri, E., Franceschi, S., La Vecchia, C., 2009. Food groups and endometrial cancer risk: A case-control study from Italy. American Journal of Obstetrics and Gynecology 200 (3), 293 e1–7. https://doi.org/10.1016/j.ajog.2008.09.015. Brennan, S.F., Woodside, J.V., Lunny, P.M., Cardwell, C.R., Cantwell, M.M., 2017. Dietary fat and breast cancer mortality: A systematic review and meta-analysis. Critical Reviews in Food Science and Nutrition 57 (10), 1999–2008. Buckland G et al. Adherence to the mediterranean diet and risk of breast cancer in the European prospective investigation into cancer and nutrition cohort study. International Journal of Cancer 132 (12), 2918–2927. Buckland, G., Travier, N., Cottet, V., Gonzalez, C.A., Lujan-Barroso, L., Agudo, A., Trichopoulou, A., Lagiou, P., Trichopoulos, D., Peeters, P.H., May, A., Bueno-de-Mesquita, H.B., Bvan Duijnhoven, F.J., Key, T.J., Allen, N., Khaw, K.T., Wareham, N., Romieu, I., McCormack, V., Boutron-Ruault, M., Clavel-Chapelon, F., Panico, S., Agnoli, C., Palli, D., Tumino, R., Vineis, P., Amiano, P., Barricarte, A., Rodriguez, L., Sanchez, M.J., Chirlaque, M.D., Kaaks, R., Teucher, B., Boeing, H., Bergmann, M.M., Overvad, K., Dahm, C.C., Tjonneland, A., Olsen, A., Manjer, J., Wirfalt, E., Hallmans, G., Johansson, I., Lund, E., Hjartaker, A., Skeie, G., Vergnaud, A.C., Norat, T., Romaguera, D., Riboli, E., 2013. Adherence to the mediterranean diet and risk of breast cancer in the European prospective investigation into cancer and nutrition cohort study. International Journal of Cancer 132 (12), 2918–2927. https://doi.org/10.1002/ijc.27958. Capurso, C., Vendemiale, G., 2017. The Mediterranean diet reduces the risk and mortality of the prostate Cancer: A narrative review. Frontiers in Nutrition 4, 38. https://doi.org/ 10.3389/fnut.2017.00038. Chao, A., Thun, M.J., Connell, C.J., McCullough, M.L., Jacobs, E.J., Flanders, W.D., Rodriguez, C., Sinha, R., Calle, E.E., 2005. Meat consumption and risk of colorectal cancer. Journal of the American Medical Association 293 (2), 172–182. da Silva, R., Bach-Faig, A., Raido Quintana, B., Buckland, G., Vaz de Almeida, M.D., Serra-Majem, L., 2009. Worldwide variation of adherence to the Mediterranean diet, in 19611965 and 2000-2003. Public Health Nutrition 12 (9A), 1676–1684. Davis-Yadley, A.H., Malafa, M.P., 2015. Vitamins in pancreatic cancer: A review of underlying mechanisms and future applications. Advances in Nutrition 6 (6), 774–802. https:// doi.org/10.3945/an.115.009456. De Kleijn, M.J., van der Schouw, Y.T., Wilson, P.W., Adlercreutz, H., Mazur, W., Grobbee, D.E., Jacques, P.F., 2001. Intake of dietary phytoestrogens is low in postmenopausal women in the United States: The Framingham study (1–4). The Journal of Nutrition 131, 1826–1832. Di Sebastiano, K.M., Mourtzakis, M., 2014. The role of dietary fat throughout the prostate Cancer trajectory. Nutrients 6 (12), 6095–6109. Published online 2014 Dec 22. https:// doi.org/10.3390/nu6126095. Dinu, M., Pagliai, G., Casini, A., Sofi, F., 2018. Mediterranean diet and multiple health outcomes: An umbrella review of meta-analyses of observational studies and randomised trials. European Journal of Clinical Nutrition 72 (1), 30–43. https://doi.org/10.1038/ejcn.2017.58. Economopoulos, K.P., Sergentanis, T.N., 2010. GSTM1, GSTT1, GSTP1, GSTA1 and colorectal cancer risk: A comprehensive meta-analysis. European Journal of Cancer 31, 1617–1631. Farnetti, S., Malandrino, N., Luciani, D., Gasbarrini, G., Capristo, E., 2011. Food fried in extra-virgin olive oil improves postprandial insulin response in obese, insulin-resistant women. Journal of Medicinal Food 14, 316–321. Faulkner-Hogg, K.B., Selby, W.S., Loblay, R.H., 1999. Dietary analysis in symptomatic patients with coeliac disease on a gluten-free diet: The role of trace amounts of gluten and non-gluten food intolerances. Scandinavian Journal of Gastroenterology 34, 784–789. Ferro-Luzzi, A., Branca, F., 1995. Mediterranean diet, Italian-style: Prototype of a healthy diet. The American Journal of Clinical Nutrition 61 (6 Suppl), 1338S–1345S. Fidanza, F., Alberti, A., Lanti, M., Menotti, A., 2004. Mediterranean adequacy index: Correlation with 25-year mortality from coronary heart disease in the seven countries study. Nutrition, Metabolism, and Cardiovascular Diseases 14 (5), 254–258. Filomeno, M., Bosetti, C., Bidoli, E., Levi, F., Serraino, D., Montella, M., La Vecchia, C., Tavani, A., 2015. Mediterranean diet and risk of endometrial cancer: A pooled analysis of three Italian case-control studies. British Journal of Cancer 112 (11), 1816–1821. https://doi.org/10.1038/bjc.2015.153. Filomeno, M., Bosetti, C., Bidoli, E., Levi, F., Serraino, D., Montella, M., La Vecchia, C., Tavani, A., 2015. Mediterranean diet and risk of endometrial cancer: A pooled analysis of three Italian case-control studies. British Journal of Cancer 112 (11), 1816–1821. https://doi.org/10.1038/bjc.2015.153. Epub 2015 May 12. Fortner, R.T., Husing, A., Kuhn, T., Konar, M., Overvad, K., Tjonneland, A., Hansen, L., Boutron-Ruault, M.C., Severi, G., Fournier, A., Boeing, H., Trichopoulou, A., Benetou, V., Orfanos, P., Masala, G., Agnoli, C., Mattiello, A., Tumino, R., Sacerdote, C., Bueno-de-Mesquita, H.B., Peeters, P.H., Weiderpass, E., Gram, I.T., Gavrilyuk, O., Quiros, J.R., Maria Huerta, J., Ardanaz, E., Larranaga, N., Lujan-Barroso, L., Sanchez-Cantalejo, E., Butt, S.T., Borgquist, S., Idahl, A., Lundin, E., Khaw, K.T., Allen, N.E., Rinaldi, S., Dossus, L., Gunter, M., Merritt, M.A., Tzoulaki, I., Riboli, E., Kaaks, R., 2017. Endometrial cancer risk prediction including serum-based biomarkers: Results from the EPIC cohort. International Journal of Cancer 140 (6), 1317–1323. https://doi.org/10.1002/ijc.30560. Gann, P.H., Ma, J., Giovannucci, E., Willett, W., Sacks, F.M., Hennekens, C.H., Stampfer, M.J., 1999. Lower prostate cancer risk in men with elevated plasma lycopene levels: Results of a prospective analysis. Cancer Research 59 (6), 1225–1230. Garcia-Closas, R., Berenguer, A., Gonzalez, C.A., 2006. Changes in food supply in Mediterranean countries from 1961 to 2001. Public Health Nutrition 9 (1), 53–60. Giacosa, A., Barale, R., Bavaresco, L., Gatenby, P., Gerbi, V., Janssens, J., Johnston, B., Kas, K., La Vecchia, C., Mainguet, P., Morazzoni, P., Negri, E., Pelucchi, C., Pezzotti, M., Rondanelli, M., 2013. Cancer prevention in Europe: The Mediterranean diet as a protective choice. European Journal of Cancer Prevention 22 (1), 90–95. https://doi.org/ 10.1097/CEJ.0b013e328354d2d7. Giampaoli, S., Krogh, V., Grioni, S., Palmieri, L., Gulizia, M.M., Stamler, J., Vanuzzo, D., Gruppo di ricerca dell Osservatorio epidemiologico cardiovascolare/health examination S, 2015. Eating behaviours of italian adults: Results of the Osservatorio epidemiologico cardiovascolare/health examination survey. Epidemiologia e Prevenzione 39 (5–6), 373–379. Gibney, M.J., Forde, C.G., Mullally, D., Gibney, E.R., 2017. Ultra-processed foods in human health: A critical appraisal. The American Journal of Clinical Nutrition 106 (3), 717–724. Gong, T.T., Li, D., Wu, Q.J., Wang, Y.Z., 2016. Cholesterol consumption and risk of endometrial cancer: A systematic review and dose-response meta-analysis of observational studies. Oncotarget 7 (13), 16996–17008. https://doi.org/10.18632/oncotarget.7913. Graff, R.E., Pettersson, A., Lis, R.T., Ahearn, T.U., Markt, S.C., Wilson, K.M., Rider, J.R., Fiorentino, M., Finn, S., Kenfield, S.A., Loda, M., Giovannucci, E.L., Rosner, B., Mucci, L.A., 2016. Dietary lycopene intake and risk of prostate cancer defined by ERG protein expression. The American Journal of Clinical Nutrition 103 (3), 851–860. https:// doi.org/10.3945/ajcn.115.118703.

Diet and Cancer

499

Gupta, K., Kshirsagar, S., Chang, L., Schwartz, R., Law, P.Y., Yee, D., Hebbel, R.P., 2002. Morphine stimulates angiogenesis by activating proangiogenic and survival-promoting signaling and promotes breast tumor growth. Cancer Research 62 (15), 4491–4498. Kapiotis, S., Hermann, M., Held, I., Seelos, C., Ehringer, H., Gmeiner, B.M., 1997. Genistein, the dietary-derived angiogenesis inhibitor, prevents LDL oxidation and protects endothelial cells from damage by atherogenic LDL. Arteriosclerosis, Thrombosis, and Vascular Biology 17, 2868–2874. Khodarahmi, M., Azadbakht, L., 2014. The association between different kinds of fat intake and breast cancer risk in women. International Journal of Preventive Medicine 5, 6–15. King, D.E., Egan, B.M., Woolson, R.F., et al., 2007. Effect of a high-fiber diet vs a fiber-supplemented diet on C-reactive protein level. Archives of Internal Medicine 167, 502–506. Kromhout, D., Keys, A., Aravanis, C., Buzina, R., Fidanza, F., Giampaoli, S., Jansen, A., Menotti, A., Nedeljkovic, S., Pekkarinen, M., et al., 1989. Food consumption patterns in the 1960s in seven countries. American Journal of Clinical Nutrition 49 (5), 889–894. https://doi.org/10.1093/ajcn/49.5.889. Kuiper, G.G., Lemmen, J.G., Carlsson, B., Corton, J.C., Safe, S.H., van der Saag, P.T., van der Burg, B., Gustafsson, J.A., 1998. Interaction of estrogenic chemicals and phytoestrogens with estrogen receptor beta. Endocrinology 139, 4252–4263. Kwan, H.Y., Chao, X., Su, T., Fu, X.Q., Liu, B., Tse, A.K., Fong, W.F., Yu, Z.L., 2015. Dietary lipids and adipocytes: Potential therapeutic targets in cancers. The Journal of Nutritional Biochemistry 26 (4), 303–311. https://doi.org/10.1016/j.jnutbio.2014.11.001. Epub 2014 Dec 3. Lampe, J.W., 1999. Health effects of vegetables and fruit: Assessing mechanisms of action in human experimental studies. The American Journal of Clinical Nutrition 70 (3 Suppl), 475S–490S. Larsson, S.C., Giovannucci, E., Wolk, A., 2006. Folate intake, MTHFR polymorphisms, and risk of esophageal, gastric, and pancreatic cancer: A meta-analysis. Gastroenterology 131 (4), 1271–1283. https://doi.org/10.1053/j.gastro.2006.08.010. Lee, C., Longo, V.D., 2011. Fasting vs dietary restriction in cellular protection and cancer treatment: From model organisms to patients. Oncogene 30 (30), 3305–3316. Lee, C., Raffaghello, L., Brandhorst, S., Safdie, F.M., Bianchi, G., Martin-Montalvo, A., et al., 2012a. Fasting cycles retard growth of tumors and sensitize a range of cancer cell types to chemotherapy. Science Translational Medicine 4, 124–127. Levy, R.M., D’Anna, R., et al., 2008. Breast safety and efficacy of genistein aglycone for postmenopausal bone loss: A follow-up study. The Journal of Clinical Endocrinology and Metabolism 93, 4787–4796. Li, Y., Li, S., Meng, X., Gan, R.Y., Zhang, J.J., Li, H.B., 2017 Jul 8. Dietary natural products for prevention and treatment of breast Cancer. Nutrients 9, E728. Macias, V., Kajdacsy-Balla, A., Lumey, L.H., et al., 2013. Effect of soy protein isolate supplementation on biochemical recurrence of prostate cancer after radical prostatectomy: A randomized trial. JAMA 310, 170–178. Menotti, A., Keys, A., Aravanis, C., Blackburn, H., Dontas, A., Fidanza, F., Karvonen, M.J., Kromhout, D., Nedeljkovic, S., Nissinen, A., et al., 1989. Seven countries study. First 20year mortality data in 12 cohorts of six countries. Annals of Medicine 21 (3), 175–179. Messina, M.J., Wood, C.E., 2008. Soy isoflavones, estrogen therapy, and breast cancer risk: Analysis and commentary. Nutrition Journal 7, 17. Review. NIH-Cruciferous Vegetables and Cancer Prevention. https://www.cancer.gov/about-cancer/causes-prevention/risk/diet/cruciferous-vegetables-fact-sheet. (Accessed 14 April 2018). Nothlings, U., Murphy, S.P., Wilkens, L.R., Henderson, B.E., Kolonel, L.N., 2007. Flavonols and pancreatic cancer risk: The multiethnic cohort study. American Journal of Epidemiology 166 (8), 924–931. https://doi.org/10.1093/aje/kwm172. Parekh, N., Chandran, U., Bandera, E.V., 2012. Obesity in cancer survival. Annual Review of Nutrition 32, 311–342. Raffaghello, L., Longo, V., 2017. Metabolic alterations at the crossroad of aging and oncogenesis. International Review of Cell and Molecular Biology 332, 1–42. Raffaghello, L., Safdie, F., Bianchi, G., Dorff, T., Fontana, L., Longo, V.D., 2010a. Fasting and differential chemotherapy protection in patients. Cell Cycle 9 (22), 4474–4476. Reddy, B.S., 2002. Types and amount of dietary fat and colon cancer risk: Prevention by omega-3 fatty acid-rich diets. Environmental Health and Preventive Medicine 7 (3), 95– 102. https://doi.org/10.1265/ehpm.2002.95. Rose, D.P., 1990. Dietary fiber and breast cancer. Nutrition and Cancer 13 (1–2), 1–8. https://doi.org/10.1080/01635589009514040. Rossi, M., Lugo, A., Lagiou, P., Zucchetto, A., Polesel, J., Serraino, D., Negri, E., Trichopoulos, D., La Vecchia, C., 2012. Proanthocyanidins and other flavonoids in relation to pancreatic cancer: A case-control study in Italy. Annals of Oncology 23 (6), 1488–1493. https://doi.org/10.1093/annonc/mdr475. Schwingshackl, L., Schwedhelm, C., Galbete, C., Hoffmann, G., 2017. Adherence to Mediterranean diet and risk of Cancer: An updated systematic review and meta-analysis. Nutrients 9 (10). https://doi.org/10.3390/nu9101063. Serra-Majem, L., Roman, B., Estruch, R., 2006. Scientific evidence of interventions using the Mediterranean diet: A systematic review. Nutrition Reviews 64 (2 Pt 2), S27–S47. Serra-Majem, L., Trichopoulou, A., Ngo de la Cruz, J., Cervera, P., Garcia Alvarez, A., La Vecchia, C., Lemtouni, A., Trichopoulos, D., International Task Force on the Mediterranean D, 2004. Does the definition of the Mediterranean diet need to be updated? Public Health Nutrition 7 (7), 927–929. Sofi, F., Cesari, F., Abbate, R., Gensini, G.F., Casini, A., 2008. Adherence to Mediterranean diet and health status: Meta-analysis. BMJ 337, a1344. https://doi.org/10.1136/ bmj.a1344. Sofi, F., Macchi, C., Abbate, R., Gensini, G.F., Casini, A., 2013. Mediterranean diet and health. BioFactors 39 (4), 335–342. https://doi.org/10.1002/biof.1096. Sofi, F., Macchi, C., Abbate, R., Gensini, G.F., Casini, A., 2014. Mediterranean diet and health status: An updated meta-analysis and a proposal for a literature-based adherence score. Public Health Nutrition 17 (12), 2769–2782. https://doi.org/10.1017/S1368980013003169. Thompson Jr., I.M., Cabang, A.B., Wargovich, M.J., 2014. Future directions in the prevention of prostate cancer. Nature Reviews. Clinical Oncology 11 (1), 49–60. https://doi.org/ 10.1038/nrclinonc.2013.211. Trichopoulou, A., Lagiou, P., Kuper, H., Trichopoulos, D., 2000. Cancer and Mediterranean dietary traditions. Cancer Epidemiology, Biomarkers & Prevention 9 (9), 869–873. Tucker, K.L., 2010. Dietary patterns, approaches, and multicultural perspective. Applied Physiology, Nutrition, and Metabolism 35 (2), 211–218. UNESCO, 2010. http://www.unesco.it/it/PatrimonioImmateriale/Detail/384. USDA Database for the Isoflavone Content of Selected Food, Release 2.1 (November 2015). https://data.nal.usda.gov/dataset/usda-database-isoflavone-content-selected-foodsrelease-21-november-2015 (accessed on 14 April 2018). USDA 2018 Choose my plate. https://www.choosemyplate.gov/vegetables (accessed on 14 April 2018). Vance, T.M., Su, J., Fontham, E.T., Koo, S.I., Chun, O.K., 2013. Dietary antioxidants and prostate cancer: A review. Nutrition and Cancer 65 (6), 793–801. https://doi.org/ 10.1080/01635581.2013.806672. Varady, K.A., 2011. Intermittent versus daily calorie restriction: Which diet regimen is more effective for weight loss? Obesity Reviews 12 (7), e593–e601. Vareiro, D., Bach-Faig, A., Raido Quintana, B., Bertomeu, I., Buckland, G., Vaz de Almeida, M.D., Serra-Majem, L., 2009. Availability of Mediterranean and non-Mediterranean foods during the last four decades: Comparison of several geographical areas. Public Health Nutrition 12 (9A), 1667–1675. Wang, T., Rohan, T.E., Gunter, M.J., Xue, X., Wactawski-Wende, J., Rajpathak, S.N., Cushman, M., Strickler, H.D., Kaplan, R.C., Wassertheil-Smoller, S., Scherer, P.E., Ho, G.Y., 2011. A prospective study of inflammation markers and endometrial cancer risk in postmenopausal hormone nonusers. Cancer Epidemiology, Biomarkers & Prevention 20 (5), 971–977. https://doi.org/10.1158/1055-9965.EPI-10-1222. WCRF/AICR, 2007. Food, nutrition, physical activity and the prevention of Cancer: A global perspective. Willett, W.C., Sacks, F., Trichopoulou, A., Drescher, G., Ferro-Luzzi, A., Helsing, E., Trichopoulos, D., 1995. Mediterranean diet pyramid: A cultural model for healthy eating. The American Journal of Clinical Nutrition 61 (6 Suppl), 1402S–1406S. World Cancer Research Fund International (WCRF). Breast cancer survivors. Available online: http://www.wcrf.org/int/research-we-fund/continuous-update-project-findings-reports/ breast-cancer-survivors (accessed on 14 April 2018).

500

Diet and Cancer

World Cancer Research Fund/American Institute for Cancer Research. 2018 Colorectal cancer. https://wcrf.org/int/continuous-update-project/cup-findings-reports/colorectalcancer (accessed on 15 April 2018). www.iarc.it. https://cancer-code-europe.iarc.fr/index.php/it. www.iarc.it. https://cancer-code-europe.iarc.fr/index.php/it/.

Relevant Websites http://www.heart.org/HEARTORG/dAmerican Heart Association. https://www.dietitians.ca/dDietitians of Canada.

Dietary Factors and Cancer* Isabelle Romieu, National Institute of Public Health, Cuernavaca, Mexico; and Emory University, Atlanta, GA, United States © 2019 Elsevier Inc. All rights reserved.

Glossary Case control study An epidemiological study in which the participants are chosen based on their disease or condition (cases) or lack of it (controls) to test whether past or recent history of an exposure is associated with the risk of disease. Cohort study An epidemiological study of a group of people whose characteristics are recorded at recruitment, followed for a period of time during which the outcomes of interest are noted. Differences in the likelihood of a particular outcome are presented as the relative risk comparing one level of exposure to another. Confidence interval A measure of uncertainly in an estimate usually reported as 95% confidence interval. Dietary fiber Constituents of plant cell walls that are not digested in the small intestine. Non-starch polysaccharides are a consistent feature and are fermented by colonic bacteria to produce energy and short chain fatty acids including butyrate. Fatty acids A carboxylic acid with a carbon chain of varying length, which may be either saturated (no double bonds) or unsaturated (one or more double bonds). Meta-analysis The process of using statistical methods to combine the results of different studies. Pooling In Epidemiology, a type of study where original individual-level data from two or more original studies are obtained, combined and reanalyzed. Randomized controlled trial (RCT) A study in which a comparison is made between one intervention (often a treatment or prevention strategy) and another (control). Groups are randomized to one intervention or the other, so that any difference in outcome between the two groups can be ascribed with confidence to the intervention. Relative risk (RR) The ratio of the rate of the disease or death among people exposed to a factor, compared to the rate among the unexposed.

Introduction Interest in the cause and prevention of cancer arose from the observation that cancer rates vary greatly between countries and are correlated to dietary factors, and that nutrition can modify cancer incidence in animals. Furthermore, ecological studies showing correlation between dietary habits and cancer incidence and mortality, and migrant studies on changing rates with change in diet and life style, have stimulated research on the role of diet in cancer prevention. To provide more specific information on human cancer, many retrospectives case-control studies, some prospective cohort studies and a few randomized controlled trials (RCT) have been conducted. The most extensive review of the existing evidence on diet and cancer has been conducted by the World Cancer Research Funds/American Institute for Cancer Research (WCRF/AICR) (2007) and the continuous update project (CUP/WCRF). Lifestyle recommendations for cancer prevention were drawn up on the basis of nutrition-related factors judged to be convincingly or probably causally related to cancer, according to predefined criteria for judging the strength of the evidence regarding causality. Results from the most recent evaluations on the association between dietary factors and alcohol consumption and major cancers are presented in Table 1. According to the recommendations, a healthy diet for cancer prevention is a diet (1) that allows a person to be as lean as possible without being underweight; (2) is rich in fruit, vegetables, whole grains and pulses; (3) contains low amounts of red meat; (4) does not contain processed meats; (5) limits salt intake; (6) avoid sugary drinks, and limits intake of calorie-rich foods. A healthy diet also limits consumption of alcoholic drinks. In this article we address the mechanisms involving nutritional factors in cancer development, and the challenges in studying these associations, then we present and discuss the most recent evidence on the link between dietary factors and cancer development.

Main Mechanisms Involving Nutritional Factors in Cancer Development Excessive caloric intake results in weight gain and ultimately obesity, which in turn is associated with an increased risk of several cancers (colorectal, endometrium, kidney, esophagus, postmenopausal breast, gallbladder, pancreas, gastric cardia, liver, ovary, thyroid, meningioma, multiple myeloma, and advanced prostate cancers). Many experimental studies have shown that calorie restriction suppresses the carcinogenic process. The interactions between cellular energetics in cancer cells and the systemic metabolic changes associated with obesity are emerging as critical drivers of obesity-related cancer. In obesity, alterations occur in

*

This work was undertaken during the tenure of Senior Visiting Scientist at the International Agency for Research on Cancer, Lyon, France.

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65036-5

501

502

World Research Cancer Funds/American Institute for Cancer Research, summary of evidence on diet and cancer preventiona (updated CUP 2017) Oropharynxnx Osesopahgus Lung Stomach Pancreas Gallbladder Liver Colorectum Breast Premenopausal 2007 2014 2007 2016 2012 2015 2015 2017 2017

Fatb Red meat Processed meatc Salt, salted foods Dairy productsd Fruitse Vegetables (nonstarchy)f Dietary fiber Glycemic load Coffee Soy Diet high in calcium Carotenoidsg Alcoholic beverageh

[ YY

Y

YY

Y

[ [ [

[[ [[

[ [ [

Breast Postmenopausal 2017

[[ [[[

Y

YY Y

Y

Y YY

Y

YY

Y Y [[

[ Y [[ YY

YY YY [[[

YY [[[

YY

[[

[

Ovary Endometrium Prostate Kidney Skin 2014 2013 2014 2015 2007

[[[ [[[

Y Y [[[

[ YY

YYY, convincing decreased risk; YY, probable decreased risk; Y, limited-suggestive evidence decreased risk. [[[, convincing increased risk; [[, probable increased risk; [, limited-suggestive evidence increased risk. AICR/WCRF evidence stratification. Convincing: Evidence strong enough to support causal relationship. Evidence from more than one study type, data from at least two cohort studies, no unexplained heterogeneity between studies with regard to the presence or absence of an association, good quality studies where random or systematic error are unlikely, presence of a dose–response relationship, and strong and plausible experimental evidence. Probable: Evidence from at least two cohort studies or at least five case-control studies, no unexplained heterogeneity between studies with regard to the presence or absence of an association, good quality studies where random or systematic error are unlikely, evidence of biological plausibility. Limited-suggestive: Evidence is too limited to permit a probable convincing judgment but there is evidence of a direction of effect. Evidence from at least two cohort studies or at least five case-control studies, some evidence of biological plausibility, and the direction of the effect is generally consistent. Limited-no conclusion: evidence limited in number of studies or yielding inconsistent results. a For unmarked items there is limited evidence and no conclusion could be drawn. b Include evidence for saturated fat for pancreas. c Includes evidence for stomach (noncardia). d Includes evidence from total dairy, milk, cheese and dietary calcium intakes for colorectum. e Includes evidence on foods containing carotenoids for oropharynx (mouth, pharynx, larynx), foods containing beta-carotene for esophagus; foods containing vitamin C for esophagus and stomach (cardia). f Includes evidence on foods containing carotenoids for oropharynx (mouth, pharynx, larynx) and breast (pre and postmenopausal); foods containing beta-carotene for esophagus; foods containing vitamin C for esophagus; stronger effect on ER
breast tumors. g Includes evidence for foods containing carotenoids for oropharynx (mouth, pharynx, larynx) and breast (stronger effect for ER
tumors). Includes evidence on foods containing beta-carotene for esophagus and lung. Increased risk for lung derived from studies using beta-carotene supplements. h Based on evidence for alcohol intake up to 30 g/day (about two drinks a day). There is insufficient evidence for intake > 30 g/day.

Dietary Factors and Cancer

Table 1

Dietary Factors and Cancer

503

circulating levels of insulin and insulin-like growth factor-1 (IGF-1); adipokines, such as leptin and adiponectin; inflammatory factors; several chemokines; lipid mediators, and vascular-associated factors. Each of these factors has a putative role in the development and progression of cancer as well as several other chronic diseases including cardiovascular disease and type 2 diabetes. In addition, nutrition influences inflammatory and immune responses, and this has led to the development of the “dietary inflammatory index” and the use of the term immune-nutrition. Obesity is strongly linked to chronic inflammation and to immunologic abnormalities with fewer cytotoxic CD8 þ T cells and natural killer cells, while fasting has been associated with antiinflammatory effects. These factors may influence cellular processes and lead to the accumulation of the well-established hallmarks of cancer cells: reprogrammed energy metabolism, sustained proliferative signaling, increased chronic inflammation, increased genome instability, enabled replicative immortality, enhanced angiogenesis, activated processes related to invasion and metastasis, and resistance to growth suppressors, cell death inducers, and immunoregulators. A growing body of evidence indicates that lowering the energy density (the amount of energy in a particular weight of food) of diet can reduce caloric intake. Energy dense diets contain less fiber-rich foods, and are usually high in fats, processed starch, and added sugars. Sugar-sweetened soft drinks play an important role in the development of obesity. Nutrition can also cause oxidative stress, augment a cascade of molecular reactions in cells and alter the metabolism of tissues. Overnutrition may generate free radicals and subsequently elevated oxidative stress and ROS-mediated modulation of various molecular pathways and chronic inflammation. Some nutrients have been linked to an increase in oxidative stress response. Carbohydrates quality can influence the plasma level of glucose and insulin resistance, leading to the formation of free radicals by an imbalance in the ratio of NADH to NAD and by non-enzymatic glycation increased by glucose in cells. Fatty acids appear to affect differentially the cancer risk, in particular through inflammatory pathways. The cyclo’oxygenase-2 (COX-2) enzyme can convert omega-6 FAs into prostaglandin E2, a proinflammatory cytokine, which enables angiogenesis and cell proliferation, whereas prostaglandin E3 is produced from omega-3 FAs with the help of COX-2 which does not facilitate mitogenic angiogenesis. The ratio between omega-6 and omega-3 polyunsaturated fatty acids (n-6 PUFAs and n-3 PUFAs) could therefore play an important role. Fatty acids could also act through other mechanisms. In breast cancer, “de novo” lipogenesis through the action of the stearoylCoA desaturase-1 (SCD-1) has been suggested to increase cancer risk. Trans fatty acids are used in industrial processed sweet and salty foods and have been associated to alterations in metabolic and signaling pathways, higher circulating levels of lipids, systemic inflammation, endothelium dysfunction and possibly increased visceral adiposity, body weight and insulin resistance. Vitamins and minerals such as carotenoids, folate, vitamin C, D, E, and B6 and selenium might reduce cancer through preventive oxidative damage, affecting immune response, inhibiting cell proliferation, inducing cell-cycle arrest, and maintaining DNA methylation. Similarly, phytochemicals such as polyphenols are natural antioxidants modulating intracellular ROS levels resulting into epigenetic alterations of essential genes in tumorigenesis. Flavonoids compounds have been linked to an anticancer activity. Isoflavone possesses a well-characterized anti-estrogenic activity; acts in intracellular steroid metabolism and has antiangiogenic, antiproliferative and proapoptotic activities in tumor cells. Lignans (enterolactone) are strong antioxidant and antiinflammatory substances. Some food compounds can act as direct mutagens and carcinogens. In processed red meat, haem iron is nitrosylated, because curing salt contains nitrate or nitrite leading to the synthesis of nitroso compounds that are mutagenic and potential carcinogens. In red meat, haem iron may act by catalyzing the formation of nitroso compounds and of lipid oxidation end products such as 4hydroxynonenal. Heterocyclic amines and polycyclic aromatic hydrocarbons are formed when cooking food at high temperatures for a long time, or exposed to a direct flame such as a barbecue. In addition, various single-nucleotide polymorphisms involved in the metabolism of these potential carcinogens may modulate the association of carcinogens formed in meat with cancer risk. Sodium chloride is a food preservative used in processed foods. In animal experiments, salt intake facilitates gastric Helicobacter pylori colonization, one of the main predisposing factors for stomach cancer development, and induces mucosal damage. It may also promote or enhance the effect of nitroso-compounds and other carcinogens. Dietary fiber from plant foods can reduce the risk of cancer by several mechanisms. Dietary fiber stimulates bacterial anaerobic fermentation in the large bowel with production of short-chain fatty acids, acetate, propionate, and butyrate, shown to reduce cell proliferation and induce apoptosis. Dietary fiber may also protect against colorectal cancer by reducing the transit time of the intestinal content. Dietary fiber interferes with the enterohepatic circulation of oestrogens and experimental studies in mice have shown that soluble fiber can reduce mammary tumor growth, angiogenesis and metastasis. In addition, fiber-rich foods are important sources of phytoestrogens, estrogen-like plant compounds that may modulate the risk of hormone-dependent cancers, especially breast cancer. Carbohydrates and carbohydrate quality could influence carcinogenesis by affecting insulin resistance and plasma levels of insulin and glucose and level of insulin-like growth factor-1. In addition to increased oxidative stress, elevated circulating insulin level could promote carcinogenesis either directly by stimulating the production of insulin receptors, or indirectly by suppressing insulin growth factor binding protein 1 and 3 affecting the bioavailability of insulin growth factor-1. Folate, originating mainly from green leafy vegetables and fruits, is an important B vitamin. Folate is a critical factor in the network of biochemical reaction referred to as “one-carbon metabolism” together with other B vitamins (B2, B6, B12) that plays critical role in DNA methylation and DNA synthesis, and in turn, facilitates the cross-talk between genetic and epigenetic processes. Thus, the 1-carbon metabolism can impact both genetic and epigenetic pro-carcinogenic processes, and these biologic roles potentially make folate and other related B vitamins important in cancer prevention. However, serum folate has been inconsistently associated with DNA methylation. The effect of diet on epigenetics could be indirect by modulating chronic inflammation, an important factor associated to epigenetic changes and cancer risk.

504

Dietary Factors and Cancer

Despite its central function in maintaining DNA integrity, the role of folate in cancer prevention seems to be affected by folate status and the timing of folate administration. In particular, data from a supplementation study suggest that folate administrated prior to the existence of preneoplastic lesions can prevent tumor development, whereas provision of folate once early lesions are established appears to increase tumorigenesis. In addition, other factors may impact the relationship between folate intake and cancer risk, such as alcohol intake, and polymorphisms in methylenetetrahydrofolate reductase (MTHFR), which codes for a key 1-carbon metabolizing protein. Further, the impact of folate may be related to the tumor type, and other B vitamin namely vitamin B2, B6, B12 that serve critical cofactors for several pivotal enzymes in 1-carbon metabolism. Activation of the vitamin D pathway with calcitriol, the active component of vitamin D, or its analogues reduced tumor development and growth in numerous animal models. Calcitriol acts by binding to the vitamin D receptor (VDR) and by functioning via both genomic and nongenomic pathways to regulate target gene expression. Calcitriol has been shown to have antiproliferative, prodifferentiation and antiinflammatory and immunomodulating activities. Alcohol dinking has been classified as carcinogenic to humans (Group 1) by the International Agency for Cancer Research. Ethanol in alcoholic beverages and acetaldehyde associated with the consumption of alcoholic beverages have also been classified as carcinogenic to humans (Group 1). Overall, there is no consistent difference in cancer risk between different types of alcoholic beverages. While the mechanisms of alcohol carcinogenesis are not fully understood, several mechanisms have been suggested: the direct carcinogenicity of ethanol and its metabolites; the interplay with folate metabolism regulated in part by methylenetetrahydrofolate reductase (MTHFR); impact on endocrine and growth factor pathway leading to an increased levels of androgen and estrogen; impacts on bioactive lipids metabolism including lipoperoxidation and generation of free-radical oxygen species, all of which would be further modulated by use patterns and genetic and environmental factors. The human microbiota is a very diverse ecosystem composed of a large number of microorganisms (close to 100 trillion). Nutrition plays a pivotal role in the regulation of the bacterial community in the gut. Excess intake or imbalance in micro or macronutrients have deleterious effects. Microbiota interacts with the host via direct contact through surface antigens or via soluble molecules which are produced by the microbial metabolism. Most of the data on the link between microbiota and cancer is derived from intestinal tumor models in mice that suggest a causal role for bacterial diversity and community changes. Patients with colorectal cancer and polyposis show dysbiosis, or altered diversity of microbiota supporting the search for tumor-promoting or tumorprotectives species. Bacterial species such as enterogenic Bacteroides fragilis could confer protumorigenic properties. Microbiota can also affect the risk of cancer through its metabolic activity and functions. For example, the metabolism of fiber is a critical one for colorectal cancer. A reduction in butyrate levels has been associated with tumorigenesis in mouse models with oncogenic K-ras activation. Metagenomic analyses have consistently shown a reduction of butyrate producers in colorectal cancer patients. In addition, the host genotype may interact with diet to alter the diversity and function of the microbiota in the gut.

Challenge in Diet–Cancer Research Foods are naturally complex and provide numerous bioactive substances that can act individually or synergically to influence processes such as cell differentiation, apoptosis and hormonal regulation of cellular functions. However, demonstrating relationship between diet and cancer risk in epidemiological studies and intervention, has been challenging because of the potential for bias and random misclassification in both exposure and disease.

Study Design Prospective cohorts are least susceptible to bias, yet require very large sample size, long term follow-up and good ascertainment of cases. Randomized controlled trials (RCTs) of dietary change or supplements provide the highest level of evidence on diet–cancer relationship. However, adherence to the intervention and sufficient change in nutritional intake are needed to detect an effect. In addition, dietary effects are likely to differ based on baseline nutritional status, and volunteers who participate in dietary intervention are likely to be more health-conscious. Dietary or nutritional supplement interventions are usually conducted over a few years which is a brief period relative to the long latency of most cancers. Therefore, depending on the proposed mechanism of action, the timing of the intervention might affect the outcome. Finally, it is becoming more evident that early life exposure and exposure during infancy and adolescence might impact the risk of cancer later in life. Greater growth/stature is a risk factor common to many cancers, including breast, ovary, endometrium, kidney, testicular, colon, and rectal cancer, and malignant melanoma, non-Hodgkin’s lymphoma, and leukemia. Studying the relevant period of exposure becomes an important challenge, and focusing only on exposure in adulthood may obscure an association.

Dietary Assessment Food frequency questionnaire (FFQs) have been mostly used to assess dietary intake in cancer studies. Being self-administered, FFQs are considered easy to use with practical application in cohort studies. Accuracy of dietary assessment is obtained through validation studies using different methods of assessment (e.g., 24-h report or 24-h recall) and assessment of biomarkers in blood and urine samples measuring recovery biomarkers such as urinary nitrogen, potassium and sodium in order to calibrate FFQ data. In addition, concentrations of biomarkers correlated to intake of food or food group are available such as carotenoids, vitamin E or fatty acids and

Dietary Factors and Cancer

505

can be used to validate a FFQ. Direct measurement of blood biomarkers has the advantage of integrating the bioavailability of the nutrients and not relying on reported food intake and food composition tables so improving dietary assessment. Recently, highthroughput technologies developments have provided tools to improve dietary assessment. Metabolomics, by measuring the full profile of small molecules metabolites in biological samples such as saliva, blood and urine, can improve our knowledge of metabolic pathways relevant to diet–cancer relationship. Moreover, because nutritional epidemiological studies rely on self-reported dietary assessment methods subject to recall bias and measurement error, and because objective biomarkers do not exist for all nutrients and foods, metabolomics can be a promising technique for objectively identifying dietary biomarkers and dietary patterns.

Cancer Phenotype The biology of cancer has shown heterogeneity within and across different organ sites and a number of carcinogenic pathways are involved through various genetic mutations; therefore, tumors originating within the same organ site are often etiologically different and are likely to be associated with different risk factors. For example, in the pooling project of prospective studies on diet and cancer, the intake of fruit and vegetables had a consistent protective effect on estrogen receptor negative (ER
) tumors, but not on estrogen receptor positive (ERþ) breast cancer across 20 cohorts. It is therefore important to classify adequately tumor subtypes.

Early Life Exposure Diet in early life can affect markers of adolescent growth and development such as age at menarche, age at peak height growth velocity, and peak height growth velocity. This association has implications for chronic diseases such as breast cancer which has some risk factors stemming from adolescent growth and development, in particular because the breast has not yet completed cell differentiation. Several longitudinal studies have shown that the intake of animal protein in young girls has been associated to earlier age at menarche, while vegetable protein has been associated to a later onset of menarche. Animal protein intake has also been related to faster growth and higher attained height. Higher peak height growth velocity and early age at menarche each increase the risk of breast cancer. Data on the contribution to breast cancer risk of childhood and adolescent dietary exposure is little documented. Soy intake particularly in childhood has been related to lower risk of breast cancer. Data from the ongoing Nurses ‘Health study II (NHS II), a large cohort study of young US nurses 24 to 44 years of age, suggest that meat intake and fatty diet during adolescence are related to an increase of premenopausal breast cancer. A significant linear association was observed with every additional 100 g of red meat consumed per day during high school (20% increase in risk; Relative risk (RR) ¼ 1.10 95% CI ¼ 1.00–1.43). Fat intake in the highest quintile (142 g/day) was associated with a 35% increase in risk when compared to fat intake in the lowest quintile (105 g/day) (RR ¼ 1.35 95% CI ¼ 1.00–1.81). Fiber intake during adolescence was also related to the risk of breast cancer; women in the highest quintiles (27.8 g/day) had 24% less risk of premenopausal breast cancer than women in the first quintile (18.2 g/day) (RR ¼ 0.76 95% CI ¼ 0.58% 1.00%). Higher total fruit intake during adolescent independently of fruit intake in adulthood was also associated to a reduced risk of premenopausal breast cancer. In addition, consumption of fish, fruit and vegetables during adolescence has been associated with a reduced risk of colorectal adenomas, a precursor lesion of colorectal cancer. “Prudent” dietary pattern, characterized by high intake of vegetables, fruits, legumes, fish and poultry, during adolescence was inversely associated with premenopausal breast cancer. Similar results were observed by scoring higher on the Alternative Healthy Eating Index (AHEI) and the association appeared to be stronger for ER
/progesterone receptor-negative (PR
) tumors. High consumption of fish, fruit and vegetables during adolescence was associated with a reduced risk of colorectal adenomas. This association is plausible since dietary fiber can modify the composition of the gut microbiota to metabolize and reduce circulating estrogen. In addition to reducing the risk of early cancers, prevention efforts that begin in early life may also provide important benefits much later in life by shifting the long-term trajectory of risk accumulation. The importance of the timing of exposure during adolescence is further emphasized by the adverse effect of alcohol consumption. Adolescent alcohol consumption is directly related to the risk of premalignant and invasive breast cancer in prospective cohort studies. In NHS II, alcohol consumption between menarche and first full-time pregnancy (FFTP) was associated with an 11% increased risk per 10 g/day (1 drink) in breast cancer (RR ¼ 1.11 95% CI ¼ 1.00–1.23) and a 16% increase in benign breast disease (RR ¼ 1.16 95% CI ¼ 1.02–1.32). In addition, a stronger effect was observed with binge drinking pattern. In the European prospective investigation into cancer and nutrition (EPIC), a large multicenter European cohort, alcohol intake was associated with both pre- and postmenopausal breast cancer. However, breast cancer risk was stronger among women who started drinking prior to FFTP. An increase of 10 g of alcohol/day was related to an 8% (RR ¼ 1.08 95% CI ¼ 1.02–1.14) increased risk of ER
tumors in women who start drinking prior to FFTP, while no association could be detected among women who start drinking after FFTP.

Association of Adult Diet and Cancer in Population Studies Fat Among dietary factors, fat has received the greatest attention due to strong international correlations with rates of several cancers common in developed countries. However, prospective studies have consistently shown little relationship of fat intake with breast

506

Dietary Factors and Cancer

cancer risk. Some epidemiological studies indicate that, rather than total fat intake, subtypes of fatty acids could be more determinant and diversely affect breast cancer risk; although, overall findings to date are conflicting. Epidemiological data on biomarkers of exposure to fatty acids and breast cancer risk are also limited. A meta-analysis of prospective studies has suggested a protective effect of n-3 PUFAs on breast cancer risk. One prospective study showed a significant association between high blood levels of industrial trans-fatty acids (ITFA) and increased risk of breast cancer. However, in general prospective studies have not shown clear associations between patterns of fatty acids and risk of breast cancer, overall and by hormonal receptor status. In two large RCT of low-fat diet, there was no significant effect on risk of breast cancer. Prospective studies of colon cancer have also not supported the positive relationship with fat intake suggested by international comparison of cancer rates. Prospective studies of dietary fat and prostate cancer are fewer but do not generally support the relation with total or specific types of dietary fats. Intake of saturated fatty acids has been related to an increase in pancreatic cancer. In a recent meta-analysis including 5 cohort studies, an 11% increase in pancreatic cancer incidence per 10 g saturated fatty acid per day (RR ¼ 1.11 95% CI ¼ 1.01–1.21) was observed while no consistent results were observed for total fat intake. Studies of lung cancer have found inconsistent evidence of an association with total fat.

Red and Processed Meat Numerous studies have shown an association between high intake of processed meat (such as ham, bacon, sausage and hot dogs), red meat (mainly beef, pork and lamb) and colorectal cancer. Processed meat has been classified as carcinogenic to human by the International Agency for Research on Cancer (IARC) (group 1) while red meat has been classified as probably carcinogenic to human (group 2A). A meta-analysis including 10 cohort studies has shown significant dose–response relationship and concluded to an18% increase (RR ¼ 1.18 95% CI ¼ 1.10–1.28) per 50 g/day of processed meat intake and a 17% increase (RR ¼ 1.17 95% CI ¼ 1.05–1.31) per 100 g/day of red meat. Processed meat cured with nitrite contains high content of nitroso compounds (NOCs), potent carcinogenic agent in animal experiment, and nitrosylated haem iron potential carcinogenic to human. Carcinogens formed during the cooking of meat heterocyclic aromatic amines (HAAs), polycyclic aromatic hydrocarbon, and benzopyrene have provided conflicting results, in part because of the difficulty in assessing dietary carcinogen intake. However, interaction between specific phenotypes of N-acetyltransferase 2 (NAT2) (rapid, intermediate or slow acetylation), and enzyme requires for the bioactivation of HAAs, and the risk of colorectal cancer has been observed. Red and processed meat intake has also been linked to stomach, pancreatic cancer and overall cancer mortality. Although consumption of meat during midlife or later has not been associated with risk of breast cancer, a positive association has been seen with intake in adolescent and early adult life. Studies on red meat and the risk of prostate cancer have been limited and results are inconsistent.

Salt Intake Animal and human studies supported the cocarcinogenic effect of salt on gastric cancer through synergic action with Helicobacter pylori infection, in addition to some independent effects such as increase in the rate of cell proliferation and of endogenous mutations. Salt intake may also promote or enhance the carcinogenic effect of nitroso compounds. The epidemiologic evidence is consistent across studies investigating the intakes of salt, pickle vegetables, dry fish and other salty foods.

Dairy Products Higher consumption of milk and dairy products have been associated with lower risk of colon cancer with a decreased risk of 13% (RR ¼ 0.87; 95% CI ¼ 0.83–0.90) per 400 g/day. Dairy products were not associated with rectal cancer. This protective effect is probably due to the calcium content, lactic-acid producing bacteria and potentially lactoferrin, vitamin D and short chain fatty acid butyrate. In contrast, higher consumption of dairy products has been associated with increased risk of total prostate cancer in many studies (RR ¼ 1.07 95% CI ¼ 1.02–1.12 per 400 g/day). The fat component of milk does not appear to account for these associations.

Fruits, Vegetables, Pulses and Phytochemicals Higher intake of essential nutrients and other biological active constituents of plants have reduced cancer incidence in laboratory animals. However, the potential protective effect of fruit and vegetables against cancer observed in case control epidemiological studies has not been strongly supported in cohort studies and is limited to some cancers. Fruit and vegetable intakes have been inversely associated to cancer of the mouth, pharynx and larynx in a large consortium of case-control studies. Fruit and vegetable intake have also been inversely associated with oesophageal squamous cell carcinoma in a large meta-analysis (31 studies). The evidence for fruit intake and stomach cancer has been confirmed in a meta-analysis of 22 cohort studies showing that individual with lower fruit intake were at higher risk of stomach cancer and an association with citrus fruit was confirmed in a further metaanalysis of prospective cohort studies. A 13% reduction of gastric cancer according to the intake of citrus fruit when comparing highest to lowest intake was observed (RR ¼ 0.87, 95% CI ¼ 0.76–0.99). The effect was stronger for cardia gastric cancer than noncardia gastric cancer. This association was later confirmed in cohort studies based on plasma vitamin C. For vegetables, no association with stomach cancer was observed, but allium vegetables (garlic, onions and shallots) showed a protective effect in a large meta-analysis.

Dietary Factors and Cancer

507

Lung cancer is the most common cancer death and fruits and vegetables containing carotenoids and other antioxidants have been hypothesized to decrease lung cancer risk. Data from large meta-analysis of cohort studies report a protective effect of fruits and vegetables on lung cancer risk. When comparing highest versus lowest intake, a 18% decrease risk was observed for fruit intake (RR ¼ 0.82 95% CI ¼ 0.76–0.89) and 8% decrease for vegetable intake (RR ¼ 0.92 95% CI ¼ 0.87–0.97). The effect was stronger in current smokers. For each increase of 100 g/day an inverse dose response was observed for vegetables (RR ¼ 0.94 95%CI 0.89–0.98 and fruits (RR ¼ 0.92, 95% CI ¼ 0.89–0.95). Vegetable intake has been associated to a small decreased risk of colorectal cancer (2% decrease; RR ¼ 0.98, 95% CI ¼ 0.96–0.99 per100g/day). No effect was observed for rectal cancer. Intake of fruits was not related to the risk of colorectal cancer. For breast cancer, results from the pooling of large cohort studies have shown an inverse association between vegetable intake and ER
breast cancer (18% decrease, RR ¼ 0.82 95% CI ¼ 0.74–0.90 comparing highest vs. lowest vegetable intake). No protective effect was observed in ERþ tumors. Recent data from the EPIC study suggest a protective effect from vegetable fiber on breast cancer risk.

Whole Grain and Cereal Fiber Whole grain is rich in fiber, phytochemicals, B vitamins and other active substances, while refined cereals retain almost only the starchy endosperm when the germ and bran are removed. The majority of studies focused on gastrointestinal cancers. Most of the evidence is related to the association of whole grain intake and colorectal cancer. Whole grain intake has been associated with a 7% decreased risk of colon cancer (RR ¼ 0.83 95% CI ¼ 0.78–0.89) per 90 g/day; no effect was observed for rectal cancer in a meta-analysis (4 studies). Association with other gastrointestinal cancer, genitourinary cancers (prostate and renal cancers), breast and endometrium and neck cancers are inconsistent. Cereal fiber intake has been associated to a decreased risk of colorectal cancer with a decreased risk of 9% (RR ¼ 0.91 95% CI ¼ 0.88–0.94) per 10 g/day intake. This was confirmed by results from the large prospective EPIC study (RR ¼ 0.89 95% CI ¼ 0.82–0.97 per 10 g/day). For other gastrointestinal cancers data are sparse and inconclusive. No effect has been observed on genitourinary cancers. For head and neck cancer there is suggestion of a protective effect. For breast cancer a borderline protective effect has been observed cereal fiber.

Carbohydrate, Glycaemic Index, and Glycaemic Load Cancer with potential links to diabetes leading to an increased risk include bladder cancer, breast cancer, colorectal cancer, endometrial cancer, liver cancer and pancreatic cancer while prostate cancer could possibly be lower among diabetic individuals. Many factors influence how rapidly carbohydrates are digested and absorbed and hence what their glycaemic and insulinemic effects will be. Refined carbohydrate such as pure sugar is rapidly absorbed. This physiologic response of carbohydrates can be quantified by the glycaemic index (GI), which compares the plasma glucose response to specific foods with that induced by the same amount of a standard carbohydrate source, usually white bread or pure glucose. The GI is therefore a measure of carbohydrate quality. However, both the quality and quantity of dietary carbohydrates need to be considered in relation to metabolic effects; the glycaemic load (GL) of a specific food, calculated as the product of GI and the amount of dietary carbohydrates in a food item, has been proposed as a global indicator of the glucose response and insulin demand induced by a serving of food. A recent meta-analysis of prospective studies reported a significant positive association between GI and GL and total diabetes-induced cancers risk; the pooled multivariable-adjusted RR of the overall diabetes related cancer risk when comparing the highest versus the lowest levels of GI and GL were 1.07 (95% CI ¼ 1.04–1.11) for GI and 1.02 (95% CI ¼ 0.96–1.08) for GL. For colorectal cancer a borderline positive association between GI and colorectal cancer risk was observed when comparing the highest with the lowest category of intake (RR ¼ 1.08; 95% CI ¼ 1.00–117). No significant association was observed between GL intake; for endometrial cancer risk no significant association was observed with GI intake while GL intake was significantly associated with a 21% greater risk of developing endometrial cancer compared with the lowest category of intake (RR ¼ 1.21 95% CI ¼ 1.07–1.37); for breast cancer RR comparing the highest with the lowest category were 1.06 (95% CI ¼ 1.01–1.10) for GI and 1.04 (95% CI ¼ 0.97–1.11) for GL. No association with GI or GL were observed for pancreatic and prostate cancers while data were very sparse for liver and bladder cancers (only 1 study for each) and no conclusion could be made.

Polyphenols and Food Containing Polyphenols Polyphenols/flavonoids Polyphenols are a large and diverse family of phytochemicals widely consumed by humans. Food sources for major classes of polyphenol include berries, citrus fruits, leafy vegetables, tea and chocolate, soybean and legumes for flavonoids; tea, berries, wine, fruits, vegetables and coffee for phenolic acid; and sesame seeds, whole grains, vegetables and olive oil for lignans. Flavonoids are the most studied subgroup of polyphenols with regard to cancer risk. In recent epidemiological studies, total flavonoid intake has rarely been associated with a reduction in cancer risk. However, isoflavones, whose main dietary source is soy foods, plausibly reduce the risk of colorectal, breast, and prostate cancers, especially in Asian countries. Soy consumption during childhood and adolescence might be the most relevant period and only studies in Asian populations have evaluated this association. In Western countries, the greatest potential is from the intake of tea catechins, although evidence is limited and still controversial for colorectal, breast, and prostate cancers. Other flavonoids require much more extensive study in relation to major cancers,

508

Dietary Factors and Cancer

especially in prospective studies. In brief, epidemiological evidence for a protective effect of polyphenol intake upon cancer risk remains weak. It should be kept in mind that many in-vitro studies have administered pharmacological doses of polyphenols in nonglycosylated form rather than as metabolites. Lignans are converted to the estrogen-like compounds enterolactone and enterodiol during digestion and could affect breast cancer.

Coffee and caffeine Coffee is among the most commonly consumed beverage worldwide and has been associated to human health effects. Several beneficial compounds are present in coffee including antioxidants polyphenols (phenolic acid). Meta-analysis of coffee consumption have shown reduced risk in several cancers including endometrial, colon, liver, prostate, pancreatic and gallstone when comparing highest versus lowest consumption. However, to date, evidence of a probable protective effect is only observed for liver and endometrial cancers.

Vitamins and Minerals Folate and B vitamins Among epidemiological studies that explore the relation between folate intake and the risk of developing cancer, the evidence has been most compelling for colorectal cancer. In a meta-analysis of cohort studies, the resulting summary risk estimate for high versus low dietary folate intake has shown a marginal decreased risk of 8% when comparing the highest to the lowest intake (RR ¼ 0.92, 95% CI ¼ 0.81–1.05). The protective effect was larger for colon cancer (RR ¼ 0.75, 95% CI ¼ 0.57–0.99). This result was confirmed by a large prospective study with a 30% lower risk of developing colorectal cancer among subjects with the highest versus lowest dietary folate intake (RR ¼ 0.81, 95% CI ¼ 0.52–0.97). The protective effect is more pronounced in moderate to heavy consumers of alcohol, a known inhibitor of folate metabolism. A role of folate in colorectal carcinogenesis is supported by the association between a genetic polymorphism in MTHFR, and enzyme involved in folate metabolism and risk of colorectal cancer. Epidemiological studies on the association between folate intake and breast cancer risk have provided mixed results. Results from a meta-analysis showed no significant association between dietary folate intake and breast cancer risk (pooled RR ¼ 0.95, 95% CI ¼ 0.87–1.03); however higher folate intake may reduce breast cancer risk for those with high alcohol intake. Results from prospective studies also suggest a U-shaped relationship for the dietary folate intake and breast cancer risk. Women with daily dietary folate intake between 153 and 400 mcg showed a significant reduced breast cancer risk compared with those < 153 mcg, but not for those >400 mg. These results were further confirmed in the EPIC study in which, a 14% reduction in breast cancer risk was observed when comparing the highest with the lowest dietary folate tertiles in women having a high (> 12 alcoholic drinks/week) alcohol intake (RR ¼ 0.86, 95% CI ¼ 0.75 to 0.98). There is no clear evidence for a significant association between blood folate levels and breast cancer risk. A protective role for adequate folate status in other cancers is not compelling but there are some observations implying a protective role for adequate folate intake in oropharynx, esophagus, stomach, pancreas lung, cervix neuroblastoma and leukemia.

Vitamin D Vitamin D is mostly synthesized in the skin by ultraviolet B radiation while dietary intake and supplements also contribute to overall vitamin D status. Several cancers have been associated with low sun exposure, in particular colorectal and breast cancers. An inverse association has also been observed between prediagnostic circulating serum 25-hydroxyvitamin D (25(OH)D) levels and risk for colorectal cancer. In a meta-analysis an inverse association of colorectal cancer risk was observed with dietary vitamin D (RR per 100 IU/day ¼ 0.95, 95% CI ¼ I 0.93–0.98; range of intake (midpoints) 39–719 IU/day) and serum/plasma 25-(OH)D (RR ¼ 0.96, 95% CI ¼ 0.94–0.97 per 100 IU/L; range 200–1800 IU/L). However, the association has been less consistent for other cancer types. In a recent large meta-analysis, the pooled RR of breast cancer for the highest (> 500 IU/day, mean) versus lowest categories of vitamin D intake (< 148 IU/day, mean) was 0.95 (95% CI ¼ 0.88–1.01). The pooled RR of breast cancer for the highest (> 31 ng/ml, mean) versus lowest categories of 25(OH)D blood levels (< 18 ng/ml, mean) provided similar results. Several lines of evidence suggest that effects of vitamin D may be stronger for cancer mortality than for incidence. Cancer patients with higher prediagnostic 25(OH)D have lower risks of dying from prostate, colorectal and breast cancer.

Carotenoids Carotenoids and retinol are considered biomarkers of fruits and vegetables intake, and have important antiinflammatory and antioxidant properties. In a meta-analysis of prospective studies of blood concentrations of carotenoids and retinol, and lung cancer risk, blood concentrations of a-carotene, b-carotene, total carotenoids, and retinol were significantly inversely associated with lung cancer risk or mortality. The summary RR was 0.66 (95% CI ¼ 0.55–0.80) per 5 mg/100 mL of a-carotene, 0.84 (95% CI ¼ 0.76–0.94) per 20 mg/100 mL of b-carotene, 0.66 (95% CI ¼ 0.54–0.81) per 100 mg/100 mL of total carotenoids, and 0.81 (95% CI ¼ 0.73–0.90) per 70 mg/100 mL of retinol. In stratified analysis by sex, the significant inverse associations for b-carotene and retinol were observed only in men and not in women. Carotenoids have also been linked to a decreased risk in cancer of the oropharynx. A systematic review and meta-analysis of cohort studies of dietary intake and blood concentrations of carotenoids and breast cancer risk reported stronger associations with blood concentrations than with dietary estimates, probably because the measurement error in the dietary assessment of carotenoid intake from fruits (such as mango, orange, tangerine, melon, papaya) and vegetables (such as carrots, pumpkin, tomato, kale, sweet potato) may have attenuated associations with breast cancer risk. Of

Dietary Factors and Cancer

509

the six dietary carotenoids assessed, only intake of b-carotene was significantly associated with a reduced breast cancer risk (RR ¼ 0.95 95% CI ¼ 0.91–0.99; per 5000 mg/day). In contrast, blood concentrations of total carotenoids, b-carotene, a-carotene, and lutein were inversely associated with breast cancer risk. In a recent report, an inverse association between plasma carotenoids and risk of premalignant breast diseases was found in younger women, consistent with inverse associations reported for invasive breast cancer suggesting that carotenoids may play a role in breast cancer early development. The relative levels of the different types of carotenoids are important to take into account as well as a potential influence from other lifestyle factors.

Alcohol The association between alcohol drinking and the risk of various cancers has been evaluated in a large number of case controls and cohort studies. Alcohol consumption has been convincingly associated to the cancer of the oral cavity, pharynx, larynx, esophagus, colorectal, liver (hepatocellular carcinoma) and female breast cancers. For the majority of these cancers, the association appears mostly linear. Among nonsmokers, reported RRs for oesophageal squamous cell carcinoma and laryngeal cancer range from no effect for light intake (1 drink per day or 10–12 g of ethanol) to a threefold increase (RR ¼ 3.09, 95% CI ¼ 1.75–5.46) for high intakes (4 or more drinks/day). Risks for oropharyngeal cancer are as high as fivefold increase (RR ¼ 5.04, 95% CI ¼ 4.49–6.50) in high intake. The combined exposure to alcohol drinking and tobacco smoking results in a synergistic effect which enhances the risk of these neoplasms up to 14-fold, among heavy-smokers and heavy drinkers (4 or more drinks/day). Summary effect estimates for colorectal cancer are about 11% (RR ¼ 1.11, 95% CI ¼ 0.90–1.38) at one drink per day (corresponding to 10–12 g of ethanol) and 40% (RR ¼ 1.41, 95% CI ¼ 1.16–1.72) at more than four to five drinks per day. Associations of alcohol consumption with hepatocellular carcinoma are generally found at high intakes with significant increased risks of about 30%–40% at consumption levels > 40 g/day. For breast cancer, the risk increases by 5% for premenopausal breast cancer (RR ¼ 1.05, 95% CI ¼ 1.02–1.08) and by 9% for postmenopausal breast cancer (RR ¼ 1.09, 95% CI ¼ 1.07–1.12) per 10 g/day of ethanol. While studies on binge drinking and cancer risk are still sparse and a standardized measure of exposure is lacking, the available evidence shows that the risk is increased between 33% (RR ¼ 1.33, 95% CI ¼ 1.11–1.59, for monthly binge drinking) and 55% (RR ¼ 1.55, 95% CI ¼ 1.07–2.26, for weekly binge drinking). For other cancers sites, kidney, bladder, stomach, ovary, small intestine, Hodgkin and non-Hodgkin lymphoma, ampulla of Vater, pancreas and thyroid, the evidence is considered “not conclusive”.

Vitamin and Mineral Supplementation The role of vitamin and mineral supplements in cancer prevention has been examined in both prospective and RCTs. Trials of beta-carotene and other single supplements, including vitamin E and selenium have not shown benefits. In some studies, adverse effects were observed: an increased risk of lung cancer with beta-carotene in heavy smokers and an increased risk of prostate cancer with high doses of vitamin E. In trials using combinations of multiple vitamins and mineral at lower doses than those in single supplements, a reduction in cancer rates has been seen in a Chinese population with multiple nutrient deficiencies as well as a decrease in overall mortality after 10 years follow up. However, no effect on overall and specific mortality was observed after 20 years follow up. Recent reports from the Physicians’ Health Study (PHS) II, a RCT of daily multivitamins among US physicians, found fewer total cancers in multivitamin recipients, but no effect on overall mortality. Overall supplements with multivitamins and minerals are likely to be beneficial in populations with multiple nutrients deficiency but not in well-nourished individuals. Randomized controlled trials of folic acid conducted among individuals with previously resected colonic adenomas provide conflicting results. Supplements of folic acid (with dose varying from 0.5 to 5 mg/day) did not reduce recurrent adenomas as a whole, while among subjects with lowest plasma folate a marginal reduction was observed. In one trial an increased risk of recurrent adenoma was seen. These results suggest that an adequate amount of folate intake is protective compared to an inadequate intake, but supplementation of those who are already folate-replete conveys no additional protection against colorectal cancer. Total calcium intake is consistently associated with a reduction of colorectal cancer in observational studies. Recent results of large prospective studies (NHS and PHS) report a decreased risk of colon cancer when comparing high versus low intake ( 1400 vs. < 600 mg) (RR ¼ 0.78, 95%CI ¼ 0.65–0.95). In subjects reporting the use of calcium supplementation, a protective effect against colorectal cancer has been observed. However, trials have shown a protective effect of calcium supplementation on colorectal adenomas, a precursor of colorectal cancer, but not against colorectal cancer itself. Several lines of evidence suggest that effects of vitamin D may act on cancer incidence and mortality. A meta-analysis of RCTs conducted over 2–7 years of duration, showed that vitamin D supplementations had little effect on total cancer incidence (400– 1100 IU/day) (RR ¼ 1.00, 95% CI ¼ 0.94–1.06, comparing the intervention vs. the control groups) but significantly reduced total cancer mortality (400–833 IU/day, summary RR ¼ 0.88, 95% CI ¼ 0.78–0.98). In addition recent review and meta-analysis of RCTs suggested that vitamin D supplementation is not associated with a reduced risk of breast cancer.

Dietary Patterns Dietary patterns are defined as the quantities, proportions, variety or combinations of different foods and beverages in diets, and the frequency with which they are habitually consumed.

510

Dietary Factors and Cancer

The dietary pattern has gained considerable attention because nutrients and foods are eaten in combination and may interact; therefore, the evaluation of intake pattern provides a better understanding and is more readily available for dietary recommendations. The impact of dietary pattern on health outcomes including cancer has been reviewed recently to define the dietary guideline for American 2015–2020.

Approaches to derive dietary patterns Several methods have been used to define dietary patterns. The first method is based on an “a priori index” based on a set of dietary recommendations for a healthy dietary pattern. An individual index/score is derived by comparing and quantifying their adherence to the criterion food/nutrient component of the index. Examples of dietary quality scores include: The Healthy Eating Index (HEI)d 2005 and 2010, the Alternate HEI (AHEI) and updated AHEId2010, the Recommended Food Score (RFS), the Dietary Approaches to Stop Hypertension (DASH) score, the Mediterranean Diet Score (MDS), and the Alternate Mediterranean Diet Score (aMed) (for a review see USDA, 2015). The second method of dietary pattern assessment is through data-driven approaches “some posteriori”, such as cluster, factor analyses of other statistical methods to define specific patterns and link these patterns to a health outcome such as cancer. Generally, the dietary patterns investigated are a prudent or healthy dietary pattern (high in vegetables and/or fruits, poultry, fish, low-fat dairy and/or whole grains) as compared to the Western diet or so-called unhealthy diet (red or processed meats, refined grains, sweets and/or high-fat dairy). The cancer sites that have been most investigated include colorectal, breast, lung and prostate. The scientific evidence has been classified as convincing, moderate and limited. For colon/rectal cancer, moderate evidence indicates an inverse association between dietary patterns that are higher in vegetables, fruits, legumes, whole grains, lean meats/seafood, and low-fat dairy, and moderate in alcohol; and low in red and/or processed meats, saturated fat, and sodas/sweets relative to other dietary patterns. Conversely, diets that are higher in red/processed meats, French fries/potatoes, and sources of sugars (i.e., sodas, sweets, and dessert foods) are associated with a greater colon/rectal cancer risk. For breast cancer, moderate evidence indicates that dietary patterns rich in vegetables, fruit, and whole grains, and lower in animal products and refined carbohydrates, are associated with a reduced risk of postmenopausal breast cancer. The data regarding this dietary pattern and premenopausal breast cancer risk point in the same direction, but the evidence is limited due to fewer studies. For lung cancer, limited evidence from a small number of studies suggests a lower risk of lung cancer associated with dietary patterns containing more frequent servings of vegetables, fruits, seafood, grains/cereals, and legumes, and lean versus higher fat meats and lower fat or nonfat dairy products. However, for lung cancer as well as prostate cancer no conclusions can be drawn due to the small number of studies and the wide variation in study design, dietary assessment methodology and prostate cancer outcome ascertainment.

Inflammatory dietary index Chronic inflammation has been linked to alteration of cellular processes and carcinogenesis. Evidence also shows that diet plays a central role in the regulation of chronic inflammation. The dietary inflammatory index DII has been developed to categorized the inflammatory (anti or pro) potential of individual foods and nutrients based on an extensive review of literature. Higher score indicates a more proinflammatory diet while a lower score indicates a more antiinflammatory diet. For example, the Mediterranean diet characterized by high consumption of plant foods, whole grain products, vegetables, fruits, nuts and legumes and regular intake of fish and seafood, low intake of red and processed meat and high-fat dairy products, olive oil and moderate alcohol consumption (preferably red wine taken with meals) has been linked to antiinflammatory proprieties. A recent meta-analysis including 24 studies reports that highest versus lowest DII categories are associated to a 25% overall increase of cancer incidence (RR ¼ 1.25 95% CI ¼ 1.16–1.35) and a 67% higher cancer mortality (RR ¼ 1.67 95% CI ¼ 1.13–2.48). For each of the cancer studies (colorectal, breast and lung) higher DII was related to a high incidence of cancer. The DII provides a useful summary measure of total inflammatory potential of multiple food items and can inform cancer prevention efforts.

Future in the Field Cancers are increasingly seen as metabolic diseases; however, the contribution of metabolic dysfunction to the onset or development of cancer remains poorly understood. Recently developed technologies including metabolomics and deep DNA sequencing have become major tools to explore the metabolism of cancer cells. Hundreds to thousands of metabolites can now be measured simultaneously in bio specimens and give a highly detailed picture of metabolites profiles in tumor tissues and biofluids. The metabolomedthe sum of low molecular-weight metabolites at a given timedis the most fundamental biochemical indicator of biological systems and an indicator of both genetic and environmental conditions. Identification of metabolites and metabolic pathways can improve our understanding on the mechanisms of carcinogenesis and related nutritional factors. In addition, both the modulating effect of diet on epigenetics and the connection between diet and the microbiota are fields of intense research as potential mechanisms of action on cancer development.

See also: Cancer-Related Inflammation in Tumor Progression. Prevention and Control: Nutrition, Obesity, and Metabolism.

Dietary Factors and Cancer

511

Further Reading Arkan, M.C., 2017. The intricate connection between diet, microbiota, and cancer: A jigsaw puzzle. Seminars in Immunology 32, 35–42. Baumann, J., Sevinsky, C., Conklin, D.S., 2013. Lipid biology of breast cancer. Biochimica et Biophysica Acta 1831, 1509–1517. O’flanagan, C.H., Bowers, L.W., Allott, E.H., Hursting, S.D., 2017. Molecular and metabolic mechanisms underlying the obesity-cancer link. In: Romieu, I., Dossus, L., Willett, W.C. (Eds.), Energy balance and obesity. International Agency for Research on Cancer/World Health Organization, Lyon. Guasch-Ferre, M., Bhupathiraju, S.N., Hu, F.B., 2018. Use of Metabolomics in Improving Assessment of Dietary Intake. Clinical Chemistry 64 (1), 82–98. https://doi.org/10.1373/ clinchem.2017.272344. International Agency for Research on Cancer, 2010. Alcohol consumption and ethyl carbamate, vol. 96. International Agency for Research on Cancer/World Health Organization, Lyon. International Agency for Research on Cancer, 2015. Red meat and processed meat, vol. 114. Lyon, International Agency for Research on Cancer/World Health Organization. Mason, J.B., Tang, S.Y., 2017. Folate status and colorectal cancer risk: A 2016 update. Molecular Aspects of Medicine 53, 73–79. Moukayed, M., Grant, W.B., 2013. Molecular link between vitamin D and cancer prevention. Nutrients 5, 3993–4021. Norat, T., Scoccianti, C., Boutron-Ruault, M.C., Anderson, A., Berrino, F., Cecchini, M., Espina, C., Key, T., Leitzmann, M., Powers, H., Wiseman, M., Romieu, I., 2015. European code against cancer 4th edition: Diet and cancer. Cancer Epidemiology 39 (Suppl 1), S56–66. Rothwell, J.A., Knaze, V., Zamora-Ros, R., 2017. Polyphenols: Dietary assessment and role in the prevention of cancers. Current Opinion in Clinical Nutrition and Metabolic Care 20, 512–521. Saha, S.K., Lee, S.B., Won, J., Choi, H.Y., Kim, K., Yang, G.M., Dayem, A.A., Cho, S.G., 2017. Correlation between oxidative stress, nutrition, and cancer initiation. International Journal of Molecular Sciences 18. Sapienza, C., Issa, J.P., 2016. Diet, nutrition, and cancer epigenetics. Annual Review of Nutrition 36, 665–681. Scoccianti, C., Cecchini, M., Anderson, A.S., Berrino, F., Boutron-Ruault, M.C., Espina, C., Key, T.J., Leitzmann, M., Norat, T., Powers, H., Wiseman, M., Romieu, I., 2016. European code against cancer 4th edition: Alcohol drinking and cancer. Cancer Epidemiology 45, 181–188. USDA Department of Agriculture, 2015. Scientific report of the 2015 dietary guidelines advisory committee. Department of Human Services, Washington, DC. Willett, W.C., Key, T., Romieu, I., 2014. Diet, obesity and physical activity. In: Stewart, B.W., Wild, C.P. (Eds.), World Cancer Report. International Agency for Research on Cancer/ World Health Organization, Lyon, p. 2014. Willett, W.C., 2013. Nutritional epidemiology. Oxford University Press, New York. Zitvogel, L., Pietrocola, F., Kroemer, G., 2017. Nutrition, inflammation and cancer. Nature Immunology 18, 843–850. World Research Cancer Funds International/American Institute for Cancer Research (WCRF/AICR) Continuous Update Project (CUP), 2017. http://www.wcrf.org/int/research-wefund/continuous-update-project-cup/cup-expert-panel [Online], [Accessed].

DNA Methylation Changes in Cancer: Cataloguing Tibor A Rauch, Rush University Medical Center, Chicago, IL, United States Gerd P Pfeifer, Center for Epigenetics, Van Andel Research Institute, Grand Rapids, MI, United States © 2019 Elsevier Inc. All rights reserved.

Glossary Bisulfite sequencing Bisulfite treatment of DNA converts cytosine residues to uracil, but 5-methylcytosine residues are unaffected, which reveals the methylation status of the region of interest at the single nucleotide level. CpG island (CGI) Genomic regions with a high frequency of CpG dinucleotides, usually over 200 base pairs in length. DNA methylation Addition of methyl-groups to cytosines, predominantly in the context of CpG dinucleotides. Genome-wide DNA methylation profiling Analysis of the methylation status of CpGs in the regions of interest or in the whole genome. High-throughput (or next-generation) sequencing Millions of DNA strands are sequenced in parallel, eliminating the need for fragment-cloning before nucleotide sequence determination. It provides single-nucleotide resolution. Microarray (DNA chip) A laboratory tool (a microscope slide) that carries thousands of regions of the investigated genome, including promoters, CGIs and any nonrepetitive loci of the genome. Restriction endonucleases Specific enzymes that cleave DNA into fragments at specific recognition sites. DNA cleavages can be affected by the methylation status at the recognition site.

Introduction DNA methylation is an epigenetic process that converts cytosines, chiefly at CpG dinucleotide sequences, into their methylated state. The reaction is carried out by three catalytically active mammalian DNA methyltransferases, DNMT1, DNMT3A and DNMT3B. All of these enzymes are required for survival in mice pointing to a critical role of DNA methylation in development and cellular physiology. 5-methylcytosine, when occurring in gene regulatory sequences is generally associated with gene inactivity and loss of methylation can occur when genes become activated. The first carcinogenesis-associated DNA methylation (i.e., 5-methylcytosine [5mC]) changes were described in the early 1980s when global DNA hypomethylation in tumors was observed. Several years later, the first gene-specific hypermethylation events were detected in tumor samples. Hypermethylation commonly affects CpG island sequences, which in some cases, may be associated with silencing of well-studied tumor suppressor genes such as CDKN2A or MLH1. Since then, a number of differentially methylated regions have been described in the context of all types of cancer. Some types of tumors, even different malignancies originating in the same tissue or organ, can be distinguished by their unique DNA methylation profiles. Tumor classification based on the methylation profile of numerous sequences within the tumor when compared to the corresponding normal tissue has only become possible after sufficiently sensitive genome-scale approaches for methylation analysis had been developed. This field has moved from conducting single gene analysis or scanning of a few selected genes to obtaining overviews of the changes occurring genome-wide (i.e., the “methylome”). To identify altered DNA methylation at the genome scale, a large number of approaches have been developed. The currently employed DNA methylome mapping methodologies can be traced back to three major principles, including DNA methylationsensitive restriction endonucleases, sodium bisulfite modification of genomic DNA and DNA methylation-specific antibodies or proteins (Table 1). In the last 15 years, significant developments have been made in genome-wide DNA methylation profile analysis, whereby the methods are based on coupling the aforementioned three principle-based approaches with various microarray or next-generation sequencing (NGS) platforms. Table 1

Commonly used methods for genome-scale analysis of DNA methylation and DNA hydroxymethylation profiles Genome-wide DNA methylation profiling approaches

DNA methylation sensitive endonuclease-based methods Differential methylation hybridization HELP assay MCAM

Biological affinity-based methods

Sodium bisulfite treatment-based methods

Methylated DNA immuno-precipitation (MeDIP) MDB-Seq MIRA hMeDIP (5hmC) T4-BGT-assisted (5hmC)

BS-sequencing (targeted and whole genome) RRBS Infinium assay TAB-Seq (5hmC) Ox-BS-Seq (5hmC)

The methods discussed in this article are subdivided into restriction enzyme-based, biological affinity-based approaches and sodium bisulfite-dependent techniques.

512

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65057-2

DNA Methylation Changes in Cancer: Cataloguing

513

The recent discovery of 5-hydroxymethylcytosine (5hmC), the first in a series of oxidative products leading to active demethylation of 5mC, in mammalian DNA challenged almost all previously used DNA methylation (i.e., 5mC) mapping methods because they could not distinguish between 5mC and 5hmC bases. Thus, by employing “classical” 5mC-mapping technology, one needs to be careful if the goal is to distinguish between these two bases. Indeed, 5mC and 5hmC have completely different biological functions with 5hmC being often associated with active genes or their control elements and 5mC being linked to repressed or silent chromatin. This can be particularly critical in cases of investigating tissues where high levels of 5hmC are present, for example, in neurons of the brain. Accordingly, new techniques were developed for selective detection of the two closely related cytosine residues. In this article, we will present the most frequently used genome-wide DNA methylation profiling methods that can be effectively employed for cataloguing DNA methylation changes in tumors.

Methylation-Sensitive Restriction Endonuclease-Based Methods This family of methylome mapping techniques relies on restriction endonucleases that have a different sensitivity towards the methylation status of cytosine residues at the cleavage site. Accordingly, the methylation status of cytosine determines the actual restriction endonuclease digestion pattern, which reflects the level of DNA methylation of the investigated genomic region. A number of the originally applied methods, including methylation-specific digital karyotyping (MSDK), methylation-sensitive representational difference analysis (MS-RDA), methylation-sensitive restriction fingerprinting (MSRF) and restriction landmark genomic scanning (RLGS), are not used very widely anymore. However, at least two of these techniques are still being used; the HELP assay (HpaII tiny fragment enrichment by ligation-mediated PCR) and the methylated CpG island amplification coupled to microarray (MCAM) method. Both methods employ adapter ligation and PCR amplification steps after the digestion of chromosomal DNA with methylation-specific enzymes. The PCR amplified and labeled fragment pools are hybridized to microarray platforms or the corresponding amplicons are sequenced by one of the available NGS platforms. Here, we present MCAM, which employs SmaI (methylation sensitive) and XmaI (methylation insensitive) restriction endonuclease pairs (Fig. 1). The unmethylated cleavage sites can be cut with the SmaI enzyme, generating blunt ended fragments. A second digestion with the XmaI enzyme can cut the methylated cleavage sites as well and generate sticky ends for the subsequent adapter ligation. PCR amplification and fluorescent labelling of amplicons is followed by hybridization of labeled fragments onto microarray platforms carrying CpG-rich regions, including promoters and CpG islands. The main drawback of the methylation sensitive enzyme-based profiling methods is the uneven occurrence of SmaI/XmaI cleavage sites in the genomes. Consequently, a significant number of potentially interesting genomic regions, such as promoters and CpG islands, cannot be investigated because of the lack of this particular cleavage site. Numerous enzyme pair combinations have been used to overcome the described limitations, but all of them have reduced genomic coverage, which partly explains why the methylation sensitive enzyme-based approach has been less popular recently. Recent studies have revealed that the DNA base N6-methyladenine (6 mA), previously considered to be a prokaryote-specific epigenetic modification, also occurs in higher organisms including mammals. This base can be mapped using restriction enzymes that are inhibited by 6 mA within the cleavage site, affinity-based approaches using 6 mA-specific antibodies, or by single molecule sequencing that distinguishes between 6 mA and other bases.

Fig. 1 The main steps of methylated CpG island amplification coupled to microarrays (MCAM). A detailed description of the method is provided in the text. Arrows point to SmaI and XmaI cleavage sites.

514

DNA Methylation Changes in Cancer: Cataloguing

Biological Affinity-Based Methods High affinity towards methylated DNA is the principle of these genome-wide DNA methylation-profiling techniques. Methylated DNA fragments can be affinity-purified by using either 5mC-specific antibodies or methyl-CpG-binding domain (MBD) proteins that specifically bind to methylated DNA sequences. Consequently, DNA fragments containing methylated CpG sites are effectively captured by these proteins (i.e., antibodies or MBDs) and can be selectively extracted and enriched. Using these approaches, methylation profiles can be generated with a resolution of approximately 100 base pairs, similar to chromatin immunoprecipitation (ChIP)-sequencing.

Methylated DNA Immunoprecipitation (MeDIP) MeDIP first utilizes a 5mC-specific antibody to capture methylated DNA fragments after nonspecific fragmentation of genomic DNA (i.e., by sonication) (Fig. 2). Next, the immunoprecipitated DNA fraction and the input (nonenriched) genomic DNA can be labeled with different fluorescent dyes (e.g., Cy5 and Cy3) and cohybridized onto microarray platforms. The ratio of the fluorescent intensity of the two dyes reflects the methylation status of the region of interest. The main limitations of the MeDIP method are that single-stranded (denatured) DNA is needed for analysis and the quality of anti-5mC antibodies can be variable. MeDIP is compatible with NGS (MeDIP-seq), which provides much higher coverage than microarray-coupled technologies.

MBD Protein-Based Affinity Pull-Down Methods A relatively small family of proteins including MBD1, MBD2, MBD3, MBD4 and MeCP2, possess a very similar methyl-CpGbinding domain. With the exception of human MBD3, each of these proteins is capable of binding specifically to methylated 5mC containing DNA. An affinity column-based method was initially developed, which employed the methyl-CpG binding domain of MeCP. However, it was recently demonstrated that MeCP2 could also bind to 5hmC with good affinity and therefore may not effectively discriminate between 5mC and 5hmC bases. This indiscriminative nature of MeCP’s MBD can compromise the accuracy of methylation profile mapping. However, the other MBD proteins’ recombinant MBD fragments can be fused to affinity tags (e.g., His- or GST-tag) and be used for capturing the methylated DNA fraction from genomic DNA fragment pools. These fragments can then be analyzed, as done earlier using microarrays or, more recently, by employing NGS.

Fig. 2

Methylated DNA immunoprecipitation sequencing (MeDIP-seq). The detailed description is provided in the text.

DNA Methylation Changes in Cancer: Cataloguing

515

Methylated-CpG Island Recovery Assay (MIRA) MBD2b, the shorter isoform of MBD2, has the highest affinity to methylated DNA and can form heterodimers with other MBD proteins via its C-terminal coiled-coil domain. One of its interacting partners is MBD3L1. MBD3L1 has no DNA binding domain itself, but can act as a hetero-dimerization partner and then bind to methylated DNA. There is no dependence on DNA sequences other than that it requires a minimum of two methylated CpGs in the captured fragment. The MBD2b/MBD3L1 heterodimer complex has a higher affinity to methylated DNA templates than MBD2b alone, a feature that is harnessed in the methylatedCpG island recovery assay (MIRA) (Fig. 3). Bacterially expressed recombinant GST-MBD2b and His-MBD3L1 proteins are mixed and incubated with the fragmented genomic DNA. The heterodimer complex captures methylated DNA fragments, which are then purified, PCR amplified, labeled and analyzed on microarray platforms (MIRA-chip). The MIRA-enriched methylated fraction can be also coupled with NGS platforms (MIRA-seq). MIRA can reliably be performed on a few hundred nanograms of genomic DNA. The MIRA technique was used to profile DNA methylation patterns at a resolution of 100 base pairs in the entire genome of human B lymphocytes, providing one of the first mammalian “methylome” data sets.

Affinity-Based 5hmC Mapping Methodologies Discovery of 5hmC in mammalian genomic DNA demanded new technologies that could distinguish between 5hmC and 5mC bases. For this purpose, two biological affinity-based methods were developed. The hMeDIP approach is based on a 5hmC-specific antibody that has been used for immunoprecipitation of DNA sequences harboring 5hmC residues. This method is similar to the detailed MeDIP technique, except that the employed high affinity antibody is specific for 5hmC in single-stranded DNA (Fig. 2). Another method(s) requires the specific modification of 5hmC using T4 bacteriophage-derived b-glucosyltransferase (T4-BGT), which can transfer a modified glucose moiety to the hydroxyl group of 5hmC. Glucosylated 5hmC residue carrying fragments can be enriched by using a specific protein (e.g., Jbp1) or can be further modified to be suitable for biotin-streptavidin pull-down (Fig. 4). NGS is conducted on the enriched fractions to annotate the genomic locations of 5hmCs. The best resolution that can be achieved by these methods is about 100 bp, which is determined by the size fragmentation and by the employed NGS platform.

Sodium Bisulfite Treatment-Based Methods The previously discussed genome-wide DNA methylation mapping techniques provide valuable information regarding the locations of 5mC and 5hmC residues, but they also have several obvious drawbacks, including the relatively low resolution of mapping,

Fig. 3

Methylated-CpG island recovery assay (MIRA-seq). Please see text for detailed description of this methodology.

516

Fig. 4

DNA Methylation Changes in Cancer: Cataloguing

Mapping of 5-hydroxymethyl cytosine (5hmC) in the genome. Detailed description is provided in the text.

and that the number of the modified nucleotides cannot be identified at the single nucleotide level. Bisulfite treatment-based profiling approaches overcome these issues inasmuch as they provide single base resolution and modified cytosine residues are countable within single molecules. Bisulfite sequencing is based on the premise that unmodified cytosine and 5mC showing different sensitivity towards sodium bisulfite treatment. Sodium bisulfite provokes deamination of cytosine and turns it into uracil, but 5mC is resistant to bisulfite-induced deamination. Thus, bisulfite treatment introduces specific changes in the DNA sequence, which depends on the methylation status of cytosine residues. Whole-genome bisulfite sequencing (WGBS) consists of several main steps: NGS library preparation, high-throughput sequencing of NGS libraries, and data analysis. NGS library preparation is a multistep process that involves the fragmentation of genomic DNA followed by adapter ligation, bisulfite conversion and amplification using adapter-specific PCR primers. The actual nucleotide sequence analysis is conducted by using Roche 454, Illumina, or ABI/ SOLiD platforms. Finally, the collected sequencing reads are aligned to “computationally bisulfite-converted” model genomes and the actual methylation status of all cytosines is revealed. Although WGBS is a powerful method that provides the most complete genome-wide DNA methylation profiling data, the associated high cost of sequencing still limits its widespread application (about $ 2000–4000 per sample). Therefore, a related method, the reduced representation bisulfite sequencing (RRBS) (Fig. 5), was developed, which is a cost-effective alternative to WGBS, although only 85% of CpG islands and 60% of promoters can be investigated. In RRBS, genomic DNA is digested with MspI, a methylation-insensitive endonuclease, which is followed by adapter ligation and fragment size selection. Adapter-ligated fragments in the several hundred bp size range are isolated and subjected to bisulfite conversion. Bisulfite-treated samples are PCR amplified and sequenced. RRBS is frequently employed when high-throughput and low-cost DNA methylation profiling is desired, such as in clinical applications. The method is very sensitive and requires low input allowing the analysis of limited numbers of cells.

Infinium Methylation Bead Arrays The Illumina Infinium methylation assay is a company-developed platform that provides quantitative array-based measurement at the single-CpG-site level (Fig. 6). The assumption behind this experimental design (i.e., focusing on a single CpG in a particular region of interest) is that neighboring CpGs are usually (but not always) at similar methylation states. The monitored CpG sequences are selected from promoter regions, first exons, gene bodies and 3’UTRs. The flagship of this methylation platform has been the Infinium 450 k methylation assay, which could be used for detecting the methylation status of
480,000 CpG of

DNA Methylation Changes in Cancer: Cataloguing

517

Fig. 5 Reduced representation bisulfite sequencing (RRBS). MspI cleavage enriches genomic fragments from CpG islands and other CpG-rich regions. See text for details.

Fig. 6

Infinium 450 k methylation bead-array.The detailed description is provided in the text.

the human genome. An improved version of this platform, the EPIC array, is suitable to methylation profiling of
850,000 CpG sites. The EPIC array now covers 99% of RefSeq gene promoters, 95% of CpG islands, many distant enhancers and miRNA promoters. However, the coverage of the large number of enhancers in mammalian genomes is still limited. Genomic DNA is bisulfite-treated and amplified, employing a special DNA polymerase (Phi29) with very low error rate. Whole genome amplification is carried out and the product of this reaction is enzymatically fragmented and applied to the Infinium chip (bead array). On the chip, there are two bead types for each CpG site per locus. Each bead type is attached to a single stranded 50-mer DNA oligonucleotide that differs in sequence only at the 30 end, which corresponds to the investigated cytosine at CpG sequences. The bisulfite converted and denatured DNA can hybridize to either the methylation-specific probe or the nonmethylation probe, and in a single-base extension reaction, labeled dideoxynucleotides can be incorporated. Through conducting appropriate labelling and

518

DNA Methylation Changes in Cancer: Cataloguing

scanning, the methylation status of the particular CpG site can be determined. This methylation bead array technology is now used very commonly in basic research and clinical settings.

5hmC Mapping at Single Nucleotide Level Soon after the discovery of 5hmC in mammalian genome, it was realized that the conventional bisulfite treatment cannot be used directly for identifying 5hmC, since both bases are resistant to this treatment. Thus, such methodologies have been developed that can distinguish between these two closely related nucleotides. Currently, there are two available methods that can be used for unambiguous identification of 5hmC at single-base resolution.

TET-Assisted Bisulfite Sequencing (TAB-Seq) The TET enzymes catalyze the oxidative demethylation of 5mC in a stepwise manner. The TETs convert 5mC to 5hmC, then 5formylcytosine (5fC), and then 5-carboxylcytosine (5caC). It is thought that 5fC and 5caC are then converted to unmodified C by base excision repair initiated by thymine DNA glycosylase (TDG). In TAB-seq (Fig. 7. left panel), 5hmC is first protected from further oxidation by using T4-BGT enzyme that transfers a glucose moiety to the hydroxyl group of 5hmC. Next, DNA is treated with recombinant TET1 enzyme or its catalytic domain that oxidizes the 5mC to 5caC, while leaving the glucosylationprotected 5hmC intact. Subsequent bisulfite treatment converts unmodified C and 5caC residues to uracils that can be read as T after sequencing. Accordingly, any C-containing reads at CpG sequences in the resulting sequence can be interpreted as a direct readout of 5hmC.

Oxidative Bisulfite Sequencing (OxBS-Seq) In OxBS-Seq (Fig. 7. Right panel), first, 5hmC is oxidized to 5fC by employing a strong oxidizing agent (KRuO4). KRuO4 treatment of DNA leaves 5mC and unmodified C residues intact. Unlike 5mC and 5hmC, 5fC is sensitive to bisulfite-induced deamination; thus, subsequent bisulfite-treatment converts the 5fC to uracil. Next, the KRuO4 and bisulfite treated DNA is subjected to NGS. In OxBS-Seq, T reads in the resulting sequence can be either originating from 5hmC or from unmodified C. Therefore, OxBS-Seq and conventional BS-Seq must be run parallel and, by comparing the two sequences, 5hmC can be identified. The

Fig. 7 TET-assisted bisulfite sequencing (TAB-seq) and oxidative bisulfite sequencing (OxBS-Seq). Left panel depicts TAB-seq, right panel describes the main steps of OxBS-Seq. OxBS-Seq displays 5mC only and requires a comparison to standard BS-seq, which shows the sum of 5mC and 5hmC. A detailed description of these approaches is provided in the text. caC, 5-Carboxylcytosine; CMS, cytosine-5-methylsulfonate; gmC, glucosylated 5hydroxymethylcytosine; hmC, 5-hydroxymethylcytosine; mC, 5-methylcytosine.

DNA Methylation Changes in Cancer: Cataloguing

519

disadvantage of OxBs-Seq that it displays 5hmC only indirectly, but the advantage is that this methodology does not require recombinant TET protein.

Prospective Vision Epigenome-focused studies have already made remarkable progress towards exploring carcinogenesis-involved genes, pathways and mechanisms. It is quite predictable that DNA profiling techniques will receive more and more applicability and credit in clinical diagnostics. Application of high-throughput technologies can contribute to better cancer management in multiple ways: (i) by exploring cancer-specific epigenetic biomarkers, these methods can help in early tumor detection and prognosis, (ii) by monitoring and evaluating the efficacy of therapies, and (iii) by revealing novel “druggable” targets. The method of choice depends on multiple factors, including the number and complexity of the investigated regions, the depth of the required information (i.e., single base resolution is needed, or lower resolution is sufficient). Cost may have a decisive role as well. It is expected that the technological development of high-throughput approaches will continue, which might allow for more cost-effective analysis. The currently employed DNA profiling techniques are multistep and still time-consuming methodologies. New technologies, such as nanopore sequencing, are on the horizon; they allow eliminating PCR amplification and fluorescent labelling from the protocol. Nanopore sequencing has the potential to discriminate between the four canonical DNA bases and their oxidized derivatives. In summary, efficient cataloguing of cancer-associated DNA methylation changes has more than academic significance inasmuch as it can provide novel inroads to diagnosis and choice of treatment options.

See also: Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate. Defective 5-Methylcytosine Oxidation in Tumorigenesis. DNA Methylation Changes in Cancer: Mechanisms. Mutations in DNA Methyltransferases and Demethylases.

Further Reading Booth, M.J., Ost, T.W., Beraldi, D., Bell, N.M., Branco, M.R., Reik, W., et al., 2013. Oxidative bisulfite sequencing of 5-methylcytosine and 5-hydroxymethylcytosine. Nature Protocols 8 (10), 1841–1851. Brinkman, A.B., Simmer, F., Ma, K., Kaan, A., Zhu, J., Stunnenberg, H.G., 2010. Whole-genome DNA methylation profiling using MethylCap-seq. Methods 52 (3), 232–236. Cross, S.H., Charlton, J.A., Nan, X., Bird, A.P., 1994. Purification of CpG islands using a methylated DNA binding column. Nature Genetics 6 (3), 236–244. Dedeurwaerder, S., Defrance, M., Calonne, E., Denis, H., Sotiriou, C., Fuks, F., 2011. Evaluation of the Infinium methylation 450K technology. Epigenomics 3 (6), 771–784. Estecio, M.R., Yan, P.S., Ibrahim, A.E., Tellez, C.S., Shen, L., Huang, T.H., et al., 2007. High-throughput methylation profiling by MCA coupled to CpG island microarray. Genome Research 17 (10), 1529–1536. Hahn, M.A., Szabo, P.E., Pfeifer, G.P., 2014. 5-Hydroxymethylcytosine: A stable or transient DNA modification? Genomics 104, 314–323. Huang, Y., Pastor, W.A., Shen, Y., Tahiliani, M., Liu, D.R., Rao, A., 2010. The behaviour of 5-hydroxymethylcytosine in bisulfite sequencing. PLoS One 5, e8888. Jin, S.G., Kadam, S., Pfeifer, G.P., 2010. Examination of the specificity of DNA methylation profiling techniques towards 5-methylcytosine and 5-hydroxymethylcytosine. Nucleic Acids Research 38, e125. Jung, M., Kadam, S., Xiong, W., Rauch, T.A., Jin, S.G., Pfeifer, G.P., 2015. MIRA-seq for DNA methylation analysis of CpG islands. Epigenomics 7 (5), 695–706. Khulan, B., Thompson, R.F., Ye, K., Fazzari, M.J., Suzuki, M., Stasiek, E., et al., 2006. Comparative isoschizomer profiling of cytosine methylation: The HELP assay. Genome Research 16 (8), 1046–1055. Kriaucionis, S., Heintz, N., 2009. The nuclear DNA base 5-hydroxymethylcytosine is present in Purkinje neurons and the brain. Science 324 (5929), 929–930. Rauch, T., Li, H., Wu, X., Pfeifer, G.P., 2006. MIRA-assisted microarray analysis, a new technology for the determination of DNA methylation patterns, identifies frequent methylation of homeodomain-containing genes in lung cancer cells. Cancer Research 66 (16), 7939–7947. Rauch, T.A., Wu, X., Zhong, X., Riggs, A.D., Pfeifer, G.P., 2009. A human B cell methylome at 100-base pair resolution. Proceedings of the National Academy of Sciences of the United States of America 106 (3), 671–678. Sun, Q., Huang, S., Wang, X., Zhu, Y., Chen, Z., Chen, D., 2015. N6-methyladenine functions as a potential epigenetic mark in eukaryotes. BioEssays: News and Reviews in Molecular, Cellular and Developmental Biology 37 (11), 1155–1162. Tahiliani, M., Koh, K.P., Shen, Y., Pastor, W.A., Bandukwala, H., Brudno, Y., et al., 2009. Conversion of 5-methylcytosine to 5-hydroxymethylcytosine in mammalian DNA by MLL partner TET1. Science 324 (5929), 930–935. Weber, M., Davies, J.J., Wittig, D., Oakeley, E.J., Haase, M., Lam, W.L., et al., 2005. Chromosome-wide and promoter-specific analyses identify sites of differential DNA methylation in normal and transformed human cells. Nature Genetics 37 (8), 853–862. Yu, M., Hon, G.C., Szulwach, K.E., Song, C.X., Jin, P., Ren, B., et al., 2012. Tet-assisted bisulfite sequencing of 5-hydroxymethylcytosine. Nature Protocols 7 (12), 2159–2170.

DNA Methylation Changes in Cancer: Mechanisms Hideyuki Takeshima and Toshikazu Ushijima, National Cancer Center Research Institute, Tokyo, Japan © 2019 Elsevier Inc. All rights reserved.

Glossary CpG island CpG sites are observed less often than expected in the genome. However, there are clusters of CpG sites, and such a cluster is named as a CpG island. If a CpG island is located at a promoter region, its DNA methylation can strongly repress transcription of its downstream gene. CpG island methylator phenotype (CIMP) In some cancers, a large number of CpG islands are simultaneously methylated, and such a phenotype is designated as the CpG island methylator phenotype. CIMP can be associated with specific cancer pathophysiology, such as patient prognosis and drug sensitivity. DNA methylation DNA modification that occurs at the 5th position of the cytosine residue (5-methylcytosine, 5-mC) within a CpG dinucleotide (CpG site). DNMT Enzymes that can catalyze DNA methylation are collectively designated as DNA methyltransferases (DNMTs), and DNMTs can be classified into two classes, de novo type (DNMT3A and DNMT3B) and maintenance type (DNMT1). DNMT3A and DNMT3B are mainly involved in the establishment of DNA methylation status during early embryogenesis. DNMT1 is mainly involved in the maintenance of DNA methylation status at DNA replication. TET An enzyme that can catalyze the conversion of 5-mC to 5-hydroxymethylcytosine (5-hmC), 5-formylcytosine (5-fC), and 5-carbocylcytosine (5-caC). 5-hmC, 5-fC, and 5-caC are removed by base excision repair. Therefore, TET enzymes are critical for the DNA demethylation process.

Introduction DNA methylation is involved in various biological processes, such as embryogenesis, tissue differentiation, genomic imprinting, and inactivation of the X chromosome. At the same time, DNA methylation status can be altered by exposure to various environmental stimuli, and alterations of DNA methylation status are deeply involved in development and progression of human cancers.

Regulation Mechanism of DNA Methylation DNA methylation is present at the 5th position of the cytosine residue (5-methylcytosine, 5-mC) within a CpG dinucleotide (CpG site) (Fig. 1). DNA methylation status is established during early embryogenesis by de novo type DNA methyltransferases (DNMTs), DNMT3A and DNMT3B. Once DNA methylation status is established, it is accurately transmitted upon somatic cell division (DNA replication) by maintenance DNMT (DNMT1) (Fig. 2). At the same time, DNA demethylation can take place at specific biological periods in preimplantation embryos and primordial germ cells. Biochemically, 5-mC can be converted to 5-hydroxymethylcytosine (5-hmC), 5-formylcytosine (5-fC), and finally 5carboxylcytosine (5-caC) by tet methylcytosine dioxygenase (TET). These modified bases are removed by the action of a DNA repair pathway, namely base excision repair (BER), and unmethylated cytosine is incorporated (Fig. 1).

Changes of DNA Methylation in Cancers Regional Hypermethylation DNA methylation status is altered in cancer cells, and the alteration is generally classified into two types; regional hypermethylation and global hypomethylation (Fig. 3). Regional hypermethylation, also called aberrant DNA methylation, represents increased DNA methylation of a CpG island (a cluster of CpG sites) unmethylated in normal cells. Aberrant DNA methylation of a promoter CpG island completely represses transcription of its downstream gene (methylation-silencing) (Fig. 4A), and can be involved in repression of various tumor-suppressor genes, depending upon cancer types. RB gene is repressed in many types of cancers, including retinoblastomas and pituitary adenomas, CDKN2A also in a variety of cancers, BRCA1 gene predominantly in breast and ovarian cancers, CDH1 gene in gastric and breast cancers, MLH1 gene in colorectal, gastric, and lung cancers, and RASSF1A gene in various cancer types, such as breast and lung cancers. In addition to tumor-suppressor genes, aberrant DNA methylation can be present at promoter CpG islands of thousands of genes. Most of such genes are not transcribed in corresponding normal tissues (as described in section “Aberrant DNA Methylation Induction by Environmental Stimuli”). Since these genes are considered not to be functional in normal tissues, their aberrant DNA methylation is not considered to be “driver methylation” but “passenger methylation”.

520

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.65058-4

DNA Methylation Changes in Cancer: Mechanisms

521

Fig. 1 DNA methylation and demethylation. DNA methylation is present at the 5th position of the cytosine residue (5-methylcytosine, 5-mC) within a CpG dinucleotide (CpG site). 5-mC can be converted to 5-hydroxymethylcytosine (5-hmC), 5-formylcytosine (5-fC), and finally 5-carboxylcytosine (5-caC) by tet methylcytosine dioxygenase (TET). These modified bases are finally removed by base excision repair (BER).

(A)

DNA replication

de novo methylation

CG GC

m

CG GC m

m

CG GC

Maintenance methylation

m

m

CG GC m (B)

DNA replication

m

CG GC

m

CG GC

Passive demethylation

m

CG GC m m DNA methylation

Fig. 2 Maintenance of DNA methylation. Once DNA methylation status is established, it can be accurately transmitted, even after cell division by maintenance DNMT (DNMT1) (A). When maintenance methylation is inhibited, passive DNA demethylation takes place (B).

522

DNA Methylation Changes in Cancer: Mechanisms

Normal

Cancer

Regional hypermethylation

Global hypomethylation

Normally unmethylated CpG islands

Repetitive elements and centromere satellite

unmethylated

Normally methylated CpG islands

methylated

CpG island Repetitive sequence/satellite Fig. 3 Alterations of DNA methylation in cancer cells. They are generally classified into two types, namely regional hypermethylation and global hypomethylation. Regional hypermethylation represents increased DNA methylation of a CpG island unmethylated in normal cells. Global hypomethylation represents decreased 5-mC content in genomic DNA, due to mainly demethylation of repetitive sequences highly methylated in normal cells.

(A)

Transcription ON

Unmethylated

Aberrant methylation

Transcription OFF

Methylated

Unmethylated CpG

Methylated CpG

(B)

Treatment with demethylating reagent

Non-annotated TSSs

Fig. 4 Regulation of gene transcription by DNA methylation. (A) Repression of gene transcription by aberrant DNA methylation. Aberrant DNA methylation of a promoter CpG island completely represses transcription of its downstream gene (methylation-silencing), and can be deeply involved in the repression of various tumor-suppressor genes. (B) Generation of aberrant transcripts by DNA hypomethylation. Treatment of cells with a DNA demethylating reagent induced aberrant transcripts from thousands of non-annotated transcription start sites via demethylation of a gene body.

DNA Methylation Changes in Cancer: Mechanisms

523

Global Hypomethylation Global hypomethylation represents decreased 5-mC content in genomic DNA (Fig. 3), mostly due to demethylation of repetitive sequences (retrotransposons, Alu and LINE, and centromeric satellite DNA, a-satellite) highly methylated in normal cells. Global hypomethylation can lead to elevation of mutation rates, thus is involved in chromosomal instability. In addition, hypomethylation of promoter CpG islands of normally methylated genes can lead to their activation, including cancer-testis antigens, such as MAGE-A1 and NY-ESO-1. Activation of oncogenes by hypomethylation has been proposed, but may need more robust analysis. Recently, it has been reported that treatment of cells with a DNA demethylating agent induces aberrant transcripts from thousands of treatment-induced non-annotated transcription start sites (TSSs) via demethylation of the gene body (Fig. 4B). These aberrant transcripts can generate aberrantly truncated proteins, which are potentially involved in the dysregulation of cellular functions and immunogenicity. Global hypomethylation in cancer cells is proposed to lead to induction of such aberrant transcripts.

CpG Island Methylator Phenotype A large number of CpG islands can be simultaneously hypermethylated in a cancer, and such status is referred to as the “CpG island methylator phenotype (CIMP)”. The presence of cancers with the CIMP was first reported in colorectal cancers. Subsequently, cancers with the CIMP have been reported in various types of cancers, such as gastric cancers, glioblastomas, hepatocellular carcinomas, lung cancers, neuroblastomas, and renal cancers. The CIMP is associated with specific cancer pathophysiology, such as patient prognosis and drug sensitivity, depending upon cancer types. In colorectal cancers, the CIMP is associated with old age, right-side location, and female patients. The association between the CIMP and patient prognosis in colorectal cancers is controversial, and associations with both good prognosis and poor prognosis have been reported. In neuroblastomas, the prognosis of patients with the CIMP is much poorer than those without. The predictive power of the CIMP is stronger than that of MYCN amplification, a clinically utilized prognostic marker. In lung cancers and renal cell carcinomas, the prognosis of patients with the CIMP is also poorer. In contrast, in glioblastomas and breast cancers, the prognosis of patients with the CIMP is better than that without. As for the mechanisms of how the CIMP is induced, there had not been known. However, IDH mutations in glioblastomas and SDH (succinate dehydrogenase) mutations in gastrointestinal stromal tumors were recently shown to be causal for the CIMP. In contrast, an association between the CIMP and a BRAF mutation has been known for a long time in colorectal cancers, but its causal role has not been demonstrated.

Expression Changes of DNA Methylation Machineries in Cancers Genes Involved in DNA Methylation All the three DNMTs have been shown to be up-regulated in various types of cancers, such as breast cancers, colorectal cancers, and liver cancers. Therefore, the upregulation of DNMTs is a possible mechanism of aberrant DNA methylation induction. In addition, UHRF1 (Np95), which is required for maintenance of DNA methylation, is also upregulated in lung cancers and bladder cancers, and may be involved in aberrant methylation induction.

Genes Involved in Histone Modification Expression changes of epigenetic machineries other than those directly involved in DNA methylation status have also been reported in various types of cancers. EZH2, a methyltransferase of histone H3 lysine 27 (H3K27), is upregulated in various types of cancers, such as prostate cancers, breast cancers, bladder cancers, gastric cancers, lung cancers, and liver cancers. The upregulation is often associated with aggressive phenotypes, such as the presence of metastasis and poor prognosis. It is known that genes aberrantly methylated in cancer cells are frequently marked with trimethylation of H3K27 (H3K27me3) in their corresponding normal cells (as described in section “Aberrant DNA Methylation Induction by Environmental Stimuli”). Therefore, changes of H3K27me3 status by EZH2 upregulation may be possibly involved in aberrant DNA methylation induction.

Genetic Alterations of DNA Methylation Machineries in Cancers Genes that Methylate DNA Cancer genome analysis has revealed that various genes encoding epigenetic machineries are mutated, depending upon cancer types (Fig. 5). As for the genes that methylate DNA, DNMT3A mutations, mainly located within its C-terminal catalytic domain, especially at Arg882 (such as R882C, R882S, and R882H), have been reported in acute myeloid leukemia (AML). However, it had been unclear whether these DNMT3A mutations are a gain-of-function mutations or not (a loss-of-function mutations). Initially, it was reported that the DNMT3A mutations caused the reduction of its enzymatic (DNA methylation) activity, and were considered as loss-of-function mutations. This reduction of enzymatic activity was shown to lead to hypomethylation of HOXB cluster genes, critical genes for leukemogenesis.

Urinary tract

Stomach

Soft tissue

Skin

Prostate

Pancreas

Ovary

Oesophagus

Lung

Liver

Large intestine

Kidney

Haematopoietic and lymphoid

Cervix

Central nervous system

DNA Methylation Changes in Cancer: Mechanisms

Breast

524

ARID1A ARID1B ARID2 PBRM1 SMARCA2 SMARCA4 SMARCB1 ATRX DNMT1 DNMT3A DNMT3B EZH2 G9A IDH1 IDH2 TET1 TET2 TET3

Mutation frequency

0-1%

1-5%

5-10%

10-20%

>20%

Fig. 5 Genetic alterations of genes encoding epigenetic machineries. Cancer genome analysis has revealed that various genes encoding epigenetic machineries are mutated in cancers. As for the genes involved in addition of DNA methylation, DNMT3A mutations mainly located within its C-terminal catalytic domain have been reported in AML. As for the genes involved in removal of DNA methylation, TET2 mutations have been reported in several hematological cancers. Also, mutations of isocitrate dehydrogenase, IDH1 and IDH2, whose gain-of-function product can inhibit TET activity has been reported in glioblastoma and AML. Therefore, mutations of these genes are considered to be involved in aberrant DNA methylation induction. Mutation frequencies were obtained from COSMIC (Catalogue of Somatic Mutations in Cancer; http://cancer.sanger.ac.uk/cancergenome/ projects/census/).

Afterwards, gain of function by the DNMT3A mutation was also reported. R882H mutant DNMT3A protein, but not its wildtype protein, was shown to interact with polycomb repressive complex 1 (PRC1), which is involved in gene repression by inducing H3K27me3. In cells with the DNMT3A R882H mutation, myeloid differentiation-related genes, such as Cebpa, Cebpe and PU.1, are down-regulated, independently of DNA methylation. As for the other DNMTs, genetic alterations have not been documented.

Genes that Demethylate DNA Mutations of genes encoding isocitrate dehydrogenase (IDH1 and IDH2) have been reported in glioblastoma and AML. IDH mutation is a typical gain-of-function mutation, and confers a novel enzymatic activity to convert a-ketoglutarate (a-KG) to (R)-2hydroxyglutarate (2-HG) in addition to the impairment of its original enzymatic activity (conversion of isocitrate to a-KG). This oncometabolite, 2-HG, produced by mutant IDH protein inhibits the enzymatic activities of TET proteins. The accumulation of 5-mC by reduced TET activities is considered to be one of the molecular mechanism of aberrant DNA methylation induction.

DNA Methylation Changes in Cancer: Mechanisms

525

In addition, TET2 is known to be mutated in several hematological cancers, and the 5-hmC levels are reduced in the bone marrow of patients with TET2 mutations. In solid tumors, 5-hmC levels are also reduced in several cancer types, such as melanoma, liver cancers, and lung cancers. TET2 and TET3 are actually mutated in these solid tumors, especially for TET2 (Fig. 5). Therefore, TET2 and TET3 mutations are considered to be involved in aberrant DNA methylation induction.

Genes Related to Other Epigenetic Mechanisms Epigenetic machineries involved in histone modifications are also mutated in cancers. Among various histone modifications, H3K27me3 in normal cells is known as a premark of aberrant DNA methylation induction (as described in section “Aberrant DNA Methylation Induction by Environmental Stimuli”). In lymphoma, EZH2 is mutated predominantly at tyrosine641 (such as Y641H, Y641N, and Y641S). This mutation enhances histone methyltransferase activity of EZH2, and disturbs cellular H3K27me3 status. In addition, inactivating mutations of UTX (KDM6A), an H3K27 demethylase, have been reported in AML, colorectal cancers, and esophageal squamous cell carcinomas (ESCCs). Activation of an H3K27 methylase and inactivation of an H3K27 demethylase can affect H3K27me3 status, and thus these alterations are potentially involved in aberrant DNA methylation induction.

Genes Related to Chromatin Remodeling A chromatin remodeling factor, such as SWI/SNF, is composed of multiple components, and regulates gene transcription via changing nucleosome occupancy. Genes encoding SWI/SNF components are frequently mutated in various types of cancers, and these mutations are considered to disrupt SWI/SNF functions, which is likely to lead to the alterations of nucleosome occupancy. The presence of nucleosome can protect genomic DNA from DNA methylation catalyzed by de novo type DNMTs, DNMT3A, and DNMT3B. Therefore, there is a possibility that SWI/SNF mutations might be one of the possible mechanisms of aberrant DNA methylation induction.

Aberrant DNA Methylation Induction by Environmental Stimuli Inducers of Aberrant DNA Methylation Aberrant DNA methylation can be detected not only in cancer cells but also in normal cells, even far before cancer development. As for the inducers of aberrant DNA methylation, aging has been known for a long time (Table 1). DNA methylation levels of genes encoding estrogen receptor in human colonic tissues were initially shown to increase, depending upon age. It has also been shown that some infectious agents can induce aberrant DNA methylation in various tissues. In the stomach, DNA methylation levels are up to 300-fold higher in people infected with Helicobacter pylori (H. pylori), an almost exclusive cause of gastric cancers, than those without. Also in the liver, aberrant DNA methylation is induced in people infected with hepatitis C virus (HCV). In addition, some chemicals (smoking) and a hormone (estrogen) are also reported to induce aberrant DNA methylation. Aberrant DNA methylation can be accumulated in normal tissues for a long time. The degree of accumulation of aberrant methylation, namely methylation burden, is closely correlated with risk of cancer development in the stomach, esophagus, and mammary glands. This strongly indicates that the accumulation of aberrant DNA methylation in normal tissues leads to the formation of an epigenetic field for cancerization (epigenetic field defect), an apparently normal tissue but predisposed to carcinogenesis

Table 1

Inducers of aberrant DNA methylation

Inducing factors

Tissues

Aging Bacterial infection Helicobacter pylori Viral infection Hepatitis B virus Hepatitis C virus Epstein-Barr (EB) virus Papillomavirus Parasitic infection Liver fluke Schistosoma Smoking Hormone Estrogen

Colon Stomach Liver Liver Stomach Uterine cervix, Oropharynx Bile duct Bladder Esophagus Mammary epithelium (under the culture condition)

526

DNA Methylation Changes in Cancer: Mechanisms

Gene

Normal tissues

H. pylori infection

1 2 3 4

Driver Passenger

Infiltration of inflammatory cells Gene

Normal tissues exposure to Chronic inflammation

1 2 3 4

Aberrant methylation

Cytokines

Driver Passenger

Nitric oxide

M Clinical tumor formation

UM

Driver Passenger

Fig. 6 Formation of an epigenetic field for cancerization. Aberrant DNA methylation can be detected not only in cancer cells but also in normal cells, even far before cancer development. It affects both driver and passenger genes. The accumulation of aberrant DNA methylation in normal tissues is associated with increased cancer risk, and produces a field for cancerization, a normal tissue predisposed to carcinogenesis.

(Fig. 6). The measurement of the degree of methylation burden is useful for cancer risk prediction, and the usefulness has been demonstrated even clinically for gastric cancer.

Importance of Chronic Inflammation in Aberrant DNA Methylation Induction Chronic inflammation is especially important in aberrant DNA methylation induction. Suppression of chronic inflammation in the stomach of Mongolian gerbils by the administration of cyclosporin A (an immunosuppressant) markedly inhibited methylation induction while it did not affect colonization of H. pylori. This clearly showed that chronic inflammation, but not the H. pylori itself, is important for aberrant DNA methylation induction.

Mechanism of Target Gene Specificity Aberrant DNA methylation is induced at specific genes, depending upon tissues and possibly methylation inducers. In gastric tissues, some genes are methylated at a high incidence (susceptible to methylation induction) in individuals with H. pylori infection while others are not (Fig. 7). As for the mechanism of the target gene specificity, methylation-susceptible genes were shown to have low transcription levels in normal cells. Then, it was also shown that genes aberrantly methylated in cancer cells are pre-marked by H3K27me3 in their corresponding normal cells (Fig. 8A). Further, methylation-resistant genes had high levels of RNA polymerase II (Pol II) binding at their nucleosome free regions (NFRs), even if they are not transcribed (Fig. 8B). Exposure to chronic inflammation can also induce an increase of H3K27me3 levels, which eventually leads to induction of aberrant DNA methylation (Fig. 8C). Tumor-suppressor genes can be silenced by aberrant DNA methylation, despite their high levels of Pol II binding at their NFRs in normal cells.

Perspective: Clinical Applications of Aberrant DNA Methylation Aberrant DNA methylation can be utilized for cancer diagnosis and therapy. Measurement of aberrant DNA methylation specifically present in cancer cells can be utilized for cancer detection. The presence of such methylation can be measured using specimens that can be obtained with minimal invasion (tumor-derived cell-free DNA, cfDNA) or without invasion (urine, sputum, and stool). Measurement of DNA methylation in cancer biopsy specimens can be utilized for the prediction of cancer pathophysiology, such as drug sensitivity and patient prognosis. Measurement of DNA methylation in normal cells can be utilized for risk prediction of cancer development. Aberrant DNA methylation can be removed using DNA demethylating drugs. Now two drugs, azacytidine (5-azacytidine) and decitabine (5-aza-20 -deoxycytidine), are clinically utilized for treatment of myelodysplastic syndrome (MDS) and a part of acute

DNA Methylation Changes in Cancer: Mechanisms

527

Normal gastric tissue (–)

(+)

1 2 3 4 5 6 7 8 9 10 11

1 2 3 4 5 6 7 8 9 10 11

H. pylori Case #

Resistant

1

Intermediate susceptible

21 22

Highly susceptible

Gene number

7 8

47

Fig. 7 Target gene specificity of aberrant DNA methylation induction. Aberrant DNA methylation is induced at specific genes, depending upon tissue types and possibly methylation inducers. For example, in gastric tissues, some genes were consistently methylated in different individuals exposed to H. pylori-triggered chronic inflammation (Green, highly susceptible to methylation induction) while the others were resistant (Red, resistance to methylation induction). Modified from Nakajima et al. International Journal of Cancer 124, 905–910, 2009. (A)

Methylation susceptible genes

RNA polymerase II (–) H3K27me3 (+) (B)

Methylation resistant genes Pol II

Pol II

RNA polymerase II (+) H3K27me3 (–) (C)

Genes whose susceptibility changes by environmental exposure Pol II

Acquisition of H3K27me3 by environmental exposure

H3K27me3 Fig. 8 Mechanisms of target gene specificity of aberrant DNA methylation induction. Genes aberrantly methylated in cancer cells are pre-marked by H3K27me3 in their corresponding normal cells (A). In contrast, methylation-resistant genes show high levels of RNA polymerase II (Pol II) binding at their nucleosome free regions (NFRs), even if they are not transcribed (B). Exposure to specific environmental factors can also induce an increase of H3K27me3 levels at a relatively early stage of the exposure, and this leads to aberrant DNA methylation after prolonged exposure (C).

528

DNA Methylation Changes in Cancer: Mechanisms

myeloid leukemia (AML). Clinical trials for solid tumors are also being conducted, especially by focusing on a combination of a DNA demethylating drug and conventional chemotherapy (or another epigenetic drug). For example, a combination of azacytidine and entinostat (a histone deacetylase inhibitor) was shown to be effective for at least some patients with recurrent metastatic non-small cell lung cancers. Also, the efficacy of a combination of decitabine and carboplatin has been reported for carboplatinresistant ovarian cancers, and a combination of decitabine and panitumumab (an anti-EGFR antibody) for metastatic colorectal cancers.

Epilogue Recent development of technologies for genome-wide DNA methylation analysis revealed DNA methylation profiles in cancer cells that can be potentially used for diagnostic purposes. As inducers of aberrant DNA methylation, genes encoding epigenetic machineries are mutated and have altered expressions in cancers. The role of chronic inflammation is well known, and may be useful as a target of cancer prevention. Further efforts are required to deepen our understanding and to accelerate clinical applications of the findings in aberrant DNA methylation.

See also: Mutations in DNA Methyltransferases and Demethylases. DNA Methylation Changes in Cancer: Cataloguing. Defective 5-Methylcytosine Oxidation in Tumorigenesis. Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate.

Further Reading Abe, M., Ohira, M., Kaneda, A., Yagi, Y., Yamamoto, S., Kitano, Y., Takato, T., Nakagawara, A., Ushijima, T., 2005. CpG island methylator phenotype is a strong determinant of poor prognosis in neuroblastomas. Cancer Research 65, 828–834. Delhommeau, F., Dupont, S., Della Valle, V., James, C., Trannoy, S., Masse, A., Kosmider, O., Le Couedic, J.P., Robert, F., Alberdi, A., Lecluse, Y., Plo, I., Dreyfus, F.J., Marzac, C., Casadevall, N., Lacombe, C., Romana, S.P., Dessen, P., Soulier, J., Viguie, F., Fontenay, M., Vainchenker, W., Bernard, O.A., 2009. Mutation in TET2 in myeloid cancers. The New England Journal of Medicine 360, 2289–2301. Hahn, M.A., Hahn, T., Lee, D.H., Esworthy, R.S., Kim, B.W., Riggs, A.D., Chu, F.F., Pfeifer, G.P., 2008. Methylation of polycomb target genes in intestinal cancer is mediated by inflammation. Cancer Research 68, 10280–10289. Issa, J.P., Ottaviano, Y.L., Celano, P., Hamilton, S.R., Davidson, N.E., Baylin, S.B., 1994. Methylation of the oestrogen receptor CpG island links ageing and neoplasia in human colon. Nature Genetics 7, 536–540. Jones, P.A., Issa, J.P., Baylin, S., 2016. Targeting the cancer epigenome for therapy. Nature Reviews. Genetics 17, 630–641. Juergens, R.A., Wrangle, J., Vendetti, F.P., Murphy, S.C., Zhao, M., Coleman, B., Sebree, R., Rodgers, K., Hooker, C.M., Franco, N., Lee, B., Tsai, S., Delgado, I.E., Rudek, M.A., Belinsky, S.A., Herman, J.G., Baylin, S.B., Brock, M.V., Rudin, C.M., 2011. Combination epigenetic therapy has efficacy in patients with refractory advanced non-small cell lung cancer. Cancer Discovery 1, 598–607. Ley, T.J., Ding, L., Walter, M.J., McLellan, M.D., Lamprecht, T., Larson, D.E., Kandoth, C., Payton, J.E., Baty, J., Welch, J., Harris, C.C., Lichti, C.F., Townsend, R.R., Fulton, R.S., Dooling, D.J., Koboldt, D.C., Schmidt, H., Zhang, Q., Osborne, J.R., Lin, L., O’Laughlin, M., McMichael, J.F., Delehaunty, K.D., McGrath, S.D., Fulton, L.A., Magrini, V.J., Vickery, T.L., Hundal, J., Cook, L.L., Conyers, J.J., Swift, G.W., Reed, J.P., Alldredge, P.A., Wylie, T., Walker, J., Kalicki, J., Watson, M.A., Heath, S., Shannon, W.D., Varghese, N., Nagarajan, R., Westervelt, P., Tomasson, M.H., Link, D.C., Graubert, T.A., DiPersio, J.F., Mardis, E.R., Wilson, R.K., 2010. DNMT3A mutations in acute myeloid leukemia. The New England Journal of Medicine 363, 2424–2433. Maeda, M., Nakajima, T., Oda, I., Shimazu, T., Yamamichi, N., Maekita, T., Asada, K., Yokoi, C., Ando, T., Yoshida, T., Nanjo, S., Fujishiro, M., Gotoda, T., Ichinose, M., Ushijima, T., 2017. High impact of methylation accumulation on metachronous gastric cancer: 5-year follow-up of a multicentre prospective cohort study. Gut 66, 1721–1723. Maekita, T., Nakazawa, K., Mihara, M., Nakajima, T., Yanaoka, K., Iguchi, M., Arii, K., Kaneda, A., Tsukamoto, T., Tatematsu, M., Tamura, G., Saito, D., Sugimura, T., Ichinose, M., Ushijima, T., 2006. High levels of aberrant DNA methylation in Helicobacter pylori-infected gastric mucosae and its possible association with gastric cancer risk. Clinical Cancer Research 12, 989–995. Morin, R.D., Johnson, N.A., Severson, T.M., Mungall, A.J., An, J., Goya, R., Paul, J.E., Boyle, M., Woolcock, B.W., Kuchenbauer, F., Yap, D., Humphries, R.K., Griffith, O.L., Shah, S., Zhu, H., Kimbara, M., Shashkin, P., Charlot, J.F., Tcherpakov, M., Corbett, R., Tam, A., Varhol, R., Smailus, D., Moksa, M., Zhao, Y., Delaney, A., Qian, H., Birol, I., Schein, J., Moore, R., Holt, R., Horsman, D.E., Connors, J.M., Jones, S., Aparicio, S., Hirst, M., Gascoyne, R.D., Marra, M.A., 2010. Somatic mutations altering EZH2 (Tyr641) in follicular and diffuse large B-cell lymphomas of germinal-center origin. Nature Genetics 42, 181–185. Niwa, T., Tsukamoto, T., Toyoda, T., Mori, A., Tanaka, H., Maekita, T., Ichinose, M., Tatematsu, M., Ushijima, T., 2010. Inflammatory processes triggered by Helicobacter pylori infection cause aberrant DNA methylation in gastric epithelial cells. Cancer Research 70, 1430–1440. Plass, C., Pfister, S.M., Lindroth, A.M., Bogatyrova, O., Claus, R., Lichter, P., 2013. Mutations in regulators of the epigenome and their connections to global chromatin patterns in cancer. Nature Reviews. Genetics 14, 765–780. Suzuki, H., Yamamoto, E., Maruyama, R., Niinuma, T., Kai, M., 2014. Biological significance of the CpG island methylator phenotype. Biochemical and Biophysical Research Communications 455, 35–42. Takeshima, H., Ushijima, T., 2010. Methylation destiny: Moira takes account of histones and RNA polymerase II. Epigenetics 5, 89–95. Toyota, M., Ahuja, N., Ohe-Toyota, M., Herman, J.G., Baylin, S.B., Issa, J.P., 1999. CpG island methylator phenotype in colorectal cancer. Proceedings of the National Academy of Sciences of the United States of America 96, 8681–8686. Turcan, S., Rohle, D., Goenka, A., Walsh, L.A., Fang, F., Yilmaz, E., Campos, C., Fabius, A.W., Lu, C., Ward, P.S., Thompson, C.B., Kaufman, A., Guryanova, O., Levine, R., Heguy, A., Viale, A., Morris, L.G., Huse, J.T., Mellinghoff, I.K., Chan, T.A., 2012. IDH1 mutation is sufficient to establish the glioma hypermethylator phenotype. Nature 483, 479–483. Ushijima, T., Hattori, N., 2012. Molecular pathways: Involvement of helicobacter pylori-triggered inflammation in the formation of an epigenetic field defect, and its usefulness as cancer risk and exposure markers. Clinical Cancer Research 18, 923–929.

DNA Methylation Changes in Cancer: Mechanisms

529

Weisenberger, D.J., Siegmund, K.D., Campan, M., Young, J., Long, T.I., Faasse, M.A., Kang, G.H., Widschwendter, M., Weener, D., Buchanan, D., Koh, H., Simms, L., Barker, M., Leggett, B., Levine, J., Kim, M., French, A.J., Thibodeau, S.N., Jass, J., Haile, R., Laird, P.W., 2006. CpG island methylator phenotype underlies sporadic microsatellite instability and is tightly associated with BRAF mutation in colorectal cancer. Nature Genetics 38, 787–793. Wilson, B.G., Roberts, C.W., 2011. SWI/SNF nucleosome remodellers and cancer. Nature Reviews. Cancer 11, 481–492. Xiao, M., Yang, H., Xu, W., Ma, S., Lin, H., Zhu, H., Liu, L., Liu, Y., Yang, C., Xu, Y., Zhao, S., Ye, D., Xiong, Y., Guan, K.L., 2012. Inhibition of alpha-KG-dependent histone and DNA demethylases by fumarate and succinate that are accumulated in mutations of FH and SDH tumor suppressors. Genes & Development 26, 1326–1338.

DNA Mismatch Repair: Mechanisms and Cancer Geneticsq William J Graham V, Christopher D Putnam, and Richard D Kolodner, Ludwig Institute for Cancer Research, San Diego Branch, La Jolla, CA, United States © 2019 Elsevier Inc. All rights reserved.

Glossary Deletion Loss of a segment of DNA from a chromosome, sometimes part of a gene or an entire gene(s). Frameshift mutation Type of mutation in a gene resulting from insertion or deletion of additional base pairs, usually 1 or 2, that changes the reading frame of the gene downstream from the mutation. Heterodimer Protein complex consisting of two different proteins. Heterozygous Having two different alleles of a gene. Homozygous Having two identical alleles of a gene. Immunohistochemistry A method to detect the presence of a specific protein in a tissue sample using protein-specific antibodies. Mispair DNA nucleotides paired in non-Watson–Crick base pairs. For example, G-T rather than the correct G-C or A-T. Mutation A change in the original DNA sequence. Neoantigen A newly formed antigen due to a mutation in a gene that results in the expression of a new protein sequence that is not present in normal tissues. Template strand The strand of DNA used by a DNA polymerase as instructions to build a new DNA strand with the same information as the template.

Introduction Mispaired DNA bases can occur as a result of misincorporation of incorrect nucleotides by DNA polymerases during DNA replication. DNA mismatch repair (MMR) is a conserved mechanism that corrects both base–base mispairs (G-T rather than G-C, for example) and small insertion/deletion loops that are caused by DNA polymerase slippage. The MMR machinery first recognizes the mispair and then excises the DNA strand containing the mispaired base followed by filling in the resulting single stranded gap in the DNA. Critically, MMR identifies the newly synthesized DNA strand and only targets this strand for excision and resynthesis. If repair was targeted to the template strand or if repair failed to occur before the next round of DNA replication, the mispaired base would result in a mutation (Fig. 1). MMR is a major pathway for suppressing DNA replication errors. Defects in MMR result in increased cellular mutation rates, which means that MMR-defective cells accumulate mutations at higher rates than cells with functional MMR. Increased numbers of mutations and increased mutation rates have been observed in many cancers. The “mutator phenotype hypothesis” posits that increased mutation rates drive tumor evolution. These mutations can directly activate proto-oncogenes or inactivate tumor suppressor genes to initiate tumorigenesis as well as increase growth advantages of tumor subclones during cancer progression. Consistent with the predictions of this hypothesis, several cancer predisposition syndromes including Lynch syndrome (also called hereditary non-polyposis colorectal cancer or HPNCC) and biallelic MMR deficiency are caused by inherited mutations in MMR genes. It is also known that somatic MMR defects can underlie the development of sporadic cancers. This article reviews our current understanding of MMR pathways and the proteins involved, while highlighting some major remaining questions in the field. Additionally, the cancer predisposition syndromes caused by inherited MMR defects and the occurrence of MMR defects in sporadic cancers are discussed. It should be noted that MMR also acts on (1) mispaired bases that are the result of chemical damage to DNA and (2) mispared bases that occur in the heteroduplex DNA intermediates formed during recombination, and MMR acts to prevent recombination between divergent but highly homologous DNAs as well as playing roles in somatic hypermutation and meiotic recombination; however, these aspects of MMR are not discussed in this article.

Mechanism of MMR Scope and Comments on Nomenclature Much of the early work on MMR mechanisms was performed using Escherichia coli as a model system. However, the MMR pathways of E. coli and other closely related Gammaproteobacteria have a number of mechanistic differences compared to the MMR pathways

q

Change History: November 2017. William J. Graham V, Christopher D. Putnam, and Richard D. Kolodner updated the text and references to this article. This article is an update of Richard D. Kolodner, and Hernan Flores-Rozas, Mismatch Repair: Biochemistry and Genetics, in Encyclopedia of Cancer (Second Edition), edited by Joseph R. Bertino, Academic Press, 2002, Pages 179–185.

530

Encyclopedia of Cancer, 3rd edition, Volume 1

https://doi.org/10.1016/B978-0-12-801238-3.96130-0

DNA Mismatch Repair: Mechanisms and Cancer Genetics

531

T

G

Replication Error

A T G T

MMR. Another round of replication.

Fig. 1

No MMR. Another round of replication.

Wild-type

A T

G C

Mutant

Wild-type

A T

A T

Wild-type

Model for the formation and repair of mispaired based during DNA replication.

Table 1

Eukaryotic MMR proteins

H. sapiens/M. musculus

S. cerevisiae

E. coli

Function in MMR

Mispair recognition MSH2-MSH3 MSH2-MSH6 Accessory factors MLH1-PMS2

Msh2-Msh3 Msh2-Msh6

MutS

Mispair recognition protein. Msh2-Msh3 is also called MutSb, and Msh2-Msh6 is also called MutSa. See text for more details

Mlh1-Pms1

MutL

MLH1-PMS1 MLH1-MLH3

Mlh1-Mlh2 Mlh1-Mlh3

Mispair activated endonuclease that also recruits downstream proteins. In some bacteria, MutL does not have an endonuclease activity. Also called MutLa Accessory factor for MMR acting through an unknown mechanism. Also called MutLb Can substitute for S. cerevisiae Mlh1-Pms1 in some cases. Acts primarily in resolution of meiotic crossovers. Also called MutLg

a

50 / 30 dsDNAb exonuclease that excises DNA around the mispair

b Clamp

DNA polymerase processivity factor and required for activating the Mlh1-Pms1 (S. cerevisiae)/ MLH1-PMS2 (H. sapiens) endonuclease and for DNA Pol d Loads the b clamp/PCNA onto DNA ssDNAb binding protein that acts in the excision and resynthesis reactions DNA polymerase that resynthesizes excised DNA Another DNA polymerase that can act during MMR Seals any nicks in the DNA

Excision reaction EXO1 Exo1 Resynthesis reaction/accessory proteins PCNA PCNA RFC RPA DNA Pol d DNA Pol ε Ligase I

RFC RPA DNA Pol d DNA Pol ε Ligase I

g complex SSB c c

DNA ligase

a

E. coli uses a different excision mechanism involving a DNA helicase and one of four different ssDNA-specific endonucleases. dsDNA, double-stranded DNA; ssDNA, single-stranded DNA. c E. coli uses DNA polymerase III, which is most analogous to DNA polymerase d because of the requirement of the b clamp/PCNA processivity factor. b

of eukaryotes and other bacteria, even though a number of MMR proteins are highly conserved between different organisms. Therefore, we will only discuss eukaryotic MMR proteins and mechanisms in this article. See Table 1 for a list of the known Saccharomyces cerevisiae and human MMR proteins and their relationship to the known E. coli MMR proteins. The discussion of these proteins will focus on the human proteins and use the names of the human proteins as much as possible. However, MMR proteins and their function are the same in S. cerevisiae and human, and since studies of S. cerevisiae MMR have contributed enormously to our understanding of MMR mechanisms the discussion below will also draw upon insights developed through the study of S. cerevisiae MMR.

532

DNA Mismatch Repair: Mechanisms and Cancer Genetics

MutS-Related Proteins Eukaryotes possess three homologs of bacterial mutS that encode proteins known to function in MMR: MSH2, MSH3, and MSH6. The protein products of these genes form two heterodimeric complexes, MSH2-MSH3 (also called MutSb) and MSH2-MSH6 (also called MutSa), which have slightly different, but partially redundant roles in recognizing mispairs (Fig. 2). Base–base mispairs are primarily recognized by MSH2-MSH6, although MSH2-MSH3 can recognize a small number of these mispairs, and MSH2-MSH3 is solely responsible for recognizing large insertion/deletions. In contrast, both complexes act in the repair of small insertion/deletions, which are the most common type of DNA replication errors. Because of this partial redundancy and the importance of MSH2-MSH6 in base–base mispairs, MSH2-MSH6 appears to play a larger role in the suppression of mutations than MSH2MSH3. Consequently deletion of MSH3 causes only a small increase in mutation rates compared to deletion of MSH6. However, disruption of MSH2 results in completely defective MMR and much higher mutation rates than disruption of either MSH6 or MSH3 because MSH2 encodes a common subunit in both complexes. The MSH2-MSH6 and MSH2-MSH3 mispair binding proteins have two composite ATP binding/hydrolysis sites. Binding of ATP, hydrolysis of ATP, and release of ADP at these sites modulates the interaction between these proteins and DNA as well with downstream MMR factors. The nucleotide-free and ADP-bound forms of MSH2-MSH6 and MSH2-MSH3 can bind to mispairs and induce a dramatic bend in the DNA at the site of the mispair. Remarkably, the structures of these complexes with mispair-containing DNAs has revealed that mispairs are recognized in a specific manner, such as the recognition of the thymidine base in a T:G mispair, which indicates that the orientation of mispair binding by MSH2-MSH6 or MSH2-MSH3 cannot be the source of strand-specificity for the downstream steps of MMR. In contrast, the ATP-bound form of these proteins cannot bind mispaired bases. Moreover, ATP binding by mispair-bound MSH2-MSH6 and MSH2-MSH3 triggers a conformational change whereby the proteins convert to a sliding clamp that releases from the mispair, eliminates the bend in the DNA, and slides along the DNA. ATP binding by MSH2-MSH6 and MSH2-MSH3 is also required for the recruitment of the MutL homolog complexes; however, some mutant versions of Msh2-Msh6 in S. cerevisiae are competent for recruitment but not sliding, suggesting that these two ATP-dependent states are not identical.

MutL-Related Proteins Eukaryotes encode four genes related to bacterial mutL that encode proteins that act in MMR: PMS1 (MLH2 in S. cerevisiae), PMS2 (PMS1 in S. cerevisiae), MLH1 and MLH3. The protein products of these genes form three heterodimeric complexes, MLH1-PMS2 (Mlh1-Pms1 in S. cerevisiae; also called MutLa), MLH1-PMS1 (Mlh1-Mlh2 in S. cerevisiae; also called MutLb) and MLH1-MLH3 (also called MutLg). MLH1-PMS2 is the major MutL-related protein complex that functions in MMR. Mispair bound MSH2-MSH6 and MSH2-MSH3 recruit and load MLH1-PMS2 onto DNA (Figs. 2 and 3). There is evidence that either multiple molecules of MLH1-PMS2 are recruited per molecule of MSH2-MSH6 (or MSH2-MSH3) or that MLH1-PMS2 has a longer residence time on DNA than MSH2-MSH6 or MSH2-MSH3. Moreover, structural and single molecule studies suggest that MLH1-PMS2 may also form a clamp on the DNA. It should be noted that the MSH2-MSH6 sliding clamp is not required for MLH1-PMS2 recruitment, suggesting that sliding clamp formation and MLH1-PMS2 recruitment may represent different aspects of ATP binding-induced conformational changes. MLH1-PMS2 is an endonuclease that makes single strand breaks in DNA, and this activity is essential for MMR. Its endonuclease activity, which can be observed under a number of different reaction conditions, is highly activated in the presence of

PMS1

PMS2

MLH3

MLH1

MLH1

MLH1

RFC

EXO

1

PMS2

MLH1 POLδ MSH2 PCNA RPA MSH6

Base:Base Mispair Complete loss of MMR

Fig. 2

POLε

MSH2 MSH3

Small Insertion:Deletion Mispair Loss of base:base MMR

Large Insertion:Deletion Mispair

Redundant/weak effects

Illustration of the known MMR proteins and their specificity for different mispaired bases.

Essential for viability

DNA Mismatch Repair: Mechanisms and Cancer Genetics

533

(A`)

MSH2-MSH6 binds a mispair MSH2-MSH6

MLH1-PMS2

(B)

PCNA

MLH1-PMS2 recruited by MSH2-MSH6 PCNA recruited or retained from replication by MSH6

Activation of MLH1-PMS2 endonuclease activity by PCNA and MSH2-MSH6.

(C)

Many nicks near the mispair.

Exo1-Dependent (D)

Exo1 recruited to a nick (pre-existing in vitro, MLH1-PMS2-generated in vivo?) and digests error containing strand

Exo1-Independent MLH1-PMS2 makes many nicks around the mispair and ssDNA fragments are lost (or displaced by strand-displacement DNA synthesis?)

EXO1

(E)

polδ or polε resynthesizes digested DNA

DNA ligase seals nicks Fig. 3

Models for the mechanisms of Exo1-dependent and Exo1-independent MMR reactions.

MSH2-MSH6 or MSH2-MSH3, the proliferating clamp nuclear antigen (PCNA), replication factor C (RFC) and ATP on mispaired base containing DNA substrates that also contain a single strand break. Under these reaction conditions, MLH1-PMS2 makes single strand breaks (called nicks) only in the DNA strand that contains the pre-existing nick, although the mechanisms by which nicking is directed to the pre-nicked strand under these conditions are not yet understood. It is still unknown how eukaryotic systems distinguish between the newly synthesized and template strands to target MMR to the newly synthesized strand; however, it has been suggested that because the newly synthesized DNA strands can contain differing levels of nicks and PCNA may remain on the DNA after replication it may mark the newly replicated strand by an unknown mechanism and that PCNA-dependent strandspecific nicking by MLH1-PMS2 might then play a role in initiating strand-specific MMR. The MLH1-MLH3 heterodimer is less important for MMR than MLH1-PMS2. Genetic studies in S. cerevisiae suggest that the MLH1-MLH3 heterodimer plays a small role in MMR that is partially redundant with that of MLH1-PMS2, and studies in mice support this idea (Fig. 2). Like PMS2, MLH3 contains an endonuclease active site; however, biochemical studies have not yet been able to demonstrate that MLH1-MLH3 has a mispair-dependent nicking activity like that of MLH1-PMS2. Other studies

534

DNA Mismatch Repair: Mechanisms and Cancer Genetics

have demonstrated that the major function of MLH1-MLH3 is in the resolution of meiotic crossing over intermediates and that the MLH1-MLH3 endonuclease may cleave Holliday junction recombination intermediates. In contrast, less is known about the role of MLH1-PMS1. PMS1 does not appear to have an endonuclease active site like that of PMS2 or MLH3. In S. cerevisiae, Mlh1-Mlh2 (human MLH1-PMS1) can be recruited to DNA containing mispaired bases by both Msh2-Msh6 and Msh2-Msh3. Deletion of S. cerevisiae MLH2 does not cause a mutator phenotype except under conditions of reduced Pms1 expression (human PMS2). These results suggest that the human MLH1-PMS1 complex functions in MMR as an MLH1-PMS2 accessory protein, though its exact function in this regard is unknown.

Identification of Other MMR Proteins Exonuclease 1 (EXO1) was first implicated in MSH2-dependent MMR because it was identified as an MSH2-interacting protein. EXO1 was later found to interact with MLH1 through a peptide in EXO1 called a MIP (for MLH1 interacting peptide) box. EXO1 is a 50 to 30 dsDNA exonuclease that can also cleave branched DNA molecules. Considerable data support the idea that EXO1 functions in the mispair excision step of MMR under some conditions and that the interaction between EXO1 and MSH2 increases the extent of excision by EXO1 in response to a mispair. In contrast, the role of the interaction between EXO1 and MLH1 is less clear. Genetic studies show that loss of EXO1 only causes weak MMR defects, suggesting that other mechanisms for DNA strand excision during MMR must exist. Other exonucleases proposed to function in MMR are the endo/exonuclease FEN1/RAD27 and the 30 to 50 editing exonuclease functions of DNA polymerases d (DNA Pol d) and epsilon (DNA Pol ε); however, a role for these exonucleases in the excision step of MMR has not yet been definitively established. A combination of in vitro MMR reconstitution studies, genetic studies and studies of the MLH1-PMS2 endonuclease established a role for DNA Pol d, DNA Pol ε, replication protein A (RPA), PCNA and RFC in MMR. Finally, recent studies identified WDHD1 and MCM9 as MSH2-interacting proteins; however, a role for these proteins in MMR has not been definitively established.

Reconstitution of MMR In Vitro A number of insights into MMR mechanisms have been obtained through in vitro reconstitution studies using different combinations of proteins to catalyze MMR reactions. Two types of mispair-dependent excision/repair reactions have been reconstituted with human and S. cerevisiae proteins. In the first type of reaction, a combination of one of the mispair recognition factors MSH2-MSH6 or MSH2MSH3, EXO1, DNA Pol d, the single-stranded DNA binding protein RPA, PCNA and the PCNA loading factor RFC catalyze the repair of a circular mispaired substrate containing a single strand break (called a nick) on the 50 side of the mispair. In this reaction, the mispair recognition factors stimulate excision by Exo1 from the nick past the mispair to produce a gap that is filled in by DNA Pol d, PCNA, and RFC to repair the mispair. DNA Pol ε can substitute for DNA Pol d in this first type of MMR reaction, although PCNA and RFC are no longer required under these conditions. In a second type of reaction, a combination of MSH2-MSH6 or MSH2-MSH3, MLH1-PMS2, EXO1, DNA Pol d, RPA, PCNA and RFC promotes the repair of a circular mispaired substrate containing a nick on the 30 side of the mispair. In this reaction, the MLH1-PMS2 endonuclease is activated in a mispair-dependent fashion by a combination of MSH2-MSH6 or MSH2-MSH3, PCNA and RFC to generate DNA nicks 50 to the mispair. Once 50 -nicks are formed, repair appears to occur as observed in the 50 -nick directed MMR reactions. DNA Pol ε can substitute for DNA Pol d in this second type of MMR reaction, although PCNA and RFC are still required due to their role in activating the MLH1-PMS2 endonuclease. MMR can also occur in the absence of EXO1; however, our knowledge of EXO1-independent MMR mechanisms is not as well developed as it is for EXO1-dependent MMR. Several potential mechanisms have been suggested based on the results of biochemical and genetic studies. Biochemical reconstitution studies have shown that a combination of MSH2-MSH6, MLH1-PMS2, DNA Pol d, RPA, PCNA and RFC can promote the repair of a circular mispaired substrate containing a nick on the 30 side of the mispair. In this reaction, the MLH1-PMS2 endonuclease makes DNA nicks 50 to the mispair and strand displacement synthesis by DNA Pol d from the 50 nick removes the nicked DNA strand from the 50 nick to the 30 nick, thus repairing the mispair. It should be noted that as yet there is no definitive genetic evidence supporting this mechanism. It is also unclear how this mechanism would work on the leading strand of the replication fork as nicks on the leading strand are much less frequent than nicks on the lagging strand, and human DNA Pol ε reportedly cannot support this reaction. A combination of biochemical, genetic and cell biology experiments showed that Exo1-independent MMR requires hyper-activation of the MLH1-PMS2 endonuclease leading to the hypothesis that iterative nicking of the DNA by MLH1-PMS2 may play a role in Exo1-independent MMR. Further study will be required to elucidate the mechanisms of Exo1-independent MMR.

Unanswered Questions About MMR Mechanisms A model for eukaryotic MMR can currently be sketched out (Fig. 3); however, multiple steps remain speculative and are derived primarily from genetic inferences or are implicated primarily by biochemical reconstitution experiments that have not been verified in vivo. But even this model does not answer some important mechanistic questions. In order for MMR to prevent mutations that occur as a result of replication errors it must: (1) detect individual mispairs among a vast excess of normal basepairs even though the mispair binding proteins only have a modest selectivity for binding mispairs relative to base pairs; (2) target MMR to the newly synthesized DNA strand; and (3) complete MMR before the next round of DNA replication. A complete picture of how these steps are accomplished is not yet available.

DNA Mismatch Repair: Mechanisms and Cancer Genetics

535

Mispaired bases occur at very low frequencies and must be recognized and repaired in the presence of a vast excess of correct base pairs. However, MSH2-MSH6 and MSH2-MSH3 have only a 10–20 fold higher affinity for mispairs relative to their affinity for base pairs, suggesting there may be some mechanism to amplify the signal of the mispair. The conversion of MSH2-MSH6 and MSH2MSH3 to sliding clamps allows additional molecules of these proteins to bind the mispair and then track along the DNA, which could be part of the mechanism by which the mispair signal is amplified to trigger repair in vivo. Another possible aspect of signal amplification may have been revealed by the formation of cytological S. cerevisiae Mlh1-Pms1 (human MLH1-PMS2) foci containing multiple molecules of S. cerevisiae Mlh1-Pms1 (human MLH1-PMS2) as a MMR intermediate. If the recruitment of MLH1-PMS2 protein involves a cycle of ATP hydrolysis by MSH2-MSH6 (or MSH2-MSH3), mispair re-recognition, and ATP binding, then the small energetic difference between mispairs and fully base paired DNA will be amplified. The mispair binding proteins, MSH2-MSH6 and MSH2-MSH3, are known to colocalize with the DNA replication machinery where they function as mispair detectors for MMR. This colocalization is mediated by the interaction between MSH2-MSH6 (or MSH2-MSH3) and PCNA and is required for MMR under some conditions. Consistent with this, MMR is also temporally coupled to DNA replication during S-phase. Other mechanisms for coupling MMR to DNA replication likely exist but have not yet been definitively established; one possible mechanism is the reported interaction between MSH2-MSH6 and an S-phase specific chromatin modification through the deterostome-specific N-terminal PWWP domain of MSH6, although further studies of this possible mechanism are needed. S-phase coupling of MMR to DNA replication proteins during the S-phase in which the mispair occurred provides a potential mechanism for enhanced mispair recognition by MMR proteins and would be expected to ensure that mispairs are recognized and repaired before the next round of DNA replication. It is less clear how MMR is targeted to newly replicated DNA strands. The unusual property of the MSH2-MSH6 (or MSH2MSH3) and PCNA-dependent MLH1-PMS2 endonuclease that targets its endonuclease activity to nicked DNA strands is intriguing in this regard, as the newly replicated DNA strands, particularly the lagging DNA strand, would be expected to be marked by both nicks and PCNA. Key unresolved questions are: (1) what are the biochemical and structural mechanisms that phase MLH1-PMS2 to target it to nicked DNA strands; and (2) why would additional nicking actually be needed for MMR given that pre-existing nicks, particularly on the lagging strand, should be substrates for EXO1-mediated excision of mispairs. This potential strand recognition mechanism seems less applicable to MMR on the leading strand given the much lower density of nicks and PCNA that are expected on the leading strand. It has also been suggested that nicking at ribonucleotides that are misincorporated during DNA replication could provide a strand specificity signal, although this seems unlikely as defects that eliminate RNAseH2, the enzyme that would make these nicks, do not cause an MMR defect. Further investigation of the mechanisms of strand specificity during MMR are clearly needed.

MMR Defects and Cancer Lynch Syndrome A large number of hereditary cancer syndromes have been identified. Among the earliest of these to be described is Lynch syndrome (also called hereditary non-polyposis colorectal cancer, or HPNCC), which was first documented in the literature by Aldred Scott Warthin in 1913. Lynch syndrome is an autosomal dominant disorder characterized by inheritance of susceptibility to develop early onset cancers relative to sporadic cancers of the same type. Colorectal and endometrial cancers are the most common type of cancer associated with Lynch syndrome, other cancers are also associated with Lynch syndrome including ovarian, stomach, urological tract, small bowel, and hepatobiliary system and pancreatic, breast, prostate, brain (usually glioblastoma) and adrenocortical cancers have also been reported. At least two variants of Lynch syndrome have been reported. Muir–Torre syndrome is a variant of Lynch syndrome that is characterized by the presence of sebaceous neoplasms and/or keratoacanthomas in conjunction with the other cancers associated with Lynch syndrome. Turcot syndrome is characterized by an association between colorectal cancer and brain tumors and is due to either Lynch syndrome or familial adenomatous polyposis. Linkage studies originally mapped Lynch syndrome to two loci, one on chromosome 2p and a second on chromosome 3p. The tumors from the patients analyzed in these studies showed an unusual phenotypedstriking instability of mononucleotide and dinucleotide repeat sequences due to the accumulation of frameshift mutations in these sequences that are ubiquitous in the human genome, often referred to as microsatellite instability high or MSI-H. This phenotype was known to be caused by MMR deficiency in E. coli and S. cerevisiae and led to the initial demonstration that Lynch syndrome is caused by inherited heterozygous mutations in the MMR genes MSH2 and MLH1. While MSH2 and MLH1 mutations are the most common cause of Lynch syndrome, candidate gene studies demonstrated that mutations in PMS2 and MSH6 are also present in Lynch syndrome, but at lower frequencies. Carriers of heterozygous PMS2 mutations tend to have a milder phenotype relative to MSH2 or MLH1 carriers, including a lower frequency and later onset of cancer whereas carriers of heterozygous MSH6 mutations often have a predominance of endometrial cancer. Convincing evidence for mutations in other MMR genes in Lynch syndrome is lacking. MSH2 or MLH1 are the only MMR genes in which mutations completely eliminate MMR, providing an explanation for the high prevalence of MSH2 or MLH1 defects in Lynch syndrome. In contrast, MSH6 is partially redundant with MSH3 and PMS2 is partially redundant with MLH3, and hence defects in MSH6 and PMS2 cause partial MMR defects providing a possible explanation for their lower prevalence in Lynch syndrome. Lynch syndrome was originally diagnosed using clinical criteria but is now diagnosed using both clinical criteria and molecular diagnostics, with the results from molecular diagnostics studies leading to a broadening of the clinical definition of Lynch

536

DNA Mismatch Repair: Mechanisms and Cancer Genetics

syndrome. There are presently three clinical criteria for Lynch syndrome, the Amsterdam criteria, the Amsterdam II criteria, and the Bethesda criteria (Table 2). The Amsterdam criteria primarily focused on a family history of colorectal cancer along with age of diagnosis. The Amsterdam II criteria, sometimes called the “Modified Amsterdam Criteria” included Lynch syndrome-associated cancers and tumor pathology in addition to colorectal cancer in the criteria for diagnosis of Lynch syndrome. Finally, the Bethesda criteria allowed for inclusion of smaller pedigrees, different Lynch syndrome-associated cancers, early age of diagnosis, multiple tumors and the presence of microsatellite instability in colorectal cancer resulting in a broader clinical definition of Lynch syndrome. The broader the definition of Lynch syndrome used, the more likely patients with MMR defects will be identified, but larger numbers of patients without MMR defects will be included in the cases identified. Modern molecular and genetic testing methods have augmented and to some extent replaced clinical criteria for diagnosing Lynch syndrome. Testing of tumors from individuals suspected to have Lynch syndrome for MSI-H using accepted microsatellite markers and testing for loss of expression of MSH2, MSH6, MLH1 or PMS2 using immunohistochemistry (IHC) can effectively identify patents with candidate MMR defects; these approaches are often used to initially screen patients for potential MMR defects. It should be noted that MSH6 defects can result in a less striking MSI-H phenotype with greater mononucleotide repeat instability than dinucleotide repeat instability compared to that caused by MSH2, MLH1 or PMS2 defects due to the redundancy between MSH6 and MSH3. If abnormalities, such as loss of expression of an MMR gene or MSI-H, are observed or no data from tumor analysis is available for a suspected Lynch syndrome patient, then genetic testing for germline alterations focusing on the relevant genes is recommended. Genetic testing for MSH2, MSH6, MLH1 or PMS2 defects involves complete gene sequencing and analysis for the presence of deletion mutations and genome rearrangements affecting these genes and can also involve testing for EPCAM gene deletions that result in silencing of MSH2 expression and MLH1 epimutations that result in silencing of MLH1 expression. Identification of a defect in an MMR gene constitutes a positive diagnosis of Lynch syndrome.

Biallelic Mismatch Repair Deficiency Syndrome (BMMR-D) BMMR-D, also called constitutional MMR deficiency (CMMR-D), is a rare cancer susceptibility syndrome characterized by tumors of the gastrointestinal tract, skin lesions, brain tumors, leukemias and lymphomas, and rarely endometrial cancers usually occurring within the first two decades of life. BMMR-D is often connected to consanguinity, with over 50% of known cases coming from consanguineous families. Additionally, patients do not always have a family history of Lynch syndrome-like cancers. BMMR-D Table 2

Clinical criteria for diagnosing Lynch syndrome

Amsterdam I criteria* A positive diagnosis of Lynch syndrome requires at least three relatives with colorectal cancer (CRC) and the presence of the following five criteria: 1.One relative must be a first degree relative of the other two 2.CRC must be present in two successive generations 3.One case of CRC must have been diagnosed when the patient was 95% in such cases.

Retinal Haemangioblastoma Definition: Retinal hemangioblastoma is a tumor derived from developmentally arrested angioblasts, generally due to a mutation in the VHL gene. Burden: Von Hippel–Lindau (VHL) disease occurs in 1/36,000 live births per year. VHL has a remarkable penetrance of over 90% by 65 years of age. Of all VHL patients, 49%–85% develop retinal hemangioblastomas, making this ocular sign the most common presentation of the disease. VHL-associated retinal hemangioblastomas usually develop at a mean age of 25 years. Risk factors: VHL which is an autosomal dominant inherited syndrome. It is caused by a germline alteration of the VHL gene, a tumor suppressor gene located on chromosome 3p25.5. Retinal hemangioblastomas are often the most common and earliest presentation of VHL disease. Pathology: Hemangioblastomas are composed of capillary sized vascular channels surrounding stromal cells. The stromal cells characteristically have a prominently vacuolated cytoplasm. In addition, small clusters of angiomesenchymal cells, tumourlet cells, have been described in ocular and cerebellar hemangioblastomas. The stromal cells and the “tumourlet” cells may express CD133, a marker present on hematopoietic, endothelial, and neural progenitor cells. Stromal cells are the cells that harbor the genetic alterations in the VHL gene. These alterations are not present in the vascular or glial components. Molecular pathology and genetics: In VHL disease there is a germline defect in one copy of the gene and with random loss of the normal allele (loss of heterozygosity, LOH) patients are predisposed to develop hemangioblastomas. Presumably, in sporadic cases there is somatic loss of both normal copies of the VHL gene, resulting in a similar but solitary tumor. Microenvironment including immune response: Surrounding the vascular channels are stromal cells with vacuolation. There are limited inflammatory infiltrates within this stroma. Staging and grading: Not applicable. Prognostic and predictive biomarkers: The type of germline mutation present in the VHL gene has an effect on the disease phenotype. Deletions and truncating mutations are associated with a Type I phenotype, characterized by renal cell carcinoma and haemangioblastoma but not pheochromocytomas. Type II phenotype with renal cell carcinoma and pheochromocytoma is associated with missense mutations. Patients with complete deletions of the VHL gene are also much less likely to develop retinal hemangioblastomas (9%), than patients with partial deletions, simple missense, and nonsense mutations (45%). The prognosis for vision is good in patients with VHL. Overall, 88% of patients will retain 20/20 or better vision in at least one eye. Laser or cryotherapy is effective treatment for small to medium sized intraocular hemangioblastomas that do not involve the optic nerve. Increasingly, surgical resection of unresponsive and large retinal hemangioblastomas is being employed with good visual outcomes. Treatment of hemangioblastomas involving the optic nerve remains challenging. The prognosis related to systemic morbidity and overall survival in VHL is dependent on the treatment of the associated cerebellar haemangioblastoma and renal cell carcinoma.

Vitreoretinal Lymphoma Definition: A high-grade malignant lymphoproliferative disease in which atypical lymphocytes infiltrate the retina and vitreous in the absence of lymph node or visceral involvement. It is frequently accompanied by involvement of the central nervous system. Burden: The incidence of primary vitreoretinal lymphoma (PVRL) is difficult to estimate because of its rarity and the effect of improvement in diagnostic techniques employed. The incidence is reported to have steadily increased from 0.23–0.48 per million population from 1990 to 2010. A slight predominance of female patients is reported. Risk factors: Based on overlap with primary central nervous system lymphoma (PCNSL), HIV and Epstein-Barr virus (EBV) infections are predisposing factors for the development of PVRL. No other environmental risk factors are known. Pathology: Most PVRL are diffuse large B cell lymphomas (DLBCL) of ABC type, according to the WHO lymphoma classification. The atypical lymphoid cells are generally medium-to-large cells with a condensed chromatin pattern and prominent nucleoli (Fig. 9). The nucleus often is irregular in shape. The neoplastic lymphocytes are usually admixed with reactive inflammatory cells, such as T-cells and macrophages. In some cases, the presence of small, mature T-lymphocytes overwhelms and can mask the neoplastic B-cells. PVRL cells are typically friable with numerous apoptotic bodies, thus vitreous samples often contain abundant lytic cell remnants or ghosts, often within macrophages. Such cell ghosts are uncommon in other diseases associated with vitreous infiltrates, so their presence is highly suggestive of PVRL. The neoplastic B-cells of PVRL have the following profile: CD79aþ, CD20 þ, PAX5þ, BCL-2 þ, BCL-6 þ, MUM1 þ, IgM þ, and a high cell proliferation. Molecular pathology and genetics: Monoclonal expansion of B lymphocytes can be assessed using immunoglobulin heavy chain (IgH) polymerase chain reaction (PCR). IgH-PCR can be performed on fluid and tissue samples (fresh-frozen or fixed), and is

Eye and Orbit Cancer: Pathology and Genetics

Fig. 9

57

Vitreoretinal lymphoma. Scattered lymphoid blasts in a cytoblock of a vitrectomy sample (H&E 20
).

very useful in PVRL diagnosis as an adjunctive test to cytomorphology and immunophenotyping. Recently, evaluation for the MYD88 mutation has confirmed its presence in 70% of cases of PVRL. Microenvironment including immune response: Typically PVRL contain admixed macrophages and T-cells, which can make the diagnosis difficult, particularly as the neoplastic B-cells are quite fragile. Staging: PVRL can be staged using the Lugano Classification (Modified Ann Arbor Classification) for Hodgkin and Non- Hodgkin Lymphoma. Prognostic and predictive biomarkers: The prognosis of PVRL patients is generally poor: it is dependent on the presence and/or subsequent development of CNSL. In patients diagnosed with PVRL, CNS involvement is present in 15% of patients at first diagnosis. Secondary CNSL has been reported to develop in a high percentage of PVRL patients during follow-up. To date, there are no predictive factors for the development of CNS involvement. Steady improvement in the survival of PNCSL patients has been observed with varying chemotherapeutic regimens. At present, the median overall survival is close to 5 years.

Tumors of the Lacrimal Gland and Orbit Adenoid Cystic Carcinoma (ACC) Definition: A rare aggressive epithelial malignant neoplasm of the lacrimal gland and other glandular tissues, most commonly the salivary glands, characterized by malignancy of modified myoepithelial and ductal (luminal) cells. Burden: ACC represents around 20% of epithelial lacrimal gland tumors and is the commonest malignant tumor accounting for approximately half or even more of malignant epithelial lacrimal gland lesions. ACC is generally known to have a wide age range (12–75 years) and a bimodal age distribution with a peak in adulthood around the age of 40 and a smaller peak in teenage years. Risk factors: Not known. Pathology: ACC is a non-encapsulated tumor, which infiltrates the surrounding orbital soft tissue and includes proliferating modified myoepithelial cells as well as luminal ductal differentiated cells. Depending on the cellular composition and cytomorphological features, the tumor shows three growth patterns seen individually or together in variable combinations (Fig. 10). The cribriform pattern is the commonest, described as “Swiss cheese” because of the formation of cystic structures where the connective tissue stroma is filled with basophilic amorphous glycosaminoglycans and/or eosinophilic hyalinized basal lamina.

Fig. 10 Adenoid cystic carcinoma. Cribriform (“Swiss cheese”) growth pattern of myoepithelial cells as well as luminal ductal cells. The stroma is filled with basophilic amorphous material. (H&E 20
).

58

Eye and Orbit Cancer: Pathology and Genetics

The tubular pattern shows small glandular structures while the solid pattern shows predominantly proliferating basaloid myoepithelial cells with little glycosaminoglycan deposits or cyst-like basal lamina glandular spaces. In most series, the solid pattern is the least common. This pattern can be associated with higher numbers of mitotic figures which may explain its worse prognosis. Histopathological confirmation of perineural and bone invasion is essential. Tumor extent and margins are also important to document histopathologically. On immunohistochemistry, most ACC are MYB (myeloblastosis gene marker) positive and all are PLAG1 (pleomorphic adenoma gene marker 1) negative, in contrast to pleomorphic adenoma. High-grade transformation, previously known as dedifferentiation, may occur. Molecular pathology and genetics: Loss of chromosome 6q is a frequent finding in ACC. This may be associated with t(6;9)(q22–23; p23–24) resulting in a MYB-NFIB fusion gene. This may be associated with t(6;9)(q22–23; p23–24) resulting in a MYB-NFIB fusion gene. KRAS, NRAS and MET mutations were reported in lacrimal gland epithelial neoplasms with highest frequency (44%) in ACC. Microenvironment including immune response: ACC are typically surrounded by a variable desmoplastic reaction, but have very limited inflammatory infiltrates around them. Staging: Staging according to AJCC TNM 8th edition. Grading: In four grades from well differentiated to undifferentiated. Prognostic and predictive biomarkers: Age at diagnosis, completeness of surgical resection and tumor stage are important determinants of disease specific survival in epithelial lacrimal gland tumors in general. Prognostic factors in ACC include: Tumor histologic features (solid pattern being associated with a worse prognosis, perineural invasion being associated with local recurrence and skull invasion, bone invasion being associated with a fatal outcome in large tumors), tumor size (with the cut-off size of 2.0 cm in greatest dimension for T1 according to the 8th edition of AJCC), and tumor stage (recurrence, metastasis are significantly more common while survival is significantly worse in patients with T3 ¼ 2.5–5.0 cm according to the 6th edition of the AJCC, which is approximately equivalent to T2 ¼ > 2.0–4.0 cm in the 8th edition).

Lacrimal Gland Lymphomas Definition: Primary lacrimal gland lymphoma is a malignant proliferation of lymphocytes arising in the tissue without evidence of concurrent systemic lymphoma. Secondary lacrimal gland lymphoma occurs in patients as part of disseminated NHL. Burden: Lacrimal gland lymphomas are rare, representing  9% of all ocular adnexal lymphomas, which in turn constitute 2% of all extranodal lymphomas. Risk factors: They occur more commonly in elderly females, particularly those with Sjögren’s syndrome. No racial predilection. Pathology: Most lacrimal gland NHL are of B-cell type and are usually low-grade, although high grade transformation can occur. The most common subtype is the extranodal marginal zone B-cell lymphomas (EMZL), followed by FL and DLBCL. Histologically, EMZL show a nodular to diffuse heterogeneous B-cell infiltrate surrounding scattered reactive germinal centres and cause effacement of the normal glandular architecture. They are variably comprised of atypical small lymphocytes, centrocyte-like cells, monocytoid B-cells, immunoblasts, lymphoplasmacytic cells and plasma cells. Plasmacellular differentiation, including Dutcher bodies, may be striking. Lymphoepithelial lesions, representing infiltration of the ductal and epithelial structures by neoplastic B cells, can be seen. EMZL cells show reactivity for B-cell markers (CD20, CD79a, PAX5), BCL-2, and surface immunoglobulin. They are negative for CD5, CD10, CD23 and Cyclin D1. Cytokeratin highlights epithelial remnants in the lymphoepithelial lesions. Ki-67 growth fraction is low,  5%. Molecular pathology and genetics: Molecular testing using IgH-PCR demonstrates clonal rearrangement of the immunoglobulin chains. The genetic profile is dependent on lymphoma subtype. Microenvironment including immune response: Variable according to lymphoma subtype. Staging and grading: According to the 8th edition of the AJCC TNM Staging 2018, and also to the Ann Arbor staging system. Prognostic and predictive biomarkers: Most patients with stage I or II lymphoma are treated with low-dose radiotherapy  chemotherapy. Patients presenting with stage > III lymphoma were treated with chemotherapy alone. The prognosis of lacrimal gland lymphomas depends on the histologic type and clinical stage; however, the 5-year overall survival is 70%.

Rhabdomyosarcoma Definition: A malignant neoplasia that arises from pluripotential mesenchymal cells with a predilection to differentiate into skeletal muscle cells. Ocular rhabdomyosarcoma (RMS) may originate from the orbit, conjunctiva, eyelid or, rarely, the uveal tract. Burden: RMS is the most common primary orbital tumor in childhood accounting for 5% of all childhood malignancies. Based on histopathologic features, RMS is subdivided into embryonal, alveolar and pleomorphic type (see below). The embryonal type is most frequently seen in the first decade of life, accounting for > 90% of published cases of ocular origin. Alveolar RMS is observed in adolescents and young adults. Pleomorphic RMS occurs almost exclusively in adults. Risk factors: None known. Pathology: Tumor cells appear in the substantia propia beneath the epithelium as loosely spindle-shaped cells, with hyperchromatic nuclei with indistinct tapered cytoplasmic processes, some with long ribbons of eosinophilic cytoplasm (“tadpole”-like cells), with variable degrees of pleomorphism and proliferative rate (Fig. 11). The botryoid type contains linear condensations of tumor cells underneath the epithelial layer (“cambium layer”) and polypoid nodules with abundant, loose, myxoid stroma. Striking

Eye and Orbit Cancer: Pathology and Genetics

Fig. 11

59

Rhabdomyosarcoma. Typical “tad-pole” like cells in a botyroid type rhabdomyosarcoma of the orbit (H&E, 40
).

anaplasia may be present. Alveolar RMS is a small blue round cell neoplasm, with a worse prognosis. The term “alveolar” refers to a growth pattern in nests with central discohesion, separated by fibrovascular septa. The presence of specific immunohistochemical markers for muscle correlates with the degree of tumor cell differentiation. Desmin and actin are acquired by developing rhabdomyoblasts. Myoglobin, myosin and creatine kinase correspond to terminal differentiation. Antibodies against MyoD1 and myogenin/Myf4 are sensitive and specific to RMS. Molecular pathology and genetics: Embryonal RMS is characterized by loss of heterozygosity on the short arm of chromosome 11 (11p15.5), suggesting inactivation of a tumor-suppressor gene. By contrast, the majority of the alveolar RMS have the reciprocal chromosomal translocations t(2;13)(q35;q14) or t(1;13)(p36;q14), the former in approximately 70% of patients. The molecular counterpart of this translocation consists of the generation of a chimeric fusion gene involving the PAX3 gene, located on chromosome 2, and a member of the fork-head family FOXO1, located on chromosome 13. Staging and grading: Staging according to AJCC 8th edition. Grading: Currently, the preferred system for grading of sarcomas is the one proposed by the French Federation of Cancer Centers Sarcoma Group (FNCLCC). It uses three independent prognostic factors to determine the grade: mitotic activity, necrosis, and degree of differentiation of the primary tumor. Each feature is scored separately and the three scores are added to obtain the grade. Grade 1 is defined as a total score of 2 or 3, grade 2 as a total score of 4 or 5, and grade 3 as a total score of 6–8. Prognostic and predictive biomarkers: Poor prognostic factors in adults include age, pleomorphic histological subtype and more advanced disease at presentation. The embryonal type has a better prognosis.

Orbital Liposarcoma Definition: Malignant mesenchymal tumor with lipomatous differentiation. Burden: Generally, liposarcoma is the most common soft tissue sarcoma in adults. Liposarcoma of the orbit is rare; it can be primary or metastatic in origin. The mean age of presentation is 42.9 years (range 15–77) with slight female preponderance. Risk factors: Patients with genetic evidence of Li-Fraumeni syndrome harbor a risk to developed liposarcoma including two reported cases of orbital myxoid liposarcoma. Pathology: Liposarcoma is characterized by the presence of lipoblasts (Fig. 12). There are four major liposarcoma subtypes: atypical lipomatous tumor/well-differentiated liposarcoma (which includes the adipocytic, sclerosing, inflammatory and spindle cell variants), de-differentiated liposarcoma; myxoid liposarcoma; and pleomorphic liposarcoma. Myxoid liposarcoma is a lowgrade tumor characterized by myxoid stroma containing chicken-wire vascular channels and lipoblasts in variable stages of differentiation. It is reported to be the most common subtype occurring in orbital region. High-grade myxoid liposarcoma (formerly round cell liposarcoma) is defined by the presence of hypercellular areas, exceeding 5% of the lesion. ALT/WDL type is the second most common type occurring in this anatomical region; it is composed of variably sized adipocytes and bands of fibrosis containing spindle cells and hyperchromatic nuclei. The dedifferentiated type has been reported in the orbit, and is composed of high-grade spindle cell component resembling fibrosarcoma in addition to the well-differentiated component. Pleomorphic liposarcoma is a high-grade tumor and is composed of large polymorphic cells with bizarre nuclei. Abundant mitotic figures and areas of necrosis are also present. Molecular pathology and genetics: Both well differentiated liposarcoma and dedifferentiated liposarcoma harbors cytogenetic aberration in the form of Ring chromosomes and giant markers (12q13–15), and show amplification of MDM2, CDK4, and HMGA2. Myxoid liposarcoma is characterized by two main karyotypic aberrations: > 95% of cases carry a specific t(12;16)(q13;p11), which fuses the DDIT3 (CHOP) gene on 12q13 with the FUS (TLS) gene on 16p11. Approx. 5% of myxoid liposarcoma harbor t(12;22)(q13;q22), which fuses DDIT3 with EWSR1 on 22q12. Pleomorphic liposarcoma in contrast has complex karyotypic aberrations. p53 mutation occur in 60% of cases and NF1 mutation occur in 5% of cases.

60

Eye and Orbit Cancer: Pathology and Genetics

Fig. 12 Liposarcoma. Hyperchromatic spindle cells in a myxoid background. Lipoblasts are seen characterized by variably sized clear cells with scalloped hyperchromatic nuclei (H&E 20
).

Staging and grading: Staging according to AJCC 8th edition. Grading: See rhabdomyosarcoma. Prognostic and predictive biomarkers: Prognosis depends on histological type and the size of tumor. Orbital liposarcoma appears to be relatively favorable, likely due to small tumor size and predominantly well-differentiated histology. Plemorphic liposarcoma has a poorer prognosis with evidence of multiple recurrences. Distant metastasis from liposarcoma happens with the lung being the most common site.

See also: Malignant Tumors of the Eye, Conjunctiva, and Orbit: Diagnosis and Therapy.

Further Reading Andreasen, S., et al., 2017. An update on tumors of the lacrimal gland. Asia-Pacific Journal of Ophthalmology 6 (2), 159–172. Araujo, I., Coupland, S.E., 2017. Primary vitreoretinal lymphomadA review. Asia-Pacific Journal of Ophthalmology 6 (3), 283–289. Arnold, M.A., et al., 2016. Histology, fusion status, and outcome in alveolar rhabdomyosarcoma with low-risk clinical features: A report from the Children’s oncology group. Pediatric Blood & Cancer 63 (4), 634–639. Farquhar, N., Thornton, S., et al., 2017. Patterns of BAP1 protein expression provide insights into the prognostic significance and biology of Uveal melanoma. Journal of Pathology: Clinical Research 4 (1), 26–38. Findeis-Hosey, J.J., et al., 2016. Von Hippel Lindau disease. Journal of Pediatric Genetics 5 (2), 116–123. https://doi.org/10.1055/s-0036-1579757. Finger, P.T., Coupland, S.E., 2017. Ophthalmic tumors (on behalf of the Ophthalmic Sites Expert Panel). In: AJCC/TNM Cancer staging manual, 8th edn. Springer, Switzerland, pp. 779–854. Chapters 64–71. Kenawy, N., et al., 2013. Conjunctival melanoma and melanocytic intra-epithelial neoplasia. Eye (London, England) 27 (2), 142–152. https://doi.org/10.1038/eye.2012.254. Kirkegaard, M.M., et al., 2016. Conjunctival lymphomadAn international multicenter retrospective study. JAMA Ophthalmology 134 (4), 406–414. https://doi.org/10.1001/ jamaophthalmol.2015.6122. Larsen, A.C., et al., 2015. A retrospective review of conjunctival melanoma: Presentation, treatment, and outcome and an investigation of features associated with BRAF mutations. JAMA Ophthalmology 133 (11), 1295–1303. https://doi.org/10.1001/jamaophthalmol.2015.3200. Madge, S.N., et al., 2010. Primary orbital liposarcoma. Ophthalmology 117 (3), 606–614. Robertson, A.G., et al., 2017. Integrative analysis identifies four molecular and clinical subsets in Uveal melanoma. Cancer Cell 32 (2), 204–220. Saleh, G.M., et al., 2017. Incidence of eyelid basal cell carcinoma in England: 2000–2010. The British Journal of Ophthalmology 101 (2), 209–212. Schmitz, E.J., et al., 2017. Sebaceous gland carcinoma of the ocular adnexadVariability in clinical and histological appearance with analysis of immunohistochemical staining patterns. Graefe’s Archive for Clinical and Experimental Ophthalmology 255 (11), 2277–2285. Soliman, S.E., Racher, H., Zhang, C., MacDonald, H., Gallie, B.L., 2017. Genetics and molecular diagnostics in retinoblastomadAn update. Asia-Pacific Journal of Ophthalmology 6 (2), 197–207. https://doi.org/10.22608/APO.201711. Sun, E.C., et al., 1997. Epidemiology of squamous cell conjunctival cancer. Cancer Epidemiology, Biomarkers & Prevention 6 (2), 73–77.

Financial Burden of Cancer Care Bernardo HL Goulart and Veena Shankaran, Hutchinson Institute of Cancer Outcomes Research (HICOR), Seattle, WA, United States; Fred Hutchinson Cancer Research Center, Seattle, WA, United States; and University of Washington, Seattle, WA, United States © 2019 Elsevier Inc. All rights reserved.

Glossary Economic perspective Individual, group, institution, or entity that bear the costs for goods or services, for example, health insurance payers, hospitals, society in general. Direct medical costs Expenses directly related to the use of medical services and goods, for example, physician fees, acute inpatient care, laboratory and imaging tests, surgery, radiotherapy, chemotherapy. Direct nonmedical costs Expenses related to the logistics of pursuing healthcare or costs that are not medically-related but result from the use of healthcare services. Examples include transportation costs to medical facilities, childcare cost for parents undergoing cancer treatment, unpaid caregiver time. Indirect costs Productivity losses that result from cancer or cancer treatment. Examples include morbidity costs from missed work days due to cancer complications or treatment; mortality costs from loss of productivity due to death. Targeted therapies Oral or intravenous antineoplastic drugs that bind and modify the activity of specific trans-membrane or intracellular proteins or enzymes, leading to downstream intracellular events that cause cell death or cell cycle arrest. Medicare The publically-funded federal insurance program in the United States Medicare is the largest insurance program in the country and reimburses medical services for all individuals age 65 year or older, or those with permanent disabilities or chronic renal insufficiency. Commercial insurance plans tend to follow the reimbursement policies and fees set by Medicare.

Introduction The economics of cancer care has become a central topic in the broader debate of healthcare delivery. A simple, undeniable fact justifies the shifting perception of cancer costs as a relatively unimportant matter to a key factor in health policy: cancer costs are globally soaring and threatening the sustainability of healthcare systems. From a simplistic point of view, the skyrocketing cancer costs represent an unavoidable consequence of the rapid pace of innovation in cancer diagnostics and therapeutics. Those in support of this view argue that the high costs represent society’s willingness to pay for innovation in cancer care. From a granular point of view, cancer costs result from complex interactions between multiple factors, including (in)efficiencies of care delivery, obsolete reimbursement models (e.g., fee-for-service), and conflicting interests among stakeholders, including patients, providers, hospital administrators, insurance payers, pharmaceutical industry, and government. Irrespective of the different opinions, a growing body of evidence supports the following concerns: (1) cancer care costs are increasing faster than the gross domestic product, at least in the United States, threatening the sustainability of healthcare programs; (2) cancer patients are directly experiencing negative consequences of high cancer care costs in the form of financial distress from unaffordable medical bills, a phenomenon known as “financial toxicity”; (3) uncontrolled costs can magnify cancer disparities by creating further barriers in access to care to those who cannot afford expensive treatments. Collectively, these areas of concern indicate the urgent need for multilevel changes in cancer care delivery that emphasize quality, lower costs, and value (i.e., better health outcomes for the unit of money spent). This article initially provides an overview of the temporal trends in cancer care costs, followed by a dissection of the healthcare components that significantly contribute to cost. Next, a discussion ensues about the market forces that drive up cancer costs, particularly for hospitalizations and oncology drugs. The final section describes several strategies that have the potential to bend the cost curve and increase the value of cancer care. Although cancer costs are context-specific and vary broadly across healthcare systems, the article focuses mostly on United States cost data for simplicity of scope and greater availability of published literature. The economic perspective will be the United States public or private healthcare payer (e.g., Medicare and commercial insurance plans), unless mentioned differently. Most data refer to direct medical costs as defined in the glossary, unless noted otherwise.

Temporal Trends in Cancer Costs Total United States spending in cancer has been continuously increasing over the past several decades, from an estimated $27 billion in 1990 to $125 billion in 2010. Although cancer costs have consistently accounted for approximately 5% of total US

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65121-8

61

62

Financial Burden of Cancer Care Table 1

Time trends in U.S. total cancer direct medical costs, including healthcare payer and patient out-of-pocket expenses

Period a

Expenditures (2014 $U.S. billions)

Growth from previous period (%)

1998–2000 2001–03 2004–06 2007–09 2010–12

104.5 118.8 135.5 140.9 143.6

– 13.7 14.1 4.0 1.9

a Data sources: NHEA, National Health Expenditure Accounts; MEPS, Medical Expenditure Panel Survey; NNHS, National Nursing Home Survey; HCUP, Healthcare Cost Utilization Project; IMS Institute for Healthcare Informatics National Sales Perspective (IMS Health, Danbury, Connecticut). Adapted from Lee, J. A., Roehrig, C. S., Butto, E. D. (2016). Cancer care cost trends in the United States: 1998 to 2012. Cancer 122, 1078–1084.

healthcare expenditures, most health economists consider cancer as an increasingly concerning area of spending, given the increasing number of cancer survivors and the rising costs of cancer treatments. In recent years, total cancer costs have been increasing at a slightly faster pace than the U.S. gross domestic product (GDP). Between 1998 and 2012, total cancer expenditures increased from $104.5 to $143.6 billion, corresponding to an average annual growth of 2.5% (Table 1). In the same period, the United States GDP grew annually at an average of 2.2%. Conservative projections indicate that US cancer costs will continue to rise at a similar pace, from $125 billion in 2010 to $154 billion in 2020, based solely on current demographic trends in cancer incidence and survival (Fig. 1). Assuming a 2% annual increase in cancer treatment costs, United States cancer expenditures will increase to $187 billion in 2020, or a 49% cost hike from 2010. Cancer costs also represent a significant portion of healthcare expenditures in other developed nations, accounting for 6.3% and 9.3% of total healthcare costs in the European Union (E.U.) and Japan, respectively. In 2009, the European Union spent an estimated V126 billion in cancer, though the figures varied broadly by country, from V184 per capita in Luxembourg, to V16 per capita in Bulgaria. The United States leads in cancer spending across all developed nations, as indicated by a study that estimated the total direct medical costs attributable to cancer to be V212 per person in the United States for the year of 2004, followed by Switzerland, which spent V199 per capita. Most studies report on direct medical costs to describe national estimates of cancer spending, as the data sources are more readily available for costs related directly to medical care. Recent studies suggest that direct nonmedical costs (e.g., unpaid caregiver time) and indirect costs (i.e., productivity losses) contribute substantially to total cancer costs, justifying the need for more robust methods to estimate expenses other than those directly related to cancer treatment. In 2009, of the estimated V126 billion spent in cancer by the European Union, V51 billion (41%) related to direct medical costs, V23 billion (18%) corresponded to unpaid caregiver time, and V52 billion (41%) accounted for productivity losses, respectively. In the United States, very few studies have reported on unpaid caregiver time or other cancer direct nonmedical costs, and no readily available data informs on costs of productivity losses from cancer morbidity. One study estimated the United States costs attributable to productivity losses from cancer mortality to be $116 billion in 2000. Collectively, the available evidence overwhelmingly confirms that cancer is a major source of healthcare spending in most countries. Healthcare policy makers are all too aware of the need to closely monitor trends in cancer expenditures at national and regional levels, as well as engage all stakeholders in a debate about strategies to implement affordable and high-quality cancer care. In order to inform discussions about strategies to implement cost-effective cancer care, the following sections will describe the components of cancer care costs in greater detail, as well as the factors driving these costs.

Components of Direct Medical Costs in Cancer Care Studies have consistently shown that acute inpatient care (hospitalizations) accounts for the largest fraction (nearly 50%) of cancer care spending, followed by antineoplastic drug therapies (15%–30%). Other services (e.g., imaging diagnostics, radiation therapy, hospice) correspond to smaller fractions of cancer expenditures (Table 2). Brooks et al. estimated Medicare spending in the first 6 months after the diagnosis of advanced breast, prostate, nonsmall cell lung, colorectal, and pancreatic cancers in a sample of over 60,000 beneficiaries (Table 2). Of the total mean cost of $35,122 per patient, acute hospital care accounted for $16,953 (48%), followed by oral and intravenous chemotherapy ($5705; 16%), and outpatient procedures ($2281; 6%). Utilizing data from the Medical Expenditure Panel Survey (MEPS), one study showed that hospitalizations represented 48%–55% of all cancer care expenditures in the United States between 1998 and 2012, followed by professional and clinical services (including chemotherapy and outpatient visits), which accounted for 33%–40%. Cost distributions are similar in the European Union: in 2009, of the V51 billion spent in cancer, hospitalizations accounted for V28.4 billion

Financial Burden of Cancer Care

63

Fig. 1 Projections of National Direct Medical Costs Attributable to Cancer, United States, 2010–20 (2010 $US). Light gray areas indicate estimates of national expenditures for cancer care in 2010. Dark gray areas show projections of national cancer expenditures under the assumptions of constant incidence, survival, and costs of treatment for major cancer sites. Ini, initial year after diagnosis; Con, continuing care (i.e., time interval between the first year after diagnosis and last year of life; Last: last year of life. Adapted from Mariotto, A. B., Yabroff, K. R., Shao, Y., et al. (2011). Projections of the cost of cancer care in the United States: 2010–2020. Journal of the National Cancer Institute 103, 117–128.

Table 2

Average Medicare spending per patient for advanced stage nonsmall cell lung, colorectal, breast, prostate, and pancreas cancer, first 6 months from diagnosis, years 2004–09

Service type

Mean cost (2011 $US)

Percentage of total cost (%)

Acute hospital care Chemotherapya Outpatient proceduresb Imagingc Radiation therapy Hospice Home health Otherd Total

16,953 5705 2281 1837 1832 1743 870 3901 35,122

48.3 16.2 6.5 5.2 5.2 5.0 2.5 11.1 100.0

Source: 2004–10 Medicare claims linked to surveillance, epidemiology, and end results (SEER) data for 61,838 individuals with common advanced solid tumors. a Intravenous and oral antineoplastic agents. b Outpatient surgery and other outpatient procedures. c X-ray, ultrasound, computed tomography, magnetic resonance imaging, and nuclear medicine. d Home health, outpatient physician fees, laboratory and pathology testing, Medicare part B medications other than antineoplastics, skilled nursing facilities and rehabilitation hospitals, durable medical equipment, other services (e.g., ambulance). Adapted from Brooks, G. A., Li, L., Uno, H. et al. (2014). Acute hospital care is the chief driver of regional spending variation in Medicare patients with advanced cancer. Health Affairs (Millwood) 33, 1793–1800.

64

Financial Burden of Cancer Care

(56%), followed by V13.5 billion (27%) spent in antineoplastic drugs, while outpatient and emergency visits accounted for less than 20%. Although hospitalizations and antineoplastic drugs are responsible for the majority of cancer care costs, spending in other services have been proportionally increasing over time and may become significant contributors in the near future. One example is the use of expensive imaging diagnostic modalities. Between 1999 and 2006, Medicare beneficiaries with lung cancer had a mean annual increase of 36% and 7% in the use of positron emission tomography (PET) and magnetic resonance imaging (MRI), respectively. The increase in utilization of advanced imaging modalities corresponded to a difference in imaging costs per beneficiary of $1482 in 1999, to $3260 in 2006, or an average 9.5% annual escalation, compared with an average 2.6% annual escalation in total costs. No readily available data describe recent trends in radiation therapy costs, but those are likely to increase, given the dissemination of newer and expensive radiotherapy modalities, including stereotactic techniques, proton beam radiation, and tomotherapy. The following sections focus on costs of hospitalizations and antineoplastic drugs, as those are the two service categories in which the greatest opportunities exist for implementation of strategies that improve the value of cancer care.

Factors Driving Hospitalization Costs Two groups of factors independently influence hospitalization costs: those associated with higher utilization of acute inpatient care, and those related to variations in the unit price that hospitals charge for inpatient services.

Predictors of Utilization of Acute Inpatient Care Limited access to outpatient care seems to substantially contribute to higher utilization of inpatient care services, and, by extension, to hospitalization costs. Population-based data from the California Cancer Registry has shown that 71% of 25,032 patients with advanced breast, prostate, colorectal, nonsmall cell lung, and pancreatic cancer underwent at least one hospitalization in the first year after the diagnosis, and 16% had three or more hospitalizations in the same period. Most of the hospital admissions (64%) originated from the emergency department (ED). Several patient and health-system characteristics predicted higher likelihood of rehospitalizations, including Non-Hispanic Black and Latino race/ethnicity, lower socio–economic status, higher number of comorbidities, for-profit hospital ownership, and lack of outpatient palliative care. In a smaller study, investigators evaluated the medical reasons for hospitalizations in a retrospective cohort of 154 patients treated for gastrointestinal cancers at Dana–Farber Cancer Institute. Of a total of 201 hospitalizations, 53% were due to cancer-related symptoms, followed by 28% attributable to complications from cancer treatment. These studies jointly provide indirect evidence that limited access to outpatient care, which disproportionally affects patients of lower socioeconomic status or from racial/ethnic minority groups, lead to higher utilization of ED and inpatient acute care services. Aggressive end-of-life (EOL) care leads to higher hospitalization rates and costs, presumably due to cumulative toxicities from consecutive lines of treatment or from poorly controlled cancer symptoms. Although definitions of aggressive EOL care differ across studies, a growing body of evidence consistently demonstrates that patients are more likely to undergo hospitalizations if they receive chemotherapy in the last months of life, do not receive timely hospice services, or do not have access to outpatient palliative care programs. In the study of 154 patients with gastrointestinal cancers treated at Dana–Farber Cancer Institute, investigators classified 39 of the 201 (19%) hospital admissions as avoidable. Patients who received third or subsequent lines of noncurative chemotherapy were significantly more likely to have an avoidable hospitalization compared with patients who received only one line of chemotherapy (36% vs. 23%). In a secondary analysis of a landmark randomized trial comparing early palliative versus standard care in advanced nonsmall cell lung cancer, patients randomized to the early palliative care group had lower hospitalization costs in the last 30 days of life (mean cost difference ¼ $2896 per patient) compared with the control group. In a retrospective matched cohort study of Medicare decedents diagnosed with cancer or other diseases, patients who received hospice care at any time had lower number of hospitalizations, shorter hospital length of stay, and lower likelihood of admissions to the Intensive Care Unit. Comorbidities and inpatient complications also contribute to higher hospitalization costs for adults with advanced cancer. In a prospective, multicentric cohort study involving 1020 hospitalized patients diagnosed with multiple solid and hematologic malignancies, the mean admission cost was US$10,364 per patient. Mean costs increased by $852, $5289, and $8267 for patients with higher number of comorbidities, or who had minor or major inpatient complications, respectively (Table 3). In the same study, consultation with inpatient palliative care was associated with a decrease in mean admission costs of $1506 when no complications occurred, and of $5617 in the setting of a minor or major complication, respectively. This body of evidence strongly suggests that health systems may reduce the rate and costs of hospitalizations by improving access and coordination of care, symptom management, and by incorporating palliative care services to the inpatient and outpatient settings.

Predictors of Higher Hospital Service Prices Very few studies have systematically evaluated the degree of variation in prices charged by hospitals for specific inpatient services, but emerging data indicates substantial differences in the dollar amount billed for the same services across healthcare systems. In

Financial Burden of Cancer Care Table 3

65

Factors driving the variation in hospitalization costs in adults with cancer treated in four large US Medical Center (Hospital perspective; N ¼ 1,020)

Factor Patient characteristics Age Comorbidities Admitting diagnosis Electrolyte disorder In-hospital complications Major Minor

Average marginal effect (2011 $US)

95% Confidence interval

53 852

99 to 6 550–1153

4759

7928 to 1590

8267 5289

4509–12,025 3480–7097

Results adjusted for other patient characteristics (sex, race, education, insurance status, presence of advance directive, and activities of daily living), and disease characteristics (tumor type, other admitting diagnosis, symptom burden, analgesic use). Adapted from May, P., Garrido, M. M., Aldridge, M. D. et al. (2017). Prospective cohort study of hospitalized adults with advanced cancer: Associations between complications, comorbidity, and utilization. Journal of Hospital Medicine 12, 407–413.

a retrospective study of claims data from privately insured individuals in the United States between 2007 and 2011, variation in hospital transaction prices was the primary driver of total spending variation across hospital referral regions, after accounting for differences in patient characteristics. The same study revealed that service prices are 15.3% higher for hospitals that hold market monopoly power in their geographic regions, that is, have few or no competitors in the area. This type of data supports the concern that the ongoing merging of large for-profit healthcare systems contribute considerably to total healthcare spending, and, by extension, to cancer costs.

The Growing Costs of Antineoplastic Drugs

Monthly Cost of Treatment (2014 dollars, log scale)

The cost of antineoplastic drugs (hereafter, cancer drugs) is the fastest growing of all cancer cost categories. Bach et al. have analyzed temporal trends of monthly acquisition costs for new cancer drugs at the time of their market approval. These costs have exponentially increased from a median of approximately US$100 in the 1970s to more than $10,000 after 2010 (Fig. 2). Medicare spending in cancer drugs has increased from $3 billion in 1997 to $11 billion in 2004, a 267% difference, partly due to increase in drug prices, partly due to higher drug utilization over time. Cancer drugs account for 7 of the 10 most expensive drugs reimbursed by Medicare part B, which is the federal public insurance program responsible for payments of injectable medicines in the United States. Recent studies show that the prices of novel target drugs have been growing at a much faster pace than the overall cancer costs in the United

$100,000

$10,000

$1,000

$100

$10 Individual drugs Median monthly price (per 5-year period)

$1 1970

1980

1990 2000 Year of FDA Approval

2010

Fig. 2 Temporal trends in monthly drug acquisition costs at the time of drug regulatory approval (1965–2015). Adapted from the American Society of Clinical Oncology, (2016). The State of Cancer Care in America, 2016: A Report by the American Society of Clinical Oncology. Journal of Oncology Practice 12, 339–383.

66

Financial Burden of Cancer Care Table 4

Sources of increase in cancer drug expenditures per beneficiary, from 2001 to 2005 and 2005 to 2010

Source

Expenditure (2013 $US)

%

Total increase between 2001 and 2005 Increase in use of targeted therapy drugs Increase in launch price of new targeted therapy drugs Increase in price of targeted therapy drugs after launch Total increase between 2005 and 2010 Increase in use of targeted therapy drugs Increase in launch price of new targeted therapy drugs Increase in price of targeted therapy drugs after launch

7765.2 6534.4 432.4 798.4 6846.5 5091.6 1016.0 738.9

100 84 6 10 100 74 15 11

Source: Analysis conducted using LifeLink Health Plan Claims Database from January 2001 to September 2011. Adapted from Shih, Y. C., Smieliauskas, F., Geynisman, D. M. et al. (2015). Trends in the cost and use of targeted cancer therapies for the privately insured nonelderly: 2001 to 2011. Journal of Clinical Oncology 33, 2190–2196.

States, with annual increase rates of 12%–14%. Finally, the prices of cancer drugs have also been rapidly increasing even after their market approval. One study estimated an average annual increase of 11% in the monthly payments for oral target drugs years after approval (Table 4; Fig. 3). Although one could be tempted to attribute the rapid increase in cancer drug prices to the development of new drugs with innovative mechanisms of action, evidence suggests that this is not necessarily the case. In a metaanalysis of 74 clinical trials that supported the United States regulatory approval of drugs for solid tumors between 2000 and 2015, the median monthly drug acquisition costs (2016 US$) were $10,583, $11,980, $13,320, and $15,240, for chemotherapy, oral target drugs, immunotherapy, and antiangiogenic drugs, respectively (P ¼ .22). Of note, this study revealed an annual increase in monthly drug prices of 13%, a finding consistent with other literature reports. In addition, the authors reported a lack of correlation between monthly drug prices with efficacy endpoints reported in the trials, including progression-free and overall survival. This and other studies indicate that high drug prices have little correlation with innovation.

Why Cancer Drugs Are Expensive The general consensus is that pharmaceutical companies set drug prices based on what “the markets can bear.” This practice would be reasonable if it were not for the several distortions that characterize the oncology drug market. These market imperfections are the driving forces of the rapidly escalating costs of cancer drugs, and the misalignment between drug prices and efficacy, as outlined below. For didactic purposes, we classify the drivers of drug costs as those that lead to high acquisition prices, followed by those associated with increased drug utilization.

Factors Associated With High Drug Prices Pharmaceutical Industry’s Monopoly Power Simply put, patent protection laws allow pharmaceutical companies to behave like monopolies and set very high prices for their drugs. The rationale for patent protection is to give companies the opportunity to obtain a favorable return on the risky investment of developing a drug. In general, patent laws prohibit other companies from manufacturing generic formulations of the same compound, which prevents competition from forcing drug prices down. Pharmaceutical company representatives argue that this monopoly power is necessary and justified given the high risks and costs involved in research and development (R&D), estimated at US$2.6 billion per approved drug if one accounts for the money spent in preclinical and clinical research of agents that fail to receive market approval. The counterpoints to the Pharma views are several: (1) some health economists challenge the accuracy of reported R&D costs; (2) pharmaceutical companies do not really incur risks if the sales of approved drugs are supposed to compensate for the money spent in the development of failed compounds; and (3) drug prices can be elastic to demand, that is, pharmaceutical companies may decrease prices if clinics are not willing to adopt their high-cost drugs when cheaper alternatives exist. The case of aflibercept (Zaltrap, Sanofi) illustrates the latter point. This drug received US regulatory approval for the treatment of metastatic colorectal cancer in August of 2012. Sanofi set the monthly price of aflibercept at roughly $11,000. In October 2012, clinicians at Memorial SloanKettering Cancer Center publically announced that they would not include aflibercept in their drug formulary after noticing that bevacizumab, a drug of similar class, offered nearly identical efficacy at a monthly cost of roughly $5000. After Memorial’s public announcement, Sanofi decreased the price of aflibercept by 50% to allow its inclusion in drug formularies. The aflibercept case indicates that pharmaceutical companies can retain profitability despite large cuts in drug prices, which confirms the notion that monopoly practices contribute to costs that have little correlation with drug efficacy.

Financial Burden of Cancer Care

67

2004 2001

(A) Imatinib

$3,031

Erlotinib

$3,214

2005

Gefitinib

$1,808

Lenalidomide

$6,781

Sorafenib

$4,825

2006

Sunitinib

$6,584

Dasatinib

$4,417

Thalidomide

$4,398

Vorinostat 2007 2009

$7,562

Lapatinib

Everolimus

$5,744

Pazopanib

$5,755

$3,156

Nilotinib

$5,589

2011

Vandetanib

$10,211

Crizotinib

$9,899

Vemurafenib

$9,695

Axitinib

$9,097

2012

Enzalutamide

$7,720

Bosutinib

$8,467

Vismodegib

$8,457

Regorafenib

$9,605

$0

$4,000

$8,000

$12,000

2004 2001

(B) Imatinib

$6,237 (15.5%)

Erlotinib

$5,231 (10.2%)

2005

Gefitinib

$1,819 (0.2%)

Lenalidomide

$8,524 (4.7%)

Sorafenib

$8,534 (12.1%)

2006

Sunitinib

$8,637 (5.6%)

Dasatinib

$8,386 (13.7%)

Thalidomide

$6,756 (9.0%)

2007

Everolimus

2011

Lapatinib

2009

Vorinostat

$7,953 (1.0%) $4,162 (5.7%)

Nilotinib

$7,926 (7.2%) $7,861 (11.0%)

Pazopanib

$6,438 (3.8%)

Vandetanib

$10,244 (0.1%)

Crizotinib

$10,259 (3.6%)

Vemurafenib

$10,493 (8.2%)

Axitinib

$9,097

2012

Enzalutamide

$7,720

Bosutinib

$8,467

Vismodegib

$8,457

Regorafenib

$9,605

$0

$4,000

$8,000

$12,000

Fig. 3 Per patient per month (PPPM) Medicare costs of oral target therapies at year of launch and in 2012. PPPM costs represent a proxy of actual monthly drug prices paid by Medicare. Panel A shows PPPM costs for oral target drugs at launch year or in 2007 (whichever was later). Panel B shows PPPM costs for the same drugs in 2012, including drugs approved before 2012 and in 2012. Percentages in parentheses indicate the annual rate of increase in PPPM costs. The figure shows that 5 out of the 16 oral drugs approved before 2012 had annual rates of increase above 10%. Source: SEER-Medicare, 2007 to 2012. Adapted from Shih, Y. T., Xu, Y., Liu, L., et al. (2017). Rising prices of targeted oral anticancer medications and associated financial burden on Medicare beneficiaries. Journal of Clinical Oncology, 35(22), 2482–2489.

68

Financial Burden of Cancer Care

Federal Legislation on Drug Pricing in the United States and Abroad US federal laws preclude any considerations of costs for regulatory and market approval of drugs. Specifically, the Food and Drug Administration (FDA) appraises the evidence supporting drug efficacy and safety for regulatory approval, but not costs. Federal law also mandates that Medicare covers all drug products approved by the FDA. The Medicare Prescription Drug, Improvement, and Modernization Act of 2003 determined that Medicare reimburses drug costs based on the price set by the manufacturer, without engaging in price negotiations. In the current format, the United States federal policies on drug pricing give unfair market advantages to the pharmaceutical industry at the expense of patients, payers, and society. Since Medicare has no bargaining power to negotiate the reimbursement for drugs, companies have yet an additional incentive to set high prices to maximize profits regardless of drug efficacy. The influence of legislation on drug prices affects not only Medicare beneficiaries, but also the commercially insured population, since commercial plans tend to adopt similar drug reimbursement schedules paid by Medicare. The ultimate consequence is that the United States legislation significantly contributes to high cancer drug costs. To quantify the impact of the United States federal legislation on cancer drug costs, investigators have compared drug prices in the United States versus other national health systems that explicitly consider costs for drug regulatory approval. One study showed that the launch prices were 42% higher in the United States compared with the United Kingdom (UK) for 10 top-selling cancer drugs. The British National Institute for Health and Care Excellence (NICE) has developed and implemented a highly transparent and standardized process of appraisal of evidence to recommend the acceptance or rejection of new drug applications for market approval in the United Kingdom. The evidence appraisal requires the conduct of formal cost-effectiveness analysis (CEAs). In order to meet criteria for market approval in the United Kingdom, a given drug cannot exceed preestablished cost thresholds per unit of clinical benefit, usually measured in quality-adjusted life-years (QALYs). The explicit consideration of cost-effectiveness has enabled the U.K. National Health System to cover cancer drugs at lower costs. Additional studies also showed higher cancer drug prices in the United States versus other developed countries, including Norway, Australia, and Canada. In response to country-specific requirements for drug market approvals, pharmaceutical companies have commonly adopted a strategy of setting lower drug prices where cost regulations exist, and higher prices in countries that do not impose price caps. This global pricing strategy essentially forces the United States healthcare system to subsidize the costs of cancer drugs abroad and contributes to the escalation of cancer costs.

Drug Development Costs Representatives from the pharmaceutical industry claim that expenditures in drug research and development (R&D) at least in part explain the high prices of approved cancer drugs. A recent study suggests that R&D costs, though considerable, explain only a minor fraction of marketed drug prices. The investigators analyzed the R&D spending and the revenue sales for 10 companies and their respective 10 drugs that received market approval in the United States. The median time and costs (2017 US$) to develop a drug was 7.3 years and $648 million, respectively. With a median time of 4.0 years since approval, the median revenue sales was $1.6 billion, resulting in a $1 billion return on investment per approved drug. For all 10 drugs, the total R&D spending was $7.2 billion, compared with a total of $67.0 billion in revenue sales. The study intentionally did not consider the R&D costs of failed drugs, and concluded that the costs of drug development do not seem to justify the postmarket costs of approved cancer drugs.

Factors Associated With Higher Drug Utilization Fee-For-Service Reimbursement Models In the United States, the most common reimbursement policy for injectable drugs follows the fee-for-service (FFS) model, also known as “buy and bill.” In FFS, insurers pay a reimbursement fee to clinics whenever physicians prescribe injectable antineoplastic drugs. Clinics derive variable profit margins from administering injectable drugs, which depend on the negotiated drug price with drug manufacturers and the actual amount covered by insurance plans. Since clinics’ total revenues increase with higher number of prescribed drugs, this reimbursement model incentivizes physicians to overutilize cancer drugs in general, and to prioritize use of more costly drugs, which contributes to higher spending. The FFS model rewards clinics for higher quantity of drugs delivered, as opposed to better value offered by these drugs. This perverse incentive became more evident after the implementation of the medicare modernization act (MMA) of 2003, which determined that the reimbursement for injectable drugs would consist of the average sales price (ASP, or the average of sales transactions between drug manufacturers and clinics purchasing the drugs) plus 6%. The goal of the MMA was to reduce drug spending by cutting reimbursement fees. One study reviewed the patterns of chemotherapy prescribing for advanced lung cancer before and after the implementation of the MMA in January 2005, and showed that chemotherapy spending actually increased with the new legislation. The authors explain the findings by a change in prescription patterns from less to more costly agents (e.g., docetaxel as a substitute to paclitaxel), as a way to maximize profit margins from the 6% fee. Clearly, new reimbursement models are necessary to reduce drug spending while preserving treatment efficacy. Such models need to link payments to clinical benefits, as opposed to the quantity or price of drugs prescribed.

Financial Burden of Cancer Care

69

Imperfect Information A perfect competitive market assumes that consumers have adequate knowledge about the features of the goods and services that they purchase. Consumer information influences the demand for goods, which ultimately affects prices. In markets characterized by high degree of consumer information (e.g., food and clothing), prices tend to closely reflect how consumers value goods and services. In markets characterized by inadequate consumer information, prices often exceed what consumers would be willing to pay had they possess appropriate knowledge about the goods. The latter is the case of the oncology drug market. If we consider the patient as the consumer of cancer drugs, a legitimate concern arises about the amount and quality of the information that patients use to make decisions to pursue or forego drug treatments, including data regarding clinical benefits, toxicities, and outof-pocket costs. Not only patients may have limited understanding about the benefits and harms of cancer drugs, but their emotions and fears can impact their ability to make rational decisions about their care. Patients frequently make treatment decisions under substantial emotional distress, and often have overly optimistic expectations about the likelihood of benefit. When options are limited and prognosis is poor, fear and despair prompt many cancer patients to pressure their physicians to initiate subsequent lines of chemotherapy despite the minimal likelihood of favorable outcomes and often high risks of harm, which increases utilization of low-value drugs, drug spending, and hospitalization costs. In fact, current evidence shows that second- or third-line chemotherapy lines are not associated with longer overall survival in “real-world” settings (i.e., practice outside clinical trials), but do contribute to cancer spending.

Concerns About High Cancer Drug Costs The previous sessions characterized cancer drugs as the fastest growing category of cancer-related costs, and provided market-based explanations for this phenomenon. Yet, drug costs account for 15%–30% of cancer spending, which raises the question of whether concerns about drug costs are justifiable or constitute a distraction. The following sessions discuss the reasons why cancer drug costs clearly represent a public health concern.

Low-Value: Marginal Benefits, Sobering Costs, Budget Constraints A wealth of data demonstrates a lack of correlation between drug efficacy and costs. Despite their soaring prices, most cancer drugs offer modest clinical benefits. In a review of the clinical trials that supported the FDA approval of 71 drugs for a variety of solid tumors between 2002 and 2014, the aggregate median increase in progression-free survival was 2.5 months, and the aggregate median increase in overall survival was 2.1 months (Fig. 4). Of the 71 drug approvals, only 30 (42%) qualify as providing a “clinically meaningful” improvement, as defined by a consensus statement published by the American Society of Clinical Oncology (ASCO). In a review of 20 cancer drugs approved by the FDA between 2009 and 2013, 1-year drug costs ranged from US$59,000 to $157,000. The authors found weak correlations between annual drug costs and improvements in patient outcomes (R2 ¼ 0.132 for progression-free survival; R2 ¼ 0.165 for overall survival). Although some drugs represent true therapeutic breakthroughs (e.g., imatinib in chronic myeloid leukemia), most recently approved agents offer low value for their cost. The adoption of expensive, marginally effective drugs results in waste of scarce healthcare resources that could be spent in other interventions that offer greater benefits to cancer patients or society in general. If the escalation of cancer drug costs continues, healthcare budgets will eventually approach their limits, and the sustainability of health programs will be at risk. This trend may result in implicit rationing of care. When faced with budget constraints, insurance plans react by imposing coverage restrictions on expensive drugs, increasing copays and deductibles, or increasing premiums. Patients who cannot afford higher co-pays or premiums will have limited access to treatments, a situation that characterizes implicit rationing. Those who oppose the notion of regulation of drug prices avoid the topic of implicit rationing, but at the same time accuse regulatory agencies such as NICE of rationing. If we characterize NICE’s processes of drug appraisals as rationing, we should define it as explicit rationing, as the appraisals are at least transparent to the public. Regardless of whether society members choose to ignore or to consider costs in the approval process of new cancer drugs, at some point, decisions will have to be made about who receives treatment and who does not, either implicitly or explicitly.

Is Immunotherapy an Exception? The emergence of immunotherapy drugs represents the newest paradigm in the treatment of several advanced solid and hematologic cancers. The most popular class of immunotherapy drugs consist of the programmed death 1 (PD1) or programmed deathligand 1 (PD-L1) checkpoint inhibitors. These are monoclonal antibodies that induce an antitumoral immune response by prompting T-cells to recognize and kill tumor cells. Immune checkpoint inhibitors have gained much attention from the oncology community for their ability to induce durable tumor responses in several advanced solid and hematologic cancers. Although media outlets often portray immune checkpoint inhibitors as miraculous drugs, trial data show that their efficacy is not as impressive. Response rates range from 15% to 40% across diseases, the overall survival gain is usually measured in a few months, and most patients do not derive a survival benefit from these drugs.

70

Financial Burden of Cancer Care

Gains in Progression-free Survival, mo

(A) 10

8

6

4

2

0 1

3

5

7

9

11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 Median

Drug No.

Gains in Overall Survival, mo

(B) 10

8

6

4

2

0 1

3

5

7

9

11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 Median

Drug No.

Fig. 4 Progression-free and Overall Survival of 71 Oncology Drugs Approved by the U.S. Food and Drug Administration from 2002 to 2014. The horizontal bars represent median increases in progression-free survival (PFS, panel A) and overall survival (OS, panel B) based on results of clinical trials that led to drug approval by the Food and Drug Administration (FDA). Across the 71 approved drugs, the aggregated gain in PFS and OS were 2.5 and 2.1 months, respectively. Adapted from Fojo, T., Mailankody, S., Lo, A. (2014). Unintended consequences of expensive cancer therapeuticsthe pursuit of marginal indications and a me-too mentality that stifles innovation and creativity: The John Conley Lecture. JAMA OtolaryngologydHead & Neck Surgery 140, 1225–1236.

Despite much advertising, PD1 and PD-L1 checkpoint inhibitors bring the same concerns of excessive costs relative to benefits. In the United States, two PD1 (nivolumab, pembrolizumab) and three PD-L1 (atezolizumab, durvalumab, avelumab) checkpoint inhibitors have received FDA approval for the treatment of several cancers. In 2017, the monthly costs for these drugs ranged from US$13.8,000 to $14.9,000, based on average wholesale prices (Table 5). Multiple ongoing cost-effectiveness analyses are estimating the value of these drugs, but published studies suggested that checkpoint inhibitors are not cost-effective for the treatment on nonsmall cell lung cancer and head and neck cancers. Some peculiarities of PD1 and PD-L1 drugs deserve attention when considering the value of checkpoint inhibitors. These drugs are clearly less toxic than conventional chemotherapy. In randomized trials of second-line therapy for metastatic nonsmall cell lung cancer, grade 3 or 4 toxicities occurred in 7%–16% versus 35%–55% of patients treated with checkpoint inhibitors versus docetaxel,

Table 5

Monthly average wholesale price (AWP) for immune checkpoint inhibitors approved in the United States

Drug

AWP

Indication

Dose

Monthly cost (2017 US$)

Nivolumab Pembrolizumab Atezolizumab Durvalumab Avelumab

$3054.18/100 mg $2628.44/50 mg $10,344/1200 mg $4174.57/500 mg $1804.8/200 mg

Melanoma NSCLC Urothelial Urothelial Merkel cell

240 mg Q14 days 200 mg Q21 days 1200 mg Q21 days 10 mg/Kg Q14 days 10 mg/Kg Q14 days

$14,660.1 $14,018.3 $13,792.0 $13,809.5 $14,925.7

NSCLC, Nonsmall cell lung cancer; Q14 days, every fourteen days; Q21 days, every twenty-one days.

Financial Burden of Cancer Care

71

respectively. In order to estimate the value of checkpoint inhibitors, further studies need to fully account for the effect of their low toxicity profile on cost-effectiveness.

Financial Toxicity In response to rapidly rising healthcare costs in the United States, many insurers have shifted excess costs to patients in the form of higher premiums, deductibles, and copayments. The Kaiser Family Foundation estimates that annual out-of-pocket costs for employer-based health plan premiums and deductibles increased by over 50% between 2008 and 2015 (from $2943 to $5138). Further, an increase in multitiered prescription drug formularies in this same period from 7% to 32% disproportionately affects cancer patients who often require specialty drugs in the highest prescription tier associated with the highest level of cost sharing. In older Medicare beneficiaries, total Medicare spending on noncancer drugs is four times higher than spending on cancer drugs, as might be expected given that other medical conditions (e.g., heart disease, diabetes) are more common in the elderly. However, average annual beneficiary cost share is substantially higher in cancer patients versus others ($7226 vs. $1286). This trend in increased cost sharing coincides with one of the largest economic crises in recent times and a harsh reality that the average family with an income at 250%–400% of the federal poverty level has less than $3000 in liquid financial assets and a recent finding by the pew charitable trust that less than 40% of Americans can come up with $500–$1000 in cash in the case of an emergency. High cost sharing for cancer drugs and other services in combination with chronicity and intensity of cancer care leading to loss of work and income contributes to a set of circumstances that puts cancer patients and their families at particularly high risk for financial hardship. A growing body of literature had led to the recognition of an emerging side effect of cancer treatment known as “financial toxicity,” a term that encompasses a range of financial hardships including difficulty meeting household expenses, debt, bankruptcy, and permanent loss of career and income. A 2011 study by Bernard et al. utilizing data from the Medical Expenditures Panel Survey (MEPS) showed that patients with cancer who were either uninsured or enrolled in private and public health plans consistently reported higher financial burden (defined as > 20% of income spent on healthcare) than individuals with other chronic medical conditions. Another analysis of bankruptcy rates in cancer cases versus matched controls in Washington State showed that individuals with cancer were 2.65 times more likely to file for bankruptcy after diagnosis than matched controls. Several other studies have shown that, across various cancer types, approximately 20%–50% of patients experience significant financial hardship or “toxicity” postdiagnosis including missed work days, disability, inability to cover medical care costs or household expenses, accumulation of debt, and bankruptcy filings. Further, a recent analysis of 2011 MEPS data in approximately 1200 cancer patients showed that approximately 23% experienced “psychological financial hardship” defined as significant anxiety or worry about the costs and affordability of cancer care. Emerging evidence suggests that financial hardship before and after cancer diagnosis can potentially lead to worse health outcomes including poorer treatment adherence, lower clinical trial enrollment, poorer quality of life, and even worse survival. A recent study reported quality of life scores using multiple measures in patients with lung and colorectal cancer; patients who reported having fewer financial reserves (money in the bank to cover household expenses for 2 months or less) scored significantly lower on all quality of life measures than cancer patients with greater financial reserves. In the previously mentioned bankruptcy study, cancer patients who filed for bankruptcy had a significantly higher risk of death than propensity score matched cancer patients who did not file for bankruptcy (HR 1.79, 95% CI 1.64–1.96). While the exact pathway that leads from poorer financial status or financial hardship to poorer survival is not well elucidated, it is clear that financial toxicity contributes to undesirable outcomes and therefore must be addressed by the oncology community. Across studies, risk factors for financial toxicity include lower income, younger age, poorer preexisting financial status, advanced stage cancer, cancers requiring more aggressive treatment, and lack of health insurance. Other potential risk factors include role as the primary household income earner and being unmarried, though to potentially a lesser degree. Professional organizations such as the American Society of Clinical Oncology (ASCO) have emphasized the importance of screening patients and families for financial distress and intervening to mitigate financial toxicity in patients who might be at high risk. Unfortunately, few cancer clinics routinely screen for financial distress and, more importantly, even fewer have the capability to prevent financial hardship in those at high risk. Efforts to improve out-of-pocket cost transparency and increase communication between patients and oncologists about cost of care issues are needed. Certainly, addressing the larger problem of value in cancer care and the high price of cancer drugs may help to address financial toxicity. However, novel solutions such as changes in insurance benefit design and provision of financial counseling and navigation services as a routine part of cancer care are also needed.

Disparities in Treatment Access High drug costs can widen ongoing disparities in access to cancer treatments. Multiple studies have consistently shown that patients of lower social economic status or who lack insurance coverage tend to present with more advanced cancer stages and are less likely to receive standard therapies. Those findings indicate that poverty represents a major access barrier to cancer treatments. New to these investigations is the recent evidence that high drug out-of-pocket costs represent a barrier to treatment initiation in and of itself. The impact of copays on treatment access is particularly relevant for oral cancer drugs. One such example regards to access to imatinib or other similar oral tyrosine kinase inhibitors for the treatment of chronic myeloid leukemia (CML). Imatinib substantially increases overall survival in CML patients, and has become a standard of care for this disease. In a study of Medicare beneficiaries diagnosed with CML, patients initiated imatinib or a similar kinase inhibitor earlier if they had drug cost-sharing subsidies

72

Financial Burden of Cancer Care

versus those who did not, suggesting that higher drug copay prevented timely initiation of therapy. Likewise, a cross-sectional survey of 174 cancer patients showed that 28% and 22% of the participants did not fill prescriptions or took fewer doses of oral cancer drugs due to costs, respectively. In this study, lower income was associated with lower adherence to oral medications. Other studies in breast cancer demonstrated an association of higher drug co-pays for tamoxifen and aromatase inhibitors with lower adherence to these medications. The rising cost of cancer drugs increases the gaps between innovation and access, and accentuates the differences in the treatment patterns of high versus low income patients. Patients who can afford expensive therapies have an increasing number of treatment options, including the latest approved high-cost drugs, while those with limited financial resources can only receive a relatively fixed number of treatments. With the rapid pace of innovation, disparities in treatment access will also intensify on a global scale, as highincome countries will continue to implement cutting edge expensive technologies, while low- and middle-income countries struggle to develop infrastructure for basic cancer treatment, such as centers that offer surgery and radiation. Despite these obvious concerns, more rigorous research and policy changes are in order to address the impact of cancer costs on access to life-prolonging therapies.

Potential Solutions: Focusing on Value Value is a construct that links the benefits of healthcare interventions to their additional costs. Value-based approaches offer appealing solutions to financially strained healthcare systems, by prioritizing the adoption of interventions that offer higher benefits when costs are similar among alternatives, or interventions that cost less but offer similar benefits compared with the alternatives. The following sessions briefly describe strategies that can improve value in cancer care, including those that focus on better quality care, innovative reimbursement models, and legislation reform.

Better Quality Care Cancer care pathways Cancer care pathways consist of software physician-oriented decision tools that provide evidence-based recommendations at the point of treatment decisions through the disease course, from first- to subsequent lines of chemotherapy. Currently available pathways mostly focus on systemic therapies for common advanced solid tumors, although their role is expanding to include early stage cancers. The goal of pathway adoption is to reduce the substantial variation in chemotherapy prescribing practices, thereby reducing costs while maintaining or improving patient outcomes. Several third-party vendors have developed and licensed pathways to clinics, but academic groups and insurance companies have occasionally created their own pathways as well. A recent study from Dana–Farber Cancer Institute showed that adoption of an internally developed pathway decreased the 1-year costs while maintaining overall survival of patients with advanced nonsmall cell lung cancer (Table 6). Studies of commercial pathways have shown similar results in early and advanced stage lung and colorectal cancer, respectively. Despite the promising data, concerns have arisen regarding the processes of pathway development and implementation, including issues of transparency about the selection of regimens included in pathways, physicians’ restrictions of regimen choices, and the extra administrative burden of reporting pathway adherence. Formal guidelines for pathway implementation, such as the one proposed by the American Society of Clinical Oncology (ASCO), are necessary to ensure that patients have access to high-quality, evidence-based treatments.

Early palliative care Early integration of palliative care holds great potential to decrease costs, prolong survival, and improve quality of life, as demonstrated in a pivotal randomized trial that assigned patients with advanced nonsmall cell lung cancer to a palliative care team at the time of first-line chemotherapy versus standard of care. Subsequent randomized studies confirmed that early palliative care improved patient outcomes in other advanced solid tumors. These studies suggest that early palliative care can decrease costs particularly by reducing the rates of hospital admissions and emergency room visits, and by facilitating timely transitions to hospice. Inpatient palliative care also decreases hospital costs by avoiding unnecessary escalation of care (e.g., ICU admissions), and by Table 6

One-Year direct medial costs of before and after implementation of the Dana–Farber Cancer Care Pathways for stage IV Nonsmall cell lung cancer (Hospital perspective; N ¼ 370)

Pathways cohort Unadjusted cost Prepathway Postpathway Adjusted cost Prepathway Postpathway

Mean cost (2014 US$)

95% Confidence interval

64,508 48,515

53,140–75,876 41,421–55,608

69,122 52,037

33,242–105,001 25,200–48,849

Adapted from Jackman, D. M., Zhang, Y., Dalby, C., et al. (2017). Cost and survival analysis before and after implementation of Dana– Farber Clinical Pathways for Patients with stage IV non-small-cell lung cancer. Journal of Oncology Practice 13, e346–e352.

Financial Burden of Cancer Care

73

shortening the length of admissions. Despite the clear benefits, several challenges prevent widespread adoption of palliative care services, including a shortage of specialists, lack of proper reimbursement, and heterogeneous acceptance of palliative care by oncologists and patients. Future efforts should focus on training a workforce of palliative care providers, and educating patients and physicians about the benefits of early integration of palliative care.

Electronic symptom monitoring Electronic symptom monitoring is a novel method of following and addressing patients’ symptoms through web-based questionnaires. Patients are asked to report their symptoms frequently in between office visits. Symptoms that are intense or worsening prompt electronic messages to the physician’s office, allowing providers to timely intervene. A recent randomized study showed that electronic symptom monitoring reduces ED visits and hospitalizations, allows earlier detection of tumor progression and changes in chemotherapy regimens, improves quality of life, and prolongs survival in patients with advanced solid tumors. Economic analyses of electronic symptom monitoring are underway, but this type of intervention will very likely help decrease cancer costs and improve patient outcomes. In Fact, many clinics in the United States and Europe are embracing different versions of electronic symptom reporting tools.

Precision oncology At least in theory, precision oncology can improve the value of cancer drugs. This concept involves the use of predictive molecular biomarkers that allow the selection of patients who are more likely to benefit from target therapies, while avoiding the unnecessary costs of treating patients that will not benefit. The current data is conflicting to support or refute the hypothesis that precision oncology is cost-effective. Some published cost-effectiveness analyses have shown prohibitive costs per QALY gained for testing lung cancer patients for epidermal growth factor receptor (EGFR) or anaplastic lymphoma kinase (ALK) mutations, and treat those whose tumors harbor these genetic abnormalities with the appropriate target drug. Other studies showed favorable costeffectiveness ratios ranging from US$30,000 to $40,000 per QALY gained for treating patients with EGFR mutated lung adenocarcinomas with tyrosine kinase inhibitors compared with chemotherapy, but these studies did not account for the money spent in testing several patients to identify one mutation. The value of precision oncology will ultimately depend on several factors, including the performance and prices of multigene testing platforms such as next generation sequencing, the price, and survival benefits of target drugs, and the prevalence of clinically validated biomarkers.

Hospice Timely referral to hospice decreases cancer costs without impacting on survival. A randomized trial from 1980s showed no differences in overall survival and decreased number of hospitalizations for cancer patients assigned to hospice versus those who did not receive hospice care. In a more recent retrospective study of Medicare beneficiaries with advanced lung cancer, hospice use was associated with a nonstatistically significant survival benefit compared with no hospice use. In a subsequent retrospective matched cohort study of Medicare patients, assignment to hospice decreased total and hospital costs compared with patients who did not receive hospice care. In this study, the decrease in costs was more pronounced with earlier hospice referrals (Table 7). Physicians should be aware that a transition to hospice adds value in cancer care, particularly when patients have exhausted treatment options or are experience a decline in performance status. Several barriers may prevent patients from receiving hospice, including the psychological discomfort involved in the discussions of transitions from active therapy to comfort care, and cultural influences on patients’ perception of hospice. Oncology providers need appropriate training, psychological support, and reimbursement to appropriately conduct conversations involving hospice care and end-of-life planning.

Decision frameworks Many organizations have developed decision frameworks to facilitate formal assessments of the value offered by cancer drug regimens, with the hope that summarized information about costs and benefits will encourage stakeholders to make value-based treatment choices. Examples of such efforts include ASCO’s Value Framework, the European Society for Medical Oncology (ESMO) Magnitude of Clinical Benefit Scale, and the Institute for Clinical and Economic Review (ICER) Evidence Rating Matrix, among Table 7

Medicare expenditures at the end of life in hospice and matched nonhospice controls (N ¼ 3069)

Hospice use before death

Mean expenditures per patient, hospice group (2008 US$)

Mean expenditures per patient, nonhospice group a (2008 US$)

Adjusted mean difference (2008 US$)

Last 105 days Last 30 days Last 14 days Last 7 days

22,083 10,383 5698 4806

24,644 16,814 10,738 7457

2561 6431 5040 2651

Data source: Health and Retirement Study; Medicare claims. a Propensity score matched. Adapted from Kelley, A. S., Deb, P., Du, Q., et al. (2013). Hospice enrollment saves money for Medicare and improves care quality across a number of different lengths-of-stay. Health Affairs 32, 552–561.

74

Financial Burden of Cancer Care

others. Frameworks provide various methods and processes that quantify clinical benefits in a transparent manner, while some of these tools also explicitly consider drug costs. Although conceptually sound, prospective studies will be necessary to confirm the impact of value frameworks on treatment patterns, quality of care, and costs. One important limitation of frameworks is that they differ in regards to target audiences and scope, which prevents accurate comparisons of their impact. For example, the ASCO Value Framework targets physicians and patients, whereas ESMO’s and ICER’s frameworks were designed to inform policy makers and payers.

New Reimbursement Models Under the pressure to contain costs, healthcare payers are developing and implementing novel treatment reimbursement models that emphasize value. Common to these models is the financial or performance accountability shared by payers, clinics, and drug manufacturers.

Oncology care model In June 2016, the Centers for Medicare and Medicaid Services launched the Oncology Care Model (OCM) initiative, which consists of payment agreements between physician practices and multiple payers, including Medicare and commercial insurance plans. In OCM, physician practices are accountable for the expenses and performance surrounding predefined episodes of care that involve chemotherapy administration. Clinics that spend less than target expenditures per episode can retain a percentage of the savings if they meet prespecified quality benchmarks for that episode. OCM is an example of performance-based reimbursement models that incentivize physician practices to provide high-quality care at lower costs. Performance-based reimbursement has stirred some controversy, as clinics may have to accept financial losses when expenses exceed the targets or their performance does not meet quality benchmarks, depending on the contents of the payment agreements. Additional time and follow up assessments will be important to determine whether the OCM model actually increases value in cancer care.

Performance-linked reimbursement Performance-linked reimbursement represents yet another model that links the payments for cancer drugs to the achievement of prespecified clinical outcomes. Different than OCM models, performance-linked reimbursement involves risk-sharing agreements between drug manufacturers and clinics or payers. If a given drug provides the expected benefits in real-world settings, clinics or payers will pay the manufacturer the full drug price. If the drug proves to be less effective than expected, manufacturers offer drug price discounts, rebates, or refunds to the purchaser. One example of this model regards to the use of bortezomib (Velcade, Johnson & Johnson) in patients with multiple myeloma. In 2006, the U.K. National Health System entered a performance-linked agreement with Johnson and Johnson, in which patients would undergo four cycles of bortezomib, followed by an assessment of serum M proteins to determine tumor response. Patients were considered to have had treatment failure if the serum M protein levels declined by less than 50%. For those patients, Johnson & Johnson agreed to refund the expenses for bortezomib, or provide the same amount of drug to other patients at no charge. Although appealing, performance-linked reimbursement requires robust infrastructure for real-world data collection, and considerable negotiations among parties to determine what constitutes the desired outcome measures. Despite these caveats, this model will likely gain traction among drug manufacturers and payers in the near future.

Episode-based payments Episode-based payment models consist of lump-sum payments made to physicians at the beginning of an episode of care for each patient. Episodes of care are discrete cancer treatment events that span a predictable time period, usually of 6 months (e.g., adjuvant chemotherapy for breast or colorectal cancer). In this model, physicians take financial risk by absorbing the losses when the expenses exceed the amount paid at the beginning of the episode. The rationale is that physicians will prioritize regimens of similar efficacy and lower costs when alternatives exist, and will optimize the management of treatment- and disease-related complications to avoid patients from undergoing costly hospitalizations. A retrospective study evaluated the impact of an Episode-based payment model among volunteer clinics that contracted with United-Healthcare, a large private insurance company in the United States. In a sample of 810 patients with breast, lung, and colorectal cancers, episode-based payments resulted in estimated total savings of US$33.4 million with no effects on overall survival compared with historical controls (Table 8). In the Episode-payment patient cohort, chemotherapy utilization and costs were curiously higher than predicted, and most savings seem to have derived from lower Table 8

Impact of an episode-based payment model on direct medical costs in five medical oncology groups (N ¼ 810)

Cost category

Predicted (2012 US$)

Actual (2012 US$)

Savings (US$)

Total Chemotherapy

98.1 million 7.5 million

64.8 million 20.9 million

33.4 million  13.4 million

Adapted from Newcomer, L. N., Gould, B., Page, R. D. et al. (2014). Changing physician incentives for affordable, quality cancer care: Results of an episode payment model. Journal of Oncology Practice 10, 322–326.

Financial Burden of Cancer Care

75

hospitalization costs. Other insurance groups are implementing episode-payment agreements, but additional data is necessary to inform the applicability of this model to other diseases and episodes of care, as well as the impact on patient survival, quality of life, and satisfaction with care.

Legislation Reform No matter how impactful are the practice changes and reimbursement models implemented by physician and payers, cancer costs will continue to threaten budgets if drug acquisition prices follow the current escalating trend. Only comprehensive legislation reform can address the current lack of control on drug prices in countries like the United States. New federal policies need to give Medicare and other public insurance programs the autonomy to reject coverage for low-value drugs, and to use bargain power to negotiate prices that reflect value. The first pillar of legislation reform should be a provision that allows Medicare to decline coverage for drugs based on price. Like in any negotiations, the “walk away” option is crucial to allow buyers and sellers to agree upon prices that are perceived as fair. Without the option to say no, Medicare does not really enter any negotiations, but acts as a price taker. On the same note, provisions need to allow the use of explicit methodologies to evaluate drug value and affordability, including cost-effectiveness analysis and budget impact models. This approach would allow the implementation of value-based drug pricing policies in a publically transparent manner. New policies should also give Medicare the flexibility to consider factors other than cost-effectiveness for drug coverage decisions, including disease severity, the specific characteristics of the patient population, equity issues, and budget constraints. Legislative changes should promote more competition among manufacturers, instead of less. Competition allows market forces to decrease oncology drug prices. Generic medications and biosimilars need to enter the market sooner, and new laws need to restrict the ability of drug companies to buy extra patent protection time. Likewise, public insurance programs need to be able to import less costly cancer drugs, a strategy that would also create competition among domestic manufacturers and force prices down. Despite the clear rationale, legislation reform would require strong political motivation from society. The pharmaceutical industry has a vetted interest to maintain current federal laws regarding drug pricing, and industry lobbyists exert powerful political influence that prevent the enactment of any meaningful reform. The prospects of legislative transformation ultimately depend on the leading role of other stakeholders, including healthcare providers, payers, and patient advocates. These groups need to jointly demand major legislative changes from politicians and counterbalance the industry lobby.

Conclusions The rising costs of cancer care pose a risk to the sustainability of healthcare programs, and this trend will likely continue for the foreseeable future. Hospitalization and oncology drugs account for the majority of cancer expenditures, and result from multiple inefficiencies and misaligned incentives that affect the healthcare system. Stakeholders are developing several strategies to transform cancer care from a high-cost to a high-value enterprise.

See also: Aging and Cancer. Oncology Imaging. Radiation Oncology.

Further Reading American Society of Clinical Oncology, 2017. American society of clinical oncology position statement on addressing the affordability of cancer drugs. Journal of Oncology Practice/ American Society of Clinical Oncology. https://doi.org/10.1200/JOP.2017.027359 (Epub ahead of print). Bennette, C.S., Richards, C., Sullivan, S.D., Ramsey, S.D., 2016. Steady increase in prices for oral anticancer drugs after market launch suggests a lack of competitive pressure. Health Affairs 35 (5), 805–812. https://doi.org/10.1377/hlthaff.2015.1145. Brooks, G.A., Li, L., Uno, H., Hassett, M.J., Landon, B.E., et al., 2014. Acute hospital care is the chief driver of regional spending variation in Medicare patients with advanced cancer. Health Affairs 33 (10), 1793–1800. https://doi.org/10.1377/hlthaff.2014.0280. Fojo, T., Mailankody, S., Lo, A., 2014. Unintended consequences of expensive cancer therapeutics-the pursuit of marginal indications and a me-too mentality that stifles innovation and creativity: The John Conley lecture. JAMA Otolaryngology. Head & Neck Surgery 140 (12), 1225–1236. https://doi.org/10.1001/jamaoto.2014.1570. Kelley, A.S., Deb, P., Du, Q., Aldridge Carlson, M.D., Morrison, R.S., 2013. Hospice enrollment saves money for Medicare and improves care quality across a number of different lengths-of-stay. Health Affairs 32 (3), 552–561. https://doi.org/10.1377/hlthaff.2012.0851. Lee, J.A., Roehrig, C.S., Butto, E.D., 2016. Cancer care cost trends in the United States: 1998 to 2012. Cancer 122 (7), 1078–1084. https://doi.org/10.1002/cncr.29883. Luengo-Fernandez, R., Leal, J., Gray, A., Sullivan, R., 2013. Economic burden of cancer across the European Union: A population-based cost analysis. The Lancet Oncology 14 (12), 1165–1174. https://doi.org/10.1016/S1470-2045(13)70442-X. Mariotto, A.B., Yabroff, K.R., Shao, Y., Feuer, E.J., Brown, M.L., 2011. Projections of the cost of cancer care in the United States: 2010–2020. Journal of the National Cancer Institute 103 (2), 117–128. https://doi.org/10.1093/jnci/djq495. Porter, M.E., 2010. What is value in health care? The New England Journal of Medicine 363 (26), 2477–2481. https://doi.org/10.1056/NEJMp1011024. Prasad, V., De Jesus, K., Mailankody, S., 2017. The high price of anticancer drugs: Origins, implications, barriers, solutions. Nature Reviews. Clinical Oncology 14 (6), 381–390. https://doi.org/10.1038/nrclinonc.2017.31.

76

Financial Burden of Cancer Care

Ramsey, S., Blough, D., Kirchhoff, A., Kreizenbeck, K., Fedorenko, C., et al., 2013. Washington state cancer patients found to be at greater risk for bankruptcy than people without a cancer diagnosis. Health Affairs 32 (6), 1143–1152. https://doi.org/10.1377/hlthaff.2012.1263. Shankaran, V., Ramsey, S., 2015. Addressing the financial burden of cancer treatment: From copay to Can’t pay. JAMA Oncology 1 (3), 273–274. https://doi.org/10.1001/ jamaoncol.2015.0423. Shih, Y.T., Xu, Y., Liu, L., Smieliauskas, F., 2017. Rising prices of targeted oral anticancer medications and associated financial burden on medicare beneficiaries. Journal of Clinical Oncology 35 (22), 2482–2489. https://doi.org/10.1200/JCO.2017.72.3742. Sullivan, R., Peppercorn, J., Sikora, K., Zalcberg, J., Meropol, N.J., et al., 2011. Delivering affordable cancer care in high-income countries. The Lancet Oncology 12 (10), 933– 980. https://doi.org/10.1016/S1470-2045(11)70141-3. Winn, A.N., Keating, N.L., Dusetzina, S.B., 2016. Factors associated with tyrosine kinase inhibitor initiation and adherence among Medicare beneficiaries with chronic myeloid leukemia. Journal of Clinical Oncology 34 (36), 4323–4328. https://doi.org/10.1200/JCO.2016.67.4184.

Relevant Websites https://www.cms.gov/dCenters for Medicare & Medicaid Services. https://www.nice.org.uk/dNational Institute for Health and Care Excellence (NICE). https://www.fda.gov/dU.S. Food and Drug Administration. https://data.worldbank.org/dWorld Bank.

Gastric Cancer: Pathology and Genetics Irene Gullo and Fa´tima Carneiro, São João Hospital, Porto, Portugal; and University of Porto (FMUP), Porto, Portugal © 2019 Elsevier Inc. All rights reserved.

Glossary Advanced gastric cancer An invasive carcinoma infiltrating into the muscular propria and beyond. CDH1 gene The human CDH1 gene [MIM#192090] is localized in the long arm of chromosome 16, comprises 16 exons and encodes E-cadherin. CTNNA1 gene The human CTNNA1 gene [MIM#116805] is localized in the long arm of chromosome 5, comprises 16 exons and encodes a-E-catenin. Diffuse carcinoma Subtype of gastric carcinoma composed of poorly cohesive neoplastic cells. Early gastric carcinoma An invasive carcinoma limited to the mucosa or submucosa, regardless of nodal status. Early onset gastric cancer Early onset gastric cancer (EOGC) is a gastric cancer presenting at the age of 45 or earlier, representing approximately 10% of all patients with stomach cancer. E-cadherin E-cadherin is a transmembrane calcium-dependent protein that is predominantly expressed at the basolateral membrane of epithelial cells, where its functions are primarily cell–cell adhesion and suppression of invasion. Epstein–Barr virus Double-strand DNA virus belonging to the herpes virus family (HHV4). Gastric adenocarcinoma Malignant epithelial neoplasms with glandular differentiation. Gastric adenocarcinoma and proximal polyposis of the stomach (GAPPS) GAPPS is a variant of familial adenomatous polyposis (FAP) and is caused by point germline mutations in the promoter 1B of the APC gene. It is characterized by fundic gland polyposis, with areas of dysplasia and carries an increased risk of gastric adenocarcinoma. Gastric carcinoma with lymphoid stroma (GCLS) GCLS is a morphological subtype of GC composed of irregular sheets, trabeculae, poorly developed tubular structures and isolated cells, embedded within a prominent lymphocytic infiltrate, with occasional lymphoid follicles. Gastric microbiota The whole microbial community within the stomach. Helicobacter pylori Helicobacter pylori are Gram-negative bacteria that colonize gastric mucosa, considered by WHO as a group I carcinogen for gastric cancer. Hereditary diffuse gastric cancer (HDGC) HDGC is an autosomal dominant cancer susceptibility syndrome characterized by signet ring cell (diffuse) gastric cancer and lobular breast cancer, caused by germline mutations of CDH1 and CTNNA1 genes. Immune inhibitory checkpoint Molecules of the immune system that inhibit the T cell signal by antagonizing costimulatory molecules necessary to the activation of the T cells. Intestinal carcinoma Subtype of gastric carcinoma with glandular structure. Tumor immune microenvironment Noncancerous cells and extracellular components constituting the tumor stroma.

Nomenclature ACRG Asian Cancer Research Group AJCC American Joint Committee on Cancer CAF Cancer associated fibroblast CagA cag pathogenicity island-encoded EBER-ISH EBV-encoded small RNA in situ hybridization EBV Epstein–Barr virus EBVaGC Epstein–Barr virus associated gastric cancer EGJ Esophagogastric junction EMT Epithelial–mesenchymal transition FAP Familial adenomatous polyposis FGP Fundic gland polyps/polyposis FIGC Familial intestinal gastric cancer GAPPS Gastric adenocarcinoma and proximal polyposis of the stomach

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65076-6

77

78

Gastric Cancer: Pathology and Genetics

GC Gastric carcinoma GCLS Gastric cancer with lymphoid stroma HDGC Hereditary diffuse gastric cancer IGCLC International gastric cancer linkage consortium IHC Immunohistochemistry ISH In situ hybridization MSI Macrosatellite instability MSS Macrosatellite stable SPEM Spasmolytic polypeptide-expressing metaplasia TCGA The cancer genome atlas TAM Tumor associated macrophages TME Tumor microenvironment VacA Vacuolating cytotoxin A

Definitions: Types and Subtypes The vast majority of stomach cancer cases are gastric carcinomas (malignant epithelial neoplasms). Nonepithelial tumors predominantly include lymphomas and mesenchymal tumors. In this article, we will focus on gastric carcinoma (GC). Most cases of GC are sporadic. Familial clustering is observed in about 10% of the cases. Hereditary GC accounts for a very low percentage of cases (1%–3%), encompassing hereditary diffuse gastric cancer (HDGC) and familial intestinal GC (FIGC). Gastric adenocarcinoma and proximal polyposis of the stomach (GAPPS) has been recently recognized as a rare variant of familial adenomatous polyposis (FAP) with a distinct gastric phenotype. Furthermore, GC can develop in the setting of other hereditary cancer syndromes (see “Molecular Pathology and Genetics” section). According to the depth of invasion, GC is classified as early and advanced. Early GC refers to an invasive carcinoma that is limited to the mucosa (pT1a) or the mucosa and submucosa (pT1b), with or without lymph-node metastases and regardless of tumor size. This classification is prognostically relevant (see “Prognostic and Predictive Biomarkers” section): early GCs have a better prognosis compared to advanced GCs, with a 5-year survival rate of 80%–95%, after surgical resection. By contrast, GCs infiltrating into the muscularis propria and beyond are defined as “advanced.” Patients with advanced GC, if the tumor is unresectable, have a dismal prognosis with an expected survival of few months, even with chemotherapy and best supportive care. Regarding the anatomic location, tumors arising in the esophagogastric junction (EGJ) are staged as esophageal or GCs, depending on the location of the tumor epicenter with respect to the EGJ. According to the 8th edition of the American Joint Committee on Cancer (AJCC) staging system, EGJ tumors with the epicenter located > 2 cm distal to the EGJ should be classified using the stomach cancer system, while EGJ tumors with the epicenter at
2 cm below the EGJ that straddle the junction should be staged as esophageal tumors. GCs represent a biologically heterogeneous group of tumors with multifactorial etiologies, encompassing environmental, genetic and molecular factors. They are characterized by broad morphological heterogeneity with respect to the architecture, pattern of growth, cell differentiation, histogenesis and molecular pathogenesis (Fig. 1). Several histopathological classifications have been proposed along the years, reflecting the heterogeneity of GC morphology (see “Pathology” section). Moreover, the advent of highthroughput molecular technologies has recently led to a better understanding of GC molecular complexity (see “Molecular Pathology and Genetics” section).

Burden GC accounts for about 7% of cancers worldwide (being the fifth most common malignancy in the world, behind cancers of the lung, breast, colo-rectum and prostate) and is the third leading cause of cancer death in both sexes. However, the distribution is uneven, from areas with high incidence (> 60 per 100,000 males) such as East Asia, central and eastern Europe and Latin America, to low incidence areas (< 15 per 100,000 population) such as North America, Northern Europe and most countries in Africa and SouthEastern Asia. More than 70% of the cases occur in developing countries. Countries with a high incidence of GC, in which there are screening programs of asymptomatic individuals, have a high incidence of EGCs, ranging from 30% to 50% in East Asia (e.g., Japan), with lower figures for the West (16%–24%). In regions with high incidence of GC, most cases are localized in the antrum and pylorus. GCs of the proximal stomach and the EGJ are more common in low incidence countries (e.g., North America and Europe) and are associated with gastro-esophageal reflux disease (GERD).

Gastric Cancer: Pathology and Genetics

79

Fig. 1 Gastric cancer is a heterogeneous disease from different standpoints: etiology and pathogenesis, morphology and molecular characteristics. HDGC, hereditary diffuse gastric cancer; FIGC, familial intestinal gastric cancer; FAP, familial adenomatous polyposis.

Notably, EGJ adenocarcinoma shares many epidemiological characteristics with cancers developing in the distal esophagus and proximal stomach, and the incidence is higher in Caucasians, namely in male old patients. The incidence of EGJ adenocarcinoma increased markedly in the second half of the 20th century both in the United States and Europe, in parallel with rising incidence of adenocarcinomas of the lower esophagus.

Time Trends Worldwide, there has been a steady decline in the incidence and mortality of GC over the last 15 years. However, the absolute incidence rate continues to rise, due to the advancing age of the global population. More specifically, there has been a shift in the proportion of some subtypes: the incidence of “intestinal” carcinoma has decreased mainly in young patients, and the incidence of “diffuse” carcinoma localized to the proximal stomach is increasing. The decline of incidence has been attributed to the decrease in the prevalence of Helicobacter pylori (H. pylori) infection, better food refrigeration, reduced intake of preserved foods and salt, and a richer and more varied diet, including a higher intake of fruit and vegetables (see diet in “Risk Factors” section).

Age and Sex Distribution Age-standardized incidence rates are about twice as high in men compared to women. In general, incidence increases progressively with age in males and females. Early onset GC (EOGC) is defined as any GC presenting at the age of 45 or earlier and represents approximately 10% of all patients with stomach cancer. In some populations, susceptibility genetic factors and the infection by virulent strains of H. pylori may be important for early-onset disease. GC occurrence before the age of 30 years is very rare (1.1%–1.6%) and patients diagnosed before 20 years are exceptional. Many cases display diffuse histology and occur in HDGC families. Several studies point to a pathogenesis of EOGC different from that of sporadic cancers occurring at a later age.

Risk Factors Gastric carcinogenesis is a multistep and multifactorial process which, in many cases, appears to involve a progression from normal mucosa through chronic gastritis (chronic inflammation of the gastric mucosa), atrophic gastritis (with loss of gastric glands) and intestinal metaplasia (substitution of gastric epithelium by intestinal epithelium) to dysplasia (intraepithelial neoplasia) and carcinoma, a sequence of events that may last several years. This sequence has been designated as the Correa’s cascade of multistep gastric carcinogenesis. However, the Correa’s model does not explain the carcinogenetic pathway of all types of GC. Actually, a proportion of GCs arises in nonintestinalized mucosa and retains gastric phenotype (and gastric differentiation is also observed in gastric

80

Gastric Cancer: Pathology and Genetics

dysplasia, the ultimate precursor lesion of GC). Another pattern of metaplasia, which is believed to represent an alternative pathway to gastric neoplasia, is spasmolytic polypeptide-expressing metaplasia (SPEM), which expresses trefoil factor family 2 (TFF2) spasmolytic polypeptide and represents the metaplastic replacement of oxyntic glands by mucin secreting antral-like glands. SPEM develops in the gastric body and fundus and is strongly associated with chronic H. pylori infection and development of GC. Moreover, as occurs in GAPPS syndrome, gastric dysplasia and GC may arise in fundic gastric polyps. Together, this evidence shows that GC may also arise from the gastric epithelium.

Helicobacter Pylori Infection Among environmental factors, H. pylori infection plays a major role. Almost all noncardiac GCs develop from a background of H. pylori infected mucosa. H. pylori are Gram-negative, spiral-shaped, microaerophilic bacteria that colonize the gastric mucosa within the mucus layer and also in contact with epithelial cells. In 1994, the WHO categorized H. pylori as a group 1 carcinogen for GC, based on results of epidemiological studies that were available at that time, and later confirmed. The bacteria can be identified in gastric mucosa by routine stains, such as hematoxylin–eosin, modified Giemsa and other ancillary methods, such as Warthin–Starry staining and immunohistochemistry (Fig. 2). The infection is commonly acquired during early childhood and persists throughout adult life unless eradicated. H. pylori virulence factors that appear to influence the pathogenicity of the bacteria, as well the risk of GC development, include the cag pathogenicity island-encoded CagA, and the vacuolating cytotoxin, VacA. CagA is present in about 60%–70% of strains and is a marker for the presence of the cag pathogenicity island. This pathogenicity island encodes a type IV secretion system that allows the translocation of CagA into the host cells. CagA can activate multiple host signaling pathways leading to cell proliferation, cytoskeletal rearrangements, and disruption of cell–cell junctions. Transgenic expression of CagA was shown to lead to GC in mice. Strains producing the cagA protein that induce a greater degree of inflammation are associated with gastric precancerous lesions and a greater risk of developing cancer of the distal stomach. The gene encoding VacA is present in all H. pylori strains, but is polymorphic and varies significantly in the signal (s)-, the intermediate (i)-, and the mid (m)-regions, each with two major types. Infections with the more virulent vacA s1, i1, and m1 strains increase the risk for peptic ulcer disease and for GC, when compared with the less virulent vacA s2, i2, and m2 strains. The more virulent strains are frequently found associated with higher levels of inflammation, epithelial damage, gastric atrophy and intestinal metaplasia. Although the risk of GC in some countries of Europe and North America has been related to vacA genotype, such relationships have not been observed in eastern Asian countries, suggesting that consequences of the variation in vacuolating activity are dependent on geographical region. In East Asia, most H. pylori strains are vacA s1, i1, and m1, and are not associated with any particular clinical outcome. Corpus-predominant gastritis with multifocal gastric atrophy, and hypo- or achlorhydria is seen in approximately 1% of subjects infected with H. pylori. The acidic pH of the stomach, ranging from 1–3, works as a primary innate defense mechanism and provides an effective barrier to microbial overgrowth. Development of atrophic gastritis, which induces elevation in gastric pH, due to parietal cell loss, may lead

Fig. 2 Helicobacter pylori detected in the lumen of gastric glands and adherent to the apical pole of epithelial cells: (A) modified Giemsa, (B) Warthin Starry, (C) immunohistochemistry.

Gastric Cancer: Pathology and Genetics

81

to a shift in the composition of the gastric microbiota (the microbial community within the stomach). This event may contribute to malignant transformation, through maintenance of inflammation, and conversion of nitrates into mutagenic N-nitroso compounds. It has been demonstrated that patients with decreased acid secretion caused by H. pylori-associated atrophic gastritis, partial gastrectomy, or long-term acid suppression, especially with proton pump-inhibitors, have significant intragastric bacterial overgrowth, increased counts of nitrate-reducing bacteria, and increased nitrite and N-nitrosamine levels. This would in part explain the lack of success of H. pylori eradication in preventing GC in patients that have atrophy and intestinal metaplasia at baseline. Recent advancements in molecular techniques and computational analysis have provided evidence that the complex microbiota colonizing the gastric epithelium may influence gastric homeostasis and disease in combination with H. pylori. In support of this hypothesis, it was recently demonstrated that the microbial community of GC patients has increased nitrosating functions and genotoxic potential, compared with patients with chronic gastritis. However, the large majority (> 90%) of the individuals do not develop GC, indicating a role for other causative agents and the genetic background (Fig. 3). Host genetic susceptibility to GC involves several genes, each one however, with small effects. In this regard, several genes have been studied, namely those involved in the protection of gastric mucosa against damaging agents, in inflammatory response, in detoxification of carcinogens, in the synthesis and repair of DNA, in folate metabolism, in the regulation of gene expression and in the cell adhesion. Having all these issues in mind, it can be stated that GC (as many other, if not all diseases) is the end product of gene (genetic susceptibility)/environment interaction (Fig. 4).

Epstein–Barr Virus Infection Epstein–Barr virus (EBV) is a double-strand DNA virus that belongs to the herpes virus family (HHV4) and has been recognized as a distinct pathogenic cause of GC. EBV associated GC (EBVaGC) is considered a distinct subtype with peculiar molecular characteristics (see “Molecular Pathology and Genetics” section) and is defined by monoclonal proliferation of carcinoma cells infected by EBV, as demonstrated by EBV-encoded small RNA (EBER) in situ hybridization (Fig. 5). Distinct latency-associated patterns have been associated with specific malignancies and, in the case of GC, a latency type I or II has been found, in which EBERs, EBNA-1, BARTs, LMP-2A, and BART miRNAs are expressed. The frequency of EBVaGC varies from 2% to 20%, with a worldwide average of nearly 10%. EBVaGC occurs more frequently in male, younger patients and in the proximal stomach, or in the gastric stump of patients submitted previously to subtotal gastrectomy. EBVaGC is generally not associated with H. pylori infection. In the rare instances of coinfection (H. pylori and EBV) EBV appears to potentiate the oncogenic effects of H. pylori infection. EBVaGC may present three histological variants: (1) gastric cancer with lymphoid stroma (see “Pathology” section) (Fig. 5), in up to 50% of cases; (2) GC with Crohn’s disease-like lymphoid reaction, characterized by the presence of numerous lymphoid follicles with active germinal centers at the advancing edge of the tumor; and (3) conventional-type GC, with scant lymphocytic infiltrate, in a minority of cases. EBVaGC has frequently a favorable prognosis, which could be explained, at least in part, by the intense immune response elicited by EBV infection.

Fig. 3 Among the individuals infected with Helicobacter pylori, the large majority (>90%) of the individuals do not develop gastric cancer, indicating a role for other causative agents and the genetic background of the host.

82

Gastric Cancer: Pathology and Genetics

Fig. 4

Gastric cancer (as many other, if not all diseases) is the end product of genetic susceptibility/environment interaction.

Fig. 5 Gastric cancer with lymphoid stroma (medullary carcinoma, lymphoepithelioma-like carcinoma) is characterized by abundant lymphocytic infiltrate, with scattered lymphoid follicles (A). This morphology is frequently associated to EBV infection, as in this case, demonstrated by EBERISH (B).

Diet Certain dietary habits are associated with an increased risk of GC. These include high intakes of salt-preserved and/or smoked foods and low intakes of fresh fruit and vegetables, particularly in combination with H. pylori infection. Intake of all meat, red meat and processed meat is also associated with an increased risk of GC in the distal stomach. The so-called “Mediterranean diet” was shown to be associated with a significant reduction in the risk of incident GC; it is defined as the high consumption of fruit, vegetables, cereals, legumes, nuts and seeds, and seafood, with olive oil as the main fat source, moderate alcohol consumption (particularly red wine), a low-to-moderate consumption of dairy products and a relatively low consumption of red and processed meat.

Smoking An association has been shown between smoking and stomach cancer that could not be explained by bias or confounding factors. Smoking also potentiates the carcinogenic effect of infection with cagAdpositive H. pylori.

Genetic Polymorphisms Polymorphisms of the interleukin 1 b (IL1B) gene (initiation and amplification of inflammatory response) and the interleukin 1 receptor antagonist gene (IL1RN) (modulation of inflammation) are associated with individual (or familial) susceptibility to carcinogenesis associated with H. pylori. It has been shown that in individuals with alleles that predispose to inflammation, infection with H. pylori may cause increased production of gastric interleukin 1 b, leading to severe and sustained inflammation that increases the risk of developing GC. Other clinic-pathologic conditions that are associated with an increased risk of GC encompass autoimmune gastritis, peptic ulcer disease, hypertrophic gastropathies, gastric stump (operated stomach), and gastric polyps.

Gastric Cancer: Pathology and Genetics

83

Precursor Lesions Gastric dysplasia represents the penultimate stage of the gastric carcinogenesis sequence. It is characterized by cellular atypia, reflective of abnormal differentiation, and disorganized glandular architecture. Determination of the correct diagnosis and grade of dysplasia is critical, because it predicts both the risk of malignant transformation and the risk of metachronous GC. According to World Health Organization, dysplasia is graded in a two-tier system: low-grade and high-grade dysplasia, on the basis of architectural and cell features. The term “indefinite for dysplasia” (IFD) was coined for epithelia that were neither unequivocally dysplastic nor unequivocally nondysplastic. An alternative designation for dysplasia is intraepithelial neoplasia (IEN). Low-grade intraepithelial neoplasia/dysplasia shows minimal architectural disarray and only mild to moderate cytological atypia. The nuclei are elongated, polarized and basally located, and mitotic activity is mild to moderate (Fig. 6A). For polypoid lesions, the term “low-grade adenoma” can be used. High-grade intraepithelial neoplasia/dysplasia displays neoplastic cells that are usually cuboidal instead of columnar, with a high nucleus to cytoplasm ratio, prominent amphophilic nucleoli, more pronounced architectural disarray and numerous mitoses, which can be atypical (Fig. 6B). Increased mitotic activity per se is not pathognomonic of intraepithelial neoplasia/ dysplasia, since it is seen in regenerating epithelium as well. Importantly, the nuclei frequently extend toward the luminal half of the cell and nuclear polarity is usually lost. For polypoid lesions, the term “high-grade adenoma” can be used. In general, the diagnosis of high-grade intraepithelial neoplasia/dysplasia is more reproducible than the diagnosis of low-grade lesions. The distinction between high-grade intraepithelial neoplasia/dysplasia and well-differentiated, tubular EGC, can be very challenging, especially in biopsy material, and diagnostic criteria are slightly different for Western and Japanese pathologists. The distinction between high-grade intraepithelial neoplasia/dysplasia and well-differentiated EGC has become of limited clinical interest, since these two conditions could be managed endoscopically. Further, gastric dysplasia may present an intestinal or gastric phenotype, according to the histomorphological profile. These two phenotypes may be distinguished also on the basis of immunohistochemical expression of lineage-differentiation markers. MUC5AC and TTF1 are markers of surface gastric epithelium (foveolar cells); MUC6 and TTF2 represent markers of mucous neck cells, pyloric glands and Brunner’s glands; MUC2, CD10, and CDX2 are intestinal goblet cells differentiation biomarkers. According to their expression, four immunophenotypes may be attributed to dysplastic lesions, both low- and high-grade: intestinal (expressing MUC2, CD10, CDX2), gastric (expressing MUC5AC/TTF1 and/or MUC6/TTF2), hybrid (expressing intestinal and gastric markers) and null phenotype. Intestinal-type intraepithelial neoplasia/dysplasia generally develops in the setting of chronic atrophic gastritis with intestinal metaplasia and is considered the precursor lesion of intestinal type of GC. Dysplastic lesions harboring gastric immunophenotype may represent a divergent carcinogenesis model, compared to the conventional Correa’s cascade. Precursor lesions of the diffuse type of GC are not well characterized, except for hereditary diffuse gastric carcinoma (see “Molecular Pathology and Genetics” section).

Pathology Macroscopy Endoscopists divide EGCs into three types, on the basis of the endoscopic appearance. These are: protruded (type I), subcategorized into pedunculated (Ip) and sessile (Is); superficial (type II), and excavated (type III). Type II accounts for 80% of EGCs and is further subdivided into elevated type (IIa), flat type (IIb), and depressed type (IIc), the last being the most common.

Fig. 6

Gastric dysplasia: (A) low-grade, (B) high-grade.

84

Gastric Cancer: Pathology and Genetics

Advanced gastric cancer Advanced GC can display various gross appearances. Borrmann’s classification remains the most widely used and divides GC into four distinct types: polypoid carcinoma (type I), fungating carcinoma (type II), ulcerated carcinoma (type III), and diffusely infiltrative carcinoma (type IV) (Fig. 7). Polypoid and fungating tumors typically consist of friable, ulcerated masses that bleed easily and project from a broad base in the gastric lumen. They tend to develop in the body of the stomach, in the region of the greater curvature, posterior wall or fundus. Ulcerated tumors can differ from benign ulcers by an irregular margin with raised borders and thickened, uneven and indurated surrounding mucosa. The ulcer base is necrotic, shaggy and often nodular. Mucosal folds radiating from the crater are irregular and frequently show club-like thickening and fusion. Malignant ulcers tend to be larger than their benign counterparts. However, many malignant ulcers lack these typical features and, therefore, endoscopic appearance is not a sufficiently reliable guide to diagnosis and should be complemented by systemic biopsies. Infiltrative tumors may spread superficially in the mucosa and submucosa, giving rise to plaque-like lesions with flattening of the rugal folds. In some cases, superficial ulceration supervenes. Frequently, though, the infiltration involves the entire thickness of the wall, usually over a limited area but sometimes extensively, to produce the so-called linitis plastica or “leather bottle” stomach. In these cases, the wall assumes a stiff consistency due to an extensive desmoplastic response to tumor cells. In such cases, there is usually no visible localized growth. A characteristic feature, at endoscopy, of this tumor type is that, because of the diffuse involvement of the stomach, it fails to inflate, in marked contrast to the normal stomach. Some GCs may secrete considerable amounts of mucin, which gives the tumor a gelatinous appearance to the naked eye. These are sometimes referred to as mucinous or colloid carcinomas.

Microscopy GC is very heterogeneous from the morphological standpoint. Several histopathological classifications have been proposed, underlying such heterogeneity. As many carcinomas show morphological heterogeneity, that is, a great variety of different patterns within the same tumor, defining the predominant component could be difficult and some concerns have been raised about accuracy and reproducibility of the current classifications. The most widely accepted and used are the World Health Organization and the Laurén classifications. Two major types of GC have been described by Lauréndintestinal and diffuse typesdwhich display different clinicopathologic profiles, molecular pathogenesis and biological behavior, often occurring in distinct epidemiologic settings. Intestinal GC occurs predominantly in elderly, male patients; diffuse GC is more common in young, female patients. Both subtypes share environmental risk factors; however, diffuse GC pathogenesis is less well understood and presents a hereditary component. According to the World Health Organization (WHO), five major types of GC are recognized: tubular, papillary (Fig. 8A), poorly cohesive (with or without signet-ring cells) (Fig. 8B and C), mixed (Fig. 8D), and mucinous (Fig. 8E). Tubular and papillary carcinomas roughly correspond to the Laurén intestinal type and the poorly cohesive carcinomas (encompassing cases constituted, partially or totally, by signet-ring cells) correspond to the diffuse type in Lauren’s classification.

Fig. 7 (A) Growth patterns of advanced gastric cancer according to the Borrmann classification: type 1dpolypoid, type 2dfungating, type 3dulcerated, type 4dinfiltrative. (B) Macroscopic appearance of advanced gastric cancer according to the Borrmann classification (surgical specimens): (A) type 1dpolypoid, (B) type 2dfungating, (C) type 3dulcerated, (D) type 4dinfiltrative.

Gastric Cancer: Pathology and Genetics

85

Fig. 8 Gastric carcinoma: (A) Tubulo-papillary (intestinal type), (B) poorly cohesive, with signet-ring cells (diffuse type), (C) poorly cohesive, without signet-ring cells (diffuse type) composed of pleomorphic cells infiltrating the muscularis propria, (D) mixed, (E) mucinous, (F) adenosquamous.

Uncommon histological variants account for about 5% of GCs, encompassing: gastric carcinoma with lymphoid stroma (GCLS, see below) (Fig. 5), adenosquamous (Fig. 8F), squamous cell carcinoma, hepatoid adenocarcinoma, choriocarcinoma, small cell carcinoma, gastric carcinosarcoma, gastroblastoma, micropapillary carcinoma, gastric carcinoma of fundic gland type, parietal cell carcinoma, Paneth cell carcinoma, mucoepidermoid carcinoma, malignant rhabdoid tumor and undifferentiated carcinoma. GCLS, also described as medullary carcinoma or lymphoepithelioma-like carcinoma (Fig. 5), is composed of irregular sheets, trabeculae, poorly developed tubular structures and isolated cells, embedded within a prominent lymphocytic infiltrate, with occasional lymphoid follicles. The lymphoid infiltrate can be so prominent that immunohistochemical study may be necessary to confirm the epithelial nature of the tumor. GCLS is frequently associated to EBV infection (in over 80% of cases) and a similar morphology can be observed in GC with microsatellite instability. As demonstrated by immunohistochemical studies, cytotoxic (CD8þ) T lymphocytes constitute the predominant component of the infiltrate, which also contains B lymphocytes, plasma cells, neutrophils and eosinophils.

86

Gastric Cancer: Pathology and Genetics

Molecular Pathology and Genetics GC is the result of accumulated genomic damage affecting cellular functions essential for cancer development, the so-called hallmarks of cancer: self-sufficiency in growth signals, escape from antigrowth signals, apoptosis resistance, sustained replicative potential, angiogenesis induction and invasive or metastatic potential. Emerging hallmarks have been recently added to the list: deregulation of cellular energetics, escape from immune destruction, tumor-promoting inflammation and genomic instability/ mutations. Numerous gene mutations, somatic copy number alterations, epigenetic changes, transcriptional changes, etc. have been detected in GC, highlighting the remarkable molecular heterogeneity of this cancer. With the recent advent of next-generation sequencing, several groups have analyzed GC molecular alterations at high resolution, using various high-throughput platforms. These studies have tried to decipher such complex molecular alterations, proposing integrated molecular classification schemes, which attempt to cluster the comprehensive molecular data obtained into clinical and biological homogenous subgroups.

The Gene Expression Profiles From the Singapore-Duke Group The Singapore-Duke group used high-throughput transcriptomic technologies to identify gene expression signatures with clinical relevance. This study led to the recognition of three subtypes: (1) the mesenchymal subtype, enriched with diffuse GCs and characterized by cancer stem cell and epithelial-to mesenchymal transition (EMT) properties; (2) the proliferative subtype, with high levels of TP53 mutations, genomic instability, and activation of oncogenic pathways; and (3) the metabolic subtype, overexpressing genes that are normally expressed in gastric mucosa. Proliferative and metabolic subtypes frequently correspond to the Laurén intestinal type. New molecular profiling technologies (e.g., NanoString nCounter technology) have demonstrated the possibility of generating similar accurate molecular information from formalin-fixed and paraffin-embedded GC tissues.

The Four-Tiered Molecular Classification From the Cancer Genome Atlas The landmark study of GC molecular-based stratification was carried out by The Cancer Genome Atlas (TCGA) research network, which proposed a four-tiered molecular classification (Fig. 9) that identifies: (1) Epstein–Barr virus-positive (EBV þ) GC, characterized by recurrent PIK3CA mutations, frequent JAK2 and PD-L1 amplification, and a high level of DNA hypermethylation; (2) GC with microsatellite instability (MSI-high) characterized by DNA hypermethylation and MLH1 silencing; (3) genomically stable GC, associated with a diffuse morphology and recurrent CDH1 and RHOA events; and (4) GC with chromosomal instability (CIN), exhibiting intestinal morphology, a high number of TP53 mutations, and amplifications of tyrosine kinase receptors (TKR).

The Four-Tiered Molecular Classification From the Asian Cancer Research Group The Asian Cancer Research Group (ACRG) described four molecular subtypes with distinct prognostic impact: (1) MSI-high tumors, with an intestinal morphology and the best prognosis; (2) MSS/EMT GC, with a diffuse morphology and the worst prognosis; and (3 and 4) MSS adenocarcinomas, with no EMT signature, either TP53-active (MSS/TP53 þ) or inactive (MSS/TP53), and with an intermediate prognosis. The MSS/TP53 subtype (which roughly corresponds to the proliferative and CIN subtypes) is frequent (36%–50% of GCs) and harbors genomic amplifications of TKR and/or RAS, some of which are potential therapeutic targets. The ACRG found that TKR and RAS amplifications tended to be mutually exclusive, emphasizing GC intertumor heterogeneity and the importance of investigating molecular alterations for targeted therapy. As shown in Table 1, the three molecular classifications, described above, overlap only partially, highlighting the need for a consensual classification. In-depth studies combining molecular features with histopathological and immunohistochemical profiles may reveal interesting associations. On this note, several groups have proposed practical algorithms, based on in situ hybridization techniques, that is, EBER-ISH, and immunohistochemistry (IHC), that is, mismatch repair proteins, E-cadherin, and p53 expression, currently available in routine diagnostic practice. In these studies, the authors accomplished to translate different molecular subgroups into specific immunophenotypes with prognostic and predictive significance.

Genetic Predisposition and Hereditary Syndromes First-degree relatives of patients with GC are almost three times as likely as the general population to develop GC. This may be partly attributable to H. pylori infection being common in families and to the potential role of IL-1 gene polymorphisms. Susceptibility to carcinogens may play a role as well. For example, polymorphisms of genes encoding for glutathione S-transferase enzymes, known to metabolize tobacco-related carcinogens and N-acetyltransferase 1, increase the risk of GC development. Evidence of familial clustering is found in about 10% of GCs; among such cases, about 1%–3% GCs are truly hereditary, encompassing hereditary diffuse gastric cancer (HDGC) and familial intestinal gastric cancer (FIGC) syndromes. Despite the growing number of FIGC families, the genetic cause underlying the disease remains to be elucidated. Clinical criteria for the definition of this syndrome vary, based on the incidence of GC in the population to which the family belongs (Table 2).

Gastric Cancer: Pathology and Genetics

87

Fig. 9 The molecular classification of gastric cancer proposed by The Cancer Genome Atlas (TCGA) recognizes four molecular subtypes with distinct characteristics. EBER-ISH, EBV-encoded small RNA in situ hybridization; EBV, Epstein–Barr virus; MMRD, mismatch repair deficiency; MSI, microsatellite instability; GS, genomically stable; CIN, chromosomal instability; CIMP, CpG island methylator phenotype; GCLS, gastric cancer with lymphoid stroma; LELC, lymphoepithelioma-like carcinoma; CLR, Crohn-like reaction.

The stomach is also affected by GAPPS syndrome (gastric adenocarcinoma and proximal polyposis of the stomach) that was recently recognized as a rare variant of familial adenomatous polyposis (FAP).

Hereditary diffuse gastric cancer (HDGC) HDGC (OMIM #137215) is an autosomal dominant cancer syndrome characterized by signet-ring cell (diffuse) GC and lobular breast cancer. The genetic basis of this syndrome was discovered in 1998 by Guilford et al., who studied three Maori kindreds in New Zealand with multigenerational, diffuse GC, in which germline mutations of the CDH1 gene were identified by linkage analysis and mutation screening. CDH1 is a tumor suppressor gene, located on chromosome 16q22.1, and encodes epithelial cadherin (E-cadherin), a transmembrane glycoprotein governing the formation of adherens junctions and cell–cell adhesion. In 1999, the International Gastric Cancer Linkage Consortium (IGCLC) defined families with HDGC syndrome, on the basis of clinical criteria. Guidelines for genetic counseling and CDH1 mutational testing were updated in 2010 and 2015. The current guidelines to select patients for CDH1 testing are described in Table 2. An alternative genetically based nomenclature has been proposed, in which the term “HDGC” is restricted to families with germline mutations in the CDH1 gene. In clinically defined HDGC, CDH1 mutations are detected in 30%–40% of cases. Most (80%) are pathogenic, truncating mutations. The remainder is missense mutations, which require in silico and in vitro functional studies to elucidate the pathogenicity. In addition to point mutations, CDH1 germline deletions and promoter methylation have been found, respectively, in about 5% and 1% of HDGC families who tested negatively for point mutations. Another mechanism that has been implicated is germline monoallelic CDH1 RNA downregulation (allelic imbalance), which is found in about 37% of HDGC families. CTNNA1 gene, encoding a-E-catenin, which is also involved in intercellular cell adhesion, has been identified as the second gene whose germline alterations cause HDGC. In several familial/hereditary gastric cancer cases, germline mutations were identified in genes associated with other hereditary cancer predisposition syndromes, namely BRCA2, TP53, STK11, SDHB, PRSS1, ATM, MSR1, PALB2, INSR, FBXO24 and DOT1L. MAP3K6 and MYD88 genes have been studied in depth, but there is no evidence to consider MAP3K6 as a GC predisposing genes and the relevance of biallelic MYD88 germline mutations remain to be established. Taking all these data into account, only 16% of HDGC families are negative for germline alterations (Fig. 10). Unlike the somatic CDH1 mutations, which occur in sporadic GCs and cluster around exons 7 and 8, CDH1 germline mutations in HDGC families span the whole length of the gene and no hot spots have been identified. In HDGC, the CDH1 wild allele can be inactivated by a number of mechanisms. Most frequently, this occurs via promoter hypermethylation (epigenetic modification) and less frequently by loss of heterozygosity (LOH) and CDH1 mutations. Intragenic deletions in the wild-type allele were also reported.

88

Molecular classifications of GC

Lei Z et al., 2013a (n ¼ 248)

Bass AJ et al., 2014b The Cancer Genome Atlas (TCGA) (n ¼ 295)

Cristescu R et al., 2015c Asian Cancer Research Group (ACRG) (n ¼ 251)

EBV (9%) EBV-CIMP CDKN2A silencing PIK3A mutations PD-L1/2 amplification JAK2 amplification EBVþ cases included in MSS/ TP53þ

MSI (22%) Gastric-CIMP MLH1 silencing PIK3A mutations HER2/3 mutations EGFR mutations MSI (23%) MLH1 loss Hypermutation (KRAS, ARID1A, PIK3A) Best prognosis Intestinal GC

Mesenchymal Low TP53 mutations Low E-cadherin mRNA CSC/EMT proprieties mTOR inhibitors

Proliferative High TP53 mutations Genomic instability Oncogene amplificationd DNA hypomethylation

Metabolic Low TP53 mutations Normal gastric mucosa gene expression 5-FU sensitive

Diffuse GC GS (20%) CDH1 mutations RHOA mutations CLDN18-ARHGAP fusion (RhoAGTPase) Diffuse GC MSS/EMT (15%) CDH1 loss

Intestinal GC (intestinal Intestinal GC (gastric phenotype) phenotype) CIN (50%) High TP53 mutations TKR-RAS amplification Amplification of cell-cycle mediators Intestinal GC MSS/TP53 (inactive) (36%) MSS/TP53þ (active) (26%) High TP53 mutations Genomic instability Oncogene amplificatione

Worst prognosis Diffuse GC

Intermediate prognosis Intestinal GC

Intermediate prognosis Intestinal GC

GC, gastric cancer; 5-FU, fluorouracil; CSC, cancer stem cell; EMT, epithelial-to-mesenchymal transition; EBV, Epstein–Barr virus; CIMP, CpG island methylation phenotype; MSI, microsatellite instability; GS, genomically stable; CIN, chromosomal instability; TKR, tyrosine kinase receptors; MSS, microsatellite stable. a Lei, Z., Tan, I.B., Das, K., et al. (2013). Identification of molecular subtypes of gastric cancer with different responses to PI3-kinase inhibitors and 5-fluorouracil. Gastroenterology 145, 554–565. b Cancer Genome Atlas Research Network; Bass, A.J., Thorsson, V., Shmulevich, I., et al. (2014). Comprehensive molecular characterization of gastric adenocarcinoma. Nature 513, 202–209. c Cristescu, R., Lee, J., Nebozhyn, M., et al. (2015). Molecular analysis of gastric cancer identifies subtypes associated with distinct clinical outcomes. Nature Medicine 21, 449–456. d ERBB2, CCNE1, MYC, and KRAS. e ERBB2, EGFR, CCNE1, CCND1, MDM2, ROBO2, GATA6, and MYC.

Gastric Cancer: Pathology and Genetics

Table 1

Table 2

Clinical criteria, recommended screening and histopathological features of major hereditary syndromes affecting the stomach

Hereditary GC syndrome

Clinical criteria

- Two or more patients with GC at any age, one confirmed diffuse GC; - Individuals with diffuse GC before the age of 40; - Families with diagnoses of both diffuse GC and lobular breast cancer (one diagnosis Hereditary diffuse gastric cancer before the age of 50) (HDGC)a - Gastric polyps restricted to the body and fundus with no evidence of colorectal or duodenal polyposis; - More than 100 polyps carpeting the proximal stomach in the index case or >30 polyps in a first-degree relative; - Predominantly FDPs, some having regions of dysplasia (or a family member with either dysplastic FDPs or GC); - Autosomal dominant pattern of inheritance; - Exclusion of another gastric polyposis syndrome and use of proton-pump Gastric adenocarcinoma and proximal inhibitors polyposis of the stomach (GAPPS)b High incidence countries:

Histopathological findings

Surveillance/risk-reduction surgery

- CDH1 mutational analysis; - Search for large CDH1 rearrangements; - Search for CDH1 allelic imbalance; - CTNNA1 mutational analysis

- Diffuse GC and precursor lesions (in situ signet ring cell carcinoma, pagetoid spread of signet ring cells); - Lobular breast cancer

- Prophylactic (risk-reducing) total gastrectomy or annual endoscopic surveillance - Breast magnetic resonance imaging

APC promoter 1B mutational analysis

- Fundic gland polyposis of the stomach with areas of dysplasia; - Hyperplastic polyps of the stomach; - Adenomatous polyps of the stomach; - Mixed polyps with FGP-like, adenomatous and hyperplastic features; - Intestinal-type GC

- Endoscopic surveillance with biopsies, polipectomies or prophylactic (risk-reduction) gastrectomy

NA

Intestinal-type GC

NA

- At least three relatives with intestinal GC, one of them a first-degree relative of the other two; - At least two successive generations affected; - In one of the relatives, intestinal GC diagnosed before the age of 50 Low incidence countries: - At least two first/second-degree relatives affected by intestinal GC, one diagnosed before the age of 50; - Three or more relatives with intestinal GC Familial intestinal gastric cancer (FIGC) at any age

GC, gastric cancer; FGP, fundic gland polyp; NA, not available. a van der Post, R.S., Vogelaar, I.P., Carneiro, F., et al. (2015). Hereditary diffuse gastric cancer: Updated clinical guidelines with an emphasis on germline CDH1 mutation carriers. Journal of Medical Genetics 52(6), 361–374. b Worthley, D.L., Phillips, K.D., Wayte, N., et al. (2012). Gastric adenocarcinoma and proximal polyposis of the stomach (GAPPS): A new autosomal dominant syndrome. Gut 61(5), 774–779.

Gastric Cancer: Pathology and Genetics

Recommended genetic testing

89

90

Fig. 10

Gastric Cancer: Pathology and Genetics

Germline molecular alterations in hereditary diffuse gastric cancer.

Clinical features of HDGC The age of onset of clinically significant diffuse GC may be extremely variable (with a range of 14–85 years), even within families. Prophylactic/risk-reduction total gastrectomy (PTG) in early adulthood is the only current curative approach. Indeed, age of onset is unpredictable and, at the time of clinical presentation, affected individuals present with advanced and frequently incurable disease, in > 90% of cases. DGC frequently spreads beneath intact gastric mucosa and white light high definition endoscopy, chromoendoscopy, or endoscopic ultrasonography, have low sensitivity for detecting early HDGC. Because of the great variability in the age of onset, the optimal timing of genetic testing/PTG is still controversial. The age at which to offer genetic testing to at-risk relatives should take into consideration the earliest age of cancer onset in that family. Testing from the late teens or early 20s is favored in families with early onset GC. Current guidelines recommend that asymptomatic carriers of CDH1 mutations be offered prophylactic gastrectomy or, for selected groups, annual endoscopic surveillance, as risk-reduction strategies. Surveillance is recommended for individuals aged less than 20 years; for those aged more than 20 years who elect to delay surgery; for those for whom prophylactic gastrectomy (biopsy-negative) is unacceptable but gastrectomy with curative intent (biopsy-positive) is acceptable; and for those with mutations of undetermined significance (e.g., missense). Total gastrectomy is recommended in at-risk family members greater than 20 years of age who have a CDH1 mutation. In biopsy-positive individuals, a curative total gastrectomy is advised, regardless of age. In women, breast magnetic resonance imaging is advised, starting at 30 years of age. Histopathology of HDGC Early-stage HDGC in CDH1 mutation carriers is characterized by multiple foci of invasive (T1a) signet-ring cell (diffuse) carcinoma in the superficial gastric mucosa (Fig. 11), with no nodal metastases. Mapping of the entire gastric mucosa performed in many

Fig. 11

Intramucosal signet ring cell carcinoma.

Gastric Cancer: Pathology and Genetics

91

stomachs from kindred with different CDH1 mutations showed that there is wide variation in the number of T1a foci observed in these stomachs, both within and between kindreds, ranging from one focus to hundreds of tiny foci. The cause of this variation in number of foci is currently unknown. Discordant results have been published regarding the anatomical localization of the cancer foci. Several authors reported a proximal clustering, which is in contrast with previous reports from New Zealand Maori kindred, where most foci were found within the body-antral transitional zone and distal stomach. The cause of this variation remains to be clarified, but genetic susceptibility and environmental factors have been suggested as possible contributing factors. The signet ring cell (diffuse) carcinomas observed in asymptomatic CDH1 mutation carriers show almost always abnormal staining for E-cadherin, in keeping with a clonal origin of the cancer foci, indicating that the second CDH1 allele has been down-regulated or lost. Aberrant E-cadherin staining patterns include absence of immunoreactivity, reduced membranous expression, “dotted,” and cytoplasmic staining, in contrast to the normal membranous, complete pattern of adjacent nonneoplastic gastric mucosa. Two precursors (Tis) to T1a signet ring cell carcinoma are recognized in HDGC: (i) Pagetoid spread of signet ring cells, below the preserved epithelium of glands and foveolae but within the basal membrane (Fig. 12); (ii) in situ signet ring cell carcinoma, corresponding to the presence of signet ring cells within the basal membrane, substituting for normal epithelial cells, generally with hyperchromatic and depolarized nuclei (Fig. 13). In cases in which in situ carcinoma is identified adjacent to diffuse-type GC, genetic testing should be considered since this is rarely, if ever, seen in sporadic cases. In these lesions, E-cadherin immunoexpression is reduced or absent. Confirmation of in situ carcinoma by an independent histopathologist with experience in this area is strongly recommended. The gross and microscopic appearances of advanced (pT > 1) HDGCs and sporadic DGC are indistinguishable. Because the neoplastic cells diffusely invade the gastric wall, “linitis plastica” is the most common gross phenotype. In contrast to the early,

Fig. 12

Pagetoid spread of signet ring cells (arrowheads), below the nonneoplastic foveolar epithelium.

Fig. 13

In situ signet ring cell carcinoma.

92

Gastric Cancer: Pathology and Genetics

“indolent” HDGC phenotype, advanced cases show frequently an aggressive phenotype and are composed of poorly cohesive, pleomorphic cells, sometimes admixed with classic signet ring cells. Glandular/tubular structures, rosettes and mucinous areas can be observed, particularly in foci of lymphovascular invasion and lymph-node metastasis. The biological events leading to disease progression from early to advanced disease are poorly understood. Humar B et al. suggested that epithelial–mesenchymal transition (EMT) could mediate such progression and demonstrated the activation of c-Src kinase and downstream targets (p-Fak, fibronectin, and Stat-3) in poorly differentiated cells of intramucosal HDGC and more advanced lesions. Still, subsequent studies failed to reproduce such findings. More recently, we found a significant association between “aggressive” morphological features and aberrant (overexpressed) p53 immunoreactivity. These findings are consistent with those observed in murine models, which require inactivation of both CDH1 and TP53 genes for DGC development, and indicate that TP53 may be involved in HDGC progression. Therefore, we suggest that p53 and Ki-67 may serve as biomarkers of prognosis and progression from indolent to widely invasive lesions, and that the finding of an “aggressive” phenotype in screening biopsies of CDH1 mutation carriers should be taken as a predictive sign of widely invasive carcinoma and require prompt clinical intervention.

GAPPS (gastric adenocarcinoma and proximal polyposis of the stomach) In 2012, a new hereditary syndrome was identified: “Gastric adenocarcinoma and proximal polyposis of the stomach” (GAPPS). GAPPS is a unique gastric polyposis syndrome with a significant risk of GC, characterized by the autosomal dominant transmission of fundic gland polyposis (FGP), with areas of dysplasia or intestinal-type GC, restricted to the proximal stomach, with no evidence of colorectal or duodenal polyposis or other heritable gastrointestinal cancer syndromes. Nine families have been reported in the literature, so far. Diagnostic criteria for GAPPS syndrome are described in Table 2. The syndrome is characterized by incomplete penetrance, with a few elderly obligate carriers having normal endoscopies. GAPPS syndrome is characterized by florid gastric polyposis. The polyps (> 100) are usually < 10 mm in diameter and carpet the gastric body and fundus, with relative sparing along the lesser curve of the stomach (Fig. 14). The esophagus, gastric antrum, pylorus and duodenum are usually normal. The age of onset of widely invasive GC is variable, ranging from 23 to 75 years (median age of 50 years). Histopathology is characterized by FGPs, including areas of dysplasia (Fig. 15). Hyperplastic, pure adenomatous and mixed polyps with FGP-like, adenomatous and hyperplastic features have also been described. GCs, detected in GAPPS patients, display the features of intestinal-type and mixed GC. The genetic alteration underlined GAPPS syndrome has been recently identified: point germline mutations in promoter 1B of the APC (Adenomatous Polyposis Coli) gene (YY1 binding motif). Therefore, GAPPS syndrome has been interpreted as a variant along the broad phenotypic spectrum of familial adenomatous polyposis (FAP). Genetic testing for point mutations in APC promoter 1B is currently advised for patients fulfilling GAPPS diagnostic criteria. Upper gastrointestinal endoscopy and colonoscopy is advised in all first-degree relatives of the probands. Endoscopic surveillance with biopsies/polypectomies or prophylactic (risk-reduction) gastrectomy may be considered.

Gastric cancer in other hereditary syndromes Several inherited cancer predisposition syndromes have been associated with increased risk of GC. These include FAP (APC germline mutations), MUTYH-associated polyposis, Lynch syndrome (germline mutations in DNA mismatch repair genes), Li–Fraumeni syndrome (TP53 germline mutations), Peutz–Jeghers syndrome (STK11 germline mutations), Juvenile Polyposis syndrome (SMAD4/BMPR1A germline mutations), hereditary breast and ovarian cancer syndrome (BRCA1/BRCA2 germline mutations) and Cowden syndrome (PTEN germline mutations).

Fig. 14

Fundic gland polyposis in the setting of GAPPS.

Gastric Cancer: Pathology and Genetics

Fig. 15

93

Dysplasia in a fundic gland polyp from a patient with GAPPS syndrome.

Microenvironment Including Immune Response The tumor stroma, also referred to as tumor microenvironment (TME) is a complex, dynamic, and heterogeneous tissue, composed of extracellular matrix, vessels, nerves and numerous cell types, including fibroblasts, immune cells and endothelial cells. The inflammatory cells of the TME contribute to its complexity, by supplying also soluble molecules, including growth factors, survival and proangiogenic factors, chemokines, cytokines, metabolites, proteases and other enzymes that may facilitate progression, invasion and metastasis. Increasing evidence demonstrated the essential role of the interplay between tumor cells and the TME in sustaining tumor cell survival and behavior, tumor growth and progression, contributing to the acquisition of hallmarks traits. In addition, the TME has been associated with response to chemotherapy, affecting drug delivery and efficacy.

Cancer Associated Fibroblasts and Extracellular Matrix Cancer associated fibroblasts (CAFs) are frequently the most common cell type in TME and play a critical role in regulating the tumor–stroma interactions. Compared to normal fibroblast, CAFs present a cancer-promoting phenotype, with proprieties that resemble the fibroblasts involved in wound healing, and are regulated by protumoral growth and inflammation factors, including, among many others, platelet-derived growth factor (PDGF), TGF-b, and fibroblast growth factor-2 (FGF-2). Available evidence suggests that an increase in CAF quantity and secretion of matrix components (collagen, fibronectin, etc.) and soluble components (hypoxia inducible factor, VEGF, fibroblast activating protein, metalloproteinases, TGF-b, osteopontin, etc.) is associated to bad prognosis and resistance to chemotherapy, and explain, at least in part, the poor prognosis of diffuse GC patients, whose tumors are frequently characterized by a prominent desmoplastic reaction. To assess the prognostic and predictive significance of CAFs/extracellular matrix, recent studies have quantified, by morphometric evaluation, the proportion of intra-tumor stroma within GC specimens. A high tumor stroma percentage, defining a “desmoplastic microenvironment,” was found to be associated with worse prognosis and rapid tumor progression. By contrast, tumors characterized by high immune infiltrate and low desmoplastic response, were related to early clinical stages and favorable prognosis. A recent molecular study, based on large transcriptomic data, identify a gene expression program of stromal-related genes (“stromal super-module”), characterized by TGF-b signaling, poor patient survival and high percentages of intratumoral stroma. Such findings suggest that the morphological, morphometric evaluation of tumor stroma may be useful in the daily pathologist’s practice, as may serve as a surrogate marker for stromal gene expression and may be predictive of therapies targeting TGF-b signaling.

Tumor Immune Microenvironment The inflammatory response against tumor has been shown to play an important role in cancer development and progression. Immunosuppression has been considered a new hallmark of cancer. As such, the tumor immune microenvironment, that is, the immune cells that populate the tumor stroma and the soluble factors produced by them, is emerging as prognostic and predictive biomarkers, and, in recent years, many efforts have been made aiming to understand the complex interplay between cancer cells and the host immune system. The immune context is a double-edge sword in cancer: immunity has not only a host-protective function, by recognizing and protecting tissues from nascent tumor cells (“cancer immune-surveillance”), but also a tumor- promoting action, resulting in tumor

94

Gastric Cancer: Pathology and Genetics

“escape” from immune-surveillance. The tumor cells and the immune cells within the TME can shape the immune response and facilitate tumor cell growth, survival and metastatic potential. Both tumor-antagonizing and tumor-promoting immune cells can be found in most neoplastic lesions in different proportions, which influence the delicate equilibrium between the phase of tumor elimination or escape. Anti-tumor immune cells include both cells of the innate immune system, for example, dendritic cells, natural killer cells, and cells of the adaptive immune system, for example, tumor-specific CD8 þ effector T cells (cytotoxic T lymphocytes), CD4 þ helper T cells, and T memory cells. Tumor associated macrophages (TAM) may present different pattern of differentiation, displaying tumor-suppressing (M1) or tumor-promoting (M2) phenotypes. The M2 macrophages increase immunosuppression and angiogenesis and are associated with unfavorable outcomes. Besides M2 macrophages, a protumorigenic immune infiltrate may be constituted of myeloid derived suppressor cells (MDSCs), neutrophils, regulatory T cells (Tregs) and T helper interleukin (Th17) cells. The impairment of antigen presentation, for example, alterations of MHC complex, the decrease of costimulating signal pathways of T cells, for example, CD28 interaction with B7-1 (CD80) and B7-2 (CD86), and the aberrant increase of coinhibitory molecules of T cells, for example, CTLA-4, PD-1, PD-L1, LAG-3, TIM-3, are other important factors that may contribute to immune evasion and that may be enhanced by a TME enriched by Tregs. The systemic inflammatory response is also emerging as having a key role in cancer progression, and may influence the TME composition. Several systemic inflammatory indicators differ between GC patients and healthy controls and have been pointed as independent prognostic factors. Such parameters include preoperative values of neutrophil/lymphocyte ratio (NLR), platelet/ lymphocyte ratio (PLR), plasma fibrinogen, plasma albumin and C-reactive protein, among others. The deeper knowledge of the tumor immunobiology enabled a direct translation into immunotherapy. Antibodies blocking the interaction between PD-1 and PD-L1 have been used with promising results in several cancer types, including GC. In September 2017 the Food and Drug Administration (FDA) approved the use of pembrolizumab, an anti-PD-1 antibody, for the treatment of patients with recurrent, locally advanced or metastatic, GC/EGJC with PD-L1 expression. Although the relation between PD-L1 expression and response to PD-1/PD-L1 immune checkpoint inhibitors is not straightforward, PD-L1 expression, evaluated by immunohistochemistry, is the current available biomarker to predict response to immune checkpoint blockade. A recent study demonstrated that MSI-high solid tumors, regardless of the cancers’ tissue of origin, respond better to immune checkpoint blockade, compared to MSS tumors, and the FDA approved pembrolizumab as a treatment also for patients with unresectable or metastatic, MSI-H or mismatch repair deficient solid tumors.

Staging and Grading Staging The American Joint Committee on Cancer (AJCC) staging system defines the staging of GC. The most recent edition (Eight Edition) was published in 2017 and is described in the next paragraphs. The staging of GC encompasses the evaluation of the tumor in the gastric wall (T stage), lymph nodes (N stage) and distant organs (M stage) (Fig. 16 and Table 3). In the gastric wall, GC is staged as T1 (invasion of the mucosa or muscularis mucosaedT1a; or infiltration of the submucosadT1b), T2 (invasion of the muscularis propria), T3 (invasion of the subserosa) and T4 (penetration of the serosadT4a; or invasion of adjacent structuresdT4b). Nodal staging is based on the number of metastatic lymph nodes: N0 (no regional lymph-node metastasis), N1 (metastasis in 1– 2 regional lymph nodes), N2 (metastasis in 3–6 regional lymph nodes), N3 (metastasis in 7–15 regional lymph nodesdN3a; or in 16 or more regional lymph nodesdN3b). Adequate N staging requires that at least 16 lymph nodes to be examined and, thus, careful examination of surgical resection specimens by pathologists. The ideal number of lymph node to be evaluated is  30. If the minimum number of lymph nodes is not reached, a pN category should be attributed nevertheless. For M staging, M0 corresponds to the absence of distant metastasis and M1 to the presence of distant metastasis. It is worthy of note that the designation of pM0 must not be used: if a suspected metastatic site is biopsied and the biopsy is negative, the disease should be classified as cM0. To assign a pM1 category, histopathological diagnosis is required.

Prognostic and Predictive Biomarkers EGCs have a low incidence of vessel invasion and lymph-node metastasis and a good prognosis (about 90% of patients survive 10 years). If untreated, 63% of EGCs progress to advanced tumors within 5 years. Lymph-node metastasis occurs in 10%–20% of all EGCs: various multivariate analyses have identified (1) submucosal invasion, (2) tumor diameter > 3.0–3.5 cm, (3) the presence of vascular invasion, (4) the presence of lymphatic permeation, (5) depressed or ulcerated subtypes and (6) undifferentiated histology as independent risk factors for nodal metastasis. For patients meeting these criteria, endoscopic resection is likely to be an ineffective therapeutic modality and surgery should be considered. However, some authors claim that small intramucosal GCs of undifferentiated histology, measuring < 20 mm in size and without lymphovascular invasion have a negligible risk of lymph-node metastasis and could also be considered for endoscopic resection.

Gastric Cancer: Pathology and Genetics

Fig. 16

95

Staging of gastric carcinoma.

Table 3 T TX T0 Tis T1 T1a T1b T2 T3 T4 T4a T4b N Nx N0 N1 N2 N3 N3a N3b M M0 M1

TNM classification of gastric carcinomaa Primary tumor Primary tumor cannot be assessed No evidence of primary tumor Carcinoma in situ: intraepithelial tumor without invasion of the lamina propria, high-grade dysplasia Tumor invades lamina propria, muscularis mucosae, or submucosa Tumor invades lamina propria or muscularis mucosae Tumor invades submucosa Tumor invades muscularis propria Tumor penetrates the subserosal connective tissue without invasion of the visceral peritoneum or adjacent structures Tumor invades the serosa (visceral peritoneum) or adjacent structures Tumor invades the serosa (visceral peritoneum) Tumor invades adjacent structures/organs Regional lymph nodes Regional lymph node(s) cannot be assessed No regional lymph-node metastasis Metastasis in one or two regional lymph nodes Metastasis in three to six regional lymph nodes Metastasis in seven or more regional lymph nodes Metastasis in seven to 15 regional lymph nodes Metastasis in 16 or more regional lymph nodes Distant metastasis No distant metastasis Distant metastasis

a

Ajani, J.A., In, H., Sano, T., et al. (2017). Stomach. In Amin, M.B., Edge, S.B., Greene, F.L., et al. (eds.). AJCC cancer staging manual. 8th edn., pp. 203–220. New York: Springer.

In Japan, the 5-year survival for T2 adenocarcinoma is 60%–80% and decreases to 50% for T3 tumors. Lower survival rates have been observed in the West. Female sex and Japanese ethnicity have been associated with a survival advantage. Higher frequency of early-stage carcinomas, accurate staging and surgical expertise has also been associated with improved survival in Japan compared to Western nations. The stage of GC with special reference to extension to the serosa and lymph nodes (summarized in the TNM staging system) remains the strongest prognostic indicator. Five-year survival is 60%–80% for patients with tumors that invade the muscularis propria, but 50% for those with tumors invading the subserosa. At the time of diagnosis most patients with advanced carcinoma already have nodal metastases. Lymphatic and vascular invasion, often seen in advanced cases, carry a poor prognosis. In patients with involvement of 1–6 lymph nodes, the 5-year survival rate is 46%, compared with 30% in patients with more than seven lymph

96

Gastric Cancer: Pathology and Genetics

nodes involved. The extent of the regional lymphadenectomy performed and the quality of lymph-node evaluation are relevant. Patients undergoing a “curative” gastrectomy but limited lymph-node dissection (D1/D0) have an overall 5-year survival of only 23%, versus more than 50% for those undergoing a more aggressive lymphadenectomy (D2). In resectable cases, complete tumor removal with negative margins (R0) is important. The depth of invasion, the number of positive lymph nodes and postoperative complications are important prognostic factors. After curative resection, recurrence is loco-regional (resection margins, surgical bed and/or regional lymph nodes) in 40% of cases and systemic (liver and peritoneum) in 60% of cases. Whether distal adenocarcinomas have a better prognosis compared with proximal carcinomas is debated. Saito and colleagues reported a 5-year survival rate of 62% in patients with carcinoma of the “cardia” versus 83% for those with carcinoma of the lower third of the stomach. In another series, however, the prognoses were equally grim, with 28% and 29% survival rates, respectively.

Histological Features and Prognosis The value of histological typing in predicting prognosis is controversial. Whether the prognosis for diffuse carcinoma (Lauren classification) is or is not worse than that for intestinal carcinoma is debated. Recently, it has been suggested that diffuse carcinomas encompass lesions with different prognoses, such as a low-grade desmoplastic subtype (with no or scarce angio-lympho-neuroinvasion) and a high-grade subtype (with anaplastic cells). The prognosis is particularly bad for children and young adults with poorly cohesive carcinoma, for whom diagnosis is often delayed. Mixed GC displays more aggressive features than “pure” intestinal and diffuse GC, including larger tumor size, deeper invasion, lymphatic invasion, and lymph-node metastases. Interestingly, mixed GC shows a dual metastatic pattern, including hematogenous metastases and peritoneal dissemination with lymph-node metastases, suggesting a cumulative effect of the adverse behaviors of intestinal and diffuse-type GC. The prognostic significance of histological typing is limited, also because of the intra-tumor heterogeneity of GC, that is, the presence of distinct morphologic patterns in the tumor. Solid, trabecular, tubular and poorly cohesive components frequently coexist in the same tumor. This limitation applies particularly to biopsy specimens.

HER2 Expression/Amplification HER2 overexpression and/or amplification are found in 7%–38% of EGJC/GC, being slightly more frequent in EGJ tumors. HER2 is a transmembrane TKR of the epidermal growth factor receptor (EGFR) superfamily, which regulates cell proliferation, survival, differentiation, migration, and other cancer-relevant cellular responses. The prognostic significance of this biomarker is debated, but HER2 expression/amplification is a predictive biomarker of response to HER2-targeted therapy. In 2010, the phase III ToGA (trastuzumab for GC) trial (ClinicalTrials. gov identifier: NCT01041404) demonstrated a significant improvement in response rate, progression-free survival, and overall survival with the addition of trastuzumab to chemotherapy compared with chemotherapy alone, in prestratified patients with HER2-positive GEJC/GC. Since then, the use of trastuzumab with conventional chemotherapy has been approved by the FDA and the European Medicines Agency (EMEA) as first line therapy in advanced (defined as unresectable loco-regional, recurrent, or metastatic disease) HER2- positive GEJC/GC patients. An HER2-positive status, evaluated by IHC (3 þ) and/or fluorescent or chromogenic ISH (HER2 þ/ISH þ) (Fig. 17), is predictive of the patient response to Trastuzumab-based treatment. Compared to breast carcinoma, HER2 expression in GEJC/GC is highly heterogeneous and membrane immunoreactivity is often incomplete. Consequently, a HER2 scoring system specific for EJGC/GCs has been proposed, that is quite different from that of

Fig. 17 Expression of HER2: (A) immunohistochemistry, displaying strong immunoreactivity at the cell membrane (IHC 3þ), (B) HER2 gene amplification confirmed by chromogenic in situ hybridization.

Gastric Cancer: Pathology and Genetics Table 4

HER2 testing in gastric cancera

Score

Criteria for resection specimens

Criteria for biopsy specimens

Interpretation

0

No reactivity or membranous reactivity in < 10% of tumor cells Faint (generally visible only at 400 magnification) complete, basolateral, or lateral membrane staining in 10% of tumor cells

No reactivity in any tumor cells

Negative

Cluster(s) of at least five tumor cells with faint (generally visible only at 400 magnification) complete, basolateral or lateral membrane staining, irrespective of percentage of tumor cells stained Cluster(s) of at least five tumor cells with weak to moderate (generally visible only at 100–200  magnification) complete, basolateral or lateral membrane staining, irrespective of percentage of tumor cells stained Cluster(s) of at least five tumor cells with strong (generally visible at 25–50  magnification) complete, basolateral or lateral membrane staining, irrespective of percentage of tumor cells stained

Negative

1þ

97

2þ

Weak to moderate (generally visible only at 100– 200  magnification) complete, basolateral, or lateral membrane staining in 10% of tumor cells

3þ

Strong (generally visible at 25–50  magnification) complete, basolateral, or lateral membrane staining in 10% of tumor cells

Equivocal: perform ISH testing

Positive

ISH, in situ hybridization. a Bartley, A.N., Washington, M.K., Colasacco, C., et al. (2017). HER2 testing and clinical decision making in gastro-esophageal adenocarcinoma: Guideline from the College of American Pathologists, American Society for Clinical Pathology, and the American Society of Clinical Oncology. Archives of Pathology & Laboratory Medicine 140(12), 1345–1363.

breast cancer (Table 4). The College of American Pathologists, American Society for Clinical Pathology, and American Society of Clinical Oncology published in 2016 new guidelines for HER2 testing in EGJC/GC. Among others (see Further Reading), the authors suggest to follow these recommendations: (1) in EJGC/GC patients who are potential candidates for HER2-targeted therapy, the treating clinician should request HER2 testing on tumor tissue, preferably before the initiation of trastuzumab therapy; (2) laboratories/pathologists must specify the antibodies and probes used for the test and ensure that assays are appropriately validated; (3) pathologists should use the Ruschoff/Hofmann method in scoring HER2 IHC and ISH results; (4) pathologists should identify areas of invasive adenocarcinoma and also mark areas with strongest intensity of HER2 expression by IHC in specimens for subsequent ISH scoring when required; (5) HER2 testing on fine-needle aspiration (FNA) specimens (cell blocks) is an acceptable alternative. Given the issue of intra-tumor heterogeneity of HER2 immunoreactivity, (1) pathologists should select the tissue block with the areas of lowest grade tumor morphology in biopsy and resection specimens. More than one tissue block may be selected if different morphologic patterns are present; (2) testing of multiple biopsy fragments (from a primary or metastatic site), or from the resected primary tumor, is preferred. For biopsy specimens, current recommendations state that, when possible, a minimum of five biopsy specimens and optimally, six to eight, should be obtained.

Other Prognostic/Predictive Biomarkers Advances in genomic and molecular research have provided large data on molecular biomarkers and genetic tests with promising prognostic and predictive significance. However, despite this rapidly increasing body of evidence, the only predictive markers currently in use for EJGC/GC are HER2 and PD-L1 and no other prognostic factors, besides T, N, M categories, have been recognized for use in the clinical practice. Nevertheless, some recent findings, on molecular biomarkers, are worth of mention. MSI-H GC is diagnosed more frequently at early stages (Stage I or II) and, among the molecular subtypes defined by TGCA/ ARGC, has the best prognosis. A study demonstrated that GC patients with MSI-H or mismatch repair protein deficiency (MMRD) have better survival compared to MSS tumors, when treated with surgery alone. Conversely, MSI-H GC patients have poor survival, when treated with perioperative chemotherapy plus surgery, compared to MSS GC patients. These findings suggest that patients with MSI-H or MMRD may not benefit from perioperative chemotherapy. The study of additional cohorts will be necessary to confirm such data for the application in clinical practice. MSI-high status and PD-L1 immunoreactivity are predictors of response to monoclonal antibodies directed to immune checkpoint inhibitors (see “Microenvironment Including Immune Response” section). FDA approved Ramucirumab, a recombinant monoclonal antibody that targets the vascular endothelial growth factor receptor 2 (VEGFR2) for use as a single agent for the treatment of patients with advanced or metastatic EJGC/GC. However, no immunohistochemical/molecular biomarker is available to predict response to this targeted therapy.

Further Reading Ajani, J.A., Jeeyun Lee, J., Sano, T., et al., 2017. Gastric adenocarcinoma. Nature Reviews. Disease Primers 3, 17036. Ajani, J.A., In, H., Sano, T., et al., 2017. Stomach. In: Amin, M.B., Edge, S.B., Greene, F.L., et al. (Eds.), AJCC cancer staging manual, 8th edn. Springer, New York, pp. 203–220. Ang, Y.L.E., Yonga, W.P., Tan, P., 2016. Translating gastric cancer genomics into targeted therapies. Critical Reviews in Oncology/Hematology 100, 141–146.

98

Gastric Cancer: Pathology and Genetics

Baraniskin, A., Van Laethem, J.L., Wyrwicz, L., et al., 2017. Clinical relevance of molecular diagnostics in gastrointestinal (GI) cancer: European Society of Digestive Oncology (ESDO) expert discussion and recommendations from the 17th European Society for Medical Oncology (ESMO)/World Congress on Gastrointestinal Cancer, Barcelona. European Journal of Cancer 86, 305–317. Bartley, A.N., Washington, M.K., Ventura, C.B., et al., 2016. HER2 testing and clinical decision making in gastroesophageal adenocarcinoma: Guideline from the College of American Pathologists, American Society for Clinical Pathology, and American Society of Clinical Oncology. Archives of Pathology & Laboratory Medicine 140 (12), 1345–1363. Carneiro, F., Charlton, A., Huntsman, D.G., 2010. Hereditary diffuse gastric cancer. In: Bosman, F.T., Carneiro, F., Hruban, R.H., Theise, N.D. (Eds.), WHO classification of tumours of the digestive system, 4th edn. IARC Press, Lyon, pp. 59–63. Ferlay J., Soerjomataram I., Ervik M., et al (2013). GLOBOCAN 2012 v1.0, cancer incidence and mortality worldwide: IARC CancerBase No. 11. Lyon, International Agency for Research on Cancer. http://globocan.iarc.fr (accessed December 3, 2017). Gullo, I., Carneiro, F., Oliveira, C., et al., 2017. Heterogeneity in gastric cancer: From pure morphology to molecular classifications. Pathobiology (Epub ahead of print). Lauwers, G.Y., Carneiro, F., Graham, D.Y., et al., 2010. Gastric carcinoma. In: Bosman, F.T., Carneiro, F., Hruban, R.H., Theise, N.D. (Eds.), WHO classification of tumours of the digestive system, 4th edn. IARC Press, Lyon, pp. 48–58. Oliveira, C., Pinheiro, H., Figueiredo, J., et al., 2015. Familial gastric cancer: Genetic susceptibility, pathology, and implications for management. The Lancet Oncology 16 (2), e60–70. Shinozaki-Ushiku, A., Kunita, A., Fukayama, M., 2015. Update on Epstein-Barr virus and gastric cancer (review). International Journal of Oncology 46 (4), 1421–1434. The Cancer Genome Atlas Research Network, Bass, A.J., Thorsson, V., Shmulevich, I., et al., 2014. Comprehensive molecular characterization of gastric adenocarcinoma. Nature 513, 202–509. van der Post, R.S., Vogelaar, I.P., Carneiro, F., et al., 2015. Hereditary diffuse gastric cancer: Updated clinical guidelines with an emphasis on germline CDH1 mutation carriers. Journal of Medical Genetics 52 (6), 361–374. van der Post, R.S., Gullo, I., Oliveira, C., et al., 2016. Histopathological, molecular, and genetic profile of hereditary diffuse gastric cancer: Current knowledge and challenges for the future. Advances in Experimental Medicine and Biology 908, 371–391. Wroblewski, L.E., Peek Jr., R.M., 2016. Helicobacter pylori, cancer, and the gastric microbiota. Advances in Experimental Medicine and Biology 908, 393–408.

Genetic Instability Meiyun Guo, Lin Mei, and Christopher A Maxwell, University of British Columbia, Vancouver, BC, Canada © 2019 Elsevier Inc. All rights reserved.

Glossary Chromosomal instability Is defined as an increased rate of change in the structure or number of chromosomal segments or whole chromosomes, including amplification, deletion, loss of heterozygosity, translocation, insertion, inversion, and homozygous deletion. Chromothripsis Is a chromosomal instability phenomenon where hundreds of chromosomal rearrangements occur during one event in a localized region of one or a few chromosomes. DNA double-strand break Essentially separates the DNA double-helix molecule into two pieces. Genetic instability Refers to alterations in the DNA of tumor cells, ranging from single nucleotide mutations to changes in chromosomal structures and numbers. Microsatellite instability Is characterized by the expansion, shortening, deletion or insertion of microsatellites and is often attributed to the impairment of the mismatch repair pathway. Nucleotide instability Is an increased occurrence of mutations, including base substitutions, deletions and insertions of one or several nucleotides.

Cancer is thought to arise as a result of continuous changes to the genome resulting from random mutation and followed by natural selection applied on the cancerous cell and its progeny by the environment, including by neighboring cancerous cells, infiltrating immune cells, energy and oxygen requirements, exposure to drugs, etc. Early observations of cancer cell division by von Hansemann (1847) and Boveri (1914) indicated that cancer cells were characterized by abnormal hereditary material. Indeed, many cancers contain cells with abnormal chromosome numbers (aneuploidy) or translocations, such as the Philadelphia translocation between chromosomes 9 and 22 in chronic myeloid leukemia. Some propose that cancer is a chromosome rather than a genetic disease due to the frequency that aneuploidy is observed in cancer. Concurrently, the cancer cell genome accumulates several different genetic alterations ranging from single nucleotide variation to insertions or deletions to focal or large scale copy number alterations. It is apparent that some of this variation contributes to a greater degree to the growth and survival advantage observed in cancers; other alterations, termed passenger, may have little function on tumor development. The term genetic instability encapsulates this impressive degree of heterogeneity and the changes to this heterogeneity over time, which is observed in the cancer cell component of the tumor and also in the neighboring cells. This article will review the classification and detection of genetic instability; the cellular and tissue mechanisms for its suppression; environmental factors that induce genetic instability; the relationship between instability, the immune system and cancer progression; and, the emerging approaches to treat a heterogeneous and unstable tumor.

Classification and Detection of Genetic Instability Genetic instability refers to alterations in the DNA of tumor cells, ranging from single nucleotide mutations to changes in chromosomal structures and numbers. It is important to note that genetic instability refers to changes over time rather than the simple appearance of a mutation or an aneuploid genome. These genetic changes can be divided into the following three categories: nucleotide instability, microsatellite instability, and chromosomal instability (Fig. 1).

Nucleotide Instability Nucleotide instability is an increased occurrence of mutations, including base substitutions, deletions, and insertions of one or several nucleotides. Replication errors and malfunction in repair pathways, such as base excision repair and nucleotide excision repair, can lead to nucleotide instability. Although nucleotide instability induces small-scale changes in DNA sequence, nucleotide instability can lead to dramatic phenotypic changes, including the development of malignancy, through altered gene expression and changes to the structure or function of the affected gene products. Importantly, nucleotide instability refers to the mutation rate rather than the presence or abundance of mutations in a particular cancer. For example, by analyzing the number of mutations per cell division in human breast cancer cell lines, spontaneous point mutation rates were found to be increased (2.9X to 12X) relative to a nonmalignant mammary epithelial cell line or a colorectal cancer cell line. More recently, single nucleus gene sequencing,

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65013-4

99

100

Genetic Instability

Histone

DNA Methylation Impaired Base Excision Repair

Nucleosome Remodeling

Impaired Nucleotide Excision Repair Nucleotide Instability

Microsatellite Instability

Impaired Mismatch Excision Repair

Genetic Instability

Chromothripsis

Centrosome

Translocations Chromosome Segregation Defects Telomere Dysfunction

Chromosomal Instability

Spindle Checkpoint Defects Spindle Assembly Defects

Fig. 1 An overview of the types of genetic instability and the pathways that generate genetic instability. The yellow inner circle displays the three main types of genetic instability. The middle blue circle displays some processes that result in the three types of genetic instability. The outermost green circle represents the epigenetic modifications that affect DNA repair pathways.

in combination with mathematical modeling, found cells comprising a triple negative breast cancer demonstrated an increased mutation rate (13.3X) relative to normal cells and this breast cancer exhibited extraordinary genetic diversity in isolated tumor nuclei.

Microsatellite Instability Microsatellites, or short tandem repeats, are DNA repeats consisting of one to six base pairs that occur throughout the genome. The distribution of microsatellites in eukaryotic genomes is not random and they display different properties in different genomic regions. Microsatellite instability is characterized by the expansion, shortening, deletion, or insertion of microsatellites and is often attributed to the impairment of the mismatch repair pathway. Changes to microsatellite sequences are thought to occur during DNA replication due to polymerase slippage that is inefficiently mended due to a deficient mismatch repair response. About 15% of colorectal cancers exhibit deficient mismatch repair and microsatellite instability and the inheritance of mutations in mismatch repair genes, such as MLH1 and MSH2, can predispose a hereditary cancer predisposition syndrome (Lynch Syndrome) that associates with an elevated risk to develop nonpolyposis colon cancer.

Chromosomal Instability and Chromothripsis Chromosomal instability is highly prevalent in human cancers with about 90% of tumors displaying chromosomal abnormalities. Chromosomal instability is defined as an increased rate of change in the structure or number of chromosomal segments or whole chromosomes, including amplification, deletion, loss of heterozygosity, translocation, insertion, inversion, and homozygous deletion. The consequence of chromosomal instability can be severe owning to the large-scale alterations that can affect the expression

Genetic Instability

101

of thousands of gene products and, as a result, can dramatically alter cancer cell phenotypes that enable progression or endow intrinsic multidrug resistance. Therefore, many cellular mechanisms are dedicated to the preservation of chromosome stability, including DNA repair pathways, telomere regulation, and checkpoints to ensure mitotic spindle assembly and chromosome segregation. Chromothripsis is a chromosomal instability phenomenon where hundreds of chromosomal rearrangements occur during one event in a localized region of one or a few chromosomes (Fig. 2). An analysis of 746 cancer cell lines revealed that more than 2%– 3% of cancers display massive genomic rearrangements on one chromosome. It is proposed that the massive rearrangements occur from one single chromosome fragmentation event followed by faulty DNA repair that joins the fragments together. DNA sequencing analysis of a patient with chronic lymphocytic leukemia revealed some characteristics of chromothripsis. First, chromothripsis occurs at specific locations of the genome while rearrangement events that characterize conventional genetic instability are assumed to occur randomly across the genome. Second, the copy number at many regions in an individual chromosome changes between one or two copies, and regions with only one copy are not generated simply by deletion but through rearrangements. Third, the breakpoints at the chromosome arm are clustered so that multiple rearrangements take place within a narrow region, and the fragments of the chromosome connected at breakpoints are originally located distal from each other. Since the locations of breakpoints are clustered, it is plausible that chromothripsis occurs during chromosome condensation related to mitosis. The occurrence of chromothripsis is especially high in bone cancers, but the process is not restricted to a specific tumor subtype.

Detection of Genetic Instability As illustrated above, genetic instability occurs across multiple genetic levels. Thus, methods that measure changes in chromosomes, microsatellites, or nucleotides are adequate to measure a component of genetic instability. Such methods include, but are not limited to, karyotyping, flow cytometry, single nucleotide polymorphism arrays, whole-genome sequencing, and polymerase chain reaction. However, the measurement of the mutation rate or the change of the mutation rate over time can be challenging. Changes to the mutation rate or karyotype can be determined for immortalized cell lines and tumor evolution can be inferred by the examination of multiple sites of the same tumor or the progression of a tumor over time; that is, from diagnosis through minimal residual disease to metastasis or first relapse or second relapse. In addition, a variety of methods are often used to calculate the frequency and extent of genetic changes for static tumor cell populations, which are frequently used as a surrogate to describe genetic instability. Intratumor heterogeneity is one surrogate marker of genetic instability. Recent developments in the field of single-cell sequencing now resolve cell-to-cell genetic heterogeneity at a base-pair level within a tumor and through tumor evolution. That is, comparative analysis of temporally and spatially distinct regions in a single tumor has revealed clonal dynamics and intratumor heterogeneity. Single-nucleus-sequencing is a method that combines flow-sorting, whole genome amplification and next-generation sequencing to produce genome-wide datasets from single cancer cells. Based on data obtained from this new approach, a punctuated model of chromosome evolution is now postulated, which challenges the model of gradual chromosome evolution. Data from singlenucleus-sequencing imply that chromosome alterations occur at initial stages of carcinogenesis. This method has also uncovered

Ionizing Radation

Chromosom end-to-end fusion

Abortive Apoptosis

Single catastrophic event

DNA repair pathways join some fragments

Chromosome Instability

Pulverization of chromosomes in micronuclei

Chromosome condensation in S phase

Lost

Fig. 2 Chromothripsis leads to chromosomal instability and is often found in aggressive tumor cells. There are several proposed insults that can induce chromothripsis, including ionizing radiation acting on condensed chromosomes, chromosome end-to-end fusion caused by telomere shortening that leads to a double strand break, abortive apoptosis, pulverization of chromosomes in micronuclei, and chromosome condensation that occurs in S phase. These insults may cause a single catastrophic event which fragments specific regions of a chromosome. During the subsequent DNA repair, some of the chromosomal fragments are lost and some are rearranged.

102

Genetic Instability

large numbers of subclonal and de novo mutations that were identified to occur at low frequency (< 10%). This illustrates perhaps that no single tumor cell is genetically identical to another in the analyzed primary tumors. Such radical diversity in an evolving tumor context illustrates the remarkable plasticity with which a tumor can respond to a variety of challenges, including those presented by neighboring cancerous cells, infiltrating immune cells, energy and oxygen requirements, and exposure to therapy.

Mechanisms That Suppress Genetic Instability Genetic instability can occur across many genetic levels, ranging from base-pair changes, to focal or large scale amplifications or deletions, to whole chromosome gains and losses. The cell possesses multiple mechanisms to dampen the development of an unstable genome including: processes that ensure the replication of the genome occurs with high-fidelity; processes that recognize and repair mutations when they arise; and, processes that ensure cell division results in genetically identical daughter cells.

High-Fidelity DNA Replication The replication of DNA must occur with high fidelity and, therefore, is guarded by many factors. A detailed review of DNA replication is beyond the scope of this article. Briefly, the fidelity of DNA replication depends upon the quality of the template, the selectivity and proofreading of the polymerase, the supply of nucleotides during the process and, finally, the repair of errors by the mismatch repair process. DNA lesions in the template can present challenges to the replicative DNA polymerase and affect the selection of the incoming nucleotide or activate a checkpoint. The MTH1 protein can remove some oxidized dNTPs that may present difficulties to the polymerases. Three DNA polymerases are responsible for eukaryotic DNA replication: DNA polymerase a (Pol a), DNA polymerase d (Pol d), and DNA polymerase ε (Pol ε). Theses enzymes individually form a complex with the DNA template and bind a nucleotide and two metal ions in the active site of the enzyme, which induces a conformational change in the polymerase (an induced fit mechanism) that imparts high selectivity to these enzymes (about 1 error per 104 to 105 nucleotides). Proofreading by DNA Pol d and Pol ε, which contain an exonuclease activity, and mismatch repair further improves replication fidelity. It is proposed that due to the combination of these processes the error rate for DNA replication is 1 error per 109–1010 nucleotides.

DNA Repair Pathways DNA repair pathways are important safeguards to maintain genetic stability and distinct pathways are responsible for the recognition and repair of different types of DNA damage. Nucleotide excision repair. Nucleotide excision repair can act on a wide range of DNA lesions with different structures, such as lesions caused by ultraviolet radiation, intrastrand crosslinks, chemical adducts, and reactive oxygen species-generated cyclopurines. Nucleotide excision repair occurs during two lesion detection mechanisms: global genome nucleotide excision repair and transcription-coupled nucleotide excision repair. Global genome nucleotide excision repair searches for disturbed base pairing and distortions of nucleotide structures whereas transcription-coupled nucleotide excision repair is triggered by the stalling of RNA polymerase II due to a lesion on the DNA template strand. Deficiency in the process of global genome nucleotide excision repair, such as in Xeroderma pigmentosum (XP), can lead to the accumulation of lesions in the whole genome and result in cancer predisposition. For example, as a consequence of deficient global genome nucleotide excision repair, XP cells are very sensitive to ultraviolet radiation and XP patients have a higher risk of developing skin cancer than normal individuals. Base excision repair. Reactive oxygen species produced from cellular metabolism are believed to underlie the majority of endogenous DNA damage and inappropriate repair of these oxidative lesions can induce oncogenic mutations. The base excision repair pathway detects and repairs DNA damage caused by oxidation, deamination, and alkylation. Base excision repair is initiated by a DNA glycosylase, which detects and cleaves the targeted base, generating a baseless site (AP site) that is edited by short-patch repair or long-patch repair. Importantly, AP sites are highly prone to mutagenesis and, as a consequence, genetic stability depends upon efficient and appropriate base excision repair. DNA mismatch repair. A DNA mismatch refers to a non-Watson–Crick nucleotide base pair in double strand DNA. DNA mismatch repair is believed to reduce the amount of DNA errors 100 to 1000-fold. Mismatch repair begins with the detection of mismatched DNA by MutS, an essential protein in mismatch repair that activates downstream repair proteins. With the recruitment of MutL a, strand excision is initiated in a PCNA-, RFC- and ATP-dependent manner. Finally, DNA Pol d or Pol ε and DNA ligase 1 complete the synthesis of the strand. Deficient mismatch repair is common in colorectal cancers (15%), endometrial cancer (30%), and hereditary ovarian cancer (10% to15%). Double-strand break repair. A DNA double-strand break essentially separates the DNA double-helix molecule into two pieces. This is frankly more severe than a single strand break wherein the genetic information is still preserved on one intact strand; thus, doublestrand breaks can lead to more severe consequences, such as fragmentation and alterations of chromosomes. Double-strand break repair mechanisms can be divided into nonhomologous end-joining and homology-directed repair. Nonhomologous end-joining can occur in all cell cycle phases whereas homology-directed repair requires a homologue and thus only occurs during S- and G2phases of the cell cycle. Double-strand break repair is a critical tumor suppressive mechanism and deficiencies in these pathways, as the result of mutation to key components such as BRCA1, can causes genetic instability and cancer development. For example, female carriers of causal BRCA1 mutations are more highly prone to develop breast and ovarian cancers and those tumors tend

Genetic Instability

103

to display a higher degree of aneuploidy and chromosomal rearrangements, which are potentially indicative of an elevated rate of genetic instability.

Epigenetics and Genetic Instability While genetic instability refers to the rate of change of alterations encoded in the genome, epigenetic alterations impact not only the expression of molecular regulators of stability but also change the compaction and access to DNA, which directly influences the stability of the genome. DNA methylation. DNA methylation refers to the attachment of a methyl group to the 50 carbon of a cytosine ring, usually at CpG dinucleotides, which will alter (often turn off) the transcription of the affected genes. In cancer, the regulation of DNA methylation is impaired and can lead to hypomethylation at repeated sequences throughout the genome and hypermethylation at CpG islands. CpG islands contain 60% of the gene promoters and the hypermethylation of some CpG islands can lead to the reduced expression of tumor-suppressor genes. For example, a hypermethylated promoter region of the mismatch-repair gene MLH1 causes microsatellite instability. On the other hand, a loss of overall DNA methylation is an epigenetic characteristic of cancer. Hypomethylation of tandem repeats at the centromere can decondense heterochromatin and enable genomic rearrangements. Whereas retrotransposons Alu and LINE-1 elements are silenced by DNA methylation in normal tissues, the overall hypomethylation that characterizes cancer can lead to their elevated activation, which, in the case of LINE-1 elements, can lead to their insertion into the genome and the induction of deletions or inversions. Histone modification. Histone modifications, such as acetylation and methylation, can either activate or repress gene transcription. Histone acetylation occurs on lysine residues with an N-terminal tail extending from the nucleosome and can induce a chromatin conformation that is more accessible for protein binding. DNA double-strand break repair, for example, is very reliant upon protein access to DNA: histone deacetylase (HDAC) 1 and HDAC2 are recruited and regulate acetylation of histone H3 at sites with DNA damage. Consequently, a deficiency in HDAC1 and HDAC2 will impair double-strand break repair and can induce genetic instability. While histone methylation can either upregulate or suppress transcription depending on specific sites of methylation, SETD2dependent H3K36 trimethylation regulates DNA mismatch repair while the lysine-specific histone demethylase 5B (KDM5B) enables DNA double-strand break repair. Consequently, cancers with impaired SETD2/HYPB methyltransferase activity may display chromosomal aberrations. Nucleosome remodeling. The structure of chromatin is dynamic and can have either an “opened” or “closed” conformation, which regulates the accessibility of DNA segments and relies upon the arrangement of nucleosomes. The nucleosome remodeling process is vital to DNA double-strand break repair. Briefly, double-strand break repair requires: (1) the detection of DNA damages in chromatin structure; (2) access to the DNA damage site; (3) a proper organization of nucleosome and DNA for repair process; and, (4) restoration of the chromatin structure after repair. Nucleosome remodeling is needed for each of these processes and it is suggested that, as a consequence, DNA repair is less efficient in heterochromatin, which may lead to the accumulation of mutations in certain heterochromatin regions.

Cell Cycle Checkpoints DNA damage is induced by many different factors, such as ultraviolet light and oxidizing agents, and these damages can trigger a pause, or checkpoint, in cell cycle progression to allow DNA repair pathways to recognize and repair the damage and prevent the damage from being inherited by progeny cells. The detection and repair of a double-strand break in G1 phase relies upon Ataxia Telangiectasia Mutated (ATM) kinase to phosphorylate and activate Chk2 kinase, which induces a G1-phase checkpoint by stabilizing p53 and, through cyclin-dependent kinase (Cdk) inhibitor p21, blocking cell cycle progression. During S-phase, DNA replication forks stall at genetically fragile sites leading to the binding of replication protein A and the recruitment of Ataxia Telangiectasia and Rad3-related (ATR) kinase, which activates Chk1 to degrade Cdc25A and induce a S-phase checkpoint. ATM and ATR oversee double-strand break repair pathways and can also arrest cells at G2-phase or during the transition from G2phase into mitosis. Mutations in ATM predispose a tumor susceptibility syndrome with heightened risks to develop lymphomas and certain types of solid tumors, including breast cancer. Chromosomal instability is suppressed by a critical checkpoint during cell division, termed the mitotic spindle assembly checkpoint. Mitosis is a process that ensures the equivalent segregation of the cell’s genetic material to two daughter cells. Briefly, this process relies upon the construction of a microtubule-based structure, termed the mitotic spindle, and the coordinated actions of microtubule-associated molecular motor proteins and their adaptors. Microtubules are largely nucleated from two poles, termed centrosomes or spindle poles, as well as near to the chromosomes themselves, termed kinetochore (K)-fibers. These microtubules are aligned and organized into a bipolar structure that contacts each chromosome and forms a bipolar attachment at the kinetochore on each chromosome. A complex of proteins located at the kinetochores monitors microtubule attachment and the establishment of interkinetochore tension, which is a cue that helps fulfill the spindle assembly checkpoint and promotes the transition to anaphase. For example, Aurora kinase B is localized to the inner centromere and will phosphorylate components of kinetochore KNL/Mis12 complex/Ndc80 complex network to destabilize kinetochore-microtubule attachments in the absence of proper interkinetochore tension. Upon bipolar attachment, tension will stretch the kinetochores and separate the outer kinetochore components from Aurora kinase B. Deficiency in the spindle assembly checkpoint pathway, as induced for example through mutation of BUB1B, induce a tumor susceptibility disorder termed mosaic variegated aneuploidy.

104

Genetic Instability

Centrosome Amplification and Clustering During cell division, it is vital that each chromosome attains bipolar attachment. However, almost all types of solid tumors display centrosome amplification, which is characterized by an abnormal number or size of centrosomes with, or without, an increased number of centrioles. Centrosome amplification can cause multipolar spindles, which can result in mitotic failure, mitotic death or aneuploidy. For instance, transgenic mice designed to express widespread centrosome amplification, through induced PLK4 overexpression, exhibit an increased tendency to generate lymphomas, squamous cell carcinomas, and sarcomas by 35 weeks of age. Cancer cells seem to have developed a mechanism to cluster supernumerary centrosomes in order to avoid catastrophic multipolar cell division. Cells with supernumerary centrosomes undergo a MAD2-dependent delay before anaphase and tension on spindle microtubules is proposed to cluster centrosomes to form pseudo-bipolar spindles. Human cancer cells often overexpress proteins involved in the centrosome clustering process, which may permit these cells to bypass potential mitotic catastrophe that may result from recurrent multipolar cell divisions.

Telomere Length Telomeres are G-rich repetitive sequences (TTAGGG) located at the end of chromosomes that maintain the integrity of replication and genetic stability by protecting the ends of chromosomes from degradation and chromosomal end fusion events. Telomeres are shortened gradually by chromosome end-processing during each cell division and telomere maintenance is required for ongoing cell division. Although telomere shortening prevents transformed cells from proliferating, it can also lead to chromosome breakage and instability. In some cancer types, mutations in TERT (encoding the catalytic subunit of telomerase) promoter regions were found to activate transcription suggesting that telomerase activation occurs during tumorigenesis. Moreover, longer telomere lengths may be associated with an increased risk to develop non-Hodgkin lymphoma or chronic lymphocytic leukemia. Telomere crisis can induce fusion of telomeres and lead to dicentric chromosomes, chromosome translocation, regional amplification, and eventually genetic instability.

Environmental Causes of Genetic Instability In addition to reactive oxygen species produced by cellular metabolism, exogenous agents, such as X-rays, ultraviolet light, and various chemicals, can cause genetic changes that can promote cancer. DNA adducts, resulting from chemical carcinogens or ultraviolet radiation, are sensed by Xeroderma pigmentosum group C (XPC), which initiates the global genome nucleotide excision repair pathway to eliminate damage and maintain genome stability. However, additional precautions that reduce exposure will also dampen the initiation of genetic instability (Fig. 3).

Radiation Ultraviolet radiation can penetrate tissue and cause biological damage. Ultraviolet radiation induces crosslinks between neighboring pyrimidines creating cyclobutane pyrimidine dimers and (6-4) photoproducts (6-4PPs). However, most organisms possess a very rapid, highly efficient, and safe enzymatic tool for the repair of photolesions in the form of DNA photolyases that, by

Ionizing Radiation

Apoptosis Chemical Carcinogens DNA Damage

DNA Repair Pathways

Accumulation of Mutations

Virus exposure Tumorigenesis Diets Fig. 3 External causes of genetic instability. Ionizing radiation, chemical carcinogens, exposure to virus and diets are the major mechanisms that cause DNA damage. In the presence of impaired DNA repair pathways, mutations in the DNA can accumulate and lead to programmed cell death or tumorigenesis.

Genetic Instability

105

light-driven catalysis, revert cyclobutane pyrimidine dimers or 6-4PPs to pyrimidine monomers without excision of bases or any deoxyribose-phosphate residues. It is believed that skin cancer, the most common cancer, is at least in part the consequence of ultraviolet damage. In melanoma, the majority of mutations are cytosine to thymine (C> T) at adjacent pyrimidine residues, which reflects the generation of thymine dimers after exposure to ultraviolet light. Protective barriers, including shade, ultravioletattenuating sunglasses or topical sunscreens, are an effective way to reduce exposure and damage. Ionizing radiation is suggested to cause 1% to 3% of cancers and exposure to ionizing radiation can arise from cosmic and terrestrial radiation, radon, or medical applications. Like X-rays and gamma rays, ionizing radiation carries enough energy to move electrons from an atom. Ionizing radiation can cause direct DNA damage in tissues, including in the germline, which can induce birth defects and tumorigenesis.

Chemical Carcinogens Genotoxic chemicals react directly with DNA while nongenotoxic carcinogens interfere with pathways that are critical for DNA repair or chromosome separation. Exposure to either of these classes of chemicals can induce DNA lesions and chromosome rearrangements. Examples of genotoxic chemicals, many of which are present in cigarettes, include polycyclic aromatic hydrocarbons (PAHs), N-nitrosamines, aromatic amines, aldehydes, benzene, and 1,3-butadiene. Exposure to PAHs, for example, can give rise to high levels of both 8-OHdG in urine and DNA damage detected by the alkaline comet assay. In addition, high levels of PAH–DNA adducts are seen in tobacco smokers that developed lung cancer. Exposure to benzene, a solvent historically used in printing inks and gasoline, may result in leukemogenesis, including acute myeloid leukemia, acute lymphocytic leukemia, chronic lymphocytic leukemia, and myelodysplastic syndrome. Exposure to nongenotoxic carcinogens, such as heavy metals, acrylamide, bisphenol A, or benomyl, disrupt DNA repair pathways, DNA damage signaling and chromosome segregation pathways and therefore impair genetic integrity. While it is not possible to absolutely avoid exposure to many of these chemicals, it is critical to limit one’s exposure as completely as possible.

Virus Human papillomaviruses (HPVs) are small DNA tumor viruses that can transform squamous epithelia. HPVs replicate their own genomes by reprogramming the host cell’s DNA replication machinery. The high-risk HPV E6 and E7 oncoproteins target negative growth regulatory signaling in the host cell to enable viral genome replication in these postmitotic cells. HPV E6 and E7 oncoproteins disturb the normal regulation of cell division and centrosome duplication, which disrupts chromosomal integrity of the host cell and promotes the proliferation of HPV infected cells. Vaccines against HPV, including HPV-16 and HPV-18, may significantly reduce the prevalence of HPV-induced cancers. In addition, HPV-associated cancers may be combated through the enhancement of host immune responses and more effective antitumor immunity, as these cancers are believed to result in part from immunosuppressive factors. Combinatory therapy, partnering HPV vaccination with immune enhancement, may alter the pro-tumor inflammatory environment and suppress HPV-associated tumorigenesis. Hepatitis C virus (HCV) infection is highly associated with hepatocellular carcinoma, one of the most common malignancy. HCV replicons can endow host cells with a variety of cancer stem cell-like traits. HCV proteins can also increase the production of reactive oxygen species by enhancing mitogenesis, blocking cell death, or inducing chronic inflammation. The oxidative stress caused by high levels of reactive oxygen species, and the accumulated DNA damage caused by increased cell proliferation and inflammation, may induce genetic instability that acts as the basis for HCV-mediated carcinogenesis.

Diet Diet is an important environmental factor that influences cancer risk and may modulate genetic stability. There are several mechanisms by which diet can influence genome stability. First, folate plays a key role in several processes related to DNA integrity, such as DNA synthesis and methylation. In vitro studies have shown that folic acid deficiency causes increased uracil incorporation in human lymphocyte DNA. Folate administration reduces DNA uracil incorporation and the presence of chromosome breaks in human cells. A low folate concentration has also been implicated as a potential promoter of tumorigenesis in, for example, colorectal cancer, lung, breast, pancreatic, gastric, esophageal, and prostate malignancies. Second, niacin, or nicotinic acid, is one of the few vitamins that has an important role in DNA synthesis, DNA repair, and cell death. Niacin is required as a substrate for poly(ADP-ribose) polymerase 1 (PARP1), which is involved in DNA repair and telomere length maintenance. The consequence of niacin deficiency is an increase in the number of DNA breaks and in the rate of chromosome damage. Epidemiologic evidence indicates that intake of foods that are naturally rich in vitamin C is associated with reduced risk of various cancers, but the extent to which vitamin C contributes to this effect remains unclear. In general, low dietary intake of folate, nicotinic acid, calcium, vitamin E, retinol, and b-carotene and high intake of pantothenic acid, biotin, and riboflavin are associated with increased genetic instability indicating the importance of a balanced diet to the maintenance of genetic integrity. Finally, a greater understanding of the interconnection between diet, gut microbiota, the immune system, and cancer (including responses to immune-based treatments and immune surveillance) is an emerging and exciting field for future research.

106

Genetic Instability

Genetic Instability and Cancer Progression Many human cancers arise in epithelial tissues and these cancers would be easily treated through surgery were they to remain in their tissues of origin; unfortunately, carcinomas and other human cancers progress and move to other sites in the body, termed metastasis. It is speculated that metastasis accounts for up to 90% of the deaths associated with human cancer. It is also speculated that the process of metastasis involves numerous selective pressures on tumor cells, including: (1) growth at primary site, which involves the escape from antitumor immune responses and intrinsic antiproliferative pathways in their premetastatic niche; (2) invasion of the surrounding parenchyma, lymphatics or blood vasculature; and, (3) survival and proliferation at the metastatic site. Studies of anatomically distinct metastatic lesions in 29 prostate cancer patients have found close clonal relationships between different metastatic lesions within the same patient, which implies the metastases share a monoclonal origin with the primary tumor. However, a high degree of genetic divergence, based upon radically different patterns of allelic loss, was found between primary prostate tumors and lymph node metastases as well as primary breast tumors and asynchronous metastases. Distinct clonal evolution, as assessed by comparative-genomic hybridization, was also evident in a subset of primary breast or renal cell tumors versus metastases. Moreover, the comparison of mutations present in primary and secondary lobular breast cancers revealed multiple mutations present only in metastases and several other mutations with increased frequency in metastatic sites. Thus, while metastases retain some genetic linkage to primary tumors there is ample evidence for genetic divergence, which may be the result of a heterogeneous primary tumor cell population, generated through genetic instability, overcoming the multiple selective pressures applied through the metastatic process. The immune system plays a significant role in pruning the growth of primary tumors and maintaining dormancy of metastasized tumor cells. However, tumor variants with reduced immunogenicity will overcome the immune system, which may promote mutated characteristics that escape immunological detection and elimination. This process, termed immunoediting, has three phases: elimination, equilibrium, and escape. Elimination is the initial phase during which the immune system reacts to and initiates the destruction of transformed cells. When the tumor remodels the local stroma and disrupts the tissue’s structure, pro-inflammatory molecules and chemokines produced during this process will recruit innate immune cells to the tumor site. Innate immune cells produce IFN-g and generate a positive feedback loop that then recruits more immune cells. This inflammation and recruitment of immune cells will lead to apoptosis, angiostasis, and cell cycle inhibition that partially eradicates the tumor. Moreover, dead tumor cells displaying tumor antigens will further activate antitumor responses from the adaptive immune system. Thus, these molecules, IFN-g, perforin, or NKG2D, and an intact lymphocyte compartment apply a major selection pressure on the tumor cells. During this period of Darwinian selection, many of the original tumor cells will be eliminated, but cells that are genetically unstable and are able to mutate rapidly may withstand immune attack, survive and establish an equilibrium with the immune system. Tumor cells that survive the elimination phase may equilibrate with the host immune system and undergo the equilibrium phase, wherein the remaining tumor population may be sculpted by the immune system. In particular, lymphocytes and IFN-g will apply pressures that can amplify genetic instability in the tumor population and enable further adaption, including the promotion of tumor cells with reduced immunogenicity. Such beneficial mutations may occur in tumor cells during the equilibrium phase, which can last for a long time with no obvious sign of tumor development. Eventually, surviving tumor cells that have acquired genetic changes will resume expansion and escape from immunologic detection and elimination. The surviving tumor cells need to escape from both innate and adaptive immune responses. So, tumor cells often escape antitumor immune responses through the production of immunosuppressive cytokines or through immunosuppressive activities of T cells. Interestingly, there is another aspect of the escape phase, termed “reverse immunoediting.” Here, the tumor cells may affect immunological response through tumor antigens. That is, the immune system may select for the reduction of strong tumor antigen presented in the context of major histocompatibility complex (MHC) 1 or MHC2, which will favor the development of tumors that are more resistant to future immune responses. In addition, tumor cells actively build an immunosuppressive microenvironment by the generation of cytokines that summon a different population of cells. Taken together, genetic instability in the tumor population enables plasticity in response to the negative selection pressure applied through the immune response, and this plasticity may enable immune suppression and tumor progression.

Cancer Therapies That Target Genetic Instability Intratumoral heterogeneity enabled by genetic instability presents a great therapeutic challenge given the plasticity with which numerous clones can respond, or evolve, depending upon therapeutic pressures. Alternatively, genetic instability is a phenotype that is likely exclusive to tumor cells and, thus, represents a unique therapeutic target.

Targeting DNA Repair Pathways Tumors can display a variety of defects in the pathways needed to recognize and repair DNA. These defective repair responses can augment the heterogeneity of the tumor clones but can also increase the tumor cell’s susceptibility to inhibition of compensating pathways. That is, a tumor cell with a defective homology-directed repair capacity may be more heavily reliant upon the

Genetic Instability

107

nonhomologous end-joining pathway; this is the basis for synthetic lethality approaches to target tumor susceptibilities due to dampened or dysfunctional repair pathways. BRCA1 and BRCA2 are essential for high-fidelity repair of DNA double-strand breaks through a homology-directed pathway. Cells with deficiencies in BRCA1 or BRCA2 have a decreased rate of homologous recombination and increased sensitivity to ionizing radiation. These observations suggest that tumor cells deficient in BRCA1 or BRCA2 may be more heavily reliant upon PARP1-directed repair and, therefore, susceptible to PARP1 inhibition. Recent phase 3 trials have been extremely encouraging for the use of PARP inhibitors against BRCA-mutated, platinum-sensitive, relapsed ovarian cancer with two companies reporting median progression free survival of approximately 20 months compared to 5.5 months for the control arm. It is proposed that patients with breast cancer may also benefit from PARP inhibition. Although only about 5% of patients carry mutations in BRCA1 or BRCA2, nearly a quarter of breast cancer patients may carry tumor mutational signatures that support the use of these drugs. These encouraging results represent a leap forward following a series of disappointing clinical trial results published in 2011 and 2012. However, acquired resistance to PARP inhibition has been observed in most patients with advanced tumors, which is a common attribute of targeted therapies. Indeed, the PARP inhibitor olaparib may also increase genetic instability in mouse embryonic stem cells leading to translocations and chromosome aberrations by increasing DNA double-strand breaks. With three PARP inhibitors approved for use against ovarian cancer, however, it is an exciting time for the evaluation of combining PARP inhibition with conventional treatments, immunotherapy, or antiangiogenesis treatments for the treatment of genetically unstable and difficult-to-treat, aggressive cancers. A similar synthetic lethality approach is proposed for the directed inhibition of MutT homologue 1 (MTH1), which regulates the incorporation of oxidized nucleoside triphosphates into DNA or RNA during replication and transcription. The level of reactive oxygen species is generally increased in cancer cells and results in not only more direct damage to DNA but also the incorporation of more damaged deoxynucleoside triphosphates. Consequently, cancer cells may be more heavily reliant upon the action of MTH1 and inhibitors of MTH1 may more effectively damage tumor cells than neighboring normal cells.

Targeting Microsatellite Instability Hereditary nonpolyposis colorectal cancer is often characterized by high-level microsatellite instability (MSI-H) and an associated increased antigenicity (markers on cell membrane that can bind to antibody). The high antigenicity of MSI-H tumors makes them recognizable by the immune system. In order to overcome host immune defense mechanisms, MSI-H carcinomas express some immune checkpoint molecules to either turn up or down a signal in the immune system and create a localized immunosuppressive environment. For example, tumor cells make proteins to bind to the programmed cell death protein 1 (PD-1) receptor on lymphocytes, which dials down the immune system’s response to self and impairs the immune response to the tumor. In this context, checkpoint inhibitors, like pembrolizumab, that inhibit PD-1 on the lymphocytes may release the immune system’s ability to target and kill cancer cells. Pembrolizumab has shown an objective response rate of 40% to 50% in patients with progressive mismatch repair-deficient metastatic colorectal cancer compared to 0% in patients with mismatch repair-proficient cancer. This remarkable specificity comes with a cost as these checkpoint inhibitors will have the potential for off-target side effects on the immune system. Moreover, while the blockade of the PD-L1/PD-1 axis is a proven immunologic approach, not all patients will respond to monotherapy. The combination of a checkpoint inhibitor and chemotherapeutic regimens may provide an improved therapeutic index; this combination has been shown to be well tolerated without unexpected toxicities and more clinical trials are ongoing.

Targeting Gene Expression of Cell Cycle Components As outlined in “Cell Cycle Checkpoints” section, cell cycle checkpoints are intended to suppress genetic instability and are often compromised in cancer cells. Moreover, in spite of an unstable genome, tumor cells continue to proliferate without control; thus, a variety of cell cycle regulators are considered potential targets in cancer therapy. There are two types of cell cycle control mechanisms: (1) a cascade of protein phosphorylation that transfers a cell from one stage to the next and, (2) a set of checkpoints that monitor completion of critical events and delay progression to the next stage, when necessary. Cyclin-dependent kinase (CDK) activity controls many critical events during cell cycle progression. Activation of these kinases generally requires a complex with a second subunit that is only expressed at the appropriate period of the cell cycle; the periodic cyclin subunit associates with its partner kinase to generate an active complex with unique substrate specificity. Regulatory phosphorylation and dephosphorylation control CDK–cyclin complexes and ensure clear transitions between cell cycle stages. Silencing individual cyclins or CDKs, or inhibition of cyclin–CDK kinase activity, in tumor-bearing mice selectively blocks tumor initiation and progression of specific cancer types without having major effects on normal tissues. Various CDK inhibitors (dinaciclib, palbociclib with PD-0325901) have shown encouraging activity against tumors that are known to be characterized by unstable genomes, such as relapsed myeloma or KRAS mutant nonsmall cell lung cancer. The second type of cell cycle regulation, checkpoint control, is more supervisory. Checkpoint kinase 1 (CHK1) and WEE1 regulate the activation of p53, which inhibits CDKs and arrests the cell cycle. Cancer cells, particularly those with loss of p53 function, often inhibit CHK1 or WEE1 to enable cell cycle progression in the presence of DNA damage; these cancer cells mainly depend on the G2 checkpoint, especially in the presence of DNA damage-inducing drugs. For this reason, inactivation of p53 selectively renders cancer cells sensitive to inhibition of CHK1 or WEE1. For example, a CHK1 inhibitor in combination with radiation is able to kill

108

Genetic Instability

p53-defective tumor cells by inhibiting double-strand break repair. An alternate CHK1 inhibitor, alone or in combination with a WEE1 inhibitor, shows activity against solid tumors. Overall, these results support possible clinical trials in the future.

Centrosome Amplification Centrosome amplification is a recurrent characteristic in tumor cells. Real-time imaging of different cell types in vitro and in vivo indicates that supernumerary centrosomes tend to cluster at the poles of mitotic spindles to induce a pseudo-bipolar spindle and prevent multipolar division. Extra centrosomes assemble in small clusters during prometaphase resulting in two ring-shaped groups at the poles. However, should supernumerary centrosomes fail to cluster, the resulting multipolar spindles often trigger mitotic failure, death or apoptosis. Because supernumerary centrosomes are rare in normal cells, the inhibition of pathways used by tumor cells to cluster centrosomes would specifically target tumor cells with potentially minimal off-target effects. Crenolanib, an inhibitor for platelet-derived growth factor receptor b, inhibits centrosome clustering and induces multipolar divisions with increased cytotoxicity by cofilin-mediated disruption of the cortical actin cytoskeleton for cells with centrosome amplification. A small-molecule screen for inhibitors of centrosome clustering identified a Stat3 inhibitor; Stat3 is a regulator of gene transcription that removes a Stathmin-dependent brake on polo-like kinase 1 activity to promote centrosome amplification. Stat3 depletion and inhibition caused significant inhibition of centrosome clustering and reduced the viability of cancer cell lines and tumors grown in animal models.

Personalized Medicine and Genetic Instability The US National Cancer Institute defined personalized medicine as “a form of medicine that uses information about a person’s genes, proteins, and environment to prevent, diagnose, and treat disease”. Such personalized clinical applications include screening, diagnosis, prognosis, prevention and disease-risk reduction, prediction of treatment response, pharmacogenomics and dose adjustments, and the early detection of recurrence and stratification of patients into specific cancer subtypes for more targeted treatment. Many cancer treatments, such as radiotherapy and chemotherapy, are mostly nonselective with strong side effects. In comparison, targeted therapies are hoped to be more specific and the survival rate for some cancers has increased from 0% to 70% due to more developed targeted treatments. In the era of personalized medicine, existing drugs and drugs used for other diseases can be reintroduced for targeted treatments on subgroups or subtypes of tumors. Individualized treatments are heavily reliant upon accurate measurements of protein, DNA, RNA, and microRNA levels that, in turn, determine the tumor subclasses, prognosis, and treatment responses. Treatment dose and safety in different subtypes of cancer patients can be determined or predicted by genetic variation in the tumor. For example, high-grade serous carcinoma of the ovary displays chromosomal instability that is caused by an impaired homologous recombination repair pathway. Analysis of loss of heterozygosity in these tumors divides patients into different subgroups, which corresponds with different levels of chemotherapy resistance and progression-free survival. Similarly, the level of microsatellite instability in tumors of patients with colon cancer correlates to the efficacy of fluorouracil-based adjuvant chemotherapy. Genetic instability presents a great challenge to the introduction of targeted therapies and personalized medicine. One of the difficulties is the identification of valid biomarkers for different types of cancers and their subtypes, which is especially difficult for low frequency events. Biomarkers are critical for the development of targeted therapy since they may reveal information of individual response to treatments. An accurate quantification of targeted treatment for one patient not only requires broad sample collection but also needs enormous resources for the analysis, ranging from academic specialists to effective data processing systems. Tumor heterogeneity and selective tumor sampling may provide a biased biomarker analysis and detrimentally affect the application of treatment or the treatment response. Moreover, the need for and generation of high-content data (e.g., whole genome sequencing) and its associated analysis and application in clinical use is still not cost effective. Standardization of testing is also very difficult when many factors need to be considered, including methods of specimen collection and storage, types of specimen, biomarker selection, the platform for experiments and the interpretation of data. Although intratumor heterogeneity is a challenge to treatment prediction by biomarkers, it is still possible to take advantage of this feature of tumors by giving sequential therapies. Drug resistance is one of the most difficult challenges in cancer therapy development. Often, a population of cancer cells will develop different adaptation mechanisms against the prescribed drug depending on the targeting mechanism and the inherent genetic instability of the tumor. However, there is a cost to tumors cells that develop an adaptation mechanism. That is, tumor cells will need more resources to sustain the development of resistant clones. Thus, there may be a limitation to the number of adaptation mechanisms the tumor population can carry out. It is possible to use this limitation in the design of more effective combinatory targeted treatments. The evolution of resistance against a specific drug in a tumor population can be modeled for three different types of therapies, including: (1) monotherapy; (2) multidrug therapy where only one adaptation is required to establish resistance; and (3) multidrug therapy that targets different mechanisms. In general, resistance to a drug will increase when the tumor cells are on treatment and will decrease when the tumor cells are off treatment. The tumor cells can become completely resistant to the drugs within on–off cycles. This observation led to the proposal by Cunningham and colleagues of an “evolutionary double blind therapy” that theoretically will use the first line of therapy to drive the tumors to a specific adaptation form which will be targeted by the second line of therapy. Since the tumor cells theoretically invest more resources to adapt to the first drug, the ability of the tumor population to resist the second drug may be reduced and the efficacy of the second line of therapy may be increased. This adaption response may

Genetic Instability

109

be informative in another avenue; that is, there is likely an optimal level of instability that affords resistance to cytotoxic therapy. As an example, breast tumors with severe chromosomal instability have a better prognosis than those with moderate chromosomal instability. Thus, drugs that can augment chromosome instability, such as those that decluster centrosomes, may push the tumors into a suboptimal level of instability and make these tumors more vulnerable to cytotoxic therapy.

Perspective Vision Genetic instability is often associated with cancer and can be indicative of a poor prognosis for some tumor subtypes. But, is genetic instability a consequence of tumor progression or a driving force for tumor evolution? Is genetic instability an early and essential development in tumorigenesis and, if so, what tumor cell-intrinsic or environmental pathways restrict its consequences prior to the observation of an overt tumor? The answers to these questions have still not been completely resolved. Intriguingly, the most common form of genetic instability observed in cancer, aneuploidy, remains in many respects to be the most mysterious. A central goal of ongoing research is to systematically identify the many genetic changes that occur in the cancer cell genome and to understand how these functionally interact to cause cancer cell phenotypes. This effort will provide new insight into the genetic landscape of cancer cells, and the novel driver mutations that are discovered may have a significant impact on future cancer therapy development. In summary, as genetic instability is an enabling attribute of cancer and a modifier of sensitivity and resistance to therapy, our improved understanding of the mechanisms that generate an unstable genome and the pathways that can alter the resulting phenotypes will provide new insights into the origins of cancer subtypes and new directions for their treatment.

See also: Cell Responses to DNA Damage.

Further Reading Cunningham, J.J., Gatenby, R.A., Brown, J.S., 2011. Evolutionary dynamics in cancer therapy. Molecular Pharmaceutics 8, 2094–2100. Ferguson, L.R., et al., 2015. Genomic instability in human cancer: Molecular insights and opportunities for therapeutic attack and prevention through diet and nutrition. Seminars in Cancer Biology 35 (Suppl), S5–S24. Ganai, R.A., Johansson, E., 2016. DNA replicationdA matter of Fidelity. Molecular Cell 62 (5), 745–755. Korbel, J.O., Campbell, P.J., 2013. Criteria for inference of chromothripsis in cancer genomes. Cell 152 (6), 1226–1236. Krokan, H.E., Bjoras, M., 2013. Base excision repair. Cold Spring Harbor Perspectives in Biology 5 (4), a012583. Leibowitz, M.L., Zhang, C.Z., Pellman, D., 2015. Chromothripsis: A new mechanism for rapid karyotype evolution. Annual Review of Genetics 49, 183–211. Lord, C.J., Ashworth, A., 2017. PARP inhibitors: Synthetic lethality in the clinic. Science 355 (6330), 1152–1158. Price, B.D., D’Andrea, A.D., 2013. Chromatin remodeling at DNA double-strand breaks. Cell 152 (6), 1344–1354. Vogelstein, B., Papadopoulos, N., Velculescu, V.E., Zhou, S., Diaz Jr., L.A., Kinzler, K.W., 2013. Cancer genome landscapes. Science (New York, N.Y.) 339 (6127), 1546–1558. Wang, Y., Waters, J., Leung, M.L., Unruh, A., Roh, W., Shi, X., Chen, K., Scheet, P., Vattathil, S., Liang, H., Multani, A., Zhang, H., Zhao, R., Michor, F., Meric-Bernstam, F., Navin, N.E., 2014. Clonal evolution in breast cancer revealed by single nucleus genome sequencing. Nature 512, 155–160.

Genome Wide Association Studies (GWAS) Giuliana Restante, Elena Arrigoni, Marzia Del Re, Antonello Di Paolo, and Romano Danesi, University of Pisa, Pisa, Italy © 2019 Elsevier Inc. All rights reserved.

Glossary Allele One of several forms of a gene arised through mutation that control a specific characteristic or phenotype. Humans, being diploid organisms, have two alleles at each genetic locus, with one allele inherited from each parent. Cancer susceptibility variants Pathogenic variants responsible of the predisposition to cancer. Based on their role in biochemical and physiological pathways of tumor cells, cancer susceptibility variants may be common or rare, with high or low penetrance. Usually, variants of genes with a potent effect on apoptosis, proliferation or DNA repair processes have a moderate penetrance and are rare. On the contrary, variants of genes that play a role in molecular and cellular processes controlling the interaction between genetic and environmental components have a low penetrance and are common. Cancer phenotypes associated with high-penetrance susceptibility variants tend to be influenced by the hereditary background, while lowpenetrance susceptibility variants could also modify the behavior of hereditary cancer. Effect size The strength of association between a SNP and risk to develop a specific disease. Genetic locus A specific position or spot on a chromosome, where the allele is located. Genetic variant or mutation A change in the DNA sequence of a particular gene with harmful, beneficial, neutral, or uncertain effect on health. It may be inherited as autosomal dominant or recessive trait, or X-linked trait, if the phenotypic effect of the variant is caused when a copy of the altered gene is located on X chromosome. Genotyping The analysis of the exact sequence of nucleotides, to identify changes in the target sequence compared to the reference sequence that is the most common among genomes in the general population. Odds ratio (OR) The ratio of the odds of disease for individuals having a specific allele and the odds of disease for individuals who do not have the same allele. Pathogenic variant or predisposing mutation, or susceptibility variant The genetic alteration that increases the risk to develop deleterious symptoms associated with a certain disease or disorder. Pathogenic variants inherited as autosomal dominant trait that limit life expectancy and reproduction are usually rare in the population (prevalence < 1%). Instead, the pathogenic variant inherited as autosomal recessive trait, may be relatively common in the general population (prevalence > 1%), because subjects carrying one copy of the altered gene are healthy people without serious disability early in life. Penetrance The likelihood that in an organism with a particular genotype the corresponding phenotype may or may not be expressed. In particular, the expression of traits with an autosomal dominant pattern of inheritance may be altered if modifier or suppressor genes exist in the rest of the genome or because of a modifying effect of the environment. Thus, penetrance is defined as the percentage of individuals with a given genotype who exhibit the phenotype associated with that genotype. High penetrance would refer to a Mendelian disease. Polygenic A trait controlled by the interaction of many genes. In contrast to Mendelian disease in which mutations in a single gene cause a specific phenotype and for which the pattern of inheritance is very clear, a polygenic disease is controlled by two or more different genes located at different loci on different chromosomes, each one usually having a relatively small effect. Risk assessment The quantitative or qualitative assessment of an individual’s risk of carrying a certain gene mutation, or developing a particular disorder done by using mathematical or statistical models incorporating selected factors such as personal health history, family medical history and ethnic background. SNP Single nucleotide polymorphism, a kind of genetic variant characterized by a change of one nucleotide in the sequence of DNA.

Nomenclature CNV Copy number variations GWAS Genome-wide association studies LD Linkage disequilibrium SNPs Single nucleotide polymorphisms OR Odds ratio RR Relative risk

110

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65106-1

Genome Wide Association Studies (GWAS)

111

Introduction In the last years several progresses in mapping complex traits in humans have revealed that, for prevalent and common diseases, the predominant pattern of hereditability is controlled by the interaction of many loci, with small effect on phenotype if taken individually, in addition to the interactions between inherited genetic and environmental factors. Advances in both genetic tools, including the mapping of single nucleotide polymorphisms (SNPs), comparative genomic hybridization (CGH) and gene expression microarrays, together with the cross-species approaches, allowed the characterization of loci and genes critical for risk prediction, diagnosis, prevention, and therapy of human diseases. Moreover, genome-wide association studies (GWAS) have detecteddwith a strong statistical significancedhundreds of genetic variants associated with a large number of diseases, through the comparison of SNPs in large numbers of affected cases compared to healthy controls. In particular, GWAS, also known as whole genome association study (WGAS), is an observational study of a genome-wide set of multiple genetic variants in different individuals, able to understand if any variant is associated with a specific trait. As reported above, these studies are focused on the association existing between one or more SNPs and characteristic features of major human diseases (Fig. 1). In contrast to methods that specifically test a small number of preselected genetic loci, GWAS studies have the capability to investigate the entire genome, using a noncandidate-driven approach, and to identify SNPs and other DNA variants associated with a specific disease. Before the advent of GWAS, linkage and candidate gene studies had limited power to detect common genetic risk factors of diseases. In order to overcome the major limitation of linkage studies, and to improve the resolution owing to the limited number of meiosis within families, the candidate gene studies ignore many causal genomic region or genes because they focus on variants in specific genes that have a priori biological involvement in disease. Instead, in GWAS, allele frequencies at thousands of polymorphic sites (i.e., SNPs) are compared in a large number of cases versus a similar number of controls. These studies have successfully identified some of the common susceptibility variants for different common diseases and traits, including cancer. The first GWAS was published in 2005, on patients affected by age-related macular degeneration. Nowadays, it is possible to count over 3000 human GWAS that have examined over 1800 diseases, including several tumors, and thousands SNP association have been discovered. During last years, many novel associations between genetic variants or chromosomal aberrations and diseases were detected by GWAS, but most of these are not strongly predictive of disease occurrence, due to the low penetrance and limited public health impact, as well as the lack of validated genetic testing. The reason why, for complex traits studied in humans, the identified genetic effect comprises less than half of the estimated trait heritability, is related both to the role of the untested rare variants and the understanding of gene–gene and gene–environment interactions requiring extremely large sample sizes and well-characterized environmental exposures. Interestingly, one might consider genetic tests based on combinations of associated SNPs and evaluate if the inheritance of a panel of SNPs can reveal a higher risk to develop the disease, as often occur for cancer. As sample sizes increase, GWAS will also be able to detect additional SNP associations that have even smaller effect sizes than those observed to date.

Steps of GWAS: From Study Design to Genetic Analysis and Result Interpretation Around 10 million of the human genome’s 3 billion nucleotides are polymorphic SNPs, meaning that they occur in two or more common forms in different subjects. GWAS reveals genetic variations that are correlated with disease, but this does not mean that the variants identified by the studies cause the disease. The idea is that if a genetic variant increases the risk to develop a disease, it should be more frequent among cases than among healthy controls (Fig. 2). Thus, the genotyping of both cases and control subgroups, could uncover regions of the genome that harbor DNA changes leading to the disease, but the functional role of the identified genetic variants in disease needs to be tested. Common questions in the area of translational medicine are: how do the mutations that underlie GWAS alter the way that genes, protein, cells, or tissue function? How do this knowledge can inform the search for new medicine targeting the molecular basis of the disease? (Fig. 3). Sequencing of the human genome together with the identification of millions of SNPs, provided the initial foundation for GWAS. In the scientific literature it has been reported that many of these identified SNPs were in linkage disequilibrium (LD) with the common genetic variations. It is known that, in order to detect with statistical power the expected modest associations (e.g., odds ratio [OR] < 1.5) between a genetic variation and a specific disease, it is needed to involve thousands of subjects and to evaluate hundreds of thousands of SNPs. Nevertheless, the large sample sizes and initial high cost of SNP arrays motivated the development and use of multistage GWAS designs. Using a multistage GWAS design, it is possible to discover the strongly associated SNPs in a subset of samples, using the genomewide SNP array and subsequently validating these SNPs with a less-expensive genotyping platform in the remaining samples (Fig. 1).

112

Genome Wide Association Studies (GWAS)

Fig. 1 Steps of genome-wide association study (GWAS). In a multistage GWAS, a large number of individuals affected by a specific type of genetic disease, such as cancer, and a suitable healthy group as a comparison, are selected. In phase I, the entire genome of each subject of a small group of cases and controls are sequenced in order to identify common genetic variants in the form of single nucleotide polymorphisms (SNPs). Stringent statistical methods are used to assess the associations between SNPs and disease. In phase II, the SNPs found to be significantly associated with risk to develop the disease, are tested in a larger group of patients and controls using arrays containing the putative SNPs of interest. The association found with this approach could have clinical utility being possible the use of the biomarker to perform a genetic test to facilitate clinical decision making and improve health outcomes.

Concomitantly with the increasing understanding of the human genome, an enormous number of SNP arrays able to capture an ever-increasing number of variants were developed, permitting to cover a high number of common genetic variations across the human genome. In the last years, the use of multistage GWAS has permitted to significantly reduce the cost of these analyses. Indeed, while the array prices were initially high, at present, they have steadily decreased, because fewer SNPs are typed in a second stage. Since the data from the first and second stages are combined, analyzed, and penalized for multiple comparisons, the single-stage GWAS has been undertaken. Today, the decreasing SNP array costs have made multistage GWAS designs less essential. Indeed, genotyping 10,000–20,000 SNPs in a follow-up stage can cost just about as much as a genome-wide SNP array. The newly developed GWAS have permitted to genotype all samples in an initial SNP array, allowing to analyze simultaneously both SNPs and other forms of genetic variation in one study sample (e.g., copy number variants, CNV), but increasing the cost of analysis and encouraging, when possible, the use of multistage design.

Genome Wide Association Studies (GWAS)

113

Fig. 2 Genetics, high-throughput genotyping technology and GWAS. GWAS is an “hypothesis-generating” tool. The researchers, through an agnostic genome-based approach and high-throughput genotyping technologies, interrogate the entire human genome to reveal whether SNPs that arise in the population are associated with a specific disease. In this figure, the green colored are the most common nucleotides in the general population, while the others are polymorphic; only those in red represent the genetic variants that increase the risk to develop a disease, and are more frequent among cases colored in black than in controls colored in white. In 2005, as an evolution of the HapMap project, the cost and the resources required to test a DNA sample for the bulk of its common variations became less expensive. The International Hap Map Consortium provides a map of SNPs across the entire human genome that are often inherited together in DNA blocks, called haplotypes. The so called “SNP chips” are cost-effective genotyping arrays produced by several companies that allow the researchers to test a single sample for hundreds of thousands of SNPs.

The primary platforms that have been used for most GWAS are the Affymetrix technology (Santa Clara, CA) that use a chip with short DNA sequence able to detectdby means of hybridizationdthe specific allele contained in the sample (Fig. 2) and the Illumina system (San Diego, CA) that prints a slightly longer DNA sequence on the surface of beads in order to detect the specific allele with somewhat better specificity.

114

Genome Wide Association Studies (GWAS)

Fig. 3 Central goals of human genetic, translational medicine and the role of GWAS. GWAS plays a central role in human genetic revolution thanks to the identification of robust associations between thousands of genetic variants and hundreds of different traits and diseases. This revolution provides not only an improvement in the understanding of common complex diseases in translational field, but also in clinical and pharmacologic areas where the use of genomic informations contributes to the optimization of patient treatment. In the figure, subjects with genetic variants associated with neither drug toxicity nor pathogenic effects, are represented in white, near a double stranded DNA associated with a green star representing a neutral variant of the DNA sequence. Instead, subjects with a genetic variant associated with an alteration of drug metabolism and toxicity are represented in black, near a double stranded DNA associated with a red star representing the susceptibility variant. The identification of patients carrier of a genetic variant strongly associated with toxicity to a drug (represented as blue circles) allows the clinicians to reduce the dose to minimize the adverse effects to drugs.

The prerequisite to provide a GWAS is to compare two large groups of individuals, comprising selected cases affected by a particular disease and healthy subjects as controls. Generally, all individuals in each group are genotyped for the majority of common known SNPs (Figs. 1 and 2). In order to increase the power to detect associations, it is possible to select those with the extreme traits or symptoms instead to study the entire group of subjects. Despite study subjects should be representative of their source population and rigorous control selection should be made, in order to reduce the costs of subject recruitment and genotyping, it is possible to use existing genotype informations among controls derived from previous studies and made available to researchers.

Genome Wide Association Studies (GWAS)

115

Once genotyping is complete, SNPs are subject to a number of quality-control checks and statistical analysis. The relationship between SNPs and disease is generally evaluated trough combinations and interaction tests and haplotypes. For each of analyzed SNPs it is then investigated if the allele frequency is significantly altered between the case and the control group, evaluated by using the OR. Additionally, a P-value for the significance of the OR is typically calculated using a simple chi-squared test (the conventional threshold is 5
10 8 to be significant in the face of hundreds of thousands to millions of tested SNPs). The effect size is in inverse relationship with sample size: studies with larger sample size have higher power than those with smaller sample size, being able to detect smaller associations. Thus, thanks to GWAS, SNPs in genes or genome regions that were not been previously identified as implicated in the initiation and development of the disease, as well as the genetic variants involved in cell adhesion, signal transduction, transport activity, and protein phosphorylation, could be identified. The GWAS findings are published in the National Human Genome Research Institute’s (see the web site of “Catalog of Published Genome-Wide Association Studies”). In the last years, many of the GWAS results have been highly replicated; nevertheless, few of these variants have been described to be associated with disease, although their causative role is not firmly established. A fine mapping and mechanistic studies could determine the pathogenetic implication of GWAS results, even if this could be complicated by the fact that  30% of the associations detected to date are not even in gene regions, posing a challenge to the understanding of the biological basis of GWAS results, and to the implementation of preventive or therapeutic measures.

Implications and Limitations of GWAS Findings The findings emerged from GWAS widely increase the interest into a mechanistic work that is aimed at validation of the GWAS data and clarification of the biological basis of disease processes. For example, in order to understand the association between a specific SNP in a given locus with risk to develop a different type of cancer, the in vitro approaches could reveal important molecular and biochemical information. If of clinical interest, one or more SNPs or a panel of variants, could became part of a genetic test prescribed by a physician or marketed directly to the consumer that pay to have information about variations in the entire genome and genetic counseling about the risk to develop a disease or to benefit from a treatment. Interestingly, in the last years, the pharmacogenetic studies evaluating the genetic basis of drug response, have widely used the GWAS approaches (Fig. 3). Indeed, starting from these tests, it is possible to obtain pharmacogenetic information related to the drug and dosage that an individual with a specific pattern of genes involved in drug metabolism should receive. However, it is important to note that the justification for genetic testing also depends on the existence of effective intervention. Most of these screening tests based on GWAS SNPs may not distinguish between individuals with low and high risk of disease, and, even if some individuals are carriers of a large number of SNPs, their screening in the global population would not be cost-effective. The use of GWAS have several issues and limitations that should be taken into account for proper quality control and study setup. Lack of well defined case and control groups, insufficient sample size, control for multiple testing and for population stratification are common problems. In addition, the high number of statistical tests performed have been reported and present a potential for false-positive results.

The Role of GWAS in the Identification of Cancer Susceptibility Genes Cancer is a highly heterogeneous disease, in terms of time of development and biological properties of each tumor. Among individuals exposed and susceptible to the same carcinogen, tumors do not appear at the same time. The origin and evolution of cancer have been proposed to be the consequence of the combined effects of a number of different alleles and variants, which may control intrinsic (i.e., apoptosis, proliferation, or DNA repair) and/or extrinsic (i.e., stroma, angiogenesis, or endocrine and immune systems) functions of cancer cells. GWAS have demonstrated that much of heritable risk of most common cancers is polygenic. Thus, every susceptibility variant contributes to a small amount of risk and cancer susceptibility originates from the additive effects of combinations of common lowpenetrance variants. Studies focused on the analysis of linkage in family pedigrees comprising of several members affected by certain cancers allowed the detection of several high-penetrance susceptibility genes: BRCA1 and BRCA2 for breast and ovarian cancers; APC, MLH1, and MSH2 for colorectal cancer; CDKN2a for melanoma. Despite the linkage studies have described candidate loci containing cancer susceptibility genes, often these loci were not statistically significant due to the low number of families affected by each locus and to the regions of interest; it is therefore advisable to replicate and confirm findings in larger populations. While family studies involve few patients, GWAS involve thousand of patients, thereby significantly increasing density markers. Anyway, these two approaches can be complementary, because the pedigree evaluation, relying on genetic transmission of diseasecausing alleles between affected family members, can offer direct and persuasive sign of genetic effects. As described above, in the last years, multiple GWAS have been reported for each of the major cancers, including breast, prostate, lung, colorectal, pancreatic, gastric, renal, and bladder cancers. In addition, GWAS have been performed for malignant melanoma,

116

Genome Wide Association Studies (GWAS)

ovarian cancer, glioma, and several hematological malignancies including acute lymphocytic leukemia (ALL), chronic myeloid leukemia (CML) and Hodgkin’s lymphoma. Additionally, common risk alleles have been identified through GWAS associated with several pediatric solid cancers.

Breast Cancer Regarding the studies focusing on family members affected by breast cancer, the inherited variations in the two main susceptibility genes, BRCA1 and BRCA2, account together for only around 20% of hereditary breast cancer, elucidating < 5% of the total breast cancer susceptibility. This encouraged extensive efforts to discover additional high penetrance susceptibility genes. With the advent of GWAS, two classes of variants that have an influence on cancer susceptibility have been detected: rare moderate-penetrance variants and common low-penetrance variants. For example, through the direct interrogation of candidate genes encoding proteins involved in the DNA-damage response pathway, in addition to BRCA1 and BRCA2, also ATM, CHECK2 and PALB2 have been associated with an increased breast cancer risk. The first breast cancer GWAS was published in 2007 by Easton and colleagues. The Authors, using a two-stage GWAS analyzing 227,876 SNPs in 4398 breast cancer cases and 4316 controls, and followed by a third stage in which 30 SNPs were tested for confirmation in 21,860 cases and 22,578 controls, they identified five novel risk loci in FGFR2 [rs2981582, rs1219648, and rs1078806], TNRC9 [rs3803662], MAP3KI [rs889312], and LSP1 [rs3817198] genes, with a strong and consistent evidence of association with breast cancer (P < 10 7; OR ranging from 1.05 to 1.41). In particular, FGFR2 and MAP3K1 are members of the RAS/RAF/MEK/ERKsignaling pathway and could be important tumor markers of breast cancer risk. In particular, it has been showed that FGFR2 rs1219648 and rs2981582 genotypes were significantly associated with breast cancer in estrogen receptor-positive (ERþ), progesterone receptor-positive (PR þ) and HER2/Neu-negative (HER2 ) tumors. Some of the other susceptibility loci identified in gene-containing regions have not been implicated in cancer previously (e.g., TOX3, STXBP4, SLC4A7, MRPS30, RAD51L1, LOC643714, and STXBP4) and are being evaluated for their potential role in carcinogenesis. Nevertheless, the risk associated with each identified SNPs is modest, with ORs ranging from 1.1 to 1.4, and with a contribution to the familial risk of breast cancer of about 8%.

Prostate Cancer With the increasing incidence of prostate cancer, the identification of common genetic variants that confer risk of developing the disease has become important. In a recent meta-analysis of 87,040 individuals, it have been identified 23 new susceptibility loci for prostate cancer (P < 5
10 8), with ORs mostly ranging from 1.2 to 2.0. It has been also demonstrated that this loci in genes like GOLPH3L (rs17599629, 1q21), NEDD9 (rs4713266, 6p25), HLA-DRB6 (rs115306967, 6p21), CDKN2B-AS1 (rs17694493, 9p21), TTC9 (rs8014671, 14q24), SLC41A1 (rs1775148, 1q32), TMPRSS2 (s1041449, 21q22), PEX14 (rs636291, 1p35), MYO6 (rs9443189, 6q14), among others, combined with the known prostate cancer risk variants, explain 33% of the familial risk for this disease. In particular, variant rs4713266 is located in intron 1 of NEDD9, a gene involved in cell adhesion, motility, cell cycle and apoptosis, and is implicated in the progression and metastasis of several cancers. Otherwise, rs9443189 on MYO6 gene, is described to be a modulator of androgen-dependent gene expression, and it has been found to be overexpressed in prostate cancer driving tumor growth and metastasis. Moreover, several studies have identified independent risk variants in the 8q24 region (HapC 14 SNPs, rs16901979, DG8S737, rs1447295, rs1016343, rs6983267, rs4242382, rs1006908, rs620861, and rs16902094). Risk SNPs at or near genes (i.e., MSMB [rs10993994], KLK3 and KLK2 [rs2735839], NKX3-1 [rs2928679], and CTBP2 [rs4962416]) with possible roles in prostate carcinogenesis have also been found. Zheng and colleagues also evaluated the impact of carrying multiple risk SNPs on prostate cancer risk, demonstrating that men carrying four out of five possible risk SNPs (rs4430796 [17q12], rs1859962 [17q24.3], rs16901979 [8q24], rs6983267 [8q24], and rs1447295 [8q24]) had a 4.5-fold increased risk of disease. No evidences highlighted that the risk SNPs were associated with disease aggressiveness, earlier age at diagnosis or presence or absence of family history. Nowadays, the validity and the clinical utility of the use of individual or multiple SNP panels as screening test for prostate cancer have not been demonstrated. Nevertheless, the identification of genetic variants able to predict early-onset or more aggressive disease are in progress.

Lung Cancer In the field of research investigating the association between SNPs and lung cancer risk, a growing number of studies have been published. In a recent work published by Liu and co-workers, evidences from meta-analyses and GWAS revealed the association between 108 variants and lung cancer. Among these, about 63 were reported to be significantly associated with lung disease, and, in particular, 15 SNPs on or near 12 genes and one miRNA showed strong evidence of association with lung cancer risk, including TERT [rs2736098], CHRNA3 [rs1051730], AGPHD1 [rs8034191], CLPTM1L [rs401681 and rs402710], BAT3 [rs3117582], TRNAA [rs4324798], ERCC2 [Lys751Gln], miR-146a2 [rs2910164], CYP1B1 [Arg48Gly], GSTM1 [null/present], SOD2 [C47T], IL-10 [ 592C/A and  819C/T], and TP53 [intron 6]. Among these variants, SNP rs2736098 is localized to telomerase gene TERT, a reverse transcriptase component of telomerase, playing an essential role in telomerase enzyme production and maintenance of telomeres. In the literature, it has been reported that TERT gene amplification is commonly described in lung

Genome Wide Association Studies (GWAS)

117

adenocarcinoma. Among other genes, CHRNA3 encodes for a member of nicotinic acetylcholine receptor family which may be a key player in nicotine-mediated suppression of apoptosis in lung cancer cells. CLPTM1L contributes to the accumulation of DNA damage, which confers susceptibility to tumorigenesis, whereas BAT3, a member of the Bcl-2-associated athanogene (BAG) family of proteins, participates in a variety of primary cellular processes, playing also an important role in apoptosis regulation. In another GWAS for squamous lung cancer, rare variants of BRCA2 (p.Lys3326X [rs11571833]) and CHECK1 (p.Ile157Thr [rs17879961]), in association with the common variation 3q28 (TP6, rs13314271) were also implicated in lung adenocarcinoma. More recently, a large-scale association analysis has identified new lung cancer susceptibility loci. In this study, 14,803 cases and 12,262 controls of European descent were genotyped and combined with existing data for an aggregated GWAS analysis of lung cancer in 29,266 cases and 56,450 controls. Eighteen new susceptibility loci in several genes (FUBP1 [rs71658797], RNASET2 [rs6920364], CHRNA2 [rs11780471], BRCA2 [rs11571833], SEMA6D [rs66759488], CHRNA5 [rs55781567], CYP2A6 [rs56113850], TP63 [rs13080835], TERT [rs7705526], NRG1 [rs4236709], CDNK2A [rs885518], OBFC1 [rs11591710], AMICA1 [rs1056562], SECISBP2L [rs77468143], RTEL1 [rs41309931], MHC [rs116822326], RAD52 [rs7953330], CHEK2 [rs17879961]) were identified, showing the striking heterogeneity in genetic susceptibility across the histological subtypes of lung cancer, with four loci associated with lung cancer overall and six loci associated with lung adenocarcinoma. The locus 15q21 is predicted to target SECISBP2L. The genetic risk allele appears to correlate with decreased expression levels of SECISBP2L, already known to be downregulated in lung cancer. rs77468143 was nominally associated with lung function, potentially implicating inflammation of lung as a part of the mechanism at this locus. Three of the identified variants associated with lung adenocarcinoma are located near gene related to telomerase regulation: rs7902587 and rs41309931 near OBFC1 and RTEL1, respectively, and rs2853677 near TERT as previously noted. The genetic susceptibility alleles described in GWASs explain approximately 12.3% of the familial relative risk. Nevertheless, ongoing studies on target genes will provide new insights into the etiology of lung cancer.

Colorectal Cancer Among the most common malignancies, colorectal cancer (CRC) has one of the largest proportions of familial cases. It is reported in the literature that approximately 30% of all CRC cases are an inherited form of the disease, and of these, only about 5% of cases are associated with highly penetrant inherited mutations (i.e., common polymorphisms in genes that regulate metabolism or genes that are regulated by environmental or other genetic factors). Inherited CRCs are also likely to be caused by alterations in multiple susceptibility loci that have additive effects. In particular, it has been described that common variants influence CRC risk. In detail, performing different meta-analysis of several GWAS, several CRC risk loci were identified, including 1q41 (rs6691170 and rs6687758), 3q26.2 (rs10936599), 6p21 (rs1321311, CDKN1A), 8q23 (rs16892766, EIF3H), 8q24 (rs6983267, LOC727677, POU5F1P1), 11q13.4 (rs3824999, POLD3), 12q13.13 (rs11169552 and rs7136702), 18q21.1 (rs4939827, SMAD7), 20q13.33 (rs4925386), and Xp22.2 (rs5934683, SHROOM2), associated approximately with 0.92 to 1.27-fold increased risk of disease. These data derived from GWAS on CRC provide important insight into the biological basis of inherited genetic susceptibility to CRC, identifying risk factors able to influence the development of CRC.

Pancreatic Cancer As CRC, pancreatic cancer is one of the most common cancers, considered as the fourth leading cause of death in the developed world. Both inherited high-penetrance mutations in BRCA2, ATM, PALB2, BRCA1, STK11, CDKN2A and mismatch-repair genes and low-penetrance loci are associated with increased risk. Conducting GWAS, several loci risk associated with pancreatic cancer have been described: 9q34 (rs505922, ABO blood group gene – OR ¼ 1.20), 13q22.1 (rs9543325 and rs9564966, KLF5 – OR ¼ 1.26 and 1.26, respectively), 1q32.1 (rs3790844, NR5A2 – OR ¼ 0.77) and 5p15.33 (rs401681, CLPTM1L-TERT – OR ¼ 1.19). Recently, to identify new risk loci, Childs and colleagues performed a GWAS on 9925 cases and 11,569 controls, identifying three newly associated regions, including 17q25.1 (LINC00673, rs11655237, OR ¼ 1.26), 7p13 (SUGCT, rs17688601, OR ¼ 0.88) and 3q29 (TP63, rs9854771, OR ¼ 0.89), with high pancreatic cancer risk penetrance. Two highly correlated variants (rs11655237 and rs7214041) were associated with pancreatic cancer risk. Variant rs7214041 is to LINC00673 (long intergenic nonprotein coding RNA 673); rs11655237, a noncoding transcript variant, shows significant DNase hypersensitivity in multiple cancer cell lines and binds transcription factors including P300, FOXA1, FOXA2, and the DNA repair protein RAD21. A significant association for two variants in high LD (rs9854771 and rs1515496) and located in an intron of TP63, is also described. p63 is a p53 homologue implicated in tumorigenesis and metastasis, playing a role in cell-cycle arrest and apoptosis.

Renal Cell Carcinoma Similarly, clear cell renal carcinoma (RCC), the most common type of kidney cancer in adults, is responsible for approximately 90%–95% of cases. GWAS approaches in RCC patients have identified several variants conferring risk of the disease. One variant, rs35252396 located at 8q24, has been described to be significantly associated with RCC (OR ¼ 1.27). The conduction of two-stage GWAS of RCC in 3772 affected individuals and 8505 controls, identified two additional risk loci, 2p21 (rs11894252 and rs7579899 mapping to EPAS1) and 11q13.3 (rs7105934, SCARB1), associated with RCC susceptibility, in addition to the more recent seven

118

Genome Wide Association Studies (GWAS)

new loci at 1p32.3 (rs4381241), 3p22.1 (rs67311347), 3q26.2 (rs10936602), 8p21.3 (rs2241261), 10q24.33-q25.1 (rs11813268), 11q22.3 (rs74911261), and 14q24.2 (rs4903064).

Bladder Cancer Bladder cancer is also one of the most common cancer in men and its frequent recurrence requires regular screening and intervention. GWAS in bladder cancer identified common variants in four genomic regions on chromosome 3q289 (TP63), 4p16.3 (TMEM129, TACC3-FGFR3), 8q24.219, and 8q24.311 (PSCA) all associated with increased risk. Interestingly, the variants on 8q24.21 map to a region centromeric to MYC that has been identified in GWAS of breast, colorectal and prostate cancers, as well as chronic lymphocytic leukemia. Association with bladder cancer risk has been suggested also for variants near the TERTCLPTM1L locus on chromosome 5p15.33, which has been described to be associated with increased risk of basal cell carcinoma, cutaneous melanoma, lung, brain, and pancreatic cancers, together with three new regions on chromosomes 22q13.1 (rs1014971), 19q12 (rs8102137, CCNE1), and 2q37.1 (rs11892031, UGT1A). These findings on common variants associated with bladder cancer risk should provide, together with further discovery, new insights into the mechanisms of carcinogenesis.

Other Malignancies Other malignancies have been subject of GWAS with the aim to identify genetic variants associated with an increased risk of cancer. Cutaneous melanoma is a disease of fair-skinned individuals, with an increased risk associated with a family history of the disease. As reported in the literature, the most common high-penetrance melanoma susceptibility gene (CDKN2A) maps to 9p21. Germline CDKN2A mutations are carried by about 2% of all melanoma cases across populations. One of the first genome-wide association study of melanoma, conducted of 317K tagging SNPs for 1650 selected cases and 4336 controls, identified three risk loci: 16q24 (MC1R, rs258322), 11q14-q21 (TYR, rs1393350) and 9p21 adjacent to MTAP and flanking CDKN2A (rs7023329). In particular, MC1R and TYR genes are associated with pigmentation and cutaneous sun sensitivity, wellrecognized melanoma risk factors. An international two-stage meta-analysis of 11 cutaneous melanoma GWAS (totaling 15,990 cases and 26,409 controls), has identified some loci not previously associated with melanoma risk (P < 5
10 8), including 2p22.2 (RMDN2/CYP1B1, rs6750047), 6p22.3 (CDKAL1, rs6914598), 7p21.1 (AGR3, rs1636744), 9q31.2 (TMEM38B/RAD23B/TAL2, rs10739221), 10q24.33 (OBFC1, rs2995264), 11q13.3 (CCND1, rs498136), and 15q13.1 (OCA2, rs4778138). In addition to this, Barret and colleagues, identified additional new regions with at least one SNPs with P < 1
10 5 in ATM (rs1801516), in MX2 (rs45430), a SNP adjacent to CASP8 (rs13016963), and a fourth locus near CCND1 (rs1485993). Ovarian cancer accounts for the most common cause of deaths than all other gynecological cancers combined, with a heritable component explaining less than half of the excess familial risk. Performing GWAS, during the years, identified several common ovarian cancer susceptibility alleles. The prevalent SNPs associated with disease risk (P < 10 8) included SNPs at 9p22 (rs3814113, OR ¼ 0.82), variants at 1p36 (nearest gene, WNT4 [rs56318008]), 4q26 (SYNPO2 [rs17329882]), 9q34.2 (ABO [rs635634]) and 17q11.2 (ATAD5) associated with epithelial ovarian cancer (EOC) risk, and at 1p34.3 (RSPO1 [rs58722170]) and 6p22.1 (GPX6 [rs116133110]) variants, specifically associated with the serous EOC subtype. In addition, the analysis of HOXD1 (rs2072590), MYC (rs10088218), TIPARP (rs2665390), SKAP1 (rs9303542) and of BNC2 at 9p22 supports a functional role for these genes in ovarian cancer development. Gliomas represent about 40% of all primary brain tumors and cause around 13,000 death in the USA each year. These cancers are heterogeneous and typically associated with a poor prognosis irrespective of clinical care, having a median overall survival of only 10–15 months. As described in literature, GWASs identified SNPs in at least eight loci influencing glioma risk: 3q26.2 (near TERC), 5p15.33 (near TERT), 7p11.2 (near EGFR), 8q24.21 (near CCDC26), 9p21.3 (near CDKN2A/CDKN2B), 11q23.3 (near PHLDB1), 17p13.1 (TP53), and 20q13.33 (near RTEL1). Moreover, a metaanalysis of four GWASs totaling 4147 cases and 7435 controls identified new glioma susceptibility loci for glioblastoma (GBM) at 12q23.3 (POLR3B, rs3851634) and non-GBM at 10q25.2 (VTI1A, rs111696067), 11q23.2 (ZBTB16, rs648044), 12q21.2 (intergenic, rs12230172), and 15q24.2 (ETFA, rs1801591). In addition, several hematological malignancies including acute lymphocytic leukemia (ALL), chronic lymphocytic leukemia (CLL) and Hodgkin’s lymphoma, have been addressed within several GWAS to determine the presence of risk loci associated with the disease. Examples included the risk loci 7q12.2 (IKZF1, rs4132601), 10q21.2 (ARID5B, rs10994982), and 14q11.2 (CEBPE, rs2239633) for ALL, the variants 2q13 (ACOXL/BCL2L11, rs17483466), 2q37.1 (SP140, rs13397985), 2q37.3 (FARP2, rs757978), 6p25-p23 (IRF4, rs872071), and 19q13.2-q13.3 (PRKD2/STRN4, rs11083846) for CLL. It is well known that the risk of Hodgkin’s lymphoma (HL), a common B-cell-derived cancer involving germinal center of lymph nodes, is influenced by HLA genotype variation within the major histocompatibility complex (MHC). Nevertheless, GWASs on this cancer have identified susceptibility loci for HL at 2p16.1 (REL, rs1432295), 8q24.21 (PVT1, rs2019960), and 10p14 (GATA3, rs501764), together with loci mapping to 3p24.1 (EOMES, rs3806624), and 6q23.3 (HBS1L and MYB, rs7745098). Recently, a metaanalysis performed on three GWASs (1816 cases and 7877 controls) identified a novel variant at 19p13.3 associated with HL (rs1860661; OR ¼ 0.81), located in intron 2 of TCF3 (also known as E2A), a regulator of B- and T-cell lineage commitment known to be involved in HL pathogenesis, suggesting a link between the 19p13.3 locus and HL risk.

Genome Wide Association Studies (GWAS)

119

In conclusion, despite GWAS provided a direct evidence of polygenic susceptibility, almost all variants identified in several cancers and other diseases have no proven biological or mechanistic relevance to the disease, or medical utility for diagnosis or therapy. Almost all low-penetrance variations discovered to date have weak effects; a meta-analysis approach showed that few loci were replicated in different studies. In this regard, the genetic heterogeneity could be the reason for the limited number of genes that could play an important role in cancer susceptibility exclusively within a limited number of families, and this effect could be lost as soon as they are diluted in the general population. Moreover, the synergistic interactions of low-penetrance genes together with the environmental exposure could characterize an important portion of the heritability of cancer. For this reason, in order to improve the knowledge regarding the causative relationship between variants and diseases, large-scale association or case-control studies are warranted.

Genome-Wide Association Studies in Pharmacogenetics Pharmacogenetics is the study of inherited genetic differences in drug metabolic pathways, which can affect the individual response to drug in term of both therapeutic and adverse effects (Fig. 3). Pharmacogenetics is often related to the oncology field, and involves the study of germline mutations, included SNPs, affecting genes coding for liver enzymes and drug carriers responsible for drug metabolism and transport across membranes. Pharmacokinetics and pharmacodynamics are thus involved. Among the most important genes, a relevant role is played by the cytochrome P450 oxidase and genes coding for drug targets like receptor, transporters or enzymes. Since 2007, GWAS have increasingly been applied in this field. For example, the immunosuppressant thiopurine drugs, used in several cancer treatments, can cause bone marrow suppression associated with inactivating polymorphisms of thiopurine S-methyltransferase (TPMT). In particular, the TPMT c.719A>G SNP has been known to cause a decrease in TPMT activity, increased intracellular thiopurines, and drug toxicities. The majority of low activity phenotypes can be explained by two common variant alleles, TPMT*3A and *3C, both of which are characterized by the nonsynonymous SNP c.719A>G (rs1142345). Similarly, an association has been also described between the uridine diphosphate glucuronosyltransferase 1A1 allele *28 (UGT1A1*28) and neutropenia induced by irinotecan. One of the first GWA study referred to the analysis of genetic and nongenetic response to the anticoagulant warfarin, was performed on 1053 patients, resulting in an association with two genes: CYP2C9, encoding the main metabolizing enzyme for both drugs, and vitamin K epoxide reductase complex 1 (VKORC1), showing genome wide significance with low P values (1
10 78). The role of GWAS in this process is evident in the evaluation of the genetic component of warfarin dosing. Another example is referred to methotrexate, a chemotherapeutic agent and immune system suppressant used to treat cancer (i.e., breast, leukemia, lung cancer, and lymphoma), and autoimmune diseases (i.e., psoriasis, rheumatoid arthritis, and Crohn’s disease). The GWA study, performed on a cohort of 434 children with leukemia, has identified the SLCO1B1 gene affecting methotrexate pharmacokinetics, thus influencing both toxicity of this drug, and possible drug interactions with widely used OATP1B1 substrates, such as statins. In conclusions, ongoing GWAS are demonstrating their potential not only for the identification of genes related to the development and/or progression of a disease, but also in pharmacogenetics and pharmacogenomics. In particular, the complex physiology of drug response highlights the need for analytical methods that address this complexity, using the wealth of information available about drug mechanisms and pathways.

Conclusions Cancer is an extremely complex phenomenon and, together with the incomplete penetrance of the inherited tumor susceptibility alleles, the interaction with environmental risk factors could substantially alter hereditary susceptibility. The results of currently published GWAS have demonstrated that much of the heritable risk of most common cancers is polygenic, and that the increased risk of cancer associated with the known genetic variants are not medically addressed. With the exception of breast and prostate cancers, the currently identified loci explain only a small proportion of the familial risk of many cancers. Nevertheless, the loci identified through GWAS have greatly expanded the existing knowledge of genes that influence cancer risk. However, determining the functional consequences of data from GWAS could be important in order to gain a deeper understanding of cancer biology and to suggest potential targets for therapeutic and preventive strategies. Therefore, despite the GWAS limitations, these analysis provide a more complete understanding of the genetic basis of disease, and a wide view of the genetic architecture of complex phenotypes.

Prospective Vision In contrast to the original hypothesis according to which a common disease may be caused in part by genetic variants (e.g., SNPs) with minor allele frequency (MAF) > 5%, selected diseases are also due to rare variants. GWAS mainly contributed to the identification of common low-penetrance risk variants associated with a low relative-risk to develop a disease. Thus, efforts are currently

120

Genome Wide Association Studies (GWAS)

being directed towards testing rare variants for association with complex traits. The development of new analytic approaches and the test of rare variants for association with complex traits using imputed variants from the publicly available 1000 Genomes Project resequencing data and from direct resequencing of clinical samples, are now ongoing. In particular, novel analytic approaches are based on the integration of in vitro and in vivo experimental approaches with genomic data. Therefore, GWAS approaches are powerful tools that have enhanced and will continue to expand our understanding of cancer genetics and will inevitably lead to the identification of novel pathways of carcinogenesis. Until now, most GWAS evaluated genetic variation depend only on SNPs. At present, the research regarding other forms of genetic variation in the human genome, including genomic structural changes and sequence variation (i.e., CNV), is just beginning to emerge and will provide further insight into the genetic basis of complex diseases. Due to rapid expansion of genomic knowledge, it will be important to build and adequate collaboration between “bench” scientists and “bedside” stakeholders.

See also: Molecular Epidemiology and Cancer Risk.

Further Reading Daly, A.K., 2010. Genome-wide association studies in pharmacogenomics. Nature Reviews Genetics 11 (4), 241–246. Han, Y.L., 2016. Data analysis in the post-genome-wide association study era. Chronic Disease and Translational Medicine 2 (4), 231–234. Houlston, R.S., 2017. Genome-wide association studies of cancer: Current insights and future perspectives. Nature Reviews Cancer 17 (11), 692–704. Klein, R.J., 2005. Complement factor H polymorphism in age-related macular degeneration. Science 308 (5720), 385–389. Lee, .J.C., 2017. Beyond disease susceptibility-leveraging genome-wide association studies for new insights into complex disease biology. HLA 90 (6), 329–334. Mao, J.H., 2011. Cancer evolution and individual susceptibility. Integrative Biology: Quantitative Bioscience from Nano to Macro 3 (4), 316–328. Motsinger-Reif, A.A., 2013. Genome-wide association studies in pharmacogenomics: Successes and lessons. Pharmacogenetics and Genomics 23 (8), 383–394. Nickerson, D.A., 2008. The environmental genome project: Reference polymorphisms for drug metabolism genes and genome-wide association studies. Drug Metabolism Reviews 40 (2), 241–261. Offit, K., 2010. Genome-wide association studies of cancer. Journal of Clinical Oncology 28 (27), 4255–4267. Raj, T., 2011. Progress and promise of genome-wide association studies for human complex trait genetics. Genetics 187 (2), 367–383. Stadler, Z.K., 2010. Genome-wide association studies of cancer. Journal of Clinical Oncology: Official Journal of the American Society of Clinical Oncology 28 (27), 4255–4267. Sud, A., 2017. Genome-wide association studies of cancer: Current insights and future perspectives. Nature Reviews Cancer 17 (11), 692–704. Witte, J.S., 2010. Genome-wide association studies and beyond. Annual Review of Public Health 31, 9–20.

Relevant Websites https://ghr.nlm.nih.gov/primer/genomicresearch/gwastudiesdNational Library of Medicine, understand genetic. http://www.genome.gov/gwastudiesdNational Human Genome Research Institute’s “Catalog of Published Genome-Wide Association Studies”. http://www.illumina.com/techniques/microarrays/human-genotyping.html; http://www.thermofisher.com/it/en/home/lifescience/microarray-analysis.htmldHigh-troughput technologies for identifying genetic variation. https://www.ncbi.nlm.nih.gov/variation/news/NCBI_retiring_HapMap/dHap Map project resources.

Germ Cell Tumors: Pathology and Genetics JW Oosterhuis and Leendert HJ Looijenga, Erasmus University Medical Center, Rotterdam, The Netherlands © 2019 Elsevier Inc. All rights reserved.

Glossary Choriocarcinoma Malignant germ cell tumor composed of trophoblastic tissue. Dermoid cyst Cystic benign tumor of the ovary composed of mature somatic tissues, predominantly from the cranial part of the body. Developmental state (synonym: pluripotency state) Determines the developmental potential of embryonic stem cells: 2Cstate, omnipotent, capable of producing a complete organism; naïve-state (synonym: ground-state), totipotent, capable of producing all lineages of the embryo proper, extraembryonic tissues (yolk sac and trophoblast), and the germ line; primedstate, pluripotent, capable of producing all somatic tissues, but not extraembryonic tissues and the germ line. (Table 1) Plasticity of developmental states allows gradual transitions between the defined states. Embryonal carcinoma Malignant germ cell tumor composed of transformed embryonic stem cells, so-called embryonal carcinoma cells, the stem cells of nonseminomas. Genomic imprinting The mechanism whereby the expression of genes may differ depending on parental origin. Germ cell neoplasia in situ (GCNIS) In situ malignant germ cell tumor confined to seminiferous tubules, composed of transformed gonocytes, located in the spermatogonial niche. Gonadoblastoma In situ malignant germ cell tumor of dysplastic gonads, composed of transformed gonocytes enveloped by supporting stromal cells, usually granulosa cells. Hydatidiform mole Benign germ cell tumor composed of trophoblastic tissue. Nonseminoma Malignant germ cell tumor, resulting from reprogramming of neoplastic primordial germ cell/gonocyte, consisting of embryonal carcinoma, embryoid bodies, teratoma, yolk sac tumor and choriocarcinoma, either pure or in various combinations, and thus representing caricatures of embryonic development. Identical tumors are referred to as nonseminoma in the testis and mediastinum, nondysgerminoma in the ovary, and nongerminoma in the brain. Parasitic twin Poorly developed, usually monozygotic twin attached to outside of the body, or included within the body. Pre-GCNIS Lesion of the seminiferous tubule at risk of progressing toward GCNIS. Seminoma Malignant germ cell tumor composed of transformed primordial germ cells (extragonadal) or gonocytes (testis, ovary, and dysgenetic gonad). Identical tumors are referred to as seminoma in the testis and mediastinum, dysgerminoma in the ovary, and germinoma in the brain. Spermatocytic tumor Benign germ cell tumor composed of postpubertal spermatogenetic cells before meiosis. Previously referred to as spermatocytic seminoma. Teratoma Germ cell tumor composed of mature somatic tissues; depending on originating germ cell tumor type (Table 1) teratomas can be benign (type I and IV), malignant (type II), or both (type VI). Undifferentiated gonadal tissue Combination of granulosa cells and gonocytes in an ambiguous gonad at risk of progressing toward gonadoblastoma. Yolk sac tumor (synonym primitive endodermal tumor) Malignant germ cell tumor composed of extraembryonic structures (allantois and yolk sac), and intraembryonic endodermal derivatives (primitive gut and liver).

Introduction Germ cell tumors (GCT) are a heterogeneous family of gonadal and extragonadal tumors derived from germ cells, and their precursors: primordial germ cells (PGC)/gonocytes (Fig. 1). The wide range of clinical manifestations of GCT is due to the different developmental potential and anatomical sites of the originating cells. Characteristics of seven defined types of GCT are summarized in Table 1.

Migration of PGC and Localization of Extragonadal GCT Types I and II In human embryos PGC can be recognized in the wall of the yolk sac from week 3 to 4 postconception (wpc). They express cKIT, the receptor for KIT ligand (KITLG), which is essential for survival, proliferation, and migration of PGC. In wpc 4–5 PGC leave the gut

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.66150-0

121

122

Germ Cell Tumors: Pathology and Genetics

Fig. 1 Unifying model of the pathogenesis of GCT based on the hypothesis that the developmental potential of GCT is determined by the developmental state (2C, naïve, or primed) of the originating cell. Shown in the figure are stages of embryogenesis (upper panel), developmental potential of stem cells in subsequent stages of embryonic development and the germline (second panel), critical features of the involved stem cells (third panel), and corresponding GCT types with gender distribution and their histology, reflecting developmental potential of originating cell (bottom panel, linking Figure 1 to Table 1). Abbreviations in order of appearance: DEV.$POT., developmental potential; PGC, primordial germ cell; iPSC, induced pluripotent stem cell; PARAM, parameters; M, male; F, female; DE-(I), first wave of demethylation; DE-(II), second wave of demethylation; RE, remethylation; glob. meth., Global methylation; GI, genomic imprinting; X-inact, X-inactivation; D, distal enhancer; P, proximal enhancer; H, human; GCT, germ cell tumor; TE, teratoma; YST, yolk sac tumor; Im, immature; SE, seminoma; NS, nonseminoma; ST, spermatocytic tumor; DC, dermoid cyst; HM, hydatidiform mole. Oosterhuis, J. W. and Looijenga, L. H. J. (2017). Germ cell tumors from a developmental perspective: Cells of origin, pathogenesis and molecular biology; emerging patterns. In F. F. Nogales and R. E. Jimenez (eds.), Pathology and biology of human germ cell tumors, Berlin: Springer Nature, pp. 23–129.

epithelium by epithelial mesenchymal transition (EMT) (Fig. 2). In the period wpc 4–6 they are present in the hindgut epithelium, the mesenchyme of the dorsal mesentery, and the developing genital ridge. KITLG activates KIT signaling and facilitates the migration of PGC. After establishment of connections between the enteric and sympathetic nervous systems, PGC follow sympathetic nerve fibers towards the gonads. Numerous PGC are still present in the nervous system by wpc14. PGC failing to exit the nerve branches at the gonadal site may continue along the sympathetic trunk along the midline of the body and may end up in other distant localizations including the retroperitoneum (suprarenal region, adrenal glands), abdomen (stomach), anterior mediastinum, heart, lungs, head and neck, and central nervous system (CNS). These so-called mis-migrated PGC may, unless eliminated by apoptosis (Fig. 3), give rise to GCT in these various extragonadal sites. Upon arrival in the genital ridges human PGC, from then on called gonocytes, enter a premeiotic stage and upregulate meiotic genes. In the male genital ridge the germ cells enter mitotic arrest as G0/G1 prespermatogonia, due to the low exposure to retinoic acid (RA) within the seminiferous tubules. In the female genital ridge, and extragonadal localizations both in male and female embryos, the germ cells enter meiotic prophase due to exposure to RA. Migrating PGC undergo progressive germline-specific global demethylation, so-called reprogramming 1. In the gonads reprogramming 2 takes place, including completion of erasure of parental genomic imprinting. Extragonadal type I GCT occur at all sites where mis-migrated PGC have been demonstrated. In the developing embryo methylated early stage PGC are more robust than late stage PGC, but nonetheless will perish through apoptosis, unless they escape this

Table 1

Characteristics of seven defined types of germ cell tumors (GCT)

Sex

Anatomical site

0

Neonates

F/M

I

Neonates and children < 6; rarely beyond childhood

F/M

II

After start of puberty. In disorders of sexual developm., Klinefelter’s- and Down’s syndrome rarely before puberty Older men, usually > 55

>>M

Retroperitoneum/ sacrum/skull/hard palate Testis/ovary/sacral region/ retroperitoneum/ anterior mediastinum/neck/ midline brain/other rare sites Dysgenetic gonad/ testis/ovary/anterior mediastinum (thymus)/midline brain (pineal gland)

III

Phenotype/ developmental potential

Developmental state

Precursor cell

Genomic imprinting; methylation

Karyotype

Animal model

Included and parasitic twins

2C-state (omnipotent)

Blastomere

Biparentental

Normal diploid

Not available

(Immature) teratoma (TE)/yolk sac tumor (YST)

Primed-state (pluripotent)

PGC/gonocyte before start of global demethylation

Biparental partially erased

Diploid (TE)/ aneuploid (YST): gain: 1q,12(p13),20q loss: 1p,4,6q

Mouse teratoma

Seminoma/ dysgerminoma/ germinoma reprogrammed to nonseminoma/ nondysgerminoma/ nongerminoma

Naïve-state (totipotent)

PGC/gonocyte undergoing global demethylation

Erased

Not available

Spermatogonium to premeiotic spermatocyte Maternally imprinted 2C-state

Spermatogonium/ spermatocyte

Partially-completely paternal

Aneuploid (þ/triploid) gain: X,7,8,12p,21 loss: Y,1p,11,13,18 In mediastinum and midline brain also (near)diploid and (near)tetraploid with gain of 12p Gain: 9

Oogonia/oocyte (gynogenote)

Partially-completely maternal

Paternally imprinted 2C-state

Empty ovum/ spermatozoa (androgenote) Somatic cell induced to pluripotency

M

Testis

Spermatocytic tumor

IV

After puberty

F

Ovary

Dermoid cyst

V

After puberty

F

Placenata/uterus

Molar pregnancy

VI

Older age, usually >60

F/M

Atypical sites for GCT

Resembling type I or Primed state or nonseminoma nonseminoma components of type II lineages of naïvestate

Canine seminoma Mouse gynogenote

Completely paternal

(Near) diploid diploid/tetraploid peritriploid gain: X,7,12,15 Diploid (XX and XY)

Mouse androgenote

Imprinting pattern of originating cell

Depending on precursor cell

Xenografts derived from iPSC

Germ Cell Tumors: Pathology and Genetics

Type GCT Age (years)

123

124

Germ Cell Tumors: Pathology and Genetics

Fig. 2 Transverse section through the hindgut and dorsal mesentery of a human embryo, 4.5 weeks postconception. Remnants of OCT4 reactivity (open arrows) are shown in the hindgut epithelium (HG), as well as many migrating, OCT4-positive PGCs (arrowheads). Mamsen, L. S., Brochner, C. B., Byskov, A. G., Mollgard, K. (2012). The migration and loss of human primordial germ stem cells from the hind gut epithelium towards the gonadal ridge. The International Journal of Developmental Biology 56, 771–778.

Fig. 3 Horizontal section through abdomen of human embryo, CRL ¼ 30 mm, 7 weeks and 6 days pc, immunofluorescent labeled against c-Kit and b-III-tubulin antibody. Survey depicted (A–C), with commencing connectivity of the enteric (ENS) and sympathetic (SNS) nervous system (B). (C) Some sympathetic nerve fibers are found in the adrenal glands (AG), the pancreas (P), and especially in the dorsal mesentery located in the middle of the section. Positive b-III-tubulin reactivity is seen in nerve fibers of ENS, in general in the plexus myentericus (PM) and similar reactivity is observed in the duodenum (D). Furthermore, c-Kit is also observed in the interstitial cells of Cajal (ICC) of PM. (D) The PGCs in the nerve fibers demonstrate c-Kit reactivity. (E) Higher magnification of (B) demonstrating b-III-tubulin reactivity of SNS. (F) Higher magnification of boxed area in (C). The larger PGCs, with strong membranous c-Kit reactivity are located in close correspondence to the periphery of the individual nerve fibers of the SNS (F, arrows). The small, densely labeled c-Kit-positive cells outside of the nerve fibers are mast cells (F, arrowheads). Scale bars: (A,B) 500 mm, (C) 200 mm, (D,E) 100 mm, (F) 50 mm. Mamsen, L. S., Brochner, C. B., Byskov, A. G., Mollgard, K. (2012). The migration and loss of human primordial germ stem cells from the hind gut epithelium towards the gonadal ridge. The International Journal of Developmental Biology 56, 771–778.

Germ Cell Tumors: Pathology and Genetics

125

fate through reprogramming into an embryonic stem cell (ESC) in the primed state, the precursor of type I GCT. Extragonadal type II GCT only occur in the thymus and midline of the brain. This more limited distribution of extragonadal type II GCT compared to type I GCT is probably due to the demethylated, partially erased genome of late stage PGC, which renders them fragile and highly apoptosis-prone. These cells can normally only survive as gonocytes in specific niches in the developing gonads. Presumably, in the thymus and the midline of the brain the conditions of the developing gonads are mimicked to the extent that late stage PGC can occasionally survive there, and give rise to type II GCT.

Pathology and Genetics of GCT Types Type 0 GCT (Included and Parasitic Twins) Included and parasitic twins have an estimated incidence of 1/500,000 births. A family history of twinning is a risk factor. Included twins are in 80% localized in the retroperitoneum. External parasitic twins are localized at the same sites where conjoined twins are attached (Fig. 4). Both are composed of highly organized, mature somatic tissues often with a vertebral axis, limbs and internal organs. Most of the cranial part of the embryo, including the brain, is usually missing. Immature components are rare; progression to yolk sac tumor (YST) is exceptional. Parasitic twins have the same genetic constitution as the host, consistent with their origin as monozygotic diamniotic twins. Epidemiological data suggest a continuum between twins, conjoined twins, external parasitic twins, acardiacs (external parasitic twins attached via the placenta) internal parasitic twins, and type I teratomas. The common pathogenesis of these conditions seems proneness of the 2C state, omnipotent blastomeres (Table 1, Fig. 1), to escape control of their developmental potential by the developing embryo. If the resulting twin fails to develop a functional heart it will die. If it succeeds in getting its circulation from the host, it will develop as a parasitic twin or, exceptionally, as a type I teratoma.

Type I GCT (Pediatric GCT; Prepubertal-type GCT) Epidemiology and risk factors The incidence of type I GCT is estimated at 1–1.5 per 100,000; more than half are benign teratomas. A family history of twinning is a risk factor, and possibly Down’s syndrome. The age distribution of GCT shows a small neonatal peak (Fig. 5), before the age of six. This represents type I GCT of the sacrococcygeal area/pelvis, retroperitoneum, anterior mediastinum, head and neck (not shown), brain, and testis. Type I GCT of the ovary occur from birth through adulthood, partially overlapping with the age range of type II and IV GCT. Rarely type I GCT occur in older individuals at other sites than the ovary. Type I GCT occur more often in girls than in boys. Multiplicity of type I GCT is rare (Table 1).

Anatomical localization Type I GCT occur in the testis, ovary, and, outside the gonads, along the midline of the body at sites where mis-migrated PGC have been demonstrated. They occur also in the thymus where PGC have not yet been identified.

Pathology Extragonadal type I GCT result from accidental reprogramming of mis-migrated, early stage, still methylated PGC that have escaped immediate apoptosis. Reprogramming, due to failure of cell intrinsic factors that check developmental potential (SOX 17, BLIMP1,

Fig. 4 Neonate with parasitic twin (Type 0 GCT) attached to sacrum; clearly recognizable foot. Oosterhuis, J. W., Stoop, H., Honecker, F., Looijenga, L. H. (2007). Why human extragonadal germ cell tumours occur in the midline of the body: Old concepts, new perspectives. International Journal of Andrology 30, 256–263.

126

Germ Cell Tumors: Pathology and Genetics

Fig. 5 Age-specific incidence rates of gonadal germ cell tumors in the United States by race, US SEER-9, 1973–2007 (Cases per 1 million) for testis (A); ovary (B); mediastinum (C); retroperitoneum (D); pelvis (E); brain (F); pineal gland (G). Note the early peak before age 6 in testis, mediastinum, retroperitoneum, pelvis, and brain representing type I GCT. This peak is missing in the ovary due the broad age range of ovarian type I GCT. In the pineal gland type I GCT do not occur, only type II GCT. The large majority of extragonadal type II GCT occur in males. Stang, A., Trabert, B., Wentzensen, N., Cook, M. B., Rusner, C., Oosterhuis, J. W., McGlynn, K. A. (2012). Gonadal and extragonadal germ cell tumours in the United States, 1973–2007. International Journal of Andrology 35, 616–625.

and OCT4), results in apoptosis resistant ESC in the primed state. In the testis and ovary preerased gonocytes are reprogrammed to ESC in the primed state due to failure of control of developmental potential by germ cell-intrinsic-and niche factors, such as GDNF. This developmental origin of type I GCT explains the rarity of driver mutations in these tumors. Most type I GCT consist of immature and/or mature teratoma, in accordance with the developmental potential of ESC in the primed state. OCT4-positive stem cells, which are negative for other pluripotency markers and CD30, are occasionally found in the immature teratoma component. Small foci of YST, often only detectable with the use of immunohistochemical markers such as AFP and glypican 3, may also be present, associated with immature teratoma. The YST component will eventually outgrow and obscure the teratoma component, resulting in pure YST. Histologically, YST can be composed of intraembryonic (primitive gut and liver) and extraembryonic structures (allantois and yolk sac). Trophoblastic extraembryonic differentiation rarely occurs (Fig. 6).

Germ Cell Tumors: Pathology and Genetics

127

Fig. 6 Histology of testicular type I GCT: immature teratoma (A, H and E, 200); appears to contain foci of YST on staining for glypican 3 (B, glypican 3 brown, 200); area of pluripotent stem cells (C, H, and E, 200); with nuclei staining positive for OCT4 (D, OCT4 brown, 200); but, apart from one lymphocyte, negative for CD30 (E, no membranous brown staining, CD30 brown, 200).

Apart from high-grade immature teratoma of the ovary, type I immature and mature teratoma is benign. A YST-component, overall recognized in 5%–10% of the cases at birth, and clinically detectable by increased levels of serum AFP, is the only predictor of malignant behavior at any site. When not, or incompletely removed, type I teratoma has a high risk of progressing to YST.

(Cyto)genetics and epigenetics A genome wide association study (GWAS) suggests that a single nucleotide polymorphism (SNP) variant in BAK1, a gene involved in suppression of apoptosis, might be associated with type I GCT. Type I immature and mature teratoma is diploid, lacking chromosomal rearrangements. YST of type I, pure or combined with teratoma, is usually near-diploid and may have gains of 1q, 3, 3p, 8q24, 12p13, 20q, 22, and X, and losses of 1p (1p36), 4, 4q, 6q (6q24-qter), 16q, and 20p (Fig. 7). Overrepresentation of the whole of 12p or the region 12p11.2–p12.1, characteristic for type II GCT, is not a feature of type I GCT. Possibly involved genes in the affected chromosomes are MYC (8q24) and the pluripotency genes STELLAR, NANOG, and GDF3 (12p13). Some of the changes, such as gain of 1q and loss of 1p and 6q, are shared by type II YST and may be related to the phenomenon of progression toward YST rather than being specific for type I GCT.

128

Fig. 7

Germ Cell Tumors: Pathology and Genetics

Karyotype of testicular type I YST, note aneuploidy with characteristic gain of 1q and loss of 6q, and absence of overrepresentation of 12p.

The Wnt/beta-catenin and TGFbeta/BMP signaling pathways are strongly expressed in type I and II YST. Methylation of APC and LOH at 5q21-22 suggests that APC might be involved in the activation of the Wnt pathway.

Type I GCT by anatomical site

•

• •

Sacral region The sacrococcygeal region is the most common site of type I GCT. The male-to-female ratio is 1:3.5. With a frequency of 1/35,000 live births they constitute about a third of all type I GCT. It is the most frequent neonatal tumor, rarely diagnosed beyond the age of 2 years; exceptionally they remain undetected until adult age, and are diagnosed as a type I GCT beyond infancy. The continuum between type 0 and type I GCT makes the distinction sometimes difficult and arbitrary. Other extragonadal sites Other localizations of type I GCT are: retroperitoneum (10%); stomach (2%–3%); mediastinum/thymus (2%–3%); heart (5%); head and neck (15%); brain (10%). Females are more often affected than males. Testis and ovary Ten percent of type I GCT occur in the testis (Fig. 8). Type I GCT of the testis are not associated with germ cell neoplasia in situ (GCNIS) and testicular dysgenesis syndrome (TDS), and do not share the risk factors of testicular type II GCT nor their increasing incidence. About 20% of all type I GCT are localized in the ovary, the second most frequent site, with a broad age range from neonates to adults.

Type I GCT beyond infancy At sites other than the ovary, type I GCT are exceedingly rare beyond the age of 6. However, recently so-called prepubertal typeteratomas have been described in the testis after puberty. It is clinically important to distinguish these type I teratomas from type II teratoma, which is malignant, notwithstanding its deceptively bland histology. Remarkably, these prepubertal type teratomas of the postpubertal testis may grossly be partly solid, partly cystic, and even present as dermoid cysts, sometimes containing hair, like type IV GCT, the dermoid cysts of the postpubertal ovary. Similar partly cystic teratomas may also occur beyond infancy at extragonadal sites, like the mediastinum and brain. Particularly in the mediastinum, mature teratomas may have the gross and microscopic appearance of a dermoid cyst, containing hair and even tooth structures (Fig. 9). Thus, type I teratomas beyond infancy may show phenotypes intermediate between typical type I and type IV GCT. In the ovary, typical type I teratomas may occur side by side with type IV teratomas, both uni- and bilaterally. The explanation may be that PGC at

Germ Cell Tumors: Pathology and Genetics

Fig. 8 testis.

129

Gross image of surgically removed testis of infant. On cross section friable yellowish tissue of type I YST; left most nodule is remnant of

Fig. 9 Surgical specimen of mediastinal type I GCT beyond infancy in a 48 year old women showing a cystic tumor containing sebaceous material and hairs, resembling a type IV GCT (dermoid cyst) of the ovary. Probably an intermediate phenotype between a type I and a type IV GCT. Oosterhuis, J. W. and Looijenga, L. H. J. (2017). Germ cell tumors from a developmental perspective: Cells of origin, pathogenesis and molecular biology; Emerging patterns. In: Nogales, F. F. and Jimenez, R. E. (eds.), Pathology and Biology of Human Germ Cell Tumors, Berlin: Springer Nature, pp. 23–129.

all sites, except in the seminiferous tubules, both in females and males enter meiotic prophase like in the ovary. These cells resemble primary oocytes (the precursors of type IV GCT), perhaps with a genotype in between that of gonocyte/oogonium and primary oocyte, with partial erasure of GI and some degree of maternal imprinting. Such cells may give rise to teratomas beyond infancy with a phenotype intermediate between type I and type IV GCT.

Type II GCT (Seminoma and Nonseminoma) Epidemiology Worldwide the average incidence is 1.5/100,000 with a 20-fold difference between areas with the lowest and highest incidence. Over 90% of type II GCT occur in the testis, the most frequent cancer of males aged 25–45 in Western white Caucasian populations. The remainder develop in dysgenetic gonads/ovary (about 4%) and in the extragonadal sites: anterior mediastinum/thymus and brain midline/pineal gland (about 3%). Type II GCT occur in patients in whom puberty has started or is completed, except for rare cases associated with disorders of sex development (DSD), Down’s syndrome, and Klinefelter’s syndrome. The peak age of type II GCT is about 30 for of the testis, 25 for ovary and mediastinum, and 15 for the brain. At each of the extragonadal sites, males greatly outnumber females. Apparently, type II GCT is very much a disease of adolescent and adult males, probably related to the presence of the TSPY gene on the Y chromosome (Table 1). Testicular and ovarian type II GCT are bilateral, respectively, in 3%–5% and over 6%. Rarely, gonadal tumors may be combined with extragonadal type II GCT in the same individual. Type II GCT of the testis have a strong familial component: over 5% of patients with a testicular type II GCT have a relative with a similar tumor; an estimated 25% of testicular cases is due to familial susceptibility. Although much rarer, familial clustering is also documented for ovarian tumors, and testicular- may cluster with ovarian type II GCT. The only difference between sporadic and

130

Germ Cell Tumors: Pathology and Genetics

familial cases is a younger age of clinical manifestation: 2–3 years for testicular tumors and about 7 years for ovarian cases. Gonadal type II GCT may cluster in families with I GCT, and very rarely with type III GCT (spermatocytic tumor).

Anatomical localization Type II GCT occur at sites that provide niches for late stage, demethylated PGC/gonocytes: dysgenetic gonads, testis and ovary. Outside the gonads they occur only in the thymus and the midline of the brain, which offer surrogate niches for late stage PGC. So-called retroperitoneal extragonadal type II GCT are not primary tumors, but metastases from unrecognized primary testicular type II GCT.

Pathology Type II GCT are derived from late stage, demethylated PGC/gonocytes. They can be transformed into seminoma cells in two different ways. Firstly, and most frequently, the “developmental pathway” involving TSPY in the GBY region on Y in concert with OCT4. Secondly, the much rarer “somatic mutation pathway,” usually in the absence of TSPY. Identical tumors are traditionally named seminoma and nonseminoma in the testis and thymus; dysgerminoma and nondysgerminoma in the ovary and dysgenetic gonad; germinoma and nongerminoma in the midline of the brain (Table 1). From here on we will only use the terms seminoma and nonseminoma. At all anatomical sites, over half of all primary type II GCT are pure seminomas. The younger the patient population, the higher the proportion of seminoma (in dysgenetic gonads and the brain about 80%, in the ovary 60%, in the mediastinum 55%, and in the testis about 50%), suggesting that reprogramming to nonseminoma is a chance event accumulating over time. Reprogramming occurs when due to chromosomal imbalances NODAL- prevails over BMP-signaling (with downregulation of BLIMP1, SOX17, OCT4), whereby niche factors (such as GDNF in the testis) fail to check the potential omnipotency seminoma cells. In invasive seminoma stromal NOGGIN may start reprogamming by inhibiting BMP in the tumor cells. Seminomas are by definition pure tumors of neoplastic late stage PGC/gonocytes. The only other cells are scattered trophoblastic cells occurring in less than 10% of cases. Reactive host lymphocytes are a consistent histological feature (Fig. 10). Nonseminomas contain combinations of embryonal carcinoma (EC), YST, choriocarcinoma, immature, and mature teratoma with or without a seminoma component. Early germ cell differentiation is occasionally encountered. The proportions vary by primary site (Fig. 11). The more aggressive biology of nonseminoma explains its earlier clinical manifestation than seminoma, at all sites. Somatic tissues in nonseminomas may progress to form somatic-type malignancies that closely resemble their somatic counterparts (Fig. 12). Seminomas and nonseminomas may metastasize to regional lymph nodes and from then on to lungs, liver, brain, and bone, socalled visceral metastasis. Choriocarcinoma has a propensity for blood–borne metastases, sometimes the first clinical manifestation of the tumor. In up to 44% of cases, seminoma cells at metastatic sites are reprogrammed to nonseminoma. In nonseminoma, EC cells are the main metastatic cell type. This is microscopically apparent as tumor emboli are virtually always composed of EC cells only (Fig. 13). At the site where these tumor stem cells eventually lodge, they may resume differentiation, mimicking the histology of the primary tumor. The level of differentiation at the metastatic site is in general less than in the primary tumor.

(Cyto)genetics and epigenetics Testicular type II GCT are virtually always peritriploid (GCNIS and seminoma hypertriploid, and nonseminoma hypotriploid), whereas in the ovary, mediastinum, and brain, type II GCT may be (near)diploid or (near)tetraploid.

Fig. 10 Histology of seminoma, a tumor composed of transformed gonocytes, large pale cells with angulated nuclei, eliciting a host response, apparent from the many lymphocytes (dark small nuclei) in between the tumor cells (A: H and E, 200). The seminoma cells show a strictly nuclear staining for OCT4 (B, OCT4 brown, 200).

Germ Cell Tumors: Pathology and Genetics

131

Fig. 11 Histology of nonseminoma with a variegated appearance on low power; the dark areas represent EC, the lighter areas mainly consist of YST, the red areas are bleedings caused by trophoblastic giant cells (A: H and E,  12.5); EC cells arranged in strands and tubular structures (B: H and E, 50); these totipotent stem cells express OCT4 in the nucleus and the cytoplasm (C: OCT4 brown, 50); EC cells may form embryoid bodies, representing early embryos (D: H and E, 200); gut-like structure in mature teratoma (E: H and E, 50); upon staining for TSPY, an early marker of germ cell differentiation, it appears that there are germ cell present in the gut epithelium and the surrounding stromal tissue (F: TSPY red, 50); YST with characteristic Schiller–Duval bodies, concentric structures centered around blood vessels, representing yolk sac elements (G: H and E, 40); choriocarcinoma, in fact, malignant placenta-like tissue with large multinucleated syncytiotrophoblastic giant cells alternating with mononuclear cytotrophobastic cells (H: H and E, 100).

132

Germ Cell Tumors: Pathology and Genetics

Fig. 12 Teratoma showing progression to somatic-type malignancy; to the left glandular tissue lacking features of malignancy, and to the right glandular tissue with malignant features, to be classified as adenocarcinoma (H and E, 200). Oosterhuis, J. W., Peeters, S. H., Smit, V. T., Stoop, H., Looijenga, L. H., Elzevier, H. W., Osanto, S. (2013). Patient with two secondary somatic-type malignancies in a late recurrence of a testicular nonseminoma: Illustration of potential and flaw of the cancer stem cell therapy concept. The International Journal of Developmental Biology 57, 153–157.

Fig. 13 EC cells, the stem cells of nonseminoma, are the cells invading lymphatics and blood vessels; shown here are numerous tumor emboli in vessels (A: H and E, 40); similar area stained for D2–40 which demonstrates that the vessels are lymphatics, and the tumor cells within the vessels are EC cells; the right half of the panel shows GCNIS cells with membranous staining for D2–40 within seminiferous tubules (B: D2–40 brown, 40).

At all anatomical sites type II GCT are characterized by gain of (parts of) the short arm of chromosome 12, most often in the form of an isochromosome of 12p (i(12p)) (Fig. 14). In the testis, it occurs in almost 100% of cases, in the ovary/dysgenetic gonad in about 75%, in the mediastinum in close to 90%, and in the midline of the brain in 60%. Especially in invasive components the more proximal parts of 12p, in particular 12p11.2-p12.1, may be amplified. The proportion of tumors with 12p gain increases with the age of clinical presentation. The very high rate of 12p gain and the peritriploidy in testicular tumors is probably due to the long period of intratubular karyotype evolution. Isochromosome 12p is of uniparental origin, arising from an erroneous centromeric division during mitotic anaphase preceded by tetraploidization. Among the genes involved in the 12p aberrations are NANOG, STELLAR, GDF3, and EDR1 for maintaining pluripotency; cyclin D2 and KRAS providing proliferative advantage; genes involved in glucose or glycolytic metabolism, including GLUT3, GAPDH, and TPI1 for energy metabolism in a low-oxygen environment; and genes involved in suppression of apoptosis such as EKI1, SOX5, and DAD-R. Expression of these genes maintains the PGC/gonocyte-phenotype of the tumor cells and allows them to survive and proliferate in the proper niches. In addition to 12p gain the tumors have numeric changes involving other chromosomes: gain of chromosomes X, 7, 8, 12, and 21 and loss of chromosomes Y, 1p, 11, 13, and 18. The pattern is explained by early tetraploidization of the tumor cells, followed by nonrandom losses and gains of (parts of) chromosomes. The large stretches involved, often entire chromosome arms, suggest that aberrant meiotic division plays a role in the evolution of these chromosomal aberrations. The resulting polyploid genome, combined with hypomethylation, renders Type II GCT chromosomally instable. Chromosomal instability likely drives tumor progression of type II GCT. Mutations and amplifications of oncogenes are rare in type II GCT, which is at 0.5 mutations per Mb lower than in any other solid cancer of adults. KIT is most frequently mutated and mainly involved in seminoma: of the testis in about 30%; of the ovary in

Germ Cell Tumors: Pathology and Genetics

133

Fig. 14 Left panel, partial karyotype showing four copies of chromosome 12 and one isochromosome 12p; middle panel, cartoon of chromosome 12 and isochromosome 12p; right panel, in situ hybridization of type II GCT cell with three copies of chromosome 12, and two isochromosomes 12p (red dots, centromeres of chromosome 12; green dots, marker on 12p). Ulbright, T. M., Amin, M. B., Balzer, B., Berney, D. M., Epstein, J. I., Guo, C., et al. (2016). Germ cell tumours. In: Moch, H., Humphrey, P. A., Ulbright, T. M., Reuter, V. E. (eds), WHO classification of tumors of the urinary system and male genital organs, 4th edn. Lyon: IARC Press, p. 189–226.

up to 50%; of the mediastinum in 40%; and of the brain in over 50%. In nonseminomas, KIT mutations are rare, less than 1.5%; a similar low figure has been reported in gonadoblastoma and derived seminomas. KIT mutations are activating and occur predominantly in the activation loop (exon 17, usually in codons 816, 820, 822, 823, and 825) of the second TK domain. In seminomas mutations of KIT appear to be only one of the mechanisms of activation of KIT signaling and its downstream pathways: KRAS/RAF/MEK/ERK and AKT/mTOR. In these tumors, KIT signaling is always activated either by upregulation of expression or genetically by mutation or amplification (Figs. 15–17). The fact that KIT mutations are predominantly found in seminomas and only rarely in nonseminomas suggests that they are in general involved in the progression of seminomas, rather than in initiation of type II GCT. Activating KRAS mutations occur in some type II GCT, more or less at the same rate in seminomas and nonseminomas. KIT and KRAS mutations are mutually exclusive in type II GCT. Due to early tetraploidization, loss of heterozygosity is an unlikely event in type II GCT. In addition, inactivation of tumor suppressor genes by promoter hyper-methylation is rare. Therefore, tumor suppressor genes probably play a minor role in initiation and progression of type II GCT. In somatic-type malignancies the same mutations may be found as in their somatic counterparts, such as 2q37 rearrangements in rhabdomyosarcoma, p53 mutations in sarcomas, and t(11;22) translocations in PNET. Type II GCT including GCNIS and gonadoblastoma are globally demethylated, with permissive histone modifications and an open chromatin structure; parental genomic imprinting is erased. In nonseminomas only Alu repeats and some CpG islands are methylated. Mature teratoma components are methylated. The developmentally important miRNAs, miR-371-373 as well as miR-302 and miR-367, which have a crucial role in development of embryonic stem cells, are highly expressed in type II GCT.

Fig. 15 KIT/RAS and AKT/mTOR pathway interactions showing frequencies of somatic alterations in key genes. Alteration frequencies are expressed as a percentage of all intracranial GCT patients. Red text, protein positively regulates signaling; blue text, protein negatively regulates signaling; green text, physically interacting protein. Wang, L., Yamaguchi, S., Burstein, M. D., Terashima, K., Chang, K., Ng, H. K. et al. (2014). Novel somatic and germline mutations in intracranial germ cell tumors. Nature 511, 241–245.

134

Germ Cell Tumors: Pathology and Genetics

Fig. 16 Nucleus of a seminoma cell case showing multiple copies of KIT (green) relative to the centromere of chromosome 4 (red). McIntyre, A., Summersgill, B., Grygalewicz, B., Gillis, A. J., Stoop, J., van Gurp, R. J. et al. (2005). Amplification and overexpression of the KIT gene is associated with progression in the seminoma subtype of testicular germ cell tumors of adolescents and adults. Cancer Research 65, 8085–8089.

Fig. 17

Seminoma with strong expression of KIT (KIT red, 100).

Treatment sensitivity and resistance Like their normal counterparts, PGC and ESC, seminoma cells and EC cells have an open chromatin structure, and a hypersensitive DNA damage response. The latter characterized by inducibility of wild-type p53 jointly with the absence of p21-induced cell cycle arrest, whereby cells with damaged DNA are not repaired but eliminated due to a low threshold for apoptosis. This mechanism is a physiologic protection against propagation of repair errors via germ cells into the next generation or via ES cells into the developing embryo. The open chromatin structure in combination with a hypersensitive DNA damage response explains the high sensitivity of type II GCT for DNA damaging agents. As a consequence, Type II GCT are the solid tumors in adults with the highest sensitivity to DNA-damaging agents with cure rates of over 80% in disseminated disease. Nonseminoma lacks the exquisite sensitivity to radiotherapy of seminoma. The chemotherapy sensitive germ cell/embryonic phenotype gets lost upon somatic differentiation into teratoma, as adult tissue stem cells are able to survive DNA damage by prolonged G1 and G2 arrest and proficient repair. This explains why characteristically the teratoma component survives chemotherapy of a Type II GCT as so-called residual mature teratoma. Primary and acquired resistance is relatively rare in type II GCT because of the low mutation rate in these tumors. In general, loss of germ cell- and embryonic phenotype and acquisition of somatic mutations are factors inducing treatment resistance. A final important mechanism causing resistance of type II GCT is further progression of teratoma- and YST-elements due to accumulation of mutations commonly found in adult cancers.

Germ Cell Tumors: Pathology and Genetics

135

Type II GCT by anatomical site Type II GCT of dysgenetic gonads in patients with disorders of sex development (DSD) Gonadoblastoma is a rare lesion that develops in the dysgenetic gonad of patients with of DSD. At high risk are 46,XY patients with mutations in WT1 (including Denys–Drash, Fraser, and WAGR syndromes), SRY, SOX9, DHH, ARX, RR5A1, or TSPYL1. These germ line mutations in the presence of Y-chromosomal sequences result in a dysgenetic testis. • Pathology: Gonadoblastoma is most often caused via the developmental pathway. Mutations in genes involved in male gonadal development in the presence of GBY/TSPY result in a hypovirilized gonad allowing sustained coexpression of OCT4 and TSPY in gonocytes in undifferentiated gonadal tissue (UGT) (Fig. 18), later accompanied by overexpression of KITLG. Initiating somatic mutations, particularly in KIT are rare in the presence of GBY/TSPY; probably more frequent in its absence. The usual precursor of type II GCT in dysgenetic gonads is gonadoblastoma (Fig. 19), and much less frequently GCNIS; rarely gonadoblastoma and GCNIS are simultaneously present (Fig. 20). Gonadoblastoma may develop in streak gonads in the abdomen, and in dysgenetic testes in abdominal, inguinal and scrotal position. In 40% of cases gonadoblastoma is bilateral. Microscopically, gonadoblastoma consists of solid sheets composed of two cell types: nonneoplastic immature granulosa cells, the niche cells, and germ cells, in various arrangements associated with small round depositions of basement membrane-like matrix, called Call-Exner bodies, or psammoma bodies when calcified (Fig. 19). Some of the germ cells resemble oogonia, others, the actual gonadoblastoma cells, have the same atypical morphology of transformed gonocytes as the cells of GCNIS and seminoma, with

Fig. 18 Undifferentiated gonadal tissue (UGT) composed of strands of small stromal cells resembling granulosa cells, enveloping large pale cells, the germ cells (A: H and E, 200); some of them, the gonocytes, express OCT4 (B: OCT4 brown, 200); all germ cells express TSPY (C: TSPY red, 200).

Fig. 19 Gonadoblastoma: areas of immature granulosa cells containing coarse, mulberry-like psammoma bodies and lager pale cells (A: H and E, 50); Gonadoblastoma: oogonia and gonadoblastoma cells embedded in area of immature granulosa cells, as well as pink-colored Call-Exner bodies (B: H and E, 200); gonadoblastoma cells express OCT4, oogonia do not (C: OCT4 brown, 200); oogonia and gonadoblastoma cells express TSPY (D: TSPY red, 200); the granulosa cells express FOXL2 (E: FOXL2 reddish brown, 200).

136

Germ Cell Tumors: Pathology and Genetics

Fig. 20 Gonadoblastoma (left side of each photograph) and GCNIS (right side of each photograph) both within abnormal testicular tubules (A: H and E, 200); both gonadoblastoma and GCNIS cells have nuclear expression of OCT4 (B: OCT4 brown, 200); both express TSPY (C: TSPY red, 200); the supportive cells express KITLG in gonadoblastoma and GCNIS (D: KITLG brown, 200); Sertoli cells supporting GCNIS express SOX9 (E: SOX9 brown, 400), whereas the granulosa cells supporting gonadoblastoma express FOXL2 (F: FOXL2 brown, 400). Hersmus, R., Stoop, H., White, S. J., Drop, S. L., Oosterhuis, J. W., Incrocci, L., Wolffenbuttel, K. P., Looijenga, L. H. (2012). Delayed recognition of disorders of sex development (DSD): A missed opportunity for early diagnosis of malignant germ cell tumors. International Journal of Endocrinology 2012, 671209.

enlarged angulated nuclei with large nucleoli and abundant clear cytoplasm. Immunohistochemically, the morphologically normal germ cells express early differentiation markers of oogonia, such as TSPY and VASA, and are negative for the pluripotency marker OCT4. The gonadoblastoma cells coexpress markers of gonocytes (e.g., OCT4, PLAP, AP-2gamma, and KIT), and of oogonia (TSPY and VASA). The immature granulosa cells express FOXL2, a key regulatory protein for ovarian development of the undifferentiated gonad (Fig. 19). In dysgenetic testes the tubules are lined by immature Sertoli cells expressing SOX9, the key regulator for testicular development (Fig. 20). Occasionally, tubules in transition zones between testicular and ovarian development may be lined by FOXL2 positive primitive granulosa cells. KITLG is also expressed in gonadoblastoma, probably both in the granulosa and gonadoblastoma cells (Fig. 20). Gonadoblastoma cells will eventually outgrow the nonneoplastic oogonia, and by way of further progression become an invasive type II GCT: seminoma (80%), and upon reprogramming, nonseminoma (20%). Most mixed nonseminomas contain a seminoma component.

Type II GCT of the testis

•

Epidemiology and risk factors Africa and parts of Asia have the lowest incidence figures of less than 0.5 for testicular type II GCT. The incidence is a factor 10–20 higher among most white Caucasian populations in Western societies. Worldwide, the incidence has increased in the last decades. In most Western and Northern European countries it has doubled. In the United States the incidence among Caucasians is 6.6 versus 1.2 in blacks. In both groups, the rates have increased in the past 30 years.

Germ Cell Tumors: Pathology and Genetics

•

•

137

The large incidence differences among ethnic groups within the same society demonstrate the importance of genetic factors. However, the worldwide increasing incidence of testicular type II GCT, and the changing incidence among immigrant populations toward the country of destination, point to an important role of environmental factors, associated with Western lifestyle, favoring hypovirilization of the developing male embryo. The strongest risk factors for testicular type II GCT are cryptorchidism (OR 4.3), previous inguinal hernia (OR 1.63), hypospadias, and impaired spermatogenesis. These conditions are considered part of the so-called testicular dysgenesis syndrome (TDS), a relatively mild disturbance of sex differentiation, due to hypovirilizing factors in utero. Other established risk factors are: previous testicular cancer, and a family history of testicular type II GCT. Genetic risk factors Familial risk is among the highest in cancers. The small size of affected families, usually a father and a son or two brothers, and the high risk in monozygotic compared to dizygotic twins are consistent with multiple autosomal recessive low-penetrance susceptibility genes. The first identified risk locus was the gr/gr deletion in azoospermia factor c region of Y, and recently a deleterious probably causative germline mutation in PDE11A was discovered in familial and sporadic cases. Both mutations explain only a few percent of the familial cases. GWAS have discovered variants in over 30 tumor-biologically plausible genes, which increase susceptibility for testicular type II GCT. The ones with the highest OR are KITLG, BAK1, SPRY4 and PDE11A, which are involved in KIT/KITLG signaling, and thus in survival and proliferation of PGC/gonocytes. DMRT1 is a niche factor involved in sex determination and regulation of meiotic division. These variants explain still only 20% of the excess familial risk compared to testicular type II GCT families. Obviously more low penetrance genes are involved, like variants in TGFBR3 and BMP7 and AR, which are associated with TDS. The different distribution of variants, such as in KITLG and AR, in Caucasian and black populations may partly explain the 20-fold ethnic difference in the incidence of testicular type II GCT. Homozygosity jointly for variants that target the PGC/gonocyte and those targeting niche cells substantially increase risk. For example, men with testicular type II GCT have a 14 times higher chance than controls to be homozygous for the two risk alleles, KITLG and DMRT1. Pathology The developmental pathway, due to hypovirilization of the niche of the gonocytes in the developing testis is the most important mechanism of initiation of type II GCT. The disturbed niche interferes with downregulation of OCT4 in gonocytes relocated from the center of the tubules to the prespermatogonial niche, thereby creating a window for coexpression of OCT4 and TSPY. In due time accompanied by overexpression of KITLG, which enhances proliferation of the maturation-disturbed prespermatogonia, and thereby promotes neoplastic transformation. The earliest changes, preceding overt GCNIS, are maturation delay, and pre-GCNIS, (Figs. 21 and 22) which have been studied in cryptorchid testes and androgen insensitivity syndromes. GCNIS is the common precursor of seminoma and nonseminoma of the testis (Fig. 23); it is bilateral in 3%–5% of the patients. This high rate of bilaterality is probably because the germ cell niche is disturbed in both testes as a consequence of TDS. Noteworthy, in DSD, where the disturbance of the niche is more severe, bilaterality of gonadoblastoma may occur in up to 40%. Supposedly GCNIS when left untreated, will always progress to an invasive type II GCT. By default GCNIS develops into intratubular seminoma (Fig. 24), whereby the lumen of the tubule is packed with GCNIS cells, often accompanied by lymphocytes, like in seminoma. Remarkably, necrosis is rare in intratubular seminoma. Apparently the tumor cells, like gonocytes, are well adapted to the low oxygen environment within the tubules. In general, intratubular seminoma becomes invasive by overextension and destruction of the seminiferous tubules.

Fig. 21 Maturation delay: gonocytes located centrally in the tubules still express OCT4, although the infant is older than 6 months (OCT4 brown, 200).

138

Germ Cell Tumors: Pathology and Genetics

Fig. 22 Pre-GCNIS is characterized by coexpression of OCT4 and TSPY by germ cells located in the prespermatogonial niches and focal expression of KITLG. Immature tubules of a young boy contain gonocytes (centrally in the tubules) and prespermatogonia (peripherally in the tubules) (A: H and E, 200); OCT4 is still expressed in the germ cells, not only centrally, but also in the periphery of tubules by germ cells in the prespermatogonial niches (B: OCT4 brown, 200); all germ cells express TSPY (C: TSPY red,  200); focal expression of KITLG (D: KITLG red, 200). Double staining for OCT4 (brown) and TSPY (blue) demonstrates germ cells in the spermatogonial niches with coexpression of OCT4 and TSPY, consistent with preGCNIS (E, 200).

Fig. 23 In the upper left corner of the photograph are tubules with GCNIS, in the lower right corner tubules with spermatogenesis. The tubules with GCNIS have a narrower than normal caliber and a thickened wall; GCNIS cells have dark, enlarged, angular nuclei (A: H and E, 200); GCNIS cells have nuclear expression of OCT4 (B: OCT4 brown,  200), and membranous and cytoplasmatic expression of TSPY (C: TSPY red,  200), as well as membranous expression of KIT (D: KIT red, 200); the Sertoli cells diffusely express KITLG in a granular fashion (E: KITLG red, 200); GCNIS cells consistently coexpress OCT4 (reddish brown, nuclear) and TSPY (bluish, membranous, and cytoplasmatic) (F, 400).

Germ Cell Tumors: Pathology and Genetics

Fig. 24

139

Intratubular seminoma: seminiferous tubule packed with seminoma cells, extending the tubule (H and E, 200).

Fig. 25 The immune reaction of the host elicited by seminoma, apparent from the inflammatory cells, is capable of destroying invasive seminoma cells, and GCNIS cells within the seminiferous tubules, leaving fibrotic tubules and interstitial tissue (A: H and E, 100); tubule with GCNIS, infiltrated by lymphocytes (B: H and E, 200).

Upon invasion, seminoma invariably elicits an inflammatory host response, usually composed of lymphocytes, macrophages, plasma cells, and often a granulomatous reaction (Fig. 25), which may destroy tumor cells, leaving scar tissue. It is probably clinically relevant as in patients with AIDS the median age of seminoma is 25, 10 years younger than in the general population. Deviation from the default development of seminoma occurs when a seminomatous cell, either a GCNIS cell or an intratubular-, invasive- or metastatic seminoma cell, is reprogrammed to an EC cell, the totipotent stem cell of nonseminoma. One in three nonseminomas has a seminoma component, suggesting that reprogramming has occurred in an invasive seminoma (Fig. 26). Probably the remaining nonseminomas are due to reprogramming of a GCNIS- or an intratubular seminoma cell. Virtually without exception intratubular nonseminoma has the morphology of EC. It is always partly necrotic (Fig. 27), apparently not adapted to the low oxygen intratubular environment. This relative anoxia may induce hypoxia factors, like MET, which triggers intratubular EC to invasion. Interestingly, upon invasion the EC cells often start to differentiate (Fig. 28). The intratubular environment seems to suppress differentiation of EC, while stromal factors stimulate differentiation. Candidate stromal factors are TGF-b, FGF, and BMP, which may derepress differentiation-promoting genes by removing polycomb repressive complexes recruited to the promotor sites of these genes by the pluripotency proteins OCT4, SOX2, and NANOG.

Type II GCT of the ovary The incidence of ovarian type II GCT is about 20-fold lower than those of the testis. Ninety-five percent of ovarian type II GCT become manifest after puberty in women with a normal 46,XX karyotype, a couple of years earlier than in males, in accordance with the earlier onset of puberty in females. The remaining 5% are diagnosed before puberty in individuals with manifest or silent DSD. Overall, 6% are bilateral. An established risk for (bilateral) ovarian type II GCT are various forms of DSD in phenotypic females who carry the GBY/TSPY region in their genome. As most studies of risk factors do not distinguish between type I and II GCT, the results have to be

140

Germ Cell Tumors: Pathology and Genetics

Fig. 26 Seminoma with reprogramming to EC, teratoma, YST, and choriocarcinoma (A: H and E, 12.5). Embedded in seminoma are EC (B: H and E, 200); teratomatous tubule lined with goblet cells (C: H and E, 200); YST (D: H and E, 200); choriocarcinoma (E: H and E, 200).

Fig. 27

Intratubular nonseminoma consists of EC only that is largely necrotic (pink staining material) (H and E, 50).

interpreted with caution. Reported risks with the highest OR, such as exogenous hormones during pregnancy, maternal obesity and early regular menstruation after menarche, are related with hormonal regulation of reproduction. An estimated 20% of ovarian type II GCT develop via the developmental pathway in women carrying GBY/TSPY: 6% with overt, and the remaining with clinically silent DSD. Ovarian seminomas in phenotypically and genotypically normal women have KIT mutations in about 50%. Only a minority of these being initiating, it leaves the pathogenesis of about three-quarters of ovarian type II GCT unexplained. It is possible that in the ovary, the majority of type II GCT develop in the context of mild dysgenesis, comparable to TDS of the male, mainly caused by imbalances of factors regulating gonadal development and maintenance, such as FOXL2 (Fig. 29). Downregulation of these factors during gonadal development, even only transiently, in the stage before oocytes become arrested in the prophase of meiosis I, would create a hypovirilized testis-like environment favoring disturbance of maturation oogonia/gonocytes and a window for the development of a type II GCT. However, due to the absence of GBY/TSPY, at a much lower rate than in males, or in DSD patients carrying GBY in their genome, who have an up to 70-fold risk of developing a type II GCT as compared to normal females.

Germ Cell Tumors: Pathology and Genetics

141

Fig. 28 Intratubular nonseminoma (right lower corner of the photograph) consisting of EC only, that starts to differentiate upon invasion (A: H and E, 50; b, OCT4 brown, 50). Oosterhuis, J. W. and Looijenga, L. H. J. (2017). Germ cell tumors from a developmental perspective: Cells of origin, pathogenesis and molecular biology; Emerging patterns. In Nogales, F.F. and Jimenez, R.E.(eds.) Pathology and Biology of Human Germ Cell Tumors, Berlin: Springer Nature, pp. 23–129.

Fig. 29 Dmrt1 silences RA-dependent feminization genes to ensure postnatal sex maintenance during fetal sex determination (1), the bipotential gonad makes a choice between male (blue) and female (pink), largely guided by the presence or absence of Sry. The sexual differentiation machinery downstream of sex determination transforms the undifferentiated gonad into a mature testis or ovary (2), manifested in the formation of Sertoli-cellcontaining seminiferous tubules in the male and granulosa-cell-containing ovarian follicles in the female. Postnatal sex maintenance within Sertoli cells (3) is achieved via the silencing of RA signaling-dependent feminization genes (such as Foxl2) by the transcriptional regulator Dmrt1. RA is thereby allowed to act in adjacent spermatogonia to promote spermatogenesis within the seminiferous tubule. In Dmrt1 mutant Sertoli cells, however, RA acting through RARa activates feminizing genes and reprograms the Sertoli cell into a granulosa-like cell through the process of transdifferentiation. DeFalco, T. (2014). DMRT1 keeps masculinity intact. Developmental Cell 29, 503–504.

Type II GCT of the mediastinum Mediastinal type II GCT constitute 50%–70% of extragonadal type II GCT. Both in whites and blacks over 95% occur in males, thus being male is the most important a risk factor. Mean age for seminomas is about 30 and for nonseminomas 25 years. Over 20% occur in individuals with Klinefelter’s syndrome, who have a 67-fold risk compared to males without this condition. They present at median age of 17 (range 4–31), substantially younger than in non-Klinefelter patients. Klinefelter cases younger than 12 virtually always have precocious puberty, due to b-HCG produced by the tumor. In Klinefelter patients, increased levels of gonadotropins in

142

Germ Cell Tumors: Pathology and Genetics

Fig. 30 Gonadoblastoma-like lesion in the thymus with seminoma cells enclosed by thymic epithelium (A: cytokeratin brown, 200); the seminoma cells are OCT4 positive (B: OCT4 brown, 200); thymic epithelium expresses KITLG (C: KITLG red, 200). Oosterhuis, J. W., Stoop, H., Honecker, F., Looijenga, L. H. (2007). Why human extragonadal germ cell tumors occurin the midline of the body: Old concepts, new perspectives. International Journal of Andrology 30, 256–263.

response to testicular atrophy may enhance malignant transformation of PGC in the thymus, explaining their 67-fold risk of these tumors compared to other males. Mediastinal type II GCT originate from demethylated mis-migrated PGC that survived in the surrogate niche offered by thymic epithelium, shown to express KITLG, (Fig. 30), the critical factor for survival and proliferation of PGC. Primary mediastinal type II GCT are only localized in the anterior or antero-superior mediastinum within, or in association with the thymus. Apart from a somewhat higher proportion of pure YST and pure choriocarcinoma, the distribution of seminoma, and nonseminoma subtypes is similar as in the testicular type II GCT. In Klinefelter’s syndrome the tumors are virtually always nonseminomas with 15% pure choriocarcinoma, and trophoblastic differentiation in 85%. Of mediastinal nonseminomas 10%–20% develop a solid somatic-type malignancy. The distribution of the various histological types is largely similar to that in the testis, but angiosarcomas are more frequent. Hematologic malignancies, developing in 2%–6%, are uniquely associated with mediastinal YST. The most frequent types are megakaryoblastic leukemia, malignant and benign histiocytosis, and myelomonoblastic leukemia. The (cyto)genetic aberrations are largely the same as in testicular type II GCT. Excess copies of chromosome X are partly due to Klinefelter cases.

Type II GCT of the central nervous system

•

•

•

Epidemiology and risk factors The combined incidence of type I and II GCT of the brain is 0.143 for males and 0.046 for females in Japan and 0.118 for males and 0.030 for females in the United States. Probably more than 90% are type II. The overall male-to-female ratio of CNS GCT is about 4:1. The incidence of nonmalignant GCT is similar in males and females, 0.029 and 0.020, respectively. The male-to-female ratio for malignant GCT is 16:1 in the pineal region and 2.1:1 in the rest of the CNS. Being male is the most important a risk factor for Type II GCT of the CNS. More than half of all malignant GCT are located in the pineal region, virtually all of them type II. Type II GCT of the CNS can be familial and may cluster with type II GCT of the gonads and the mediastinum, sometimes in Klinefelter’s and Down’s syndrome. In Klinefelter’s syndrome Type II GCT are the most frequent malignant CNS tumors, suggesting that it is a risk factor for these tumors. Anatomical localization Over 80% of type II GCT of the brain are located in the pineal gland, suprasellar region (neurohypophysial axis; occasionally within the neurohypophysis), hypothalamus, and the wall of the third ventricle. In these midline structures, the large majority are in male patients, malignant, and seminomas. Seminomas occur also in the basal ganglia, cerebral hemispheres, and in the posterior fossa. However, in these and other nonmidline anatomical sites, the proportion of females, nonseminomas and benign GCT (the latter probably of type I) is higher. In fact, type I GCT of the brain have an anatomical distribution resembling that of the type II nonseminomas. The anatomical distribution of type II GCT in Klinefelter’s syndrome is similar to that in the general population. In Down’s syndrome, where there is a high percentage of nonseminomas, the tumors lie most often outside the typical midline sites. Pathology The midline of the brain, in particular the pineal gland and the supra-sellar region, offers a niche where some mis-migrated, totipotent PGC can survive long enough to undergo transformation to a seminomatous precursor cell. Secretion of gonadotropins in the diencephalic centers at the inception of puberty may stimulate neoplastic transformation, and explain the young age of tumor manifestation. Eighty percent of type II GCT of the CNS are seminomas; most nonseminomas are mixed, followed by pure mature and immature teratoma, EC, YST and choriocarcinoma. Of the mixed tumors 75% have a seminoma component. Somatic-type malignancies rarely develop.

Germ Cell Tumors: Pathology and Genetics

•

143

The strong male predominance suggests initiation via the developmental pathway with a crucial role for coexpression of OCT4 and TSPY. Probably only the minority of the tumors is initiated via somatic mutation of one of the genes involved in KIT signaling and downstream pathways (Fig. 15). Eighty percent of the tumors developing in the midline is seminoma. Away from the midline type II GCT are rarer and more often nonseminomas. In these nonmidline sites, the conditions are supposedly less suitable for neoplastic PGC, favoring precursors that have undergone reprogramming to an ESC-like precursor, with a developmental potential in between the totiand the pluripotent state. Indeed, the spectrum of histologies of type II nonseminomas of the brain (and to a lesser degree of the mediastinum) resembles that of type I GCT: relatively high proportions of pure (immature) teratoma, pure YST, and pure choriocarcinoma and rarely pure EC. Yet these tumors occur most often beyond the age of six and have the (cyto)genetic characteristics of type II GCT. In fact, in the brain, particularly away from the midline, there seems to be a gray area with a gradual transition between GCT of type I and II, featuring tumors that are type II by genotype and age but resembling type I by phenotype. Klinefelter patients have the same distribution of type II GCT as non-Klinefelter patients, completely different from mediastinal type II GCT in Klinefelter’s syndrome, where seminomas are virtually lacking. In Down’s syndrome, type II GCT of the brain are less frequently seminomas than in the general population (55%). The anatomical localization of the tumors is atypical, with only one in the pineal gland. Of five nonseminomas, four were YST and one teratoma. (Cyto)genetics Activating mutations in KIT (47%), KRAS (18%), and NRAS (6%) and inactivating mutations in CBL (6%), a negative regulator of KIT expression, are all mutually exclusive and occur most often in seminomas and mixed GCT that lack gain of 12p. The complementary character of these genetic events and their preferred occurrence in seminomas and mixed GCT indicate that they are probably not initiating but engaged in the progression of seminoma. The AKT/mTOR signaling pathway is activated in about 20% of cases, mostly by focal amplification of 14q32.33, containing the AKT1 locus, often in tumors lacking gain of 12p (Fig. 15).

Type III GCT (Spermatocytic Tumor) Epidemiology The incidence of spermatocytic tumor is 0.4 per million, without a significant increase over the past 20 years. The median age of clinical presentation is 55. In about 9% the tumor is bilateral.

Anatomical localization Spermatocytic tumors occur only in the postpubertal testis, virtually always in scrotal position, and sporadically in a maldescended testis (Table 1).

Pathology Spermatocytic tumor is a benign neoplasm, of which the cells resemble postpubertal germ cells with early, premeiotic differentiation. The nuclei appear in three size classes corresponding to A-dark spermatogonia (reserve spermatogonial stem cells) with the smallest nuclei with dense chromatin; A-pale spermatogonia (self-renewing stem cells) with intermediate and large paler nuclei with finely granular filamentous chromatin; B spermatogonia, and leptotene spermatocytes (Fig. 31). The tumor cells express

Fig. 31 Spermatocytic tumor with small dark nuclei, intermediate nuclei, and large nuclei with finely granular filamentous chromatin (A: H and E, 400); the nuclei express DMRT1 (B: DMRT1 brown, 400).

144

Germ Cell Tumors: Pathology and Genetics

markers of these cell types, such as DMRT1, which is a diagnostically useful immunohistochemical marker (Fig. 31). Markers of gonocytes, such as OCT4 and PLAP have been switched off, and markers of meiosis are not yet expressed. Spermatocytic tumor lacks the characteristic host response of seminoma. Adjacent seminiferous tubules may contain intratubular spermatocytic tumor. The occasional finding of an exclusively intratubular spermatocytic tumor without an invasive component proves that the tumor has its origin within the seminiferous tubules (Fig. 32). Exceptionally, the tumor is associated with a sarcomatous component, usually undifferentiated sarcoma, and rarely rhabdomyoor chondrosarcoma. These are highly malignant and readily metastasize to regional lymph nodes or, blood–borne, to visceral organs. Disturbance of the niche- and spermatogonial factors involved in the regulation of spermatogenesis are the factors driving tumorigenesis (Figs. 33 and 34). Like in the other types of GCT, in spermatocytic tumors, initiation is primarily due to

Fig. 32 Intratubular spermatocytic tumor (A: H and E, 200); expressing DMRT1 (B: DMRT1 brown, 200). Oosterhuis, J. W. and Looijenga, L. H. J. (2017). Germ cell tumors from a developmental perspective: Cells of origin, pathogenesis and molecular biology; Emerging patterns. In Nogales, F.F. and Jimenez, R.E.(eds.), Pathology and biology of human germ cell tumors, Berlin: Springer Nature, pp. 23–129.

Fig. 33 Signaling pathways triggered by GDNF in spermatogonial stem cells. GDNF dimerizes and binds to the GFRA1/RET receptor complex. (A) Binding of GDNF activates RET, which triggers SRC-kinase phosphorylation and the downstream activation of PI3K/AKT. Ultimately the transcription factor MYCN is upregulated. (B) Binding of GDNF also can activate the RAS-mediated signaling pathway, which triggers ERK1/2 phosphorylation and upregulation of the transcription factor FOS. Waheeb, R., and Hofmann, M.-C. (2011). Human spermatogonial stem cells: A possible origin for spermatocytic seminoma. International Journal of Andrology 34, e296–e305.

Germ Cell Tumors: Pathology and Genetics

145

Fig. 34 A simplified scheme of the spermatogonial stem cell niche showing the main extrinsic factors driving spermatogonial stem cell (SSC) maintenance and self-renewal. Sertoli cells and spermatogonial stem cells (SSCs) are both attached to the basement membrane (BM). Sertoli cells provide for structural support and produce glial cell line-derived neurotrophic factor (GDNF) and basic fibroblast growth factor (bFGF) which are crucial for SSC self-renewal in vitro and in vivo. Leydig cells (L) and peritubular cells (PE) produce colony-stimulating factor-1 (CSF-1), also essential for selfrenewal. Waheeb, R. and Hofmann, M-C. (2011). Human spermatogonial stem cells: A possible origin for spermatocytic seminoma. International Journal of Andrology 34, e296–e305.

a developmental deregulation. Somatic mutations, and whole chromosome-aneuploidy are probably mainly progression related. The tumor promoting niche factors explain the high percentage of bilaterality of spermatocytic tumor.

(Cyto)genetics and epigenetics Most spermatocytic tumors are (near)diploid, the second largest group is (near)tetraploid, and a few are peritriploid. The only consistent cytogenetic aberration is extra copies of chromosome 9 (Fig. 35). The candidate gene on this chromosome involved in tumorigenesis is DMRT1 (Fig. 36). A small number of tumors, usually in the oldest half of the patients, have mutually exclusive, paternal age-related mutations in FGFR3, HRAS or NRAS. In 80% of spermatocytic tumors p53 is expressed. Spermatocytic tumors have heterogeneous pattern of DNA methylation and histone modification, different from the regular patterns in normal spermatogenesis, probably because the tumor cells are deprived of regulatory niche factors.

Type IV GCT (Dermoid Cyst of the Ovary) Epidemiology and risk factors The dermoid cyst of the ovary is the most frequent GCT, with an estimated incidence between 10 and 15, occurring at the reproductive age (median age 30, range 10–90) (Fig. 37). Over 10% of the patients have bilateral tumors.

Anatomical localization Dermoid cysts occur exclusively in the ovaries. On rare occasions, they become detached from the ovary and reimplanted in either omentum, fallopian tube, or Douglas’ pouch. Tumors in the testis and extragonadal sites resembling dermoid cysts are type I GCT beyond infancy, with a developmental potential in between that of type I and type IV GCT.

Pathology Dermoid cysts are derived from parthenogenetically activated primary oocytes (Fig. 38) that have escaped the normal, postpubertal regulation of meiotic arrest, and started uncontrolled meiosis, which explains why they occur at reproductive age. Genetic variants in the factors regulating meiotic arrest may be involved in proneness for bilaterality and familial occurrence of type IV GCT. Thus, it seems probable that, as in other GCT, the origin of type IV GCT is mainly determined by developmental factors, with a minor role for genetic events. Half of the dermoid cysts have the molecular characteristics of primary oocytes that have been parthenogenetically activated after meiosis I (Type-I), a quarter after meiosis II (Type-II), and the remaining quarter after endoreduplication (Type-III) (Fig. 39). Dermoid cysts are completely mature teratomas, which present as a thin-walled cyst lined with epidermis with attached appendages and filled with sebaceous material and hairs. Usually one solid nodule (Rokitansky’s protuberance) protrudes from the wall into the lumen of the cyst. It is mainly composed of cranial tissues: skin covered with hair (scalp), bone (skull), choroid plexus and glial tissue (intracranial tissues), retinal epithelium (eyes), teeth (jaw), and thyroid (neck). In fact, a dermoid cyst represents the rostral part of an attempted embryo turned inside out (Fig. 40). In addition, virtually any adult tissue can be present. Exceptional tumors are predominantly solid with highly organized structures resembling a fetus, however, lacking extraembryonic tissues, as do

146 Germ Cell Tumors: Pathology and Genetics

Fig. 35 Combined CGH data of four spermatocytic tumors; underrepresentation of chromosomes is indicated lines on the left of the chromosome-ideograms, overrepresentation on the right. Chromosome 9 is overrepresented in each of the four cases. Rosenberg, C., Mostert, M. C., Schut, T. B., van de Pol, M., van Echten, J., de Jong, B., Raap, A. K., Tanke, H., Oosterhuis, J. W., Looijenga, L. H. (1998). Chromosomal constitution of human spermatocytic seminomas: Comparative genomic hybridization supported by conventional and interphase cytogenetics. Genes Chromosomes Cancer 23, 286–291.

Germ Cell Tumors: Pathology and Genetics

147

Fig. 36 Detailed representation of chromosome 9 pattern showing a subchromosomal amplification of the 9p21.3-pter region containing DMRT1. Looijenga, L. H., Hersmus, R., Gillis, A. J., Pfundt, R., Stoop, H. J., van Gurp, R. J., Veltman, J., Beverloo, H. B., van Drunen, E., van Kessel, A. G., Pera, R. R., Schneider, D. T., Summersgill, B., Shipley, J., McIntyre, A., van der Spek, P., Schoenmakers, E., Oosterhuis, J. W. (2006). Genomic and expression profiling of human spermatocytic seminomas: primary spermatocyte as tumorigenic precursor and DMRT1 as candidate chromosome 9 gene. Cancer Research 66, 290–302.

Fig. 37 Age distribution of patients with type IV GCT (dermoid cysts) showing that these GCT occur in postpubertal women. Comerci, J. T., Licciardi, F., Bergh, P. A., Gregori, C., Breen, J. L. (1994). Mature cystic teratoma: A clinicopathologic evaluation of 517 cases and review of the literature. Obstetrics & Gynecology 84, 22–28.

148

Germ Cell Tumors: Pathology and Genetics

Fig. 38

Two cell-stage embryo due to parthenogenetic activation of an ooctye in a dysgenetic ovary (H and E, 640).

Fig. 39 The postulated mechanism of human ovarian teratoma formation. The source cells are proposed to be primary oocytes that escaped from meiotic arrest. Subsequent uncontrolled meiotic division could produce ovarian teratomas. PGC, primordial germ cell. Kaku, H., Usui, H., Qu, J., Shozu, M. (2016). Mature cystic teratomas arise from meiotic oocytes, but not from pre-meiotic oogonia. Genes Chromosomes Cancer 55, 355–364.

dermoid cysts. Benign tumors, such as struma ovarii and carcinoids, may arise from organ anlages within a dermoid cyst. Somatictype malignancies develop reportedly in 0.2%–3% of dermoid cysts, of which about 90% are squamous cell carcinomas. Dermoid cysts may sporadically contain immature foci, even more exceptionally combined with YST.

(Cyto)genetics and epigenetics Virtually all dermoid cysts are diploid, a few percent have chromosomal abnormalities including trisomy for chromosomes 7, 8, 12, 15, and X. Trisomy for chromosomes 8, 12, and X may be shared by immature teratomas of the ovary. Somatic-type malignancies have the same genetic changes as their somatic counterparts. For example, malignant struma ovarii has the same BRAF point mutations as papillary carcinomas of the thyroid. The genomic profile of dermoid cysts reflects the stage of meiosis of the oocytes from which they have originated. Most significantly, the chromosomes of dermoid cysts have an exclusively maternal imprinting (Fig. 41), inhibiting the development of trophoblast, like in the mouse gynogenote (Fig. 42). (Table 1).Thus, the developmental potential of type IV GCT corresponds to the 2C state, whereby, due to the exclusively maternal imprinting, only somatic tissues are formed.

Germ Cell Tumors: Pathology and Genetics

Fig. 40

149

Gross image of type IV GCT (dermoid cyst), fully developed hair on a scalp-like covering of Rokitansky’s protuberance (dermoid hill).

Fig. 41 Cycle of genomic imprinting (GI). Upon fertilization, the zygote acquires a haploid set of paternally imprinted chromosomes from the father and a haploid set of maternally imprinted chromosomes from the mother; the cells of the embryo therefore have a biparental GI pattern. In the germline, GI is erased; during spermatogenesis and oogenesis, respectively, paternal and maternal imprinting is reestablished. Hirasawa, R. and Feil, R. (2010). Genomic imprinting and human disease. Essays in Biochemistry 48, 187–200.

Type V GCT (Molar Pregnancies) Epidemiology and risk factors The incidence of gestational trophoblastic disease, mostly complete hydatidiform moles, is 1 in 120 pregnancies in some parts of Asia and South America, a factor 10 higher than in Western societies. Risk factors are pregnancy at young or old age, prior gestational trophoblastic disease, Asian ethnicity, and possibly dietary deficiencies and low socioeconomic status. Genetic risk factors are maternal mutations of NALP7/NLRP7 on 19q13.4, causing abnormal imprinting with overexpression of the paternal genome, which results in recurrent familial biparental complete hydatidiform moles and reproductive wastage.

Anatomical localization Molar pregnancies occur virtually always in the uterus, occasionally in a fallopian tube as an ectopic pregnancy.

Pathology Complete hydatidiform moles are caused by androgenesis with two, rarely four, haploid sets of paternal chromosomes, arising from fertilization of an anuclear empty ovum by one 23,X sperm that replicates its chromosomes. Those with a 46,XY karyotype result from fertilization of an empty ovum by two sperm. Both mechanisms create a zygote with an exclusively paternal imprint, which

150

Germ Cell Tumors: Pathology and Genetics

Fig. 42 Compare development of control embryo (A) with that obtained from eggs with two maternal genomes, gynogenotes (B), in which a small but well-advanced 25-somite embryo was the maximum development but with poor extraembryonic tissues. The eggs with two paternal nuclei, androgenotes (C), developed maximally to about the 6- to 8-somite stage but with extensive trophoblast development. YS yolk sac, TB trophoblast. Surani, M. A., Barton, S. C., Norris, M. L. (1986). Nuclear transplantation in the mouse: Heritable differences between parental genomes after activation of the embryonic genome. Cell 45, 127–136.

gives rise to placental tissue only and no somatic tissues of the embryo proper, similar to experimentally produced mouse androgenotes (Fig. 42) (Table 1) The precursor cell of the hydatidiform mole has the 2C-state developmental potential, except for the ability to form somatic tissues of the embryo proper. Consistent with this pathogenesis complete hydatidiform moles, type V GCT, are composed of hypertrophic, dysplastic placental tissue only, lacking somatic tissues of the embryo proper (Fig. 43). Thus, these growths are the mirror image of dermoid cysts, which are composed of somatic tissues only, and lack trophoblastic tissue. They progress to choriocarcinoma in 2%–3% of cases, possibly driven by hypomethylation-associated genomic instability.

Genetics and epigenetics Complete hydatidiform moles are generally diploid with a 46,XX (90%) or 46,XY (10%) karyotype; rare cases are tetraploid.

Fig. 43 Gross image of hydatidiform mole, resembling a bunch of grapes, shown in an opened uterus. The individual grapes consist of hypertrophic, edematous placental villi.

Germ Cell Tumors: Pathology and Genetics

151

Type VI GCT Definition Type VI GCT are defined as neoplasms resembling GCT as to their developmental potential. However, they are not derived from germ cells, but from somatic cells induced to pluripotency (Table 1).

Epidemiology and risk factors As type VI GCT are a new concept, epidemiological data are scarce. With a median age of over 50, the patients are old for a germ cell tumor. Most type VI GCT develop in advanced cancers, in particular common epithelial cancer of the ovary; few arise de novo.

Anatomical localization The anatomical distribution follows the various types of cancer in which of GCT components may develop, in particular of the ovary and stomach. Sinonasal teratocarcinosarcomas are virtually always located in the nasal cavity and/or ethmoid sinus. The few published de novo type VI GCT with balanced translocations were localized at atypical sites for GCT, certainly in view of the old age of the patients: retroperitoneum, posterior mediastinum, and inside the sacrum.

Pathology Type VI GCT developing in somatic cancers may contain all elements of GCT apart from seminoma, either pure or in various combinations. Probably the most frequent types are YST and immature teratoma. Thus their developmental potential resembles that of type I GCT and the nonseminoma variant of type II GCT, like genetically engineered human induced pluripotent stem cells (iPSC) (Fig. 44).

(Cyto)genetics Little is known on the genetics of GCT originated in somatic cancers. Overrepresentation of 12p has not been demonstrated. In a subpopulation of tumor cells of a nasal teratocarcinosarcoma an extra copy of 12p13 (a feature of type I GCT) was demonstrated by ISH. The published de novo type VI GCT showed complex balanced translocations, with 6p21 being a common breakpoint in each of them; chromosome 12 was not involved. (Fig. 45) A nasal immature teratoma was diploid with a balanced translocation t(1;11)(q12;p15). In type VI GCT developing in somatic cancers, various genetic and epigenetic changes may de-repress pluripotency genes. MYC might play a crucial role, since it is a central player in oncogenesis and pluripotency. Indeed, more aggressive cancers may express both the core pluripotency genes (OCT4, NANOG, SOX2, and KLF4) and MYC-centered networks. The few cytogenetically characterized de novo type VI GCT suggest that breakpoints in certain chromosomal regions might activate the pluripotency program. The breakpoint in 6p21–22 in these tumors could involve OCT4. The overrepresentation of 12p13 in a sinonasal teratocarcinosarcoma may have led to overexpression of the pluripotency cluster NANOG, STELLAR, and GDF3.

Fig. 44 Nonseminoma component of a type VI GCT developed in a clear cell carcinoma of the ovary with EC cells expressing OCT4. Nogales, F. F., Prat, J., Schuldt, M., Cruz-Viruel, N., Kaur, B., D’Angelo, E., Matias-Guiu, X., Vidal, A., McCluggage, W. G., Oosterhuis, J. W. (2018). Germ cell tumour growth patterns originating from clear cell carcinomas of the ovary and endometrium: A comparative immunohistochemical study favouring their origin from somatic stem cells. Histopathology 72, 634–647.

152

Germ Cell Tumors: Pathology and Genetics

(A)

(B)

Fig. 45 Karyotype (A), and cartoon of the translocations (B) of a de novo type VI GCT arisen in posterior mediastinum, showing breakpoints in 6p21, 6p22, 6q23, and 11q13, with the chromosomal regions, 6p22::6q23 and 6p21::11q13, involved in fusions. Van Echten, J., de GJong, B., Sinke, R. J., Weghuis, D. O., Sleijfer, D. T., Oosterhuis, J. W. (1995). Definition of a new entity of malignant extragonadal germ cell tumors. Genes Chromosomes Cancer 12, 8–15.

GCT With Intermediate Phenotypes Consistent with the plasticity of the developmental states of embryonic stem cells, there are intermediate phenotypes between the seven defined types of GCT. There is continuum between multiple pregnancies, conjoined twins, parasitic twins, and type I GCT, with intermediate types between type 0 and type I GCT, which could arbitrarily be classified in either type. Type I GCT and type II GCT have gradual transitions, in particular among prepubertal GCT of the mediastinum in Klinefelter’s and among GCT of the brain in patients with Down’s syndrome. Such tumors are genotypically type II but phenotypically resemble type I, suggesting that the genomic changes typical for type II, particularly gain of 12p, have occurred in a PGC that is still too heavily methylated to allow the full spectrum of the totipotent developmental potential of a type II GCT.

Germ Cell Tumors: Pathology and Genetics

153

The type I GCT beyond infancy, composed of (immature) teratoma combined with dermoid cysts likely represent a transitional phenotype between type I and type IV GCT. It may be hypothesized that such tumors are derived from a precursor cell somewhere in between an oogonium and a type I oocyte, in which maternal GI is not yet completed. There is at least one published case of a spermatocytic tumor that is intermediate between a type II and a type III GCT, in terms of morphology, chromosomal composition, and behavior: a seminomatous morphology, lacking lymphocytes, gain of 12p and chromosome 9, and metastasis. Type VI tumors have developmental characteristics with features of type I GCT and type II nonseminomas, and intermediate phenotypes.

Conclusions GCT are indeed one family of gonadal and extragonadal tumors derived from germ cells and their precursors. Their developmental potential/pathology is related to a developmental state rather than a specific cell of origin, as a particular originating cell may assume different developmental states, and different cell types may have the same developmental potential, in agreement with developmental state plasticity. A new concept is the type VI GCT, derived from somatic cells, most often cancer cells, induced to pluripotency. Apart from type VI, most GCT are caused via the so-called developmental pathway: disturbance of niche-, and cell intrinsic factors that control the developmental potential of germ cells and their precursors. The somatic mutation pathway is much rarer; most mutations occur during tumor progression. This developmental pathogenesis explains the strong familial component of most GCT, including clustering of different types of GCT, their frequent bilaterality, and their intact p53 response, which renders malignant GCT highly sensitive to DNA-damaging agents. Having TSPY, located in the GBY region of the Y chromosome, is the most important risk factor for a type II GCT, regardless of anatomical site and phenotypic gender. Coexpression of OCT4 and TSPY in maturation delayed PGC/gonocytes emerges as a plausible factor for initiation of type II GCT via the developmental pathway. Clinically it is important to differentiate between type I and IV teratomas, which are benign, and type II teratoma, which may look identical microscopically but behaves malignant.

Prospective Vision For the advancement of the field of GCT it is important to reconsider the traditional site-oriented approach, which has led to the present confusing nomenclature, and blurred entities. GCT should rather be considered as one disease manifesting in different organs, like for example soft tissue tumors and lymphomas. This change will stimulate the development of a unifying, biologically plausible, and clinically relevant classification of existing and emerging entities, including tumors in the wake of clinical application of iPSC. Finally, it will provide a reliable basis for genotype/phenotype correlations carried out in joint efforts of basic scientists and clinicians.

Further Reading Ben-David, U., Benvenisty, N., 2011. The tumorigenicity of human embryonic and induced pluripotent stem cells. Nature Reviews Cancer 11, 268–277. Cools, M., Drop, S.L., Wolffenbuttel, K.P., Oosterhuis, J.W., Looijenga, L.H., 2006. Germ cell tumors in the intersex gonad: Old paths, new directions, moving frontiers. Endocrine Reviews 27, 468–484. De Backer, A., 2006. Surgical treatment and outcome of extracranial germ cell tumors in childhood [Doctoral Thesis]. Erasmus University Medical Center, Rotterdam, ISBN 908559-166-X. Giannoulatou, E., Maher, G.J., Ding, Z., et al., 2017. Whole-genome sequencing of spermatocytic tumors provides insights into the mutational processes operating in the male germline. PLoS One 12 (5), e0178169s. https://doi.org/10.1371/journal.pone.0178169. Greene, M.H., Mai, P.L., Loud, J.T., et al., 2015. Familial testicular germ cell tumors (FTGCT)dOverview of a multidisciplinary etiologic study. Andrology 3, 47–58. Hackett, J.A., Surani, M.A., 2014. Regulatory principles of pluripotency: From the ground state up. Cell Stem Cell 15, 416–430. Kaku, H., Usui, H., Qu, J., Shozu, M., 2016. Mature cystic teratomas arise from meiotic oocytes, but not from pre-meiotic oogonia. Genes, Chromosomes & Cancer 55, 355–364. Killian, J.K., Dorssers, L.C.J., Trabert, B., et al., 2016. Imprints and DPPA3 are bypassed during pluripotency and differentiation-coupled methylation reprogramming in testicular germ cell tumors. Genome Research 26, 1490–1504. Lindeman, R.E., Gearhart, M.D., Minkina, A., et al., 2015. Sexual cell-fate reprogramming in the ovary by DMRT1. Current Biology 25, 764–771. Mamsen, L.S., Brochner, C.B., Byskov, A.G., Mollgard, K., 2012. The migration and loss of human primordial germ stem cells from the hind gut epithelium towards the gonadal ridge. The International Journal of Developmental Biology 56, 771–778. Oosterhuis, J.W., Looijenga, L.H.J., 2005. Testicular germ-cell tumors in a broader perspective. Nature Reviews Cancer 5, 210–222. Oosterhuis, J.W., Looijenga, L.H.J., 2017. Germ cell tumors from a developmental perspective: Cells of origin, pathogenesis and molecular biology; emerging patterns. In: Nogales, F.F., Jimenez, R.E. (Eds.), Pathology and biology of human germ cell tumors. Springer Nature, Berlin, pp. 23–129. Skakkebaek, N.E., Rajpert-De Meyts, E., Main, K.M., 2001. Testicular dysgenesis syndrome: An increasingly common developmental disorder with environmental aspects. Human Reproduction 16, 972–978. Spencer, R., 2001. Parasitic conjoined twins: External, internal (fetuses in fetu and teratomas), and detached (acardiacs). Clinical Anatomy 14, 428–444. Surani, M.A., Barton, S.C., Norris, M.L., 1986. Nuclear transplantation in the mouse: Heritable differences between parental genomes after activation of the embryonic genome. Cell 45, 127–136.

Glioblastoma: Biology, Diagnosis, and Treatment Akanksha Sharma and Maciej M Mrugala, Mayo Clinic, Scottsdale, AZ, United States Jan Buckner, Mayo Clinic, Rochester, MN, United States © 2019 Elsevier Inc. All rights reserved.

Glossary Akt/PI3K signaling pathway A signal transduction pathway involving the phosphatidylinositol 3-kinase (P13K) and protein kinase B (Akt) proteins that promotes survival and growth in response to extracellular signals. Cytotoxic T-lymphocyte associated protein 4 (CTLA-4) A protein receptor that functions as an immune checkpoint and can downregulate immune response. It is over-expressed in tumor cells, thus helping them evade the body’s immune system. Epidermal growth factor receptor (EGFR) A transmembrane protein that is a receptor for members of the epidermal growth factor family of extracellular protein ligands. Mutations that lead to its over-expression are associated with various cancers including glioblastoma. Glial fibrillary acidic protein (GFAP) A protein expressed by numerous cell types of the central nervous system including astrocytes, and involved in cell communication and the blood brain barrier. It is heavily expressed in glioblastoma. Isocitrate dehydrogenase (IDH) An enzyme that catalyzes the oxidative decarboxylation of isocitrate, producing alphaketoglutarate and CO2. Mutations in the isocitrate dehydrogenase gene IDH1 have been found in several brain tumors including glioblastoma and hold prognostic implications. Mammalian target of rapamycin (mTOR) kinase A protein kinase that is involved in regulating cell growth, cell proliferation, cell motility and transcription, amidst other functions, and is implicated in the development of glioma. Mitogen-activated protein (MAP) kinase A protein kinase involved in regulation of cell functions including proliferation, differentiation, mitosis, cell survival and apoptosis, and is implicated in the development of glioma. O-6-methylguanine-DNA methyltransferase (MGMT) An enzyme encoded by the MGMT gene and is crucial for genome stability through its mechanism of repairing mutagenic DNA lesions and preventing mismatch and errors during DNA replication and transcription. Phosphatase and tensin homolog (PTEN) gene The PTEN protein encoded by this gene acts as a tumor suppressor by negatively regulating the Akt/PKB signaling pathway. Programmed cell death ligand 1 (PD-L1) A protein that suppresses the immune system by reducing proliferation and inducing apoptosis of T cells. Cancer cells upregulate expression of this protein and this may help them evade the body’s immune system. Telomerase reverse transcriptase (TERT) Subunit of the enzyme telomerase which comprises an important unit of the telomerase complex. Tumor treating fields (TTF) Electromagnetic field therapy that applies low-intensity electrical fields to target areas containing proliferating tumor cells, impairing mitosis in these cells and thus selectively causing tumor cell death. Vascular endothelial growth factor (VEGF) An important signaling protein that stimulates the formation of blood vessels. World Health Organization (WHO) classification Classification of tumors of the central nervous system

History In the mid-1800s German pathologist Rudolf Virchow coined the term “glioma” and became the first to attempt to classify this group of neoplasms. Glioblastoma, the most aggressive of these gliomas, was introduced by its name in 1926 by Percival Bailey and Harvey Cushing in their book “A classification of the tumors of the glioma group on a histogenetic basis with a correlated study of prognosis.” The term was inspired by the fact that the tumor appeared to originate from primitive precursorsd“glial cells”dand had a highly variable “multiform” appearance due to necrosis, hemorrhage, cysts, and blood vessels. Cushing and Bailey attempted to organize the gliomas into 10 groups, correlating with patient survival. Over the next few decades, various different classification systems were used until 1993, when the World Health Organization (WHO) introduced its grading system. Morphologic features continue to form the basis for classifying gliomas. Four distinct grades of glioma exist, with grades I representing circumscribed, generally slow-growing tumors and grades II, III, and IV representing diffusely infiltrating tumors with variable rates of growth. This classification was recently revised in 2016 and is discussed in greater detail below.

Epidemiology Primary brain tumors have an incidence rate of 7 per 100,000 with a prevalence of almost 222 per 100,000 individuals. The incidence of primary brain tumors appears to be increasing over the last 30–40 years, especially in the elderly. Almost 35,000 new

154

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65616-7

Glioblastoma: Biology, Diagnosis, and Treatment

155

diagnoses will be this year, and in 2017 alone, the condition will lead to almost 17,000 deaths. Of the primary brain tumors, 33% are malignant in nature. Glioblastoma is the most common form of malignant glioma, accounting for up to 70% of the cases. It is also the most aggressive and has the highest histological grade of the gliomas (WHO Grade IV). The typical age at presentation is between 45 and 55 years of age. Malignant gliomas including glioblastoma are almost twice as likely to occur in men than women, and twice as common in whites as in blacks.

Classification The 2016 WHO classification of brain tumors categorizes glioblastoma by isocitrate dehydrogenase (IDH) status into IDH-wild type, IDH-mutant, and not-otherwise-specified subtypes. The IDH-wild type glioblastoma generally presents de novo (primary glioblastoma) in older (> 50 years of age) patients, and 90% of glioblastomas fall into this subtype. These tumors exhibit epidermal growth factor receptor (EGFR) amplification and mutations, loss of heterozygosity of chromosome 10q, deletion of the phosphatase and tensin homolog (PTEN) on chromosome 10, and p16 deletion. The IDH-mutant glioblastoma typically present in younger patients with or without a previous diagnosis of a lower grade glioma that transforms to an aggressive, grade IV secondary glioblastoma. These have mutations in the p53 tumor suppressor gene, overexpression of the platelet derived growth factor receptor (PDGFR), abnormalities in the p16 and retinoblastoma pathways, and loss of heterozygosity of chromosome 10q. Primary and secondary glioblastomas differ at the molecular level, with differences in DNA copy number and their transcriptional patterns, but on a morphological level appear essentially identical. The glioblastoma NOS (not otherwise specified) is reserved for cases where the IDH status cannot be clarified further (Table 1). A new variant to the 2016 classification is the epithelioid glioblastoma, which falls under the IDH-wildtype category. These tumors typically occur in children and young adults and are located in the cerebrum or diencephalon. They are characterized by large epithelioid cells with abundant eosinophilic cytoplasm and frequently have BRAFV600E mutations.

Etiology The only environmental risk factor that has consistently demonstrated brain tumor risk has been exposure to therapeutic doses of ionizing radiation. Several studies have raised concern about cell phone usage and glioma, but to date the evidence overall has been inconsistent. These studies may be influenced by recall and selection bias. Diseases such as diabetes mellitus, hyperlipidemia, obesity, etc., have been evaluated and are not associated with any increased risk of developing glioblastoma. Epidemiological evaluations have noted that women have a lower glioma risk, and this protective effect is more obvious premenopause. However, the protective role of female hormones remains unclear and has still to be established conclusively. Head injury, diet, smoking, alcohol consumption, exposure to electromagnetic fields, prenatal exposure to drugs or medications, are some of the many factors that have been studied and the evidence has been minimal or inconclusive at best. Genetic factors have also been evaluated. While specific genetic polymorphisms are associated with specific types of gliomas, and may impact DNA repair and cell cycle regulation, patterns of occurrence are not always consistent with hereditary disease. Certain hereditary syndromes such as neurofibromatosis, Turcot and Li–Fraumeni syndromes are associated with increased brain tumor risk, but these account for a very small proportion of incident cases. There is evidence that immunological factors that are related to allergic conditions and infections may have an impact on glioblastoma risk. Many studies evaluating risk of glioma in patients with allergic conditions (such as asthma or hay fever) or auto-immune diseases have found that risk of developing a glioma or glioblastoma is actually decreased in these conditions. Prior infection with certain viruses (herpes virus or varicella zoster virus, for instance) may also be associated inversely with adult glioma risk.

Table 1

Classification of glioblastoma under the 2016 WHO Guidelines Glioblastoma

Type Age distribution Molecular characteristics Prognosis

IDH Wildtype

IDH Mutant

Spontaneous, de novo Generally >50 years of age EGFR amplification Loss of heterozygosity of chromosome 10q Deletion of PTEN Worse

Transformation of lower grade glioma Generally younger adults p53 Tumor suppressor Gene mutations Overexpression of PDGFR Better

156

Glioblastoma: Biology, Diagnosis, and Treatment

Pathological Features As already mentioned, glioblastoma is classified as WHO Grade IV, and thus has the highest grade of the malignant gliomas. Anaplastic astrocytomas, which are WHO Grade III, demonstrate high cell density, nuclear atypia, and mitotic activity on pathology. Glioblastoma is distinguished from the grade III pathology by the addition of abundant areas of endothelial proliferation or necrosis. Pseudopalisading necrosis may or may not be seen. IDH evaluation is routinely performed on all suspected glioblastomas, since the new classification system relies on knowing the wildtype or mutant status of the IDH gene. The tissue may also be evaluated for presence or absence of O-6-alkylguanine DNA alkyltransferase (MGMT) promoter methylation. Methylation of the MGMT promoter prevents gene transcription, reducing tumor cell DNA damage repair. Thus, methylation of the MGMT promoter makes the tumor cell more susceptible to chemotherapeutic agents that cause DNA damage, and subsequently, cellular death.

Cell of Origin and Signaling Pathways Origins of glioblastomas and malignant gliomas have been explored since the first time the tumor was described by Dr. Harvey Cushing. Neural stem cells are proposed to be the primary origin of these tumor cells, primarily due to similarities in protein and cell surface marker expression. Neural stem cells have been known to express nestin, glial fibrillary acidic protein, and also a CD-133 cell surface marker that may lend properties of self-renewal to tumor cells. Epidermal growth factor receptor (EGFR) and platelet derived growth factor receptor (PDGFR) are oncogenes that are often overexpressed in 40%–50% of primary glioblastomas. These likely create an autocrine loop that stimulates tumor proliferation. Moreover, several signal transduction mediators may impact the phosphatidylinositol-4,5-bisphosphate 3-kinase (PI3-K) pathways in these cellsdthese kinases are involved in cellular growth, proliferation, differentiation, and survival. These include the Ras-mitogen activated protein (MAP) kinase pathway and the mammalian target of rapamycin (mTOR) pathway. Mutation in or deletions of phosphatase and tensin homolog (PTEN), a tumor suppressor gene, may result in increased activity of the P13-K pathway also. Molecular therapeutics are currently targeting these pathways in an attempt to inhibit cell growth and encourage apoptosis of tumor cells.

Clinical Features and Diagnostic Approach Primary brain tumor patients can present with either focal or generalized signs and symptoms. Headaches, nausea, vomiting, fatigue, confusion or altered/depressed mental status are more generalized and fairly nonspecific symptoms. Focal seizures, weakness and/or sensory loss of a particular side or limb, language and visuospatial difficulties are focal symptoms that may be more alarming and prompt earlier care. Headache has been known to be the most common initial presenting symptom, as evaluated in many studies of patients with high grade primary brain tumor. A new diagnosis of headache or a distinct change in headache pattern in a patient older than 50 should be considered a red flag. Chronic, persistent headache with protracted nausea, vomiting, especially with positional worsening should also be evaluated carefully. This is also true for headaches that present in the morning, or wake the patient up from sleep, or are provoked by Valsalva maneuvers, coughing or straining. When concerned, evaluation can generally be initiated with a contrast enhanced computed tomography (CT) scan of the head. At times it may be reasonable to go directly towards a gadolinium enhanced magnetic resonance imaging (MRI) scan if index of suspicion is high enough. An MRI can also provide many clues toward the nature of the tumor, based on location, presence and degree of enhancement, diffusion restriction, and perfusion, but definitive diagnosis is still made by pathological examination of tissue after excision or biopsy. Imaging in glioblastomas often reveals irregularly shaped borders, and a ring-shaped zone of contrast enhancement around a dark central area of necrosis. Significant peritumoral edema may be seen on fluid attenuation inversion recovery (FLAIR) sequences. Necrotic tumors will also have a high apparent diffusion coefficient (ADC). ADC may be also helpful in estimating cellularity of a tumor (Fig. 1). Tumor perfusion, as measured by relative cerebral blood volume (rCBV), tends to be elevated in glioblastoma. Proton magnetic resonance spectroscopy or MRS can detect levels of metabolites. Areas with malignant gliomas show an increase in the choline peak and a decrease in the N-acetyl aspartate or NAA peak when using the MRS. Positron emission tomography (PET) is being increasingly investigated, but thus far the data are insufficient to support its use in routine evaluation of central nervous system tumors (Fig. 2).

Prognosis Overall median survival is still less than 15 months in unselected patients and further reduced in the elderly population. Two-year survival rate in patients 65–74 years of age is only 8.9%, and only 3.2% for those greater than 75 years of age. Over the years, several factors have been demonstrated to be related to prognosis in glioblastoma patients. In addition to age, these include the Karnofsky Performance Status score, and the extent of resection.

Glioblastoma: Biology, Diagnosis, and Treatment

157

Fig. 1 Panel A and B Postcontrast T1-weighted and FLAIR MRI images showing left frontal lobe IDH wildtype glioblastoma; Panel C and D Postcontrast T1-weighted and FLAIR MRI images of a different patient with multifocal glioblastoma.

Fig. 2

PET/MRI of recurrent high grade glioma. Arrow points toward areas of increased uptake of radioactive label suggesting active disease.

158

Glioblastoma: Biology, Diagnosis, and Treatment

Tumor IDH mutation and MGMT promoter methylation are extremely strong prognostic factors. Median survival of patients with IDH-mutant glioblastoma (which is a much smaller percentage) can be as high as 66.8 months in some cases. Similarly, patients with methylation of the MGMT promoter had a median survival of 21.7 months on standard of care versus 15.3 months for those with unmethylated MGMT promoter. A combination of MGMT promoter methylation and IDH1 mutation may have the best prognosis, thus, with an overall survival of up to 70 months in some series. It has also been noted that glioma patients with telomerase reverse transcriptase (TERT) mutations have poorer overall survival (Figs. 3 and 4).

Fig. 3 Survival in glioblastoma by molecular subtype. Glioma patients with TERT mutations have poorer survival, while IDH mutations confer a survival benefit. Mutations of both TERT and IDH appears to be helpful for survival. From Eckel-Passow, J. E., Lachance, D. H., Molinaro, A. M., Walsh, K. M., Decker, P. A., Sicotte, H., et al. (2015). Glioma groups based on 1p/19q, IDH, and TERT promoter mutations in tumors. The New England Journal of Medicine 372(26), 2499–2508.

Glioblastoma: Biology, Diagnosis, and Treatment

159

Fig. 4 Survival in glioblastoma by MGMT methylation status. Survival is longer with the combination of radiation therapy plus temozolomide in patients with either methylated or unmethylated tumors, but the treatment effect is found to be greater in patients with methylated tumors. From Stupp, R., Hegi, M. E., Mason, W. P., van den Bent, M. J., Taphoorn, M. J., Janzer, R. C., et al. (2009). Effects of radiotherapy with concomitant and adjuvant temozolomide versus radiotherapy alone on survival in glioblastoma in a randomised phase III study: 5-year analysis of the EORTC-NCIC trial. Lancet Oncology 10 (5), 459–466.

Treatment There is currently no cure for glioblastoma. Whenever possible, patients should be offered a clinical trial as a part of their therapy plan. Maximal safe resection has been recommended as the standard of care for high grade gliomas such as glioblastoma. Various retrospective studies and meta-analyses have established that obtaining a gross total resection (GTR) when possible can improve outcomes including overall survival when compared to a subtotal resection or biopsy alone. Tumor location often dictates safety and feasibility of the resection, since morbidity of the surgery has to be balanced with the benefit from resection. Radiation is the second pillar of therapy in glioblastoma. It should be initiated as soon as it is safe after surgical intervention, and is generally started between 3 and 6 weeks from surgery, barring any complications or infection. For patients who are younger than 65 with Karnofsky performance status greater or equal to 60, optimal dose fractionation for external beam radiation therapy after resection or biopsy is 60 Gy in 2-Gy fractions, delivered over 6 weeks. This schedule has been repeatedly demonstrated to have the maximal benefit. Elderly patients greater than 65 with a Karnofsky greater or equal to 50 are also advised external beam radiation therapy, but recent studies have demonstrated that hypofractionated radiotherapy has similar survival, lesser side effects, including

160

Glioblastoma: Biology, Diagnosis, and Treatment

Fig. 5 Overall and progression-free survival in elderly patients treated with hypofractionated radiation in combination with chemotherapy. From Perry, J. R., Laperriere, N., O’Callaghan, C. J., Brandes, A. A., Menten, J., Phillips, C. et al. (2017). Short-course radiation plus temozolomide in elderly patients with glioblastoma. The New England Journal of Medicine 376(11), 1027–1037.

lower steroid requirement during treatment. This is a dosing of 40 Gy in 15 fractions over 3 weeks instead of the schedule described above. Patients with poor Karnofsky performance status may receive hypofractionated radiotherapy alone, chemotherapy alone, or best supportive care. Partial brain radiation therapy is recommended, since no benefit has been seen with whole-brain radiation therapy. Radiation techniques involve using image guided radiotherapy and intensity modulated radiotherapy to deliver a high dose of radiation to the target area, while delivering a lower dose to the margin, thus sparing healthy surrounding tissue. While the evidence is weak, reirradiation to focal areas of recurrence with stereotactic radiosurgery or hypofractionated radiotherapy has been tried and may improve outcomes (Fig. 5). Chemotherapy is the third and integrally important component of treatment. Prior to 2005, standard therapy consisted of maximal safe resection followed by radiotherapy and adjuvant carmustine, a nitrosourea. Significant survival benefit was not established with this regimen. In 2005, reported results using the “Stupp protocol” changed the management of glioblastoma patients. Patients receiving temozolomide, an oral methylating agent, with concurrent radiotherapy, followed by six monthly cycles of adjuvant temozolomide experienced prolongation of median survival by 2.5 months. Two year survival with this regimen

Glioblastoma: Biology, Diagnosis, and Treatment

161

improved by 16%. In addition, temozolomide is taken orally, and is generally well tolerated. The more common side effects included fatigue, headache, nausea, gastrointestinal discomfort. Grade 3 or 4 cytopenias are uncommon. Following the results of this trial, standard of care for newly diagnosed glioblastoma transitioned to maximal surgical resection followed by radiotherapy, generally at a dose of 60 Gray given in 30 fractions over 6 weeks, with concomitant temozolomide at 75 mg/m2 per day for the 42 days of radiotherapy. Four weeks after radiation, adjuvant temozolomide is given at 150–200 mg/m2 on days 1–5 of each 28 day cycle, for 6 cycles. Treatment effect is greatest in patients with MGMT promoter-methylated tumors. Carmustine wafers are another Food and Drug Administration (FDA) approved chemotherapeutic approach that may be undertaken with newly diagnosed glioblastoma at time of resection. Small polymers containing carmustine are implanted into the tumor bed, with the idea that slow release of the drug over the course of several weeks may kill residual tumor cells. Median survival has been seen to improve by approximately 2 months with integrated use of these polymers in newly diagnosed glioblastoma. These wafers do have significant limitations, and cannot be offered to every patient. Adverse effects in most studies include brain edema, seizures, wound-healing problems, infection, thrombosis, and intracranial hypertension. The NovoTTF-100A system known as Optune was recently approved by the FDA for patients with recurrent and newly diagnosed glioblastoma. This is a portable, noninvasive device that generates low intensity, intermittent frequency alternating electric fields and delivers them via transducer arrays that are placed directly on the patient’s scalp. The electrical fields, also called tumortreating fields, have antimitotic effects by interfering with mitotic spindle cell formation and chromosomal segregation during tumor cell division. Usage compliance rates of > 18 h a day are required for the best outcomes. Treatment begins in conjunction with temozolomide following completion of radiation therapy. When combined with temozolomide, median survival has been shown to improve to 20.5 months from 15.6 months with chemotherapy alone. Recent data from Stupp and colleagues indicate a 2 year survival rate of 43% in the TTF-treated group versus 31% in the group treated with standard of care alone. At 5 years, the difference remainedd13% in the treatment group and 5% with standard of care alone. Unfortunately glioblastoma almost always recurs. When it does recur, treatment options vary. Reoperation may be feasible and is often offered depending on the functional status of the patient and tumor (recurrence) location, size, and safety of the resection. There are no definitive data yet that the second surgery provides survival benefit, but it may be helpful to confirm the lesion is indeed recurrence and not radiation necrosis or inflammation. Radiotherapy for recurrent glioblastoma is controversial at this time and though frequently used, has not been established as having clear survival benefit. This is being evaluated in clinical trials. Stereotactic radiosurgery may be an option that can benefit these patients that have a small relatively well-circumscribed pattern of recurrence. Carmustine wafers have been used in patients with recurrent glioblastoma with modest increase in time to progression. Several other chemotherapy agents have been assessed, including lomustine, procarbazine with lomustine, and vincristine (PCV), carboplatin, irinotecan, etoposide, and others. However, benefits of these agents are modest at best, with no evidence that any of the agents improves survival. Rechallenge with temozolomide has been studied, with inconclusive results. Bevacizumab, a vascular endothelial growth factor (VEGF) inhibitor was approved for use in recurrent glioblastoma in 2009. It can be used as a single agent or in combination with the drugs mentioned above. Clinical trials thus far have demonstrated that bevacizumab may improve progression free survival, but does not improve overall survival. It does prove to be very useful in treating peritumoral edema and radiation necrosis in the appropriate patient, thus improving quality of life. The NovoTTF-100A device has also been evaluated in patients with recurrent glioblastoma and compared with physician’s choice of chemotherapy. There was no difference in outcomes between the two treatments.

General Medical Management Apart from treating the tumor, patients with glioblastoma also require management of morbidity associated with their condition and/or its treatment. Seizures, peritumoral edema, venous thromboembolism, fatigue, headaches, and cognitive dysfunction are some of the most common problems encountered in care of the glioblastoma patient. Patients may also develop significant functional disability over the course of their disease. Prevalence of seizures can be as high as 50% in glioblastoma patients. Patients with temporal lobe tumors and those with hemorrhage in the bed of the tumor may be at a higher risk of seizures. No statistically significant benefit has been seen thus far for the prophylactic use of antiepileptics in patients with tumors, even in those treated with a craniotomy for resection. Antiepileptics should only be started when patient has a clearly witnessed seizure or provides history that is concerning enough for a seizure. A single event is usually enough to start an antiepileptic in this high risk population, and an EEG is not always required if the clinical history is convincing. The choice of antiepileptics is based on many considerations, including ease of dosing and tolerance, the side effects and the drug interactions. Generally, drugs that induce the hepatic cytochrome p-450 enzymes (such as phenytoin or carbamazepine) are avoided since they can increase the metabolism of many chemotherapeutic agents. Phenytoin should also be avoided in patients once patients have received cranial radiation, given higher risk of rash in this population. Levetiracetam, lamotrigine, pregabalin, and lacosamide are generally more preferred in this population. They appear to be better tolerated and have fewer drug interactions. Levetiracetam is perhaps the most frequently prescribed antiepileptic for this patient population. Side effects in a small percentage of people include irritability, aggression, hostility, suicidality, and depression. Lacosamide also has few drug interactions, and has fewer of these psychiatric side effects, but can have cardiac side effects and is quite expensive. Valproic acid in recent years has been

162

Glioblastoma: Biology, Diagnosis, and Treatment

investigated for antiglioma properties in addition to anticonvulsant properties, as it may function as a radiosensitizer in glioblastoma patients. However, it does have side effects of teratogenicity, hyperammonemia, alopecia and weight gain, and thus has to be used in the appropriate patient. Patients often complain of fatigue more than any other symptom, and can impact their overall sense of well-being. The disease as well as radiation and chemotherapy all contribute to this symptom. Management should include addressing reversible factors such as medications (anticonvulsants, opioids, antiemetics), diet and vitamin intake, sleep quality, anemia, and nutritional deficiencies. Aerobic exercise and steroids may improve fatigue. Psychostimulants such as modafinil and methylphenidate have been studied with mixed results, but may be appropriate for some patients. Headache is experienced by up to 50% of patients with brain tumors and can get worse toward the end of life. Management of headache in glioblastoma depends on the stage of disease and overall functionality. Steroids may alleviate symptoms when edema is contributing but cannot be used indefinitely due to side effects. Antiinflammatory agents may be helpful when used in a limited manner to avoid overuse and kidney or gastrointestinal injury. Prophylactic agents such as gabapentin and topiramate may be beneficial but have side effects that may outweigh the benefits in some patients. At the end of life, opioids may be preferred, weighing the risk of sedation with pain control. Peritumoral edema can be quite disabling, resulting in focal weakness and even coma. Steroids can treat and control edema in the glioblastoma patient, rapidly alleviating the symptoms. Steroids also work in alleviating nausea, treating headaches, and improving appetite. However, steroids also have significant side effects such as hyperglycemia, weight gain, myopathy, psychosis, delirium, and irritability. In an acute setting, the glioblastoma patient may be disabled by edema from the tumor. The goal should be to start with the lowest possible dose and taper down as quickly and safely as possible. Calcium/vitamin D supplementation and a proton pump inhibitor or histamine receptor blocker should be prescribed with the steroids for bone health and gastrointestinal prophylaxis, respectively. Bevacizumab may be another agent that can decrease peritumoral edema in these cases and reduce the need for steroids. Wider use for this indication, however, is limited due to high cost. Glioblastoma patients are at an increased risk of venous thromboembolism. Anticoagulation therapy is recommended and safe for venous thrombo-embolism unless there is obvious systemic or intracerebral hemorrhage. Intratumoral hemorrhage rates in anticoagulated GBM patients are relatively low. Low molecular weight heparin can be safe and more effective than warfarin in this population, and is thus frequently used. Oral thrombin inhibitors have not yet been studied in this specific patient population; in fact, data on their safety and efficacy in cancer patients receiving chemotherapy at large is lacking. Neurocognitive problems may be a presenting symptom of brain tumors or present later in the course of the disease. Impairment in executive functioning, memory, and attention are the areas where deficits are most commonly noted. Memantine has been trialed in a small group of patientsdwhen initiated within 3 days of radiotherapy for 24 weeks, cognitive function over time may be better preserved. No other preventive interventions have been clearly established thus far. Several small studies have noted that methylphenidate at 10 mg twice a day can improve cognition and mood as well as functional status in high grade glioma patients, and can be taken with minimal adverse effects. Modafinil has also been seen to have a beneficial effect in speed of processing and executive function requiring attention in some studies. Fatigue and mood may also improve with this drug.

Prospective Vision The neuro-oncology field, similar to the cancer therapy field at large, is in the midst of a variety of exciting new research that hopes to create new targeted therapy for primary brain tumors. One of these areas is immunotherapydagents developed for advanced systemic tumors have demonstrated significant clinical benefit but this has not yet been true for glioblastoma despite decades of research in the field. The T cell response is key in this processdantigens released from tumor cells at their death are presented to T cells by antigen presenting cells, which in turn prime the T cells against these antigens. Checkpoint pathways between the antigen presenting cells, tumor cells, and T cells, provide signals that activate or inhibit T cells and thus regulate this immune response. The brain, despite being highly protected by the blood brain barrier, does communicate with the immune system since we now know that activated cells are able to traffic through the barrier and the CNS lymphatic system also facilitates immune surveillance. The rationale is that antibodies to checkpoint inhibitorsdspecifically to cytotoxic T lymphocyte-associated antigen-4 (CTLA-4), programmed cell death protein 1 (PD-1), and programmed cell death-ligand (PD-L1)dcould result in increased susceptibility of glioblastoma to an immune response. These are being investigated in clinical trials but at this point, a single phase III trial using one of these agents showed no benefit in the setting of recurrent glioblastoma, but numerous other clinical trials are in early stages and hold promise. Studies are also investigating the combination of these immunotherapies with established therapies, and with radiotherapy. One limitation is that we are still working on understanding how these specific pathways work in the brain and how glioblastoma specifically evades the immune system. It is important to ensure that these treatments will be safe in the CNS, where an intense immune response could have the potential of inflammatory damage to the brain. Assessing response in glioblastoma, and differentiating inflammation from progression, will also be key challenges. Various vaccine-based therapies are also under investigation. One approach utilizes a mixture of autologous tumor cells with allogeneic glioblastoma cell populations that have been inactivated with irradiation. Vaccines with antigen-presenting dendritic cells primed ex vivo with glioblastoma antigens and peptide vaccines to enhance the immune response are in clinical trials. For

Glioblastoma: Biology, Diagnosis, and Treatment

163

example, rindopepimut, an EGFRvIII peptide combined with an adjuvant, failed to demonstrate survival benefit in patients with newly diagnosed glioblastoma, but is currently being studied in combination with bevacizumab in patients with recurrent disease. In recent years T cell therapy has gained a lot of attention in the field of oncology in general, and neuro-oncology is no different. Tumor infiltrating lymphocytes can be harvested, expanded in vitro and re-infused into the patient with the hope that a cytokine immune response will lead to cell death of the tumor. Engineered T cellsdhigh avidity tumor specific transgenic T cells or TCR, and chimeric antigen receptor or CAR T cellsdare being investigated specifically in glioblastoma. Viruses that can infect tumor cells and lead to host cell death are an attractive concept. Various viruses are being investigated in phase I and II trials at this point, including herpes simplex virus-1, adenovirus, poliovirus, zika virus, measles virus, and parvovirus. However, many challenges must be overcome at this point. One is safetydonly malignant tumor cells must be infected and healthy brain tissue surrounding the tumor margins must be spared. In addition, the likely immune response in the brain to the virus must protect the brain without irreversible inflammatory damage, and also not diminish the impact of the viral agent itself. This area continues to carry much excitement and hope for the future, as researchers continue to troubleshoot these aspects. Advances in surgical and imaging techniques also hold promise for the future. While it has been established that maximal safe resection improves prognosis in glioblastoma, achieving that gross total resection has been a challenge on the operating table given limitations in visualizing tumor margins. To this end, advances in imaging have allowed for improved volumetric analyses and surgical planning. Fluorescence guided surgery (“tumor paint”) has been noted to help improve extent of resection and in turn overall survival in advanced glioma. Intraoperative MRI guidance has also been used in several centers to achieve maximal resection with subsequent improvement in progression free survival. Other technologies increasingly available to the surgeon include cortical and subcortical stimulation mapping and intraoperative ultrasonography, techniques that can help preserve normal and essential brain tissue while helping resection of tumor. The field of radiation therapy has had many exciting advances in recent years and we have learned more about what is both efficacious and safe for our patients. Proton beam radiotherapy has received significant attention in recent years for several malignancies, including glioblastoma. The hope is that proton therapy will offer the same or improved benefit with less toxicity since the radiation exposure to normal tissue is less than with conventional photon beam therapy. Trials investigating this in glioblastoma are on going. The future also holds promise with tractography and functional mapping to better diagnose and delineate anatomy. Diffusion tensor imaging tractography may be able to help differentiate between gross tumor and infiltrative margins, allowing for safer treatment. [18F]-fluoromisonidazole has been used to identify hypoxic areas in tissue and is being explored as a means to identify specific different regions within the tumor and tailor treatment doses accordingly. Improved imaging techniques will continue to enable the radiation oncologist to provide the best treatment for the tumor bed, thus impacting overall and progression free survival.

See also: Glioblastoma: Pathology and Genetics.

Further Reading Bernstein, M., Berger, M.S., 2014. Neuro-oncology: The essentials, 3rd edn. Thieme, Stuttgart. Brown, C.E., Alizadeh, D., Starr, R., Weng, L., Wagner, J.R., Naranjo, A., et al., 2016. Regression of glioblastoma after chimeric antigen receptor T-cell therapy. New England Journal of Medicine 375 (26), 2561–2569. D’Amico, R.S., Englander, Z.K., Canoll, P., Bruce, J.N., 2017. Extent of resection in gliomadA review of the cutting edge. World Neurosurgery 103, 538–549. Diaz, R.J., Ali, S., Qadir, M.G., De La Fuente, M.I., Ivan, M.E., Komotar, R.J., 2017. The role of bevacizumab in the treatment of glioblastoma. Journal of Neuro-Oncology 133, 455–467. Eckel-Passow, J.E., Lachance, D.H., Molinaro, A.M., Walsh, K.M., Decker, P.A., Sicotte, H., et al., 2015. Glioma groups based on 1p/19q, IDH, and TERT promoter mutations in tumors. New England Journal of Medicine 372, 2499–2508. Farrell, C.J., Plotkin, S.R., 2007. Genetic causes of brain tumors: Neurofibromatosis, tuberous sclerosis, von Hippel-Lindau, and other syndromes. Neurologic Clinics 25, 925–946. Fisher, J.L., Schwartzbaum, J.A., Wrensch, M., Wiemels, J.L., 2007. Epidemiology of brain tumors. Neurologic Clinics 25, 867–890. Gardeck, A.M., Sheehan, J., Low, W.C., 2017. Immune and viral therapies for malignant primary brain tumors. Expert Opinion on Biological Therapy 17, 457–474. Halani, S.H., Babu, R., Adamson, D.C., 2017. Management of glioblastoma multiforme in the elderly: A review of the literature. World Neurosurgery 105, 53–62. Hegi, M.E., Diserens, A.C., Gorlia, T., Hamou, M.F., de Tribolet, N., Weller, M., et al., 2005. MGMT gene silencing and benefit from temozolomide in glioblastoma. New England Journal of Medicine 352, 997–1003. Louis, D.N., Ohgaki, H., Wiestler, O.D., Cavenee, W.K., Ellison, D.W., Figarella-Branger, D., Perry, A., Reifenberger, G., Deimling, A., 2016. WHO classification of tumors of the central nervous system, 4th edn. World Health Organization, Geneva. Mrugala, M.M., 2013. Advances and challenges in the treatment of glioblastoma: A clinician’s perspective. Discovery Medicine 15, 221–230. Ostrom, Q.T., Gittleman, H., Xu, J., Kromer, C., Wolinsky, Y., Kruchko, C., et al., 2016. CBTRUS statistical report: Primary brain and other central nervous system tumors diagnosed in the United States in 2009–2013. Neuro-Oncology 18, v1–v75. Preusser, M., Lim, M., Hafler, D.A., Reardon, D.A., Sampson, J.H., 2015. Prospects of immune checkpoint modulators in the treatment of glioblastoma. Nature Reviews Neurology 11, 504–514. Schiff, D., Lee, E.Q., Nayak, L., Norden, A.D., Reardon, D.A., Wen, P.Y., 2015. Medical management of brain tumors and the sequelae of treatment. Neuro-Oncology 17, 488–504. Stupp, R., Mason, W.P., van den Bent, M.J., Weller, M., Fisher, B., Taphoorn, M.J., et al., 2005. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. New England Journal of Medicine 352, 987–996. Stupp, R., Taillibert, S., Kanner, A.A., Kesari, S., Steinberg, D.M., Toms, S.A., et al., 2015. Maintenance therapy with tumor-treating fields plus Temozolomide vs Temozolomide alone for glioblastoma: A randomized clinical trial. Journal of the American Medical Association 314, 2535–2543. Walbert, T., Chasteen, K., 2015. Palliative and supportive care for glioma patients. Cancer Treatment and Research 163, 171–184.

164

Glioblastoma: Biology, Diagnosis, and Treatment

Relevant Websites http://www.abta.orgdAmerican Brain Tumor Association. http://www.braintumor.orgdNational Brain Tumor Foundation. http://www.cbtrus.org/dCentral Brain Tumor Registry of the US (CBTRUS). http://www.nccn.orgdNational Comprehensive Cancer Network Central Nervous System Guidelines. http://www.sciencedirect.com/science/referenceworks/9780080450469dEncyclopedia of Neuroscience. https://seer.cancer.gov/dSurveillance, Epidemiology, and End Results Program. http://www.uptodate.comdUpToDate.

Glioblastoma: Pathology and Genetics Christian Hartmann, Medical School Hannover, Hannover, Germany Pieter Wesseling, VU University Medical Center/Brain Tumor Center Amsterdam, Amsterdam, The Netherlands; Princess Máxima Center for Pediatric Oncology, Utrecht, The Netherlands; and University Medical Center Utrecht, Utrecht, The Netherlands © 2019 Elsevier Inc. All rights reserved.

Introduction Gliomas are a very diverse group of glial neoplasms and account for the great majority of tumors originating in the parenchyma of the central nervous system (CNS). Two main categories of gliomas are recognized: diffuse gliomas, by far the most frequent gliomas in adult patients and characterized by extensive infiltrative growth into the surrounding CNS parenchyma, and gliomas with a much more circumscribed growth pattern (“nondiffuse gliomas”). Traditionally, diffuse gliomas are further classified based on their microscopic similarities with (precursors of) glial cells and then designated as astrocytoma, oligodendroglioma, or mixed glioma/oligoastrocytoma. After assessment of the glioma type, a malignancy grade is assigned to these tumors based on especially the following histologic features: marked mitotic activity, florid microvascular proliferation, and necrosis. Diffuse astrocytic tumors showing marked mitotic activity, as well as necrosis and/or florid microvascular proliferation are diagnosed as glioblastoma. Glioblastomas account for about 15% of all primary CNS tumors and for over 50% of all gliomas. Even though extracranial metastases of gliomas are very rare, most glioblastoma patients die within 1–2 years after diagnosis, despite today’s standard of care (surgery, radiotherapy, and chemotherapy). Many patients with a lower grade (WHO grade II or III) diffuse astrocytoma survive (much) longer, but virtually all these patients will eventually experience progression to glioblastoma resulting in fatality. For a long time, microscopic evaluation has provided the gold standard for the diagnosis of gliomas, assessment of prognosis and formed the basis for therapeutic management. Multiple studies, however, showed that a purely histopathologic classification suffers from considerable inter- and intraobserver variability. Especially in the course of the last decade it became clear that molecular characteristics provide a more robust and objective basis for subtyping of diffuse gliomas. Indeed, in the revised 4th edition of the WHO classification of CNS tumors (published in 2016), classification of diffuse gliomas has fundamentally changed as for the first time presence/absence of particular molecular aberrations is now part of the definition of these tumors. After providing some essentials of the traditional, histopathology-based classification of glioblastomas, this review discusses novel insights in molecular characteristics of (different subgroups of) these neoplasms and how this information is now used for a combined histomolecular diagnosis. Furthermore, information is provided on how based on tools like expression profiling and methylation profiling additional subgroups can be recognized, while in the last part some future perspectives are given with regard to where the field is or may be heading.

From Histopathologic to Histomolecular Classification For over a century, microscopic analysis of tumors of the CNS has been performed on histochemically, especially hematoxylin-andeosin (H&E) stained sections, later on supplemented by electron microscopy and immunohistochemistry. Diffuse gliomas represent the vast majority of glial neoplasms in adult patients and are microscopically characterized by diffuse infiltrative growth within the CNS parenchyma, with tumor cells extensively invading individually or in small groups in the neuropil along myelinated fiber tracts and along blood vessels. This invasive growth may well extent over several centimeters outside the abnormal area as seen on magnetic resonance imaging (MRI) scans. Also, not infrequently the tumor cells of diffuse gliomas cross the corpus callosum, resulting in a “butterfly glioma” pattern on MRI. Especially in the less cellular areas of diffuse gliomas the matrix may consist of relatively intact preexistent gray and white matter with aggregation of neoplastic cells around neurons (perineuronal satelitosis), blood vessels, and under the pial membrane. Such “secondary structures of Scherer” are nearly pathognomonic of diffuse glioma. Occasionally, a diffuse glioma may present radiologically as multiple lesions (multifocal or multicentric glioma) resembling brain metastases or abscesses while it in fact concerns multiple foci of high cellularity, microvascular proliferation and/or necrosis in a widespread diffuse glioma. Based on their histopathologic phenotype, diffuse gliomas are traditionally typed as astrocytic, oligodendroglial, or mixed oligodendroglial-astrocytic, and graded as WHO grade II (low grade), III (anaplastic), or IV (glioblastoma). Diffuse gliomas without mitotic activity, microvascular proliferation, and necrosis are graded as low-grade (WHO grade II). With increased mitotic activity, the diagnosis of anaplastic astrocytoma or anaplastic oligodendroglioma (WHO grade III) is rendered, while in astrocytic tumors the presence of necrosis and/or microvascular proliferation leads to a diagnosis of glioblastoma (WHO grade IV). In contrast, pure oligodendroglial tumors with necrosis and florid microvascular proliferation are still graded as WHO grade III. Necrosis in glioblastomas often consists of irregular, serpiginous foci surrounded by densely packed, somewhat radially oriented tumor cells (“palisading necrosis”). Glioblastoma is by far the most frequent and most malignant diffuse glioma and was previously called “glioblastoma multiforme” because of the often striking intratumoral cytologic and histologic heterogeneity. Furthermore, several histologic subtypes of glioblastoma are listed in the WHO 2016 classification, including the small cell variant of glioblastoma (predominantly showing small, relatively monomorphous tumor cells with little cytoplasm which may resemble anaplastic oligodendroglioma),

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65070-5

165

166

Glioblastoma: Pathology and Genetics

gliosarcoma (with a sarcoma-like, reticulin-rich component), giant cell glioblastoma (characterized by the presence of many large, multinucleated tumor cells), epitheloid glioblastoma (often apparently more circumscribed tumors showing prominent epitheloid or even rhabdoid morphology and thereby showing resemblance with metastastic carcinoma or melanoma), glioblastomas with primitive neuronal component (with nodular foci of primitive neuronal cells resembling embryonal tumor), and granular cell glioblastoma (with histiocyte-like cytoplasmic granularity and vacuolation). (Fig. 1). Until the most recent World Health Organization (WHO) classification of CNS tumors published in 2016, the histologic diagnosis was the gold standard for classification of gliomas. In daily clinical practice, however, unequivocal typing and grading of gliomas can be difficult. For instance, because diffuse gliomas often show marked phenotypic heterogeneity with spatial differences in cellular phenotype, degree of anaplasia and mitotic activity there is a danger of underestimation of malignancy grade because of sampling error. Furthermore, the minimum diagnostic criteria of, for example, florid microvascular proliferation are not clear. Even identification of necrosis may be troublesome in biopsy samples that are small or poorly preserved, and in recurrent gliomas discrimination of native tumoral from therapy-induced necrosis may be difficult. Meanwhile, since 2008 multiple studies showed that based on the presence or absence of mutations in the isocitrate dehydrogenase 1 (IDH1) or IDH2 gene and of complete, combined loss of the short arm of chromosome 1 and of the long arm of chromosome 19 (complete 1p/19q-codeletion) three major, clinically relevant subgroups of diffuse glioma can be defined. Indeed, the International Society for NeuropathologydHaarlem Consensus Guidelines and the subsequently published, revised 4th edition of the WHO classification of CNS tumors published in 2016 embraced this notion of an integrated histomolecular classification of diffuse gliomas, and the WHO classification nowadays recognizes the following three major molecular diffuse glioma subgroups:

• • •

IDH-wildtype: most of these histologically represent astrocytic tumors, a large percentage belonging to the highest malignancy grade, that is, glioblastomas; IDH-mutant and 1p/19q-non-codeleted: these tumors also generally have an astrocytic phenotype, but a much larger percentage is at first diagnosis histologically lower grade/WHO grade II or III; IDH-mutant and 1p/19-codeleted: most of these are characterized by a prominent oligodendroglial phenotype of the tumor cells.

In this new classification, the presence of complete 1p/19q-codeletion in an IDH-mutant diffuse glioma is considered pathognomonic for oligodendroglioma, while IDH-mutant oligodendroglioma-appearing tumors lacking this event are reclassified as astrocytoma, IDH-mutant. Following this strategy, the diagnosis of oligoastrocytoma can be expected to largely disappear, except when additional molecular tests cannot be performed or do not provide unequivocal results; in that situation, not-otherwise-specified (NOS) should be added to the diagnosis to indicate that ideally such samples require further workup. Similarly, molecular studies have now also revealed that most cases previously diagnosed as glioblastoma with oligodendroglioma component (GBM-O) fit in one of the three categories just mentioned (i.e., can be diagnosed as glioblastoma IDH-wildtype, glioblastoma IDH-mutant, or as anaplastic oligodendroglioma IDH-mutant and 1p/19q-codeleted) and GBM-O has therefore been removed from the WHO 2016 classification as a diagnosis. In the following paragraphs three molecularly defined groups of glioblastoma will be discussed in more detail: IDH-wildtype glioblastoma, IDH-mutant glioblastoma, and diffuse midline glioma H3 K27M-mutant. This last category was added in the

(A)

(B)

(C)

(D)

(E)

(F)

(G)

(H)

(I)

(J)

Fig. 1 Examples of histopathology of glioblastoma, including some histologic variants (¼commonly accepted subtypes associated with particular clinical features) and patterns (¼ histological appearances without any currently identified clinical significance). (A) Florid microvascular proliferation (arrowhead); (B) palisading necrosis (arrowhead); (C) immunohistochemical demonstration of IDH1 R132H mutant protein in glioblastoma, IDHmutant (molecularly defined entity); (D) small cell glioblastoma (pattern); (E) gliosarcoma (variant); (F) giant cell glioblastoma (variant); (G) epitheloid glioblastoma (variant); (H) glioblastoma with primitive neuronal component (pattern); (I) adenoid glioblastoma (pattern); (J) granular cell glioblastoma (pattern). Images A, B and D–J are from hematoxylin-and-eosin stained sections, apart from the right half of image E where a reticulin stain highlights the widespread presence of reticulin fibers in between the sarcomatoid tumor cells.

Glioblastoma: Pathology and Genetics

167

WHO 2016 classification as a separate entity. H3 K27M-mutant diffuse midline glioma generally occurs in children and should be located in the “midline” of the CNS (defined as brainstem, thalamus, cerebellum and/or spinal cord). Diffuse intrinsic pontine glioma (DIPG) is a frequent representative in this category.

Glioblastoma, IDH-Wildtype The current WHO classification thus separates glioblastomas according to their IDH and H3 K27M status. Approximately 90%–95% of adult patients with glioblastoma have an IDH-wildtype (and H3 K27-wildtype) tumor. In most of these patients the tumor presents as a “primary glioblastoma,” that is, a glioblastoma without clinical, radiological and/or pathological evidence of a lower grade precursor lesion. Data from large epidemiological cohorts are currently not yet available for IDH-wildtype versus IDH-mutant glioblastomas. The most recent CBTRUS report (published in 2017) on the occurrence of CNS tumors in the USA in the period 2010–14 still presents glioblastomas as one group. In the majority of patients IDH-wildtype glioblastoma is diagnosed in their 6th or 7th decade of life. Of note, many glioblastomas in childhood are also IDH-wildtype but the molecular underpinnings of these tumors is generally different from that in adult patients (see also paragraph on diffuse midline glioma, H3 K27M-mutant). Large consortia such as the International Cancer Genome Consortium (ICGC) (https://icgc.org), The Cancer Genome Atlas (TCGA) (https://cancergenome.nih.gov), and the Chinese Cancer Genome Consortium (CCGC) (http://cancerdb.genomics.org. cn) have provided a wealth of new information on molecular alterations in different subgroups of glioblastoma. Different molecular classification schemes have been proposed over the last decade, some of these using a restricted set of molecular markers (such as IDH, H3 K27 and 1p/19q status), others using information on additional markers and/or on data obtained by global gene expression or methylation profiling analysis. Salient molecular aberrations in IDH-wildtype glioblastomas are: loss of chromosome 10 combined with gain of chromosome 7 (þ 7/10), amplification of the epidermal growth factor receptor (EGFR) gene, mutation of the promoter of the telomerase reverse transcriptase (TERT) gene. These molecular markers will now be discussed in more detail. Additionally, in this paragraph some information will be provided on the methylation of the promoter of MGMT gene (encoding O6-methylguanine-DNA methyltransferase). Of note, MGMT promoter methylation status is an epigenetic marker with predictive value for response to alkylating chemotherapy in patients with glioblastoma rather than a marker of use for assigning a particular WHO diagnosis to a glioma. The vast majority of IDH-wildtype glioblastomas show (often whole chromosome) gain of one copy of chromosome 7 and loss of one copy of chromosome 10 (þ 7/10). (Fig. 2) This aberration is considered as a very early event in the oncogenesis of glioblastomas. Several oncogenes that have been implicated in gliomagenesis are located on chromosome 7, for example, Cyclin Dependent Kinase 6 (CDK6), MET Proto-Oncogene (MET) and EGFR, while several tumor suppressor genes are located on chromosome 10, including Tet family member Tet Methylcytosine Dioxygenase 1 (TET1) and Phosphatase and Tensin Homolog (PTEN). It is still unclear, however, how exactly þ 7/10 plays a role in gliomagenesis. With a reported mutation frequency of

Fig. 2 Chromosomal copy number variation (CNV) plot of an IDH-wildtype glioblastoma based on Illumina EPIC/850k methylation profiling analysis. This glioblastoma shows þ7/10 in the form of whole chromosome 7 gain (thin arrow) combined with whole chromosome 10 loss (thick arrow). Additionally, amplification of CDK4 and MDM2 is present (asterisk), but high copy EGFR amplification is lacking (arrowhead). Of note, demonstration of þ7/10 and/or EGFR amplification in an IDH-wildtype diffuse astrocytic tumor of histologically WHO grade II or III is a very strong indication that the tumor will clinically behave as glioblastoma (WHO grade IV).

168

Glioblastoma: Pathology and Genetics

54%–84%, TERT promoter mutation is the most frequent mutation in IDH-wildtype glioblastomas. This gene encodes the telomerase, which induces the extension of the telomeres. Interestingly, the TERT-related mutations are thus not located in the coding area of the gene, but in the promoter, with approximately 80% of the C228T type, and 20% of the C250T type. The consequence of these mutations is an increased expression of telomerase by a factor of five. Glioblastoma patients with a TERT promoter mutation are older and tend to have a worse prognosis. Of note, IDH-mutant and 1p/19q-codeleted oligodendrogliomas also show a very high frequency of TERT promoter mutations but in those tumors this marker does not have a negative prognostic significance. In contrast, diffuse, IDH-mutant astrocytomas, including IDH-mutant glioblastomas, typically do not have TERT promoter mutation but an a-thalassemia/mental retardation syndrome X-linked (ATRX) mutation resulting in alternative lengthening of telomeres (Fig. 3). Approximately 40% of all IDH-wildtype glioblastomas show amplification of the EGFR gene, and one-third to half of these EGFR-amplified tumors additionally show the EGFRvIII mutation variant. In this variant, exons 2–7 are deleted, and the consecutive loss of part of the extracellular domain of the receptor tyrosine kinase causes constitutive autophosphorylation and permanent activation of, for example, the PIK3 pathway, resulting in increased cell proliferation and resistance to apoptosis. Although the EGFRvIII mutation variant is often present in only a subset of the tumor cells, it has been demonstrated that cells expressing this variant can stimulate neighboring tumor cells by a paracrine mechanism. Furthermore, the in-frame deletion of the EGFRvIII mutation generates a neoepitope that can be detected with specific antibodies. Unfortunately, attempts to use this neoepitope for therapeutic vaccination have so far failed. In the TCGA RNAseq analysis of EGFR, glioblastomas with deletions of exons 12–13 and 14–15 were also found, but it remains to be seen if these variants are the consequence of aberrant splicing. In about 10% of the glioblastomas in the

(A)

5p15.33

(B)

1

2

3

(C)

Ets

E-Box

Ets

4

5

6

7 8

9

10

C250T (pos. 1.295.250)

C228T (pos. 1.295.228)

Ets/TCF

Ets/TCF

SP1

Ets

Ets

(D)

SP1

11

12

13 14 15

E-Box

(E)

Telomerase expression ~ 5x ↑

Fig. 3 Schematic illustration of TERT promoter mutation and its consequences. (A) The TERT gene is located on chromosome 5p15.33. (B) The TERT gene has 15 exons and a promoter (blue line), which partially extends into exon 1. (C) The TERT promoter has various transcription factor binding motifs (below the blue line); in IDH-wildtype glioblastomas, an additional Ets/TCF motif is frequently inserted by a C250T or a C228T mutation in the TERT promoter region. (D) this mutation results in increased translation of telomerase. (E) As a consequence, tumor cells are able to continue proliferating without critical telomere shortening forcing them into senescence or apoptosis.

Glioblastoma: Pathology and Genetics

169

TCGA series the mutation variant EGFRvIV (affecting exon 25–27) was found, and sporadically yet another mutation variant occurred as fusion between EGFR and adjacent genes. In somewhat over 10% of glioblastomas EGFR point mutations occur. In contrast to pulmonary cancer, EGFR point mutations in glioblastomas usually do not involve the kinase domain but the extracellular domain of the receptor tyrosine kinase. (Fig. 4) This difference may explain why studies with receptor tyrosine kinase inhibitors were negative in patients with glioblastoma. Neither EGFR amplification nor the combination with the EGFRvIII mutation variant has a clear prognostic significance in IDH-wildtype glioblastomas. However, demonstration of EGFR amplification, as well as of þ 7/10 and/or TERT promoter mutation is a serious indication of high-grade malignancy in adults with IDHwildtype diffuse astrocytic tumors to which based on histologic examination not (yet) a WHO grade IV can be assigned. The MGMT gene encodes the O6-methylguanine-DNA methyltransferase, a DNA repair enzyme which can transfer a methyl group from the O6 position of a guanine nucleotide irreversibly to itself, thereby repairing the mutated DNA. The O6-methylation or alkylation of guanine is a typical DNA mutation event induced by various mutagenic events. Application of alkylating drugs like temozolomide induces such O6-guanine methylation as well, the therapeutic intention being to induce DNA mismatch, DNAdouble-strand breakage and ultimately apoptosis in glioblastoma cells. MGMT expression is under control of a large CpG island in the gene promoter. In case of methylation of these CpG sites, MGMT expression is strongly reduced and cells have a limited capacity to repair drug-induced DNA damage. In approximately 40% of patients with IDH-wildtype glioblastomas the tumors demonstrate MGMT promoter methylation, thereby rendering this subgroup of patients relatively responsive to treatment with alkylating drugs. Because of this predictive value, recent guidelines of the European Association for Neuro-Oncology (EANO) recommend MGMT promoter methylation testing for certain groups of patients with high-grade gliomas. Diffuse gliomas with a H3F3A, HIST1H3B/C, or HIST2H3C K27M mutation are typically found in midline structures in children (see separate paragraph below). Since the first description of these mutations it became obvious that such histone H3 mutations not affect only codon 27 but also codon 34. Glioblastomas with a H3 G34R or, less frequently, H3 G34V mutation are rare, occur mostly in the cerebral hemispheres, affect predominantly older children and young adults, and are associated with a somewhat better prognosis than patients with ordinary glioblastomas. H3F3A G34R/V-mutant tumors can also have the phenotype of a CNS embryonal

(A)

(B)

EGFR 7cen

EGFR 7cen

EGFR amplification in ~ 40% of IDH-wildtype glioblastomas Point mutation

(C)

~ 13%

xxxxxxxxx xxxxxxxxxxxx xxxxxxxxxxxxxxxxx

x

xxxxx xxxxx

Extracellular domain

EGFRvIII

xx

Membrane

Del. Exon 12-13 Del. Exon 14-15

Del. Exon 2-7

~ 29%

~ 3%

~ 20%

Splice variant?

Splice variant?

x Kinase domain

x Cytoplasmatic

EGFRvIV Del. Exon 25-26 Del. Exon 25-27 Del. Exon 25-28 < 10%

Fig. 4 Schematic illustration of involvement of the epidermal growth factor receptor (EGFR) in glioblastomas. (A) The EGFR protein normally has an extracellular domain (yellow), a domain in the membrane (orange), and an intracellular part (kinase domain and cytoplasmic end). (B) About 40% of all IDH-wildtype glioblastomas show high copy amplification of EGFR, here illustrated by FISH images of two different tumors; the centromere of chromosome 7 is marked by a green signal, and the fact that more than two green signals are visible indicates polyploidy of chromosome 7; the multitude of red dots represent the signal of the EGFR probe and reflect high copy amplification of this gene. (C) Schematic illustration of other mutation variants affecting the EGFR gene. Up to half of all glioblastomas with EGFR amplification additionally have a deletion of exon 2–7, leading to the EGFRvIII mutation variant which is constitutively active. The mutation variant EGFRvIV affects exon 25–27 and is found in about 10% of cases. The importance of the potential deletions of exons 12–13 and 14–15 is still unclear, these may in fact be mRNA splicing variants rather than genetic deletions. Approximately 13% of IDH-wildtype glioblastomas show point mutations in EGFR, mainly in the extracellular domain, and partly in combination with EGFR amplification.

170

Glioblastoma: Pathology and Genetics

tumor (formerly known as (supratentorial) CNS PNET), but no prognostic difference was found between high grade glioma versus embryonal histology. In contrast, the lack of amplification of oncogenes, and the presence of MGMT promoter methylation in such tumors was reported to be associated with better outcome. As these tumors seem to represent a discrete group, one can expect that they will be listed as a separate entity in a next edition of the WHO classification of CNS tumors.

Glioblastoma, IDH-Mutant The establishment of IDH-mutant glioblastomas as a separate entity by the WHO classification was significantly influenced by the clinical concept of the existence of primary (“de novo”) glioblastomas (diagnosed in patients without clinical, radiological and/or pathological evidence for a lower grade precursor lesion) and secondary glioblastomas (that apparently evolved from such a lower grade precursor lesion). After it was shown that the vast majority of primary glioblastomas are IDH-wildtype, while most secondary glioblastomas are IDH-mutant, and given the fact that “primary” glioblastomas with IDH mutation might represent cases in which the lower grade precursor lesion was missed, nowadays the terms primary and secondary glioblastoma are considered to be largely synonymous with IDH-wildtype and IDH-mutant glioblastoma, respectively. In different studies the relative frequency of IDH-mutant glioblastomas varies between 6% and 13%. Patients with an IDHmutant glioblastoma are on average significantly younger than patients with an IDH-wildtype glioblastoma, in a populationbased study the mean age was 48 versus 61 years, respectively. In MRI imaging, patients IDH-mutant glioblastomas showed more rarely large central necrosis, less pronounced edema, more frequently cystic changes and more diffuse tumor infiltration than patients without this genetic alteration. IDH-mutant glioblastomas usually show the same genetic alterations as diffuse, lower grade (WHO grade II and III) astrocytomas, with especially TP53 and ATRX mutations being present in a high percentage of cases. While TP53 mutations also occur quite frequently in IDH-wildtype glioblastomas, ATRX mutations are typical for IDH-mutant astrocytomas/glioblastomas and functionally partially substitute TERT promoter mutations. IDH-mutant glioblastomas typically do not exhibit þ 7/10 or EGFR amplification. The gene IDH1 (located on chromosome 2q33.3) encodes the cytoplasmatic (type 1) form of isocitrate dehydrogenase (IDH1), while the IDH2 gene (located on 15q26.1) encodes IDH2 that is located in the mitochondria. Both proteins catalyze as homodimers isocitrate and NADP by an oxidative decarboxylation to a-ketoglutarate (a-KG) and NADPH. Only the much more complex IDH3, but not IDH1 and IDH2, has a functional role in the tricarboxylic acid cycle. IDH mutations only affect one of the two alleles (heterozygous deletion) and are generally “hotspot” mutations of IDH1 codon 132 or of IDH2 codon 172. IDH1 mutations account for over 90% of the IDH mutations found in diffuse gliomas, with IDH1 R132H representing up to 90% of all IDH1 mutations. The availability of a specific antibody against the IDH1 R132H mutant protein that can be applied on formalin-fixed, paraffinembedded tissue thus enables reliable detection of the vast majority of IDH-mutant gliomas/glioblastomas by a simple immunohistochemical procedure in routinely processed glioma tissue (Fig. 1C). Mutant IDH protein utilizes the physiological product a-KG as an educt and catalyzes it to 2-hydroxyglutarate (2-HG). As a result, the so-called “oncometabolite” 2-HG is present in excessive concentrations in affected tumor cells and induces a competitive inhibition of a large group of more than 60 different a-KG-dependent dioxygenases. From this point on, various dysfunctional pathways result of which the individual significance for neoplastic transformation is unclear (Fig. 5). The physiological function of TET2 comprises a large-scale demethylation of CpG sites in promoter regions of various genes, and impaired function due to IDH-mutant status results in a global hypermethylation of the genome, also described as the glioma CpG island methylator phenotype (gCIMP). This situation may explain why the vast majority of IDH-mutant gliomas show hypermethylation of the MGMT promoter. The genomic hypermethylation also affects cohesin and CCCTC-binding factor (CTCF)-binding sites, thereby compromising binding of this methylation-sensitive insulator protein. This in turn leads to a disrupted chromosomal topology with enhanced PDGFRA activity. Furthermore, excessively increased 2-HG results in inhibition of KDM4A, whose functional failure leads to an increase in DEPTOR activity, which in turn causes mTOR pathway activation. As a further consequence of massively increased concentrations of 2-HG, competitive inhibition of the a-KG-dependent prolylhydroxylase was demonstrated, resulting in inhibition of the degradation of HIF1a.

Diffuse Midline Glioma, H3 K27M-Mutant While in adult patients glioblastomas mostly are located in supratentorial parts of the CNS, in children glioblastomas occur relatively frequently infratentorially. A particular subset of such (mainly) pediatric tumors, the diffuse midline glioma, H3 K27M-mutant, was added to the list of diffuse gliomas in the WHO 2016 classification. Diffuse midline glioma, H3 K27M-mutant carries a mutation in either the H3F3A, HIST1H3B/C, or HIST2H3C gene and represents a WHO grade IV tumor. By definition, these tumors with a H3 K27M mutation should be (a) gliomas with (b) diffuse growth and (c) located in the midline (brain stem, thalamus, cerebellum, and/or spinal cord). Tumors that were previously diagnosed as diffuse intrinsic pontine glioma (DIPG) are a frequent representative of this group, with about 80% of the previously diagnosed DIPGs indeed carrying the H3 K27M mutation. Patients with a diffuse midline glioma, H3 K27M-mutant in the pons are on average younger than those with such a tumor in the thalamus (around 7 years and 11 years, respectively). In both children and adults, diffuse midline gliomas, H3 K27M-mutant that show WHO grade II or III phenotype generally still carry a grim (WHO grade IV-like) prognosis. Importantly, other glial CNS tumors (especially nondiffuse

Glioblastoma: Pathology and Genetics

171

Fig. 5 Schematic drawing of mechanisms involved in the isocitrate dehydrogenase (IDH) pathway. (A) Normal IDH protein catalyzes isocitrate to a-ketoglutarate (a-KG); (B) mutant IDH protein utilizes the physiological product a-KG as educt and catalyzes it to 2-hydroxyglutarate (2-HG); (C) 2HG occurs in excessive concentrations in affected tumor cells and is competitively inhibiting a large number of different a-KG-dependent dioxygenases (D–F); (D) by blocking the activity of the DNA demethylase TET2 a global CpG island methylator phenotype (g-CIMP) results; (E) by inhibiting the activity of histone demethylases for example H3 histone protein becomes trimethylated; (F) by inhibition of KDM4A, the activity of DEPTOR increases, which in turn causes mTOR pathway activation; (G) frequently occurring alternative route for mTOR pathway activation in cancer.

gliomas such as ependymomas and pilocytic astrocytomas, and nonmidline gliomas) may occasionally carry the H3 K27M mutation as well, but the prognostic impact of this mutation in such a nondiffuse and/or nonmidline glioma is unclear. Up to now four different histone-encoding genes carrying H3 K27M mutations have been identified. By far the most frequently mutated gene is H3F3A (mutant in about 80% of H3 K27M-mutant diffuse midline gliomas and encoding histone H3.3), followed by HIST1H3B (mutant in about 20% of these cases and encoding histone H3.1). In less than 1% of the H3 K27M-mutant diffuse midline gliomas a mutation is found in HIST1H3C (encoding histone H3.1) or HIST2H3C (encoding histone H3.2). The functional consequences of H3 K27M mutations are not yet completely understood. A H3 K27M mutation results in an altered histone protein that cannot become trimethylated at position 27 anymore. When the DNA-trimethylating polycomb repression complex 2 (PCR2) encounters such a mutant oncohistone, it stops functioning and detaches itself from the DNA. As a result, there is a global reduction in histone trimethylation (loss of H3K27me3) in tumor cells with a H3 K27M mutation. A global change of the epigenome then results, with as a consequence abnormal cell cycle control, inhibition of autophagy and presumably also an increased resistance of the tumor cells to radiotherapy (Fig. 6). Like for the IDH R132H mutant protein, a highly specific antibody is available for H3 K27M mutant protein that allows for detection of the H3 K27M-mutant status of a (glial) tumor by a simple immunohistochemical test on formalin-fixed, paraffin-embedded tumor tissue. In this staining a H3 K27M-mutant glioma shows positive staining of tumor cell nuclei (with the negative nuclei of nonneoplastic (e.g., vascular) cells representing a negative internal control in the same slide). Interestingly, H3F3A K27M mutations were reported to occur in diffuse midline gliomas in all anatomical structures of the midline, while HIST1H3B K27M mutations occur mainly in those originating in the pons. First data indicate that patients with H3F3A K27M mutation in their diffuse midline glioma usually have a worse prognosis than those with an HIST1H3B mutation. Furthermore, HIST1H3B mutations are very often combined with ACVR1 mutations which have been reported as a slightly better prognostic factor as well. PDGFRA and PIK3CA mutations are interpreted as progression-associated genetic changes in diffuse midline gliomas with a H3F3A K27M mutation. PIK3CA mutations were reported as progression-associated changes in HIST1H3B-mutant tumors as well.

172

Glioblastoma: Pathology and Genetics

(A)

K27me3

Compaction of chromatin, transcription suppression

(B)

Decompaction of chromatin, enhancedtranscription

Fig. 6 H3 K27M mutation results in decompaction of DNA and promotion of transcription. (A) The polycomb repression complex 2 (PRC2, with EZH2, SUZ12, EED, RBAP as components) moves gradually along histones associated with DNA and establishes tri-methylation at the H3 K27 site of the histone tail (H3 K27me3), resulting in DNA compaction and transcription suppression. Presence of a H3 K27M mutation in the H3F3A or HIST1H3A/B gene obstructs this effect of PRC2 not only at the site of the mutant histone, but also at subsequent histones as further movement of the PRC2 complex is impaired. The focal presence of H3 K27M may thus cause more widespread blockage of H3 K27 trimethylation and, consequently, more widespread decompaction of DNA and promotion of transcription.

Glioblastoma Classification by Next Generation Molecular Technologies Especially in the last two decades, technologies have become increasingly developed that not only enabled the investigation of individual molecular markers, but also the analysis of complex correlations on a molecular basis. In the field of neurooncology, this was achieved for the first time with the comparative genomic hybridization (CGH) technology, which was significantly improved by the array-CGH (aCGH) method. These assays allowed the detection of chromosomal gains or losses in gliomas. Another step forward was expression array analysis, for which RNA is extracted from tumor tissue and converted into cDNA. Subsequently, the expression pattern of a large number of genes can be determined on this basis. Furthermore, different methods for next-generation sequencing (NGS) were increasingly used in neuro-oncology. While initially only small panels of genes were investigated, further development of the technique allowed for wider availability and higher efficiency of NGS and now analysis of entire (glioma) exomes and genomes are performed. The NGS technology also has been extended to the transcriptome of gliomas (whole transcriptome shotgun sequencing; RNAseq). Using RNAseq it is possible to determine both the expression and the mutations of the (glioma) genome. Another advantage of RNAseq is that fusion genes can be detected more effectively than with any other technology. Probably the most productive technology for classifying gliomas at present is the 450 K, or more recently, 850 K BeadChip methylation analysis. This platform generates detailed information on the epigenetic profile of a tumor as well as on chromosomal gains and losses. On the basis of these molecular high-throughput techniques, new subgroups of glioblastomas were identified. The following paragraphs present some examples of such newly recognized glioblastoma subgroups (Fig. 7).

Expression Profiling Initial expression profiling studies revealed that glioblastomas can be distinguished from pilocytic astrocytomas, anaplastic astrocytomas or oligodendrogliomas, and that “primary” glioblastomas can be discriminated from “secondary” glioblastomas. Ordering the results of expression profiling using unsupervised clustering algorithms indicated patterns that correlated with diagnoses and malignancy grades assigned to the tumors based on histopathologic analysis. Meanwhile, several studies reported that the prognosis of patients with (astrocytic) gliomas could be more accurately predicted based on expression profiling than on morphologic evaluation. Various attempts have been made to use expression profiling for the recognition of clinically relevant subgroups of histopathologically diagnosed glioblastomas. At first, three groups were described: proneural, proliferative and mesenchymal. In

Glioblastoma: Pathology and Genetics

astrocytoma, IDH-mutant

anaplastic astrocytoma, IDH-mutant

glioblastoma, IDH-mutant ‘proneural’

diffuse midline glioma, H3 K27M -mutant

diffuse astrocytoma, IDHwildtype

glioblastoma, IDH-wildtype RTK I

malignant glioma, H3 G34 -mutant

anaplastic astrocytoma, IDH-wildtype

RTK II RTK III

‘mesenchymal’ ‘classical’

173

‘proneural’ giant cell glioblastoma gliosarcoma epitheloid glioblastoma

anaplastic ganglioglioma

anaplastic pleomorphic xanthoastrocytoma Fig. 7 Scheme of histologic and molecular categories of glioblastoma. The current WHO classification of CNS tumors lists three, partly molecularly defined groups of grade IV diffuse gliomas/glioblastomas (encircled by a solid line): (A) IDH-wildtype, by far the most frequent group in adults; (B) IDH-mutant, generally occurring in younger adults than IDH-wildtype glioblastomas and often the result of malignant progression of a previously diagnosed lower grade IDH-mutant astrocytoma; (C) H3 K27M-mutant diffuse midline gliomas (that are IDH-wildtype as well), typically but not only occurring in children. Of note, many tumors that are histologically diagnosed as lower grade (WHO grade II and III) IDH-wildtype diffuse astrocytomas (encircled by the dashed line in this figure) have molecular characteristics of and in fact clinically behave like IDH-wildtype glioblastomas (WHO grade IV). Meanwhile, the IDH-wildtype glioblastoma group is still a very heterogeneous one with different subgroups that can be recognized based on histology (e.g., gliosarcoma, giant cell glioblastoma, epitheloid glioblastoma), expression profiling (classical, mesenchymal, or proneural subtype), or methylation profiling (e.g., receptor tyrosine kinase type I (RTK I), RTK II, RTK III, mesenchymal). In the IDH-wildtype glioblastoma group, the H3 G34-mutant tumors already stand out as a distinct subgroup that may well be designated as a separate entity in the next WHO classification. Discrimination of high-grade “nondiffuse” gliomas like anaplastic pleomorphic xanthoastrocytoma or anaplastic ganglioglioma from especially the epitheloid variant of glioblastoma may be difficult not only because of histopathologic similarities but also because these tumors share the frequent occurrence of BRAF V600E mutation.

the proneural group, patients were generally younger and the tumors lacked EGFR amplification and PTEN mutation. In the proliferative and mesenchymal groups, on the other hand, the tumors typically showed the genetic profile of glioblastoma in (older) adult patients, including gain of chromosome 7 and loss of chromosome 10. Later on, based on in depth analysis of the large and complex data sets of the TCGA project, modification of expression profiling-based subgrouping of glioblastomas was suggested with recognition of four transcription types: proneural, neural, classic, and mesenchymal. These four subtypes were associated with different frequencies of mutations in especially TP53, EGFR, and NF1. More than 90% of IDH-mutant glioblastomas belonged to the proneural expression signature, and about 30% of glioblastomas with a proneural expression signature were IDH-mutant. Diffuse lower grade astrocytomas and oligodendrogliomas also typically had a proneural signature. Later on, the neural transcription profile could at least partly be explained by the presence of relatively large amounts of nonneoplastic CNS tissue in the glioblastoma samples in this group. In other words, identification of this neural group may partly have been based on “selection-bias” towards the use of samples with relatively low tumor cell percentage, for example, from the periphery of a glioblastoma, rather than have a basis in a real profile of the tumor itself. Furthermore, using single cell analysis it was demonstrated that different transcription types can occur within one and the same glioblastoma. Also in a human glioblastoma xenograft model cells could be transformed from a proneural to a mesenchymal expression type by defined conditions. Such observations raised doubts about the stability of classification of glioblastomas by RNA-based transcriptional profiling. Yet another problem with expression profiling as a diagnostic tool in clinical practice is that in routinely processed, formalin-fixed, paraffinembedded tumor tissue the integrity of RNA is significantly impaired. Ideally, for expression profiling tumor tissue should thus be fresh frozen.

174

Glioblastoma: Pathology and Genetics

Methylation Profiling Analyzing the epigenome, and in particular the methylome of tumor cells has proved to be a robust and reproducible technology allowing for the analysis of even small, formalin-fixed, paraffin-embedded tissue samples of morphologically poor quality. The analysis of the methylome can be used to evaluate which epigenetic changes are evolving due to the induction and progression of tumors. On the other hand, part of the epigenetic imprinting remains stable from the original cell to the tumor cell, so that the analysis of the methylome provides information on the cell of origin of the tumor and can be used to classify tumors. Various technologies for global methylation analysis are available. The commercial Infinium BeadChip 450 K assay, later on replaced by the Infinium MethylationEPIC BeadChip 850 K assay, has found the broadest distribution and has been successfully used for subclassification of glioblastomas as well. With 850 K analysis, over 850,000 CpG sites are examined, which corresponds to about 3.5% genomic CpG coverage. An alternative approach is whole genome bisulfite sequencing (WBGS), allowing investigation of more than 26 million CpG sites, corresponding to a CpG coverage of more than 90%. The results of tumor DNA analysis using the 450 K and the WBGS method showed comparable results, indicating that different methods can be applied for robust characterization of the tumor methylome. One of the first observations obtained with methylation profiling analysis of gliomas was the existence of a glioma CpG island methylator phenotype (g-CIMP) cluster. This “cluster #1” with g-CIMP signature encompassed most of the lower-grade diffuse infiltrating gliomas as well as a fraction of glioblastomas and was strongly associated with presence of an IDH mutation. For example, in the TCGA dataset 30 of the total number of analyzed glioblastomas (6%) carried an IDH mutation, largely overlapping with the 42 tumors that exhibited a g-CIMP phenotype (7.9%). In addition, two other methylation clusters of glioblastoma were found (#2 and #3) in which IDH-wildtype glioblastomas were grouped. Comparison of these three methylation clusters with the four established expression profiles of glioblastomas revealed that most glioblastomas with a proneural expression signature were found in methylation cluster #1 (g-CIMP), while glioblastomas with a neural, mesenchymal or classical expression signature were assigned to the other two methylation clusters without a particularly clear distribution pattern. Furthermore, tumors with a 1p/19q-codeletion (“gCIMP-A”) had a clearly different methylation profile compared to TP53-mutant gliomas (“g-CIMP-B”). A systematic classification of CNS tumors including glioblastomas on the basis of their methylation profile became possible due to the availability of the 450 K/850 K assay in combination with newly developed software, which is capable of unsupervised analysis of the complex results. Several closely collaborating research groups from Heidelberg have now established a comprehensive CNS tumor classifier based on methylation profiling and thereby became leaders in this field. In a first publication six glioblastoma subgroups were reported: (1) the IDH subtype, characterized by IDH mutations with a g-CIMP phenotype; (2) the H3 K27 subtype, consisting of tumors with, for example, an H3F3A K27M mutation; (3) the H3 G34 subtype of tumors with an H3F3A G34R/V mutation; (4) the RTK I or PDGFRA subtype; (5) the mesenchymal subtype; and (6) the classic, RTK II subtype. This RTK II subtype overlapped with methylation cluster #3 as described in the previous paragraph, subtypes 2–5 with methylation cluster #2, and subtype 1 with methylation cluster #1. The CNS tumor classifier was further refined by the analysis of a large number of additional tumors, and the current classifier (02/18) with version number 11b4 (https://www.molecularneuropathology.org/mnp/classifier/2) now lists the following nine glioblastoma methylation classes (ordered from highest to lowest median age of the patients in these classes):

• • • • • • •

Glioblastoma, IDH-wildtype, subclass RTK I PDGFRA: these tumors are typically found in the cerebral hemisphere; the patients have a median age of 64 years; genetically, gain of chromosome 7 with or without EGFR amplification, loss of chromosome 10, and homozygous deletions of CDKN2A are typically present; also, amplification of PDGFRA is more common in this methylation class and the expression profile corresponds to the proneural subtype. Glioblastoma, IDH-wildtype, subclass RTK II classic: in this methylation class the tumors have a supratentorial location; the patients have a median age of 61 years; some gliosarcomas are grouped in this class; genetically, like in the previous methylation class, these tumors also show gain of chromosome 7 with or without EGFR amplification, loss of chromosome 10, and homozygous deletions of CDKN2A; additionally, around 40% of these tumors also have gains of chromosome 19 and 20. Glioblastoma, IDH-wildtype, subclass mesenchymal: these tumors typically grow supratentorially and the patients have a median age of 59 years; histologically this group includes gliosarcomas; genetically, gain of chromosome 7 with or without amplification of EGFR, loss of chromosome 10, and homozygous deletions of CDKN2A are observed; NF1 mutations are enriched in this methylation class, which overlaps with the mesenchymal expression profile. IDH glioma, subclass high-grade astrocytoma: this class encompasses supratentorially located astrocytomas with an IDH mutation, which are often derived from a low-grade lesion and are associated with the g-CIMP phenotype; the median age of the patients in this class is 38 years; this methylation class corresponds to glioblastoma, IDH-mutant (and partially anaplastic astrocytoma, IDH-mutant) in the WHO classification. Glioblastoma, IDH-wildtype, H3.3 G34-mutant: this typically concerns supratentorial tumors with the morphological appearance of a glioblastoma or an embryonal tumor; the median age of the patient is 20 years; the tumors are H3F3A G34R/V-mutant. Glioblastoma, IDH-wildtype, subclass midline: this class represents glioblastomas growing in the midline regions of the brain and showing an epigenetic overlap with the methylation class diffuse midline glioma, H3 K27M-mutant but the tumors do not have the H3 K27M mutation; the median age of the patients in this group is 13 years. Diffuse midline glioma, H3 K27M-mutant: this class overlaps with diffuse midline gliomas, H3 K27M mutantas listed in the WHO 2016 classification; the median age of patients in this class is 12 years.

Glioblastoma: Pathology and Genetics

• •

175

Glioblastoma, IDH-wildtype, subclass MYCN: these tumors are mostly located supratentorially or in the posterior fossa; the median age of the patients is 11 years, and amplification of MYCN can be detected in 20%–30% of these tumors; Glioblastoma, IDH-wildtype, subclass RTK III: these tumors are found mainly in the cerebral hemisphere; the median age of the patients in this class is 9 years, and in this group EGFR amplification and deletions of chromosome 10 are frequent.

Further analysis of diffuse infiltrating gliomas in children with the morphological appearance of a glioblastoma allowed the identification of two further subtypes: tumors with the epigenetic profile pleomorphic xanthoastrocytoma (PXA)-like and low-grade glioma (LGG)-like. The tumors in this latter class are characterized by the absence of an IDH mutation. Patients from both subgroups showed a significantly better prognosis. A further analysis of supratentorial pediatric glioblastomas without H3 and without IDH mutation resulted in the identification of three methylation classes: (1) pediatric glioblastoma MYCN; (2) pediatric glioblastoma RTK1; and (3) pediatric glioblastoma RTK2. Patients with tumors in methylation class RTK1 and RTK2 showed a relatively good prognosis. The epigenetic pattern of the pediatric methylation classes RTK1 and RTK2 was different from the profile of the adult methylation classes RTK I and RTK II. Pediatric patients in methylation class MYCN, on the other hand, showed an exceptionally poor prognosis. MYCN amplification was relatively frequently found in this methylation class, but the majority of tumors did not exhibit such genetic alteration. In addition to the actual classification as part of a methylation analysis, information on chromosomal gains and losses (e.g., complete 1p/19q codeletion versus partial deletion of (one of) these chromosome arms) and on the methylation profile of (the promoters of) individual genes such as MGMT is also provided. Methylation analysis also facilitates that different neuropathologists generate identical diagnoses, and it was reported that compared to traditional histopathologic evaluation the classification by methylation analysis enabled a better estimation of the prognosis associated with the investigated tumors. As the “Heidelberg CNS tumor classifier” is a self-learning system, identification of new, rare subgroups can be expected by adding methylation data of additional tumors. In other words: The more CNS tumors are examined by methylation profiling in this system, the better the classification of these tumors will be. Disadvantages of this approach are, however, that the necessary laboratory infrastructure is costly and that processing individual samples is relatively expensive as well. Furthermore, using the 450 or 850 K technology as described above creates a dependence on a single manufacturer, as it has not yet been conclusively proven whether transfer to another global methylation assay is possible without problems with preserving the established classifier.

Integrated TCGA Approach As the Cancer Genome Atlas research network (TCGA) focused already in an early phase on glioblastomas, most of the data sets on these tumors are now completely available. An essential concept of the TCGA network was the integrated analysis of glioblastomas using various high-throughput assays. For example, many hundreds of clinically annotated glioblastomas were examined with regard to their protein expression, mRNA expression profile, methylation profile, mutation status and chromosomal alterations and the resulting data were combined. Fortunately, the results confirmed data from many previous studies with regard to the molecular characteristics of glioblastomas. Potential downsides of the TCGA approach are the retrospective approach, the heterogeneous therapy of the patients of which material was included, the potential bias resulting from the inclusion criterion of sufficient tissue for performing multiple high throughput analyses, the focus on adult patients and the bias towards patients that were treated predominantly in academic centers. Fusion of the TCGA datasets from different high-throughput assays resulted in only a few relevant and new association patterns. It was not possible to identify individual molecular markers that could clearly be assigned to one of the different subgroups. At best, singular molecular markers were more common in one subgroup than in the other. For example, NF1 mutations were frequently observed in glioblastomas of the mesenchymal expression profile. PDGFRA amplifications were found mainly in glioblastomas of the proneural signature, which did not correspond to the g-CIMP methylation profile. ATRX mutations and MYC amplifications were detected mainly in glioblastomas of the g-CIMP methylation profile, CDK4 and SOX2 amplifications in tumors of the proneural subtype, and amplifications of chromosomes 19 and 20 in the classical subtype. Also, within the TCGA project, an attempt was made to characterize epigenetic glioblastoma subtypes at the methylation level. Six methylation groups of glioblastomas were identified (M1–6, with M5 being the g-CIMP subtype). In comparison to the Heidelberg glioblastoma methylation classes discussed above, the Heidelberg IDH-mutant subtype was found in group M5, the mesenchymal subtype especially in groups M1 and M2, the RTK II subtype in groups M3 and M4, and the RTK I subtype in group M6. Since only adult glioblastoma patients were included in the TCGA cohort, the Heidelberg subgroups K27 and G34, which mainly affect children and young adults, could not be identified. When considering the frequencies of molecular aberrations in the cohort of glioblastoma patients investigated by the TCGA project, the TERT promoter mutation is the most frequent aberration (present in about 80% of the glioblastomas analyzed). A homozygous deletion of CDKN2A and an EGFR alteration were found in about 60% of the tumors, PTEN was altered in about 40%, and TP53 in 35% of cases. Furthermore, amplifications in CDK4 and mutations in NF1 were found in about 14%, PIK3CA and PIK3R1 mutations as well as PDGFRA amplifications in slightly more than 10% of the tumors. A surprisingly broad spectrum of EGFR alterations was detected (amplification, deletion, point mutation, fusion). Furthermore, it was noted that frequently several genes for tyrosine kinase receptors (EGFR, PDGFRA, MET, FGFR) were altered in a single tumor. While most of these observations confirmed the knowledge obtained in previous studies on the genetic changes in glioblastomas, the relatively high frequency of NF1

176

Glioblastoma: Pathology and Genetics

mutations came as a sort of surprise. However, since NF1 is a very large gene, complete analysis of this gene using classic Sanger sequencing is complex, explaining that until the TCGA data became available limited reliable data on the involvement of NF1 in glioblastomas were published. When mapping the broad spectrum of mutations identified in the TCGA glioblastoma cohort into different tumor-relevant signaling pathways, it appears that these pathways are all frequently affected. Alterations in the PIK3CA/AKT and the MAPK signaling pathway were found in 90% of the investigated glioblastomas. In 66% of the tumors, combined alterations at the upper end of the signaling pathways were found by alteration of the receptor tyrosine kinase, and in almost 70% of the tumors either a PTEN or a PIK3CA/PIK3R1 mutation was detected. Senescence/apoptosis regulation was affected in 86% of glioblastomas. The homozygous deletion of CDKN2A/B was predominant, followed by TP53 mutations and deletions as well as amplification of MDM2 and MDM4. In 79% of the investigated glioblastomas, the genes involved in cell cycle control were also impaired. Again, CDKN2A/B proved to be the leader in this respect, while other relevant genes of this signaling pathway were altered at significantly lower frequencies (Fig. 8).

Telomeres maintenance

Apoptosis/ Senescence

AKT/MAPK pathway

Cell cycle control

80

Frequency of alterations

70

60 Especially in IDH-wildtype glioblastoma

50 Especially in IDH-mutant glioblastoma (IDH mutation = defining feature!)

40

Present in both IDH-wildtype and IDH-mutant glioblastomas Frequently present in epitheloid glioblastoma

30

20 MGMT promoter methylation is an epigenetic marker, predictive for response to alkylating chemotherapy

10

TERT CDKN2A/B EGFR MGMT PTEN TP53 CDK4 NF1 PIK3CA PIK3R1 SPTA1 PDGFRA RB1 MDM2 MDM4 TCHH ATRX CDKN2C KEL IDH1 ABCC9 STAG2 COL1A2 HEATR7B2 NLRP5 GABRA6 SEMA3C SCN9A CDH18 LZTR1 ABCB1 DRD5 ADAM29 SEMG1 CXorf22 RPL5 FGFR CARD6 BRAF QKI MET CDK6

0

Fig. 8 Genes most frequently showing aberrations (mutation, loss, amplification, fusion) in glioblastomas according to The Cancer Genome Atlas consortium (TCGA; see https://cancergenome.nih.gov). Because the TCGA study focused on glioblastomas in adult patients, the H3 K27- and H3 G34-mutant tumors are not represented in this graph. Examples of genes (almost) exclusively affected in the IDH-wildtype versus IDH-mutant category of glioblastoma in adults are indicated by a light red versus light pink square (with presence versus absence of an IDH mutation in fact being the criterion based on which these two groups are discriminated). TP53 is frequently involved in both molecular groups of glioblastomas (indicated by square showing both light red and light pink color). BRAF V600E mutation (highlighted by a gray square) is frequent in epitheloid glioblastoma as well as in histologic look-alikes such as anaplastic pleomorphic xanthoastrocytoma and anaplastic ganglioglioma (present in about 50% of these neoplasms). Many of the genes listed in this graph can be assigned to the cellular systems of telomere maintenance (black dashed lines), apoptosis/ senescence (green dashed lines), the AKT/MAPK pathways (blue dashed lines) and the cell cycle control pathway (red dashed lines). CDKN2A/B belongs functionally to both the apoptosis/senescence and cell cycle control pathway. MGMT is also also included in this figure (highlighted in light green); about 50% of glioblastomas in the TCGA cohort showed MGMT promoter (hyper)methylation which serves as a marker indicating better response to alkylating chemotherapy; as a result of their glioma CpG-island methylator phenotype (g-CIMP) status, IDH-mutant glioblastomas generally have a methylated MGMT promoter, while in IDH-wildtype glioblastomas this is found in about 40% of the cases.

Glioblastoma: Pathology and Genetics

177

Prospective Vision The introduction of molecular markers in the WHO 2016 definition of CNS tumors is a paradigm shift that allows for a more robust recognition of clinically relevant subgroups of glioblastoma. This change necessitates further study of what exactly the clinical behavior of these subgroups is and how additional molecular markers can be optimally used for the diagnosis of these newly defined categories. It is already evident that apart from assessment of IDH and H3 K27M mutation status, which are part of the definition of glioblastoma subgroups, analysis of other molecular markers such as mutation of ATRX and/or TP53 (especially present in adult IDH-mutant glioblastomas in adults) or TERT promoter mutation (present in a large percentage of adult IDHwildtype glioblastomas) may be helpful as well. For instance, in a situation where molecular tools for further characterization of glioblastomas are lacking, many cases can still be put in a particular molecular category based on immunohistochemical analysis of (surrogate) markers. Cytoplasmic staining of glioblastoma cells for IDH1 R132H mutant protein is, if the procedure was adequately performed, proof of an IDH1-mutant tumor. Obviously, lack of IDH1 R132H staining does not completely rule out an IDH-mutant tumor as other IDH1 and IDH2 mutant proteins (i.e., about 10% of all IDH-mutant gliomas) are not recognized by this antibody. Similarly, in the right context H3 K27M-mutant protein staining in tumor cell nuclei (with negative nuclei of nonneoplastic cells as an internal control) allows for unequivocal identification of a diffuse midline glioma, H3 K27M-mutant. As ATRX-mutant and TP53-mutant diffuse gliomas generally show lack of ATRX staining of tumor cell nuclei versus strong and extensive p53 staining of the nuclei these immunohistochemical markers may help to put a glioblastoma in a particular, molecular subgroup as well. It can be expected that the development of novel immunohistochemical (surrogate) markers will further facilitate making a “molecular diagnosis” without performing genetic testing. Meanwhile, in many centers high-throughput technologies for sophisticated molecular classification become increasingly available in a routine diagnostic setting. Improved methods for extraction of RNA in sufficient quality and quantity from formalin-fixed, paraffin-embedded tissue may help to create a niche for expression profiling as a routine tool for the neuropathological diagnosis of glioblastomas. Application of such tools will provide more information on the presence and clinical importance of fusion genes (i.e., chromosomal alterations that are easily “lost” in DNA NGS). With TERT promoter mutations as a first example, another aspect that needs further elucidation is the importance of mutations in noncoding DNA segments. For some tumors there is already data on mutated super enhancers which may be located on chromosomes other than the relevant gene. Furthermore, it can be assumed that the classification of CNS tumors including glioblastomas will become increasingly better the more such tumors are examined at methylation level. As the “Heidelberg CNS tumor classifier” is a self-learning system, identification of new, very rare subgroups can be expected by adding further methylation data of additional tumors. Time will tell how exactly this classifier will eventually be used in the neurooncology community as a support tool for (or even alternative for) a histomolecular CNS tumor classification. Much more information on the (epi)genetic, molecular, and protein characteristics of glioblastomas will certainly be acquired in the years to come and will help to acquire a more complete understanding of these tumors. For instance, it is now possible to interrogate individual tumor cells for their DNA, RNA and methylome characteristics, and the results so far underscore a high level of intratumoral heterogeneity. Another aspect that requires a better understanding is the relationship between tumor cells and their microenvironment (or of the host more in general), not in the least because of the so far disappointing results of antiangiogenic and immune-therapy approaches in patients with glioblastoma. Elucidation of how tumor cells change under pressure of (chemo-, radio-, immuno-)therapy may help to improve the efficacy of treatment of recurrent glioblastomas. One can expect that sooner or later a plateau phase will be reached in our understanding of glioblastoma pathobiology. Meanwhile, it is increasingly possible to make a precise, sometimes even molecular diagnosis based on minimally invasive diagnostic approaches such as innovative imaging and liquid biopsies. Unfortunately, against the background of all the progress that has been made in elucidating the pathology and genetics of glioblastomas, the sobering reality so far is that the prognosis for glioblastoma patients has not substantially improved. It is thus time that the increased knowledge described in this chapter can be translated in therapeutic strategies that do improve the outcome for patients suffering from these tumors.

See also: Glioblastoma: Biology Diagnosis, and Treatment.

Further Reading Barthel, F.P., Wesseling, P., Verhaak, R., 2018. Reconstructing the molecular life history of gliomas. Acta Neuropathologica 135, 649–670. Cancer Genome Atlas Research Network, Brat, D.J., Verhaak, R.G, et al., 2015. Comprehensive, integrative genomic analysis of diffuse lower-grade gliomas. New England Journal of Medicine 372, 2481–2498. Capper, D., Jones, D.T.W., Sill, M., et al., 2018. DNA methylation-based classification of central nervous system tumours. Nature 555, 469–474. Ceccarelli, M., Barthel, F.P., Malta, T.M., et al., 2016. Molecular profiling reveals biologically discrete subsets and pathways of progression in diffuse glioma. Cell 164, 550–563. Eckel-Passow, J.E., Lachance, D.H., Molinaro, A.M., et al., 2015. Glioma groups based on 1p/19q, IDH, and TERT promoter mutations in tumors. New England Journal of Medicine 372, 2499–2508. Korshunov, A., Ryzhova, M., Hovestadt, V., et al., 2015. Integrated analysis of pediatric glioblastoma reveals a subset of biologically favorable tumors with associated molecular prognostic markers. Acta Neuropathologica 129, 669–678.

178

Glioblastoma: Pathology and Genetics

Louis, D.N., Perry, A., Burger, P., et al., 2014. International Society of Neuropathology-Haarlem consensus guidelines for nervous system tumor classification and grading. Brain Pathology 24, 429–435. Louis, D.N., Ohgaki, H., Wiestler, O.D., et al. (Eds.), 2016. WHO classification of tumours of the central nervous system. IARC Press, Lyon. Louis, D.N., Aldape, K., Brat, D.J., et al., 2017. Announcing cIMPACT-NOW: The consortium to inform molecular and practical approaches to CNS tumor taxonomy. Acta Neuropathologica 133, 1–3. Ostrom, Q.T., Gittleman, H., Liao, P., et al., 2017. CBTRUS statistical report: Primary brain and other central nervous system tumors diagnosed in the United States in 2010–2014. Neuro Oncology 19 (Suppl. 5), v1–v88. Perry, A., Wesseling, P., 2016. Histologic classification of gliomas. Handbook of Clinical Neurology 134, 71–95. Reuss, D.E., Sahm, F., Schrimpf, D., et al., 2015. ATRX and IDH1-R132H immunohistochemistry with subsequent copy number analysis and IDH sequencing as a basis for an “integrated” diagnostic approach for adult astrocytoma, oligodendroglioma and glioblastoma. Acta Neuropathologica 129, 133–146. Schwartzentruber, J., Korshunov, A., Liu, X.Y., et al., 2014. Driver mutations in histone H3.3 and chromatin remodelling genes in paediatric glioblastoma. Nature 482, 226–231. Sturm, D., Witt, H., Hovestadt, V., et al., 2012. Hotspot mutations in H3F3A and IDH1 define distinct epigenetic and biological subgroups of glioblastoma. Cancer Cell 22, 425–437. Wesseling, P., Capper, D., 2018. WHO 2016 classification of gliomas. Neuropathology Applied Neurobiology 44, 139–150. Yan, H., Parsons, D.W., Jin, G., et al., 2009. IDH1 and IDH2 mutations in gliomas. New England Journal of Medicine 360, 765–773.

Relevant Websites https://icgc.orgdInternational Cancer Genome Consortium. https://cancergenome.nih.govdThe Cancer Genome Atlas/TCGA. http://cancerdb.genomics.org.cndChinese Cancer Genome Consortium/CCGC. https://www.molecularneuropathology.orgdMethylation profiling of CNS tumors. https://www.eortc.orgdEuropean Organization for Research and Treatment of Cancer. https://www.rtog.orgdRadiation Therapy Oncology Group. http://www.cbtrus.orgdCentral Brain Tumor Registry of the United States. http://www.pedbraintumor.orgdICGC PedBrainTumor.

Glutamine Metabolism and Cancer Nianli Sang, Drexel University, Philadelphia, PA, United States; and Sidney Kimmel Cancer Center of Philadelphia, Philadelphia, PA, United States Dan He, Drexel University, Philadelphia, PA, United States Shuo Qie, Medical University of South Carolina, Charleston, SC, United States © 2019 Elsevier Inc. All rights reserved.

Glossary Anabolism The biochemical processes in living systems to synthesize complex molecules from simpler molecules. An anabolic process usually requires energy input in the form of ATP or GTP. Anaplerosis The biochemical processes that provides or replenish intermediate metabolites into a metabolic pathway. A common example is TCA cycle. During active biosynthesis, intermediate metabolites of the TCA cycle may be utilized and consumed for biosynthesis. Various mechanisms exist to replenish the metabolites of the TCA cycle. Catabolism The biochemical processes in living system that break down of complex molecules to form simpler molecules. Catabolic reaction usually releases energy, which may be used to generate ATP and heat. Chirality of amino acids Most amino acids assume either L- or D- chirality. Except for glycine which is achiral, all other proteinogenic amino acids are L-a-amino acids. To simply the discussion, we will omit the L-a prefix in the text unless other configuration is involved. Cystine A dimer made of two cysteine molecules which form a disulfide bond between the sufhydro groups. This amino acid is usually imported into the cell by the cystine/glutamate antiporter (known as system Xc- or xCT) encoded by the SLC7A11 gene in humans. After being uptaken into cells, a cystine molecule is converted into two cysteine molecules by reductive reactions. Endoplasmic reticulum stress response An evolutionarily conserved cell response to the disturbances of the normal functions of the endoplasmic reticulum. The most common causes and consequences are accumulation of unfolded proteins in the ER lumen, also known as the unfolded protein response, and a disruption of calcium homeostasis. Essential amino acids The amino acids that are required to support the metabolism of the living system, but the living system cannot synthesize them. Accordingly, essential amino acids have to be obtained by diet. Homeostasis The ability of the living system to maintain a condition of equilibrium or stability within its internal environment when dealing with external changes. mTOR Mechanistic target of rapamycin, also known as mammalian target of rapamycin, is the catalytic subunit of two protein kinase complexes (mTORC1 and mTORC2) that serve as the central sensors and integrators of multiple stimuli inputs. The best known stimuli are extracellular nutrient status, intracellular energy status and availability of growth factors. Based on the integration of the stimuli, mTORCs may regulate a variety of biological processes including cell survival, metabolism, proliferation, migration, autophagy and apoptosis. Nonessential amino acids The amino acids that can be made by a living system, thus not being required to present in the diet. Oncometabolite A molecule or metabolite produced by mutated metabolic enzymes and contribute to oncogenesis and tumor progression. An oncometabolite is commonly resulted from gain of function mutations which alter the normal function of a metabolic enzyme.

Abbreviations AARE Amino acid responsive element ALT Alanine aminotransferase AST Aspartate aminotransferase ATF4 Activating transcription factor 4 BCAA Branched chain amino acids ER Endoplasmic reticulum GCN2 General Control Non-derepressible 2 GDH Glutamate dehydrogenase GLS Glutaminase GSH Glutathione IDH Isocitrate dehydrogenase LAT1 Large neutral amino acid transporter 1

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.64997-8

179

180

Glutamine Metabolism and Cancer

NF-kB Nuclear factor kappa B TCA Tricarboxylic acid cycle a-KG a-ketoglutarate

Introduction Glutamine is one of the 20 amino acids used in protein translation processes. In plants, glutamine can be synthesized by glutamine synthetase, which uses glutamate and free ammonia as substrates and ATP as the energy source. Biochemically, the glutamine synthesis process is the major way to introduce inorganic nitrogen into organic molecules in the form of amino acids, which eventually are converted into proteins and other nitrogenous molecules and flux into higher organisms through food-chain or food-web. In addition to the dietary sources, in animals, multiple tissues can synthesize glutamine, and often use this synthesis as a way to dispose free ammonium ion, a metabolic waste. The most relevant glutamine-producing tissue is the skeletal muscle, accounting for about 90% of all glutamine synthesized. As one of the three amino acids that shuttle the toxic ammonium ion from extrahepatic tissues to hepatocytes, glutamine serves a special role in disposal of the nitrogenous waste through urea cycle. Since glutamine can be synthesized, at organism level glutamine is classified as a nonessential amino acid, and as such, its importance in anabolic pathways other than protein translation and nucleotide biosynthesis are often overlooked. However, rapid proliferating cells including cancer cells, pathogen stimulated immune cells and various progenitor or stem cells have increased demand for glutamine, and often show glutamine dependent growth and proliferation. This increased demand usually excesses the needs for protein translation and nucleotide synthesis. In fact, most types of cancer cells cultured in vitro depend on high levels of supplemented glutamine. In those cells, most of the glutamine uptaken by cells are converted into glutamate through glutaminolysis catalyzed by glutaminase. More and more published studies support the notion that proliferating cancer cells use glutaminolysis to maintain the intracellular glutamate homeostasis to support a metabolic reprogramming associated with malignant transformation. The amido and amino groups from glutamine are crucial for the active nitrogen anabolism in proliferating cells, and the carbon skeleton from glutamine catabolism eventually contributes to the carbon pool, supporting ATP production, anaplerosis and lipid biosynthesis. In this article, we will summarize the glutamine metabolism in cancer cells, focusing on biosynthetic roles of glutamine in proliferating cells as both a nitrogen source and a carbon source and the regulatory roles of oncogenic signaling.

Direct Roles of Glutamine and Major Enzymes Glutamine is the most abundant one among all amino acids in circulation, including the proteinogenic or nonproteinogenic amino acids. Cells uptake glutamine through glutamine transporters, a transmembrane protein family consisting of multiple members. The structural–functional relationships and specific roles of these transporters under defined physiological or pathological conditions remain not well studied. However, SLC1A5 (ASCT2) may have special importance for cancer cells to uptake glutamine. Upon uptaken by cells, glutamine can be directly utilized for protein translation and nucleotides biosynthesis, two processes are described in details in most molecular biology textbooks. Intracellular accumulation of glutamine also provides a chemical potential for cells to uptake other nutrients. For example, leucine is an essential amino acid, and the uptake of leucine by the large neutral amino acid transporter 1 (LAT1) antiporter requires the simultaneous release of glutamine to the outside of cells. A summary of glutamine utilization in cells are outlined in Fig. 1. Importantly, a major portion of intracellular glutamine will be converted into glutamate and an ammonium ion primarily in the mitochondria by the enzyme glutaminase. Part of the mitochondrial glutamate will be exported to cytosol to join the glutamate pool, and the other portion of glutamate is further converted into a-ketoglutarate by glutamate dehydrogenase and release ammonia. This interconversion between glutamate and a-ketoglutarate can also be catalyzed by aspartate transaminase (AST), but instead of releasing ammonia, this reaction transfers the amino group from glutamate to oxaloacetate and generates aspartate, another amino acid with special importance in nucleotide biosynthesis. There are two genes encoding glutaminase, GLS and GLS2. GLS encodes for the kidney-type glutaminase (KGA), which is primarily expressed in the brain and kidney and plays a key role in synthesizing the neurotransmitter glutamate and maintaining acid–base balance in the kidney. GLS2 encodes for the liver type glutaminase (LGA), which was originally considered liver-specific, but later found to be expressed in other tissues and cancer cells as well. Alternative splicing may result in multiple transcript variants which generate various glutaminase isoforms in different tissues and cancer cells. Aspartate transaminases, better known clinically as glutamate-oxaloacetate transaminase (GOT), also have two isoenzymes in human: the cytosolic isoenzyme (cAST or GOT1) which primarily expresses in muscle and red blood cells, and the mitochondrial AST (mAST or GOT2) which primarily expresses in hepatocytes. The expression of both isoenzymes have been reported in different types of cancers.

Glutamine Metabolism and Cancer

Gln

Mitochondria

Gln

Glutaminase

Transporter

Transporter

Cytosol

181

Glutamate

Gln

ASP

GDH

AST

α-KG Ammonia

Protein translation Nucleotide biosynthesis Asparagine biosynthesis Glucosamine biosynthesis Nutrient uptake

Fig. 1 Schematic drawing shows glutamine uptake, intracellular dynamics and direct roles in cell metabolism. Transporters and enzymes involved are marked with color text. Dotted fine lines show intracellular diamics or flux, and solid lines indicate biochemical conversion. Gln, glutamine; Asp, aspartate; AST, aspartate transaminase; GDH, glutamate dehydrogenase.

Indirect But Important Roles of Glutamine in Cancer Metabolism Replenishing the pool of cytoplasmic glutamate by glutaminolysis is the most crucial role of glutaminolysis in cancer metabolism, particularly in nitrogen anabolism. In normal cells, various vitamin B6-dependent transaminases operate a glutamatea-KG cycle, in which a-KG collects amino groups from excess amino acids doomed to be catabolized; on the other hand, when an NAEE becomes scarce, glutamate may donate the amino group to synthesize it (Fig. 2A). In cancer cells and other proliferating cells, to support proliferation, active biosynthesis, including many nitrogen anabolic processes in addition to protein translation, is necessary and sometimes can be rate-limiting for cell division. Particularly, these nitrogendependent biosynthetic processes consume large quantity of several amino acids that provide the nitrogen source, or carbon

(A)

Normal Cells

(B) Proliferating Cells

Glutamine

Glutamine

α-KA(x)

Glutamate

α-KA(y)

Glutamate

α-KA(y)

Aminotransferases Aminotransferases α-KG α-AA (excess)

α-KG

NEAA (limiting) ATP production NADH production Anaplerosis

NEAA (limiting) Aspartate Alanine Serine etc

Fig. 2 Central role of glutamate in the homeostasis of nonessential amino acids in normal cells and proliferating cells. (A) In normal cells, the glutamate and a-KG cycle plays the key role to maintain a dynamic homeostasis of the intracellular pool of amino acids. Any amino acids in excess will be catabolized by transamination with a-KG first prior to feeding the carbon skeletons (ketoacids) into TCA cycle. Glutamate then can be used to synthesize any nonessential amino acid in need. (B) In rapid proliferating cells, active biosynthesis will consume most of the amino acids, particularly, some amino acids are consumed in large quantity. In this case, glutamine plays the important role in replenishing the glutamate pool consumed directly or indirectly in biosynthesis of nitrogenous molecules. Metabolites in blue indicate the flux of nitrogen.

182

Glutamine Metabolism and Cancer

source as well in some cases. Glutamate derived from glutaminolysis either directly participates in the biosynthesis, or indirectly provides the nitrogen source or carbon source by synthesizing these amino acids required in large quantity for the active biosynthesis (Fig. 2B). In turn, glutaminolysis provides the ultimate mechanism to maintain the cytosolic homeostasis of glutamate. First of all, in addition to glutamine, glutamate, aspartate and glycine are well-known for their roles in de novo biosynthesis of nucleotides. Moreover, glycine is an important substrate for biosynthesis of heme, a prosthetic factor required for a variety of redox enzymes; and glutathione, an important molecule in maintaining the intracellular redox homeostasis. Similarly, serine plays multiple roles in proliferating cells, thus being required in large quantity. Serine cleavage provides important source of one carbon units for various reactions, including nucleotide synthesis, protein methylation, epigenetic control of gene expression etc. Serine also is a substrate for the biosynthesis of biomembrane by either serving as an alcohol head of phospholids, or the core structure of sphinosine (2-amino-4-octadecene-1, 3-diol). Collectively, to meet the needs for cell growth and division will consume considerable amount of serine in biosynthesis based on the stoichiometry. As such, glutamate is usually consumed as a substrate in large quantity through transamination to synthesize aspartate, glycine, serine and other nonessential amino acids actively participating in proliferative biosynthesis (Fig. 2B). Most human cells maintain high levels of intracellular glutamate, and the cross-membrane glutamate gradient serves as a chemical potential to facilitate the uptake of key metabolites. The human SLC7A11 gene encodes for a sodium-independent but chloride dependent cystine-glutamate antiporter well-known as system Xc- or xCT. SLC7A11 is highly expressed in astrocytes as well as in some types of cancer cells including lung, head and neck carcinomas. Cystine-glutamate antiporter uses the cross-membrane gradient of glutamate as an energy source, and couples the uptake of one molecule of cystine with the release of one molecule of glutamate to the outside of cells. The uptaken cystine in cells will be converted to two molecules of cysteine. In addition to participate in protein translation, as a special amino acid with thiol group, cysteine can be converted into methionine and participate in one carbon unit metabolism and in the biosynthesis of polyamines, a group of amines interacting with newly synthesized DNA and facilitating chromatin condensation. More importantly, cysteine is one of the critical substrate for the biosynthesis of glutathione, participating in the maintenance of redox homeostasis. The important role of glutamine metabolism in supporting glutathione biosynthesis is summarized in Fig. 3.

Glutamine

Glutaminase

Glutamate Xc-/Cystine Serine Cysteine Glutamyl-cysteine synthetase Glutamyl-cysteine

Glycine Glutathione synthetase

Glutathione Fig. 3 Biochemical importance of glutamine in the biosynthesis of glutathione. Glutathione (GSH) is a tripeptide plays important role in protecting cells from damage caused by reactive oxygen or reactive nitrogen species, which may arise from cell metabolic processes, radiation or therapeutic drugs. Rapid proliferating cells require the biosynthesis of a large quantity of GSH to prepare for the cell division. The biosynthesis of GSH requires glutamate, cysteine and glycine as substrates. Glutamine plays important roles in maintaining the intracellular homeostasis of all three amino acids. Key amino acids are boxed. Enzymes and transporters involved are in red. Intermediate metabolites are indicated in italics.

Glutamine Metabolism and Cancer

183

Glutamine, Glutamate and a-Ketoglutarate Pathway in Carbon Metabolism As the carbon skeleton derived from glutamine catabolism, a-KG joins the pool of reduced carbons and contributes to the carbon metabolism. As one of the metabolites of the TCA cycle, a-KG can be oxidized in TCA cycle as a fuel to generate NADH and FADH2, which transfers the electrons to electron transfer chain for ATP production. Moreover, a-KG also plays roles in anaplerosis to support the synthetic function of TCA cycle. Three metabolites of the TCA cycle have particular importance in supporting the proliferative biosynthesis thus being commonly withdrawn from the TCA cycle. Succinyl CoA is required as a substrate for the biosynthesis of Heme, an important cofactor needed for a variety of oxidoreductases. Oxaloacetate can be transaminated into aspartate in the mitochondria and exported to cytosol to support active protein translation and nucleotide biosynthesis. The citrate can be exported to the cytosol to replenish the cytosolic pool of acetyl-CoA and malate. Malic enzyme-catalyzed subsequent reaction utilizes malate as a substrate to produce cytosolic NADPH. Thereafter, both acetyl-CoA and NADPH are utilized for the biosynthesis of fatty acids, cholesterol and other molecules derived from the mevalonate pathway. In addition, a-KG can also directly undergo reductive carboxylation in the cytosol to generate citrate by the cytosolic NADPþ-dependent isocitrate dehydrogenase 1 (IDH1), providing another metabolic pathway generating acetyl-CoA and NADPH for lipogenesis in the cytosol. As an excitatory neurotransmitter and the precursor of the inhibitory neurotransmitter gamma-aminobutyric acid (GABA), glutamate is the most abundant amino acids in human brains. Although the role of glutamate in the physiology of central nerve system is apparently beyond the scope of this article, glutamate metabolism has a special importance in astrocytoma, glioma or glioblastoma, the major forms of brain malignancies derived from astrocytes or glial cells. These tumors frequently carry mutations in the gene encoding IDH1 or the mitochondrial IDH2, which render new enzymatic activity; for example, a common gain of function of IDH1 is to convert a-KG into D-2-hydroxoglutarate, an oncometabolite which competitively inhibits a-KG-dependent enzymes. Similar IDH1/2 mutations and oncogenic metabolism were found in other types of malignancies, including sarcomas, hematologic malignancies and colon cancers, adding a new level of twist of glutamine metabolism in cancer cells.

Glutamine in Maintaining mTORC Activity and Active Biosynthesis As a crucial nutrient, glutamine maintains the activity of mTORC signaling pathways. It is well-known that the mTORC signaling pathways serve as master regulators of cell survival, proliferation and proper functioning. Proper mTOR signaling is important to suppress degradative processes such as autophagy and apoptosis. The mTOR kinase forms two multiprotein complexes, mTOR complex 1 (mTORC1) and mTOR complex 2 (mTORC2). In addition to upstream regulators, the availability of nutrients and growth factor signaling are crucial to maintain mTORC activity. mTORC1 is required for various biosynthetic activities including protein translation and lipogenesis, whereas mTORC2 facilitates cell survival, migration and invasion. Adequate mTORC activity is also important to limit degradative processes including autophagy. According to the above discussion of the metabolic roles of glutamine, it is obvious that glutamine is required in multiple aspects to maintain the mTORC1 activity; particularly glutamine utilization is critical to ensure the homeostasis of the intracellular amino acids pool, redox status and energy which are monitored by mTORC1. Knocking down glutamine transporter SLC1A5 not only inhibits glutamine uptake, but also suppresses mTORC1 signaling.

Glutamine Utilization Promoted by Oncogenic Activation and Loss of Tumor Suppressors In cancer cells, active anabolism requires increased glutamine utilization, which is driven by the activation of oncogenes, the loss of tumor suppressors or other genetic alterations (Fig. 4). Altered expression of glutamine transporters, GLS and glutamate dehydrogenase has been linked to tumor cell growth and proliferation. The first oncogene demonstrated to promote glutamine utilization is c-Myc. MYC encodes a protein containing the basic helixloop-helix leucine zipper (bHLHZip) motifs. With this motif, c-Myc contributes to the genesis of human malignancy via a network of protein–protein interaction and DNA binding. In this network, c-Myc binds to Max, another bHLHZip family member, forming a heterodimer. The structure of Myc protein contains an unstructured N-terminal region which functions as transcriptional regulatory domain, and the C-terminal bHLHZip domain, which dimerizes with Max. The Myc-Max heterodimer recognizes the DNA sequence CACGTG which belongs to the E-box sites. The DNA binding recruits coactivator or corepressor complexes that modify the chromatin, thus regulating the expression of the downstream genes with E-box in their promoters. These genes regulate a wide range of cellular processes, including cell growth, proliferation, and tumorigenesis. Importantly, c-Myc lies at the downstream of, and is regulated in response to some upstream oncogenic signals. Studies showed that the upregulation of MYC drives glutamine utilization via upregulating glutamine transporters and metabolism enzymes such as GLS. The second reported oncogenic signaling pathway that drives glutamine utilization is HER2 (ErbB2). Activation of ErbB2 is frequently involved in breast cancer, ovarian cancer, gastric cancer and other cancers. The formation of homodimer or heterodimer of ErbB2 with other forms of ErbBs triggers its downstream signaling including MAPK/ERK, PI3K-Akt pathway, or nuclear factor kappa B (NF-kB). It has been demonstrated that ErbB2 drives glutamine utilization by activating GLS in breast cancers. Finally, AP1 (c-Jun/Fos) has been reported to promote glutaminolysis. c-Jun, as a transcription regulator, is derived from the proto-oncogene JUN. c-Jun modulates oncogenic Rho-GTPase signaling to enhance the expression and enzymatic activity of mitochondrial GLS in breast cancer, which facilitates cancer cell

184

Glutamine Metabolism and Cancer

MYC

c-Jun

NF-kB

Rho

NH3

Gln transporter Gln

HER2

Glutaminase Gln

Glutamate

P53 mut Cytosol

PKCε Mitochondria

Fig. 4 Common oncogenic signaling pathways that promote glutamine utilization in cancer cells. Major oncogenic signaling pathways that promote glutamine utilization in cancer cells are shown schematically. Oncogenes or tumor suppressors were shown in red boxes.

metabolism and progression. More recently, a truncated isoform of glutaminase (GAC) expressed in some cancer cells was found to be activated by phosphorylation catalyzed by PKCε, providing a posttranslational mechanism to enhance glutamine utilization in cancer cells. As a tumor suppressor, p53 suppresses the production of reactive oxygen species (ROS) through upregulating GLS2, which enhances the production of reduced glutathione. Exogenous GLS2 expression inhibits tumor growth in vivo, consistent with the tumor suppressive activity of p53. However, in some tumor cells, GLS2 is upregulated to promote tumor growth. In addition, another tumor suppressor, retinoblastoma protein (Rb) also regulates glutamine metabolism, and the loss of Rb drives tumor cells’ dependency on glutamine, which also exemplifies a proliferation enhanced utilization of glutamine.

Responses of Cancer Cells to Glutamine Depletion During tumorigenesis and tumor progression, cancer cells are frequently exposed to a poorly vascularized microenvironment which is often characterized by the lack of oxygen, glucose and glutamine. To survive in such a stressful microenvironment, tumor cells have obtained strategies to sense the lack of a specific nutrient and respond to the status by reprogramming cellular processes and metabolism. The general responses to stresses usually include the induction of autophagy, inhibition of mTOR activity, cell cycle arrest, endoplasmic reticulum (ER) stress response and activation of the heat shock system. Like other stress conditions, glutamine depletion triggers the above general responses as well (Fig. 5). In addition to these general responses to stresses, cancer cells also develop metabolic reprogramming more specific to glutamine depletion (Fig. 5). These reprogramming include the decreased utilization of glucose, increased uptaking and utilization of other amino acids and enhanced recycling of ammonia. The decreased utilization of glucose is manifested by the decreased expression and activity of enzymes involving in both glycolytic and pentose pathways. The increased uptaking and utilization of other amino acids is highlighted by the enhanced expression of transporters for amino acids, such as branched chain amino acids (BCAA). Moreover, glutamine depletion also promotes the expression of transaminases for alanine (ALT) and branched chain amino acids. Recycling of ammonia in cancer cells is catalyzed by glutamate dehydrogenase, whose activity can be easily reversed by the alteration of substrate and/or product concentrations. Collectively, those metabolic reprogramming events triggered by glutamine depletion may either utilize alternative sources of nitrogen, or slow down global cell metabolism to conserve the nitrogen sources. How cells monitor the glutamine availability remain illusive. Initially, the lack of essential or conditionally essential amino acids including cysteine and leucine induces an “integrated stress response” to amino acid depletion and redox stress, in which GCN2 senses the increased amount of nonaminoacylated tRNA. Nonaminoacylated tRNA binds with and activates GCN2. Like PERK in the ER lumen, activated GCN2 phosphorylates eIF2a, attenuating general protein synthesis but selectively translates ATF4, a transcription factor that binds to amino acid responsive elements (AARE) and regulates gene transcription involved in relieving the stresses caused by amino acid depletion. Later, it is found that glutamine depletion is sufficient to induce eIF2a phosphorylation and thus activates ATF4. This process may involve both GCN2 activation and the PERK kinase activity as a part of the ER stress responses. Considering glutamine’s crucial role in biosynthesis of nonessential amino acids, glutamine depletion could lead to insufficiency of nonessential amino acids such as serine, glycine, aspartate which are required in large quantity in proliferating cells. In addition, glutamine depletion disrupted glutamate homeostasis may directly compromise GSH synthesis, and thereafter, the accumulation of redox stress facilitates the activation of ER stress, providing additional mechanism to activate ATF4. Activation of ATF4 may lead to transcriptional reprogramming and the relieving of stresses including nutrient depletion. In this aspect, it seems that the availability of multiple amino acids as a pool, instead of glutamine alone, is collectively sensed by multiple cellular mechanisms.

Glutamine Metabolism and Cancer

185

Glutamine Depletion

General Stress Responses

Gln Specific Responses

mTOR inhibition

Gln Uptake

Autophagy

Uptake of other A.A.

Heat Shock System

NEAA synthesis

ER stress response

Recycling ammonia

Cell Cycle Arrest

AA-tRNA loading

Apoptosis

Glucosamine synthesis

Fig. 5 Glutamine depletion triggeres general stress responses and glutamine depletion specific, adaptive metabolism. Glutamine depletion is a common condition that cancer cells may frequently encounter in vivo. Current studies indicate that glutamine depletion triggered cellular responses can be classified into two major categories: those not specific to defined stresses, and those specific to glutamine depletion. The responses specific to glutamine depletion generally tend to increase the ability or enzyme activities to reach a new homeostasis of a metabolite in scarce supply, indicative of compensatory strategies of cancer cells.

Prospective Vision Cancer metabolism is characterized by increased glycolysis and active glutaminolysis, representing the enhanced utilization of glucose and glutamine respectively. Targeting cancer metabolism as potential novel therapies has been explored at various stages including basic research, preclinical and clinical stages in recent years. The initial success has been demonstrated by various trials targeting the glucose utilization pathways in cancers. Preclinical studies on the use of inhibitors of glutamine transporters or metabolic enzymes in order to block glutamine utilization also result in initial excitement. However, cancer cells may usually evolve the ability to adapt to nutrient depletion, which negatively affects the efficacy of various strategies to “starve” cancer cells. Accordingly, identification of critical adaptive strategies of cancer cells induced by nutrient depletion will potentially provide novel therapeutic targets. A combination of targeting both cancer metabolism and its adaptive strategy may provide therapeutic advantages and improve the efficacy. Existing data from cultured cells indicate that whereas glutamine depletion suppresses glucose utilization, and that glutamine utilization may be limited by the availability of other nutrients such as glucose and oxygen, indicating a coordinated utilization of glutamine, glucose and molecular oxygen. Our unpublished observations show that glucose depletion suppresses c-Myc/Max, which indicates limited glucose supply may suppress glutamine utilization driven by Myc oncogene. Hypoxia-inducible factors (HIF-1, 2, collectively HIFs) are heterodimeric transcription factors which are activated primarily by hypoxia and serve as drivers of ATP production. Accordingly, HIFs are often activated by hypoxia and other biological stresses such as injury and infection to promote cell repair and defense. It has been reported that there is a functional interaction between the c-Myc and HIF pathways, and HIF activation counteracts the cell cycle promoting effect of c-Myc. A better understanding of the coordination of cellular response to nutrient depletion may provide new insight into the molecular and biochemical mechanisms of cancer cells’ adaptation to a combined nutrient depletion which is more relevant to cancer metabolism in real patients. Mechanistically, while the enzymes involved in glutamine utilization are well studied, and the metabolic flux of glutamine also has been extensively studied by combined isotope-tracking and mass-spectrometric analyses, the expression and regulation of glutamine transporters in cancers remain to be further clarified. Altered metabolic pathways or metabolites in cancers caused by gain-offunction mutations of metabolic enzymes remain to be further elucidated.

Acknowledgments Relevant research in Dr. Sang’s laboratory are supported by grants from the National Cancer Institute of United States and a research contract grant from Lipogen Inc., United States.

186

Glutamine Metabolism and Cancer

Further Reading Coloff, J.L., Murphy, J.P., Braun, C.R., Harris, I.S., Shelton, L.M., Kami, K., Gygi, S.P., Selfors, L.M., Brugge, J.S., 2016. Differential glutamate metabolism in proliferating and quiescent mammary epithelial cells. Cell Metabolism 23, 867–880. Gao, P., Tchernyshyov, I., Chang, T.C., Lee, Y.S., Kita, K., Ochi, T., Zeller, K.I., De Marzo, A.M., Van Eyk, J.E., Mendell, J.T., Dang, C.V., 2009. C-Myc suppression of miR-23a/ b enhances mitochondrial glutaminase expression and glutamine metabolism. Nature 8, 762–765. Harding, H.P., Zhang, Y., Zeng, H., Novoa, I., Lu, P.D., Calfon, M., Sadri, N., Yun, C., Popko, B., Paules, R., Stojdl, D.F., 2003. An integrated stress response regulates amino acid metabolism and resistance to oxidative stress. Molecular Cell 11, 619–633. Locasale, J.W., 2013. Serine, glycine and one’carbon units: Cancer metabolism in full circle. Nature Reviews. Cancer 13, 572–583. Lukey, M.J., Greene, K.S., Erickson, J.W., Wilson, K.F., Cerione, R.A., 2016. The oncogenic transcription factor c-Jun regulates glutaminase expression and sensitizes cells to glutaminase-targeted therapy. Nature Communications 7, 11321. Medina, M.A., 2001. Glutamine and cancer. The Journal of Nutrition 131, 2539S–2542S. Meng, M., Chen, S., Lao, T., Liang, D., Sang, N., 2010. Nitrogen anabolism underlies the importance of glutaminolysis in proliferating cells. Cell Cycle 9, 3921–3932. Possemato, R., Marks, K.M., Shaul, Y.D., Pacold, M.E., Kim, D., Birsoy, K., Sethumadhavan, S., Woo, H.K., Jang, H.G., Jha, A.K., Chen, W.W., Sabatini, D.M., 2011. Functional genomics reveal that the serine synthesis pathway is essential in breast cancer. Nature 476, 346–350. Qie, S., Liang, D., Yin, C., Gu, W., Meng, M., Wang, C., Sang, N., 2012. Glutamine depletion and glucose depletion trigger growth inhibition via distinctive gene expression reprogramming. Cell Cycle 11, 3679–3690. Qie, S., Chu, C., Li, W., Wang, C., Sang, N., 2014. ErbB2 activation upregulates glutaminase 1 expression which promotes breast cancer cell proliferation. Journal of Cellular Biochemistry 115, 498–509. Spinelli, J.B., Yoon, H., Ringel, A.E., Jeanfavre, S., Clish, C.B., Haigis, M.C., 2017. Metabolic recycling of ammonia via glutamate dehydrogenase supports breast cancer biomass. Science 358, 941–946. Wise, D.R., DeBerardinis, R.J., Mancuso, A., Sayed, N., Zhang, X.Y., Pfeiffer, H.K., Nissim, I., Daikhin, E., Yudkoff, M., McMahon, S.B., Thompson, C.B., 2008. Myc regulates a transcriptional program that stimulates mitochondrial glutaminolysis and leads to glutamine addiction. Proceedings of the National Academy of Sciences of the United States of America 105, 18782–18787.

Helicobacter Pylori-Mediated Carcinogenesis Matthew G Varga, University of North Carolina, Chapel Hill, NC, United States Meira Epplein, Duke University, Durham, NC, United States © 2019 Elsevier Inc. All rights reserved.

The Discovery of H. pylori as an Infectious Agent Within the Stomach Helicobacter pylori has colonized and coevolved with its human host for millennia. The earliest direct evidence of H. pylori colonization can be seen from gastric biopsies taken from the Iceman, a 5300-year old mummy from the Copper Age, yet H. pylori colonization dates back at least 100,000 years. Indeed, conserved domains within the H. pylori genome can be used to mirror the paths of human migration out of Africa. Although H. pylori was not officially discovered until 1983, spiral-shaped bacteria had been observed in human stomachs by German physicians as early as 1875. In 1899, a Polish physician named Walery Jaworski observed spiral shaped bacteria in sediments from gastric washings and named the organism Vibrio rugula. In his book entitled “Handbook of Gastric Diseases”, Jaworski even implicated V. rugula as a causative agent of gastric disease. However, the observations of these scientists were forgotten until 1939–40 when physicians again observed “spirochaetes” within the human gastric mucosa of patients with and without gastric disease. Finally, in 1954 Palmer et al. examined over 1000 gastric biopsies and found no bacteria with spirochetal structure from their samples. They concluded that the stomach was a sterile organ and that any observed bacteria were oral contaminants that multiplied in postmortem specimens. Nonetheless, observations of bacteria within the human stomach persisted into 1975 when Steer and Colin-Jones observed Gram-negative bacilli in 80% of patients with gastric ulcer, yet they dismissed their observation as a possible contaminant of Pseudomonas (Fig. 1). In 1983 Barry J. Marshall and John Robin Warren were the first to both observe and culture back an unspecified bacteria from the gastric epithelium among patients with chronic gastritis. To accomplish what their predecessors could not, Marshall and Warren had used a silver stain, which was unconventional for gastric specimens at the time, to observe the spiral-shaped bacteria. Furthermore, the conventional amount of time for bacteria recovered from the gastrointestinal tract to grow in the lab was only 2 days, after which the agar plates would be discarded if there was no bacterial growth. However, due to a fortunate accident whereby Marshall and Warren had accidentally left their agar plates in the incubator for 6 days over the Easter weekend, they were then able to recover their unspecified bacterium. They classified the bacterium not as a spirochete as others had described, but rather a species of Campylobacter due to its morphology. In 1984, they again published that this unspecified bacterium was recoverable from chronic active gastritis as well as peptic ulcers. The pair named this unspecified organism Campylobacter pyloridis due to its predominance within the gastric antrum. In 1989, 16S ribosomal RNA sequencing of Campylobacter pyloridis revealed that it was not a Campylobacter species, but rather a new bacterial species and subsequently renamed to Helicobacter pylori. In 1983, the hypothesis that H. pylori could be the causative agent of peptic ulcer was not well received in the medical community despite a clinical trial demonstrating gastritis could be healed after bacterial eradication. One of the main obstacles in proving that H. pylori was the etiological agent of gastritis was the lack of a proper animal model to prove Koch’s postulates. Therefore, after multiple failures in trying to infect rats, mice, and pigs, Barry Marshall chose to drink a pure culture of the bacterium and track his disease progression. Within 5 days he developed acute gastritis associated with the presence of many H. pylori until day 14 when his symptoms and the Helicobacter spontaneously subsided. By 1988, two independent follow-up studies published outside

60,000- 100,000 Years Ago Earliest hypothesized H. pylori colonization in humans 20,000-50,000 Years Ago Human migration out of Africa

Pre-History

5,300 Years Ago H. pylori found in the mummified stomach of Oetzi the Iceman dating back 5,300 years

1875

1875 German physicians observe curved bacteria in human gastric specimens

1885

1895

1905

1915

1899 Polish physician Walery Jaworski implicates Vibrio rugula as causative agent of gastric disease

1939-1940 “Spirocheates” observed in gastric specimens with and without gastric disease

1925

1935

1945

1955

1954 Palmer et al. conclude no bacteria are present in the stomach after analyzing over 1,000 gastric biopsies

1975 Steer and Colin-Jones et al. observe Gram negative bacili in 80% of ulcer patients but conclude Pseudomonas 1994 contamination IARC declares H. pylori as a Class I carcinogen

1965

1975

1985

1995

1983-1987 Marshall and Warren “rediscover” H. pylori, prove it’s the etiological cause of peptic ulcer and that antibiotics can be used to resolve symptoms by eradicating the infection

2005

2005 Marshall and Warren receive Nobel Prize in Medicine for their discovery

Fig. 1 Timeline of major events in the history of H. pylori from its coevolution with humans in prehistory to the revelation of its contribution to peptic ulcer disease and gastric adenocarcinoma.

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65172-3

187

188

Helicobacter Pylori-Mediated Carcinogenesis

the United States had shown that patients with duodenal ulcer and concurrent H. pylori infection could be cured with a regimen of antibiotics combined with bismuth. By 1991 the first study demonstrating the cure of duodenal ulcer was published in the United States. These clinical trials, combined with Marshall’s proof of Koch’s postulates, proved H. pylori as a causative infectious agent leading to gastroduodenal disease. Since Marshall and Warren’s discovery, H. pylori has been associated with a range of gastroduodenal disorders including gastric and duodenal ulcers, gastric adenocarcinoma, mucosa-associated lymphoid tissue (MALT) lymphoma, as well as non-Hodgkin lymphoma of the stomach. In 1994, the International Agency for Research on Cancer officially classified H. pylori as a class I carcinogen. Additionally, as a result of their work and subsequent scientific advancement of the field of gastroenterology, Marshall and Warren were awarded the Nobel Prize in Physiology or Medicine in 2005.

Bacteriology of H. pylori Helicobacter pylori is the archetypal species of its genus. H. pylori is a Gram-negative, curved-rod or spiral-shaped microaerophilic organism that resides in the mucus gel layer above the gastric epithelium (Fig. 2). The genome size is approximately 1.7 Mbp with a G þ C content of 35%–40%. Every strain of H. pylori is genetically distinct, possibly as an adaptation to the unique gastric conditions and immune response of each human host. In vitro, H. pylori colonies are translucent and uniformly sized ( 1 mm). Each bacterium measures approximately 2–4 mm in length 0.5–1 mm in width, however in response to environmental stress, H. pylori may temporarily assume a smaller, coccoid shape until ideal growth conditions are restored. H. pylori harbor two to seven unipolar, sheathed flagella that measure approximately 3 mm in length and typically carry a distinctive bulb at the terminus. H. pylori strains are characteristically catalase, oxidase, and urease positive organisms. Urease productivity is critical for H. pylori survival, representing nearly 15% of the total expressed proteome. Urease is of particular importance because it is the foundation of both noninvasive and invasive clinical diagnostics for active H. pylori infection.

Virulence Factors In order to sustain a persistent infection, H. pylori must liberate nutrients from the host while also evading the host immune response. To accomplish this task, H. pylori maintains an armamentarium of virulence factors which serve to both strategically

Gastric Fluid

Acidic (50

Fig. 6 Potential outcomes of H. pylori infection. H. pylori infection can persist for the lifetime of the host, causing alterations in the gastric environment and histology. Approximately 85%–90% of infected individuals will remain asymptomatic, only developing nonatrophic gastritis. However, a minority of infected patients can progress over the course of decades to either antral- or corpus-predominant gastritis, which can lead to peptic ulcer disease or gastric cancer. *Percent representative of patients progressing to peptic ulcer disease, including both gastric and duodenal ulcer.

While the majority of infected individuals remain asymptomatic with only chronic gastritis, some may progress further to a permanent loss of epithelial glands termed atrophic gastritis, a recognized risk factor for gastric cancer. Another determinant of disease outcome involves the pattern of gastritis within the stomach, although the drivers of each particular pattern remain unknown. Antral-predominant gastritis is associated with duodenal ulceration in contrast to corpus-predominant or pan-gastritis which are both associated with gastric ulcer and gastric adenocarcinoma (Fig. 6).

Peptic Ulcer Disease Gastric and duodenal ulcers are mucosal defects that penetrate the muscularis mucosa and typically appear in areas where mucosal inflammation is most severe. Peptic ulcer disease is a chronic, relapsing disease that causes significant morbidity and mortality due to pain, bleeding and perforation of the gastric mucosa. Nearly 70% of all gastric ulcers and 95% of duodenal ulcers are attributable to H. pylori infection. However, eradication of H. pylori allows most peptic ulcers to heal and prevents further relapse. Mechanistically, gastric ulcers originate from prolonged, intimate contact between H. pylori and the gastric epithelium. This interaction leads to continuous inflammation in the gastric antrum, thereby perpetuating mucosal breakdown, erosive gastritis, and eventually gastric ulceration. Conversely, the mechanism by which H. pylori promotes duodenal ulceration is not completely understood. The leading hypothesis is that H. pylori-induced high levels of gastric acid secretion in the antrum can eventually lead to replacement of intestinal-type tissue in the duodenum with gastric tissue. Subsequently, H. pylori can then colonize the areas of gastric metaplasia and together with elevated acid levels, ultimately promote ulceration in this locale.

MALT Lymphoma Normal gastric mucosa does not typically contain lymphoid tissue, however colonization with H. pylori recruits mucosa associated lymphoid tissue (MALT). In very rare instances, the chronic expression of cytokines released by T-cells involved in combating H. pylori infection can ultimately support the uncontrolled growth and proliferation of a monoclonal population of B lymphocytes leading to MALT lymphoma. Almost all MALT lymphoma patients are H. pylori-positive, however this disease manifests in less than 1% of H. pylori infected individuals. If found early, H. pylori eradication can lead to complete remission in more than 80% of patients.

Gastric Adenocarcinoma Infection with H. pylori is the single greatest risk factor for the development of noncardia gastric adenocarcinoma, with nearly 89% of cases attributable to infection. Due to its association with gastric carcinogenesis, the World Health Organization has classified H. pylori as a type I carcinogen. Among infected individuals, the risk of developing gastric cancer is between 1% and 2%. The

194

Helicobacter Pylori-Mediated Carcinogenesis

Normal Gastric Mucosa

Fig. 7

Non-Atrophic Gastritis

Multifocal Atrophic Gastritis

Intestinal Metaplasia

Dysplasia

Gastric Adenocarcinoma

Schematic of the Correa Cascade representing the histological progression from normal gastric mucosa to gastric adenocarcinoma.

progression from superficial gastritis to adenocarcinoma is a lengthy process taking 15–20 years to move through a series of progressive histological steps termed the Correa Cascade (Fig. 7). H. pylori eradication can reduce the risk for gastric cancer if therapy occurs prior to the onset of premalignant histological changes. The chronic inflammatory process resultant from infection is the primary mechanism of H. pylori-induced carcinogenesis. H. pylori can decrease apoptosis and promote cell survival, which, in a highly proinflammatory environment rich in free radicals, can subsequently increase the likelihood of DNA damage and DNA mutations within host cells. In addition, the TH17 immune response elicited by H. pylori can enhance carcinogenesis because the secreted cytokine profile favors angiogenesis and tumor invasiveness. H. pylori also induces a robust TH1 response which does have antitumorigenic properties, however the tolerogenic TReg response also elicited by H. pylori to dampen the proinflammatory response can simultaneously inhibit antitumor activity.

Extragastric Diseases Helicobacter pylori has been implicated in a variety of extragastric diseases due to its coevolutionary history with humans, its immunomodulatory properties, and its antigenic molecular mimicry. As of 2017, studies have been conducted investigating the role of H. pylori infection in diseases ranging from neurological disorders (including Alzheimer’s and Parkinson’s diseases) cardiovascular diseases, skin diseases, metabolic disorders (including type II diabetes, obesity and nonalcoholic fatty liver disease), and cancers occurring outside of the stomach (including colorectal and pancreatic), as well as a potential protective effect of H. pylori infection against allergy and autoimmune diseases (including inflammatory bowel disease and asthma) and esophageal diseases (including gastroesophageal reflux disease and esophageal adenocarcinoma). The epidemiological evidence linking the role of H. pylori to these diseases is not yet definitive and requires significant further investigation.

Treatment and Prevention Strategies H. pylori Detection Strategies The diagnosis of H. pylori was pioneered by Marshall and Warren in the 1980s, including serology and urease testing. Since then further tests have been developed and refined, each with specific advantages and disadvantages. These tests can be categorized into two groups: invasive tests based upon gastric biopsies collected during endoscopy to be used for histological analysis, bacterial culture, or molecular/biochemical assays; and noninvasive tests based on peripheral samples such as blood, stool, or saliva (Table 1). Typically in clinical practice one type of test is sufficient for diagnosis, however in laboratory research-based settings a combination of at least two tests are used to determine the status of H. pylori infection.

Noninvasive tests Noninvasive tests for H. pylori include serology, urea breath test, and stool antigen testing. Serological methods are preferred for large epidemiological studies and can be used clinically for primary diagnosis of H. pylori infection due to its high sensitivity and specificity. However, serological methods are not useful in assessing successful eradication as H. pylori antibody titers fall gradually and can remain long after the infection is cleared. In contrast, the urea breath test (UBT) can be used to detect primary infection and follow-up testing for successful eradication. The UBT involves ingestion of carbon labeled (13C or radio-labeled 14C) urea, and if H. pylori are present, the urea will be hydrolyzed yielding labeled carbon dioxide which can be detected in breath samples. The sensitivity and specificity range from 88% to 95% and 95% to 100%, respectively. False negatives can arise in patients who have recently taken proton pump inhibitors (PPIs), bismuth, or antibiotics or patients with active ulcer bleeding. Stool antigen testing (SAT) is based upon an enzymatic monoclonal immunoassay that detects H. pylori antigens in feces. The detection of H. pylori antigens is representative of an active infection and can thus be used as tool for both primary diagnosis as well as confirmation of successful eradication. Similar to the UBT, the sensitivity and specificity are high (94% and 97%, respectively) and the results can be complicated by use of PPIs, bismuth, antibiotics, and peptic ulcer bleeds. In regions of low to intermediate H. pylori prevalence, the SAT is the most cost-effective diagnostic tool compared to serological assays and the UBT.

Invasive (endoscopy-based) tests Endoscopy is not required solely for the purpose of H. pylori diagnosis, however if a patient has undergone an upper endoscopy and gastric biopsies retrieved, the biopsies can be used to assess H. pylori infection. Clinical diagnostic tests for H. pylori infection based on biopsies include urease testing, histology, or bacterial culture. Similar to the UBT, biopsy samples can be placed in medium

Helicobacter Pylori-Mediated Carcinogenesis Table 1

195

Summary of diagnostic methods for the detection of Helicobacter pylori infection

Diagnostic method

Specimen type

Application

Remarks

Mucosal biopsy (at least 2)

Primary diagnosis

Culture

Mucosal biopsy

Primary diagnosis

Rapid urease test

Mucosal biopsy

Primary diagnosis

Noninvasive methods Urea breath test (UBT)

Pros: Also provides patient histology data (i.e., Inflammation, atrophy) Cons: Requires high expertise, multiple biopsies, high intra-observer variability Pros: Can be used to quantify antibiotic resistance, high specificity Cons: Low sensitivity and slow detection time Pros: Rapid results (1 h), least expensive endoscopy-based method, kits commercially available Cons: Requires additional tests to confirm H. pylori

Breath sample

Primary and follow-up diagnoses

Stool antigen test (SAT)

Stool

Primary and follow-up diagnoses

Serology

Serum

Primary diagnosis or indication of ever infection

Gastric juice, stool sample, mucosal biopsy

Research only

Invasive methods Histology

Research-based methods PCR

Pros: “Gold standard” of noninvasive tests. Cons: False negatives frequently occur in patients who have recently taken antibiotics or PPIs, or those who have an ulcer bleed. Pros: Most cost-effective option in populations with low to intermediate H. pylori prevalence Cons: False negatives frequently occur in patients who have recently taken antibiotics or PPIs, or those who have an ulcer bleed. Pros: Most cost effective for large population (epidemiological) studies Cons: Cannot discriminate between active and prior infection Pros: Highly specific, diversity of samples available for analysis, yields high amounts of microbe-specific data (i.e., Virulence factors, antibiotic resistance) Cons: Cost prohibitive for clinical use

containing urea and a pH reagent. The enzymatic breakdown of urea by H. pylori urease yields ammonia and bicarbonate resulting in a change in pH that is visualized by a color change of the pH reagent. Commercially available test kits yield a sensitivity and specificity of 90% and 95% respectively with results in about an hour. This test is the least expensive compared to other invasive tests and is thus the preferred method when endoscopy is used. Alternatively, biopsies can be assessed by histological examination for H. pylori with a hematoxylin and eosin stain (H&E) however the implementation of Giemsa, Steiner, or Warthin-Starry stains or through specific immunohistochemical means can make detection easier. Multiple biopsies from both the antrum and corpus should be examined as the density of H. pylori can be highly variable at different sites. Lastly, biopsies may be used for direct bacterial culture to diagnose H. pylori infection. Although this modality is highly specific and allows for simultaneous antibiotic resistance measurements, it is the least common diagnostic test because of its low sensitivity. The low sensitivity is attributable to the highly fastidious nature of H. pylori, which makes it difficult to recover and grow from biopsy specimens. Other uncommonly used diagnostic tests include PCR-based strategies on biopsy tissue or stool samples, however those are typically limited to research-only settings due to the high cost of the assay.

Treatment Strategies Current guidelines in the United States state that when a physician chooses to test a patient for H. pylori, they should be unequivocally committed to eradication therapy if the patient tests positive. Although H. pylori is susceptible to many antibiotics in vitro, the gastric niche protects the bacteria from efficacious single-treatment therapies due to the low pH and the high viscosity of the mucus layer. In general, treatments for H. pylori involve an acid reducer such as a proton pump inhibitor (PPI) to reduce acid output with the addition of two antibiotics. First line treatment options for H. pylori include (1) clarithromycin triple therapy, (2) levofloxacin triple therapy, (3) Bismuth quadruple therapy (4) concomitant therapy (5) sequential therapy, (6) fluoroquinolone sequential therapy, or (7) hybrid therapy (Table 2). The most important predictors of successful H. pylori eradication are the choice of treatment regimen, patient adherence, and the sensitivity of the H. pylori strain to the combination of antibiotics. Unfortunately, the frequency of treatment failures is growing due to increasing antimicrobial resistance to clarithromycin and metronidazole. Four to six weeks after treatment, patients should be retested for H. pylori via UBT, SAT, or endoscopy-based methods. If the H. pylori infection is not successfully eradicated, second line therapies should be selected based upon which antibiotics were used in firstline therapies, as well as the local antibiotic resistances in the area. Metronidazole resistance is the most common, with 75% of strains in Africa and 46% in Asia and 30% of strains in Western countries demonstrating resistance. Recently, a new class of acid reducers termed potassium competitive acid blockers (P-CAB) have emerged as a more potent alternative to PPIs and may benefit

196

Helicobacter Pylori-Mediated Carcinogenesis Table 2

Summary of Helicobacter pylori treatment regimens

H. pylori treatment regimen

Drug(s)

Duration

Triple therapy

(1) Either clarithromycin or levofloxacin (2) Amoxicillin (or other metronidazole) (3) PPI (1) Bismuth (2) Tetracycline (3) A Nitroimidazole (4) PPI (1) Clarithromycin (2) Amoxicillin (3) A Nitroimidazole (4) PPI (1) Amoxicillin (2) PPI Followed by: (1) Amoxicillin (or fluoroquinolone) (2) Clarithromycin (3) A Nitromidazole (4) PPI (1) Amoxicillin (2) PPI Followed by: (1) Amoxicillin (2) Clarithromycin (3) A Nitromidazole (4) PPI

10–14 days

Bismuth quadruple therapy

Concomitant therapy

Sequential therapy

Hybrid therapy

10–14 days

10–14 days

5–7 days 5–7 days

7 days 5–7 days

H. pylori eradication strategies. To date, P-CAB has been tested in triple therapy regimens and has shown to increase H. pylori eradication rates compared to PPIs, however its efficacy in quadruple, sequential or bismuth-containing therapies have not yet been studied.

Prospects of a Vaccine As the complexity of H. pylori treatment increases and antimicrobial resistance continues to grow, a therapeutic vaccine could be a possible long-term solution to manage an H. pylori infection. Since H. pylori infection is acquired in early childhood and the possibility exists that infection may actually be beneficial in childhood, a therapeutic vaccine to eradicate the bacteria is the preferred strategy as opposed to prophylaxis. Formulations for vaccines have been attempted using whole cell lysates, outer membrane vesicles, as well as single or multiple purified antigens such as CagA, VacA, NapA, urease, and catalase. Some studies conducted in animals have shown that H. pylori vaccines can cure infection, while other studies have shown that H. pylori vaccination can augment the efficacy of antibiotic therapy and prevent recurrence. To date, progress on a human vaccine has been slow. The most recent study involved a phase 3 trial in China which demonstrated an approximately 70% protective effect of an orally administered H. pylori vaccine (based upon a recombinant urease B protein) in children aged 6–15 years, however the efficacy began to deteriorate after one year.

Population Screening Interventional strategies such as eradication therapy, in conjunction with improved living conditions and hygiene across much of the world, have led to a significant decline in H. pylori prevalence over the recent decades. However, despite the decreasing prevalence of H. pylori, gastric cancer incidence is projected to rise as a result of changing population demographics. As such, there remains a need to reduce the number of future gastric cancer cases through eradication of H. pylori. Universal test and treat strategies among healthy, asymptomatic individuals within high gastric cancer-risk populations could reduce the incidence of gastric cancer in a costeffective way. Furthermore, the cost-benefit of H. pylori eradication is increased when considering the benefit of reducing the incidence of other diseases such as peptic ulcer and dyspepsia. However, arguments against universal test and treat strategies include: (1) findings that H. pylori infection has been associated with other gastrointestinal and nongastrointestinal diseases where it may serve a beneficial role to the host; and (2) the impact of large-scale eradication therapy on antimicrobial stewardship remains unknown. Nonetheless, the growing consensus in the field suggests that population screening for H. pylori is a worthwhile endeavor to prevent premalignant and malignant gastric lesions.

Helicobacter Pylori-Mediated Carcinogenesis

197

Prospective Vision H. pylori is the most successful bacterial pathogen worldwide, coevolving with its human host for millennia and infecting over 50% of the global population. Yet, it was not identified as a stomach pathogen until the late 20th century by the Nobel Prize-winning physicians Barry Marshall and Robin Warren. Their landmark discovery ultimately revolutionized the field of gastroenterology, providing an etiological cause for both peptic ulcer disease and gastric cancer. Laboratory investigations of this Gram-negative, spiral-shaped, microaerophilic organism have yielded the identification of numerous and highly specific virulence factors that enable its initiation and colonization in the mucus gel layer between the gastric lumen and the epithelial cell surface. Research has also illuminated many of the strategies employed by H. pylori to both persist in the highly inhospitable environment of the human stomach and to evade innate and adaptive immune responses. Simultaneously, in concert with improved hygiene and sanitary conditions around the world, clinical investigations have borne effective detection and treatment strategies that have contributed to the declining prevalence of this bacterium over the past several decades. Despite these advancements, H. pylori prevalence remains very high in developing nations and, as the global population continues to age, it is expected that we may still see an increase in lives lost from H. pylori-induced gastric diseases in the years ahead. As the diseases that result from H. pylori infection include not only those of the stomach but also potentially numerous extragastric diseases, noninvasive testing followed by eradication therapy for those in high-risk populations could significantly reduce the morbidity and mortality from this common bacterial pathogen.

See also: Cancer Disparities. Gastric Cancer: Pathology and Genetics.

Further Reading Backert, S., Yamaoka, Y. (Eds.), 2016. Helicobacter pylori research: From bench to bedside. Springer, Tokyo. Marshall, B.J. (Ed.), 2002. Helicobacter pioneers: Firsthand accounts from the scientists who discovered Helicobacters, 1892–1982. Blackwell, Victoria, Australia/Malden, MA. Tegtmeyer, N., Backert, S. (Eds.), 2017. Molecular pathogenesis and signal transduction by Helicobacter pylori. Springer, Cham.

Hepatocellular Carcinoma: Pathology and Genetics Luigi M Terracciano, Salvatore Piscuoglio, and Charlotte KY Ng, University Hospital Basel, Basel, Switzerland © 2019 Elsevier Inc. All rights reserved.

Glossary Cell cycle checkpoints Regulatory mechanisms at various phases of the cell cycle to ensure the correct completion of cellular processes crucial to the survival of a cell. Checkpoints are important for faithful DNA replication and the prevention of uncontrolled cell division. Chromosomal instability A higher than normal rate of missegregation of chromosomes or parts of chromosomes during mitosis due to defective cell cycle quality control mechanisms. Dysplastic nodule A focal nodular region (
1 mm) showing nuclear atypia with increased fat or glycogen in a cluster of dysplastic cells. Dysplastic nodules are often associated with cirrhosis but without definite evidence of malignancy. Edmondson-Steiner grading A 4-point scale describing the extent of differentiation of hepatocellular carcinoma cells and their resemblance of normal hepatocytes. It is the most widely used system for histologic grading of hepatocellular carcinoma. Telomerase reverse-transcriptase A catalytic subunit of the enzyme telomerase, which maintains telomere ends by the addition of the telomere repeat TTAGGG. Reactivation of telomerase activity allows cells to overcome replicative senescence and to escape apoptosis. WNT signaling pathways A series of signal transduction pathways responsible for passing signals into a cell through cell surface receptors. The three WNT signaling pathways described are the canonical WNT pathway, the non-canonical planar cell polarity pathway, and the non-canonical Wnt/calcium pathway.

Introduction Liver cancer was responsible for an estimated 782,000 new cancer cases and nearly 746,000 deaths in 2012. Unlike most other malignancies, mortality from liver cancer has increased significantly over the past 20 years and the medical and economic burden of liver cancer will likely increase significantly in Western populations over the next decades. Of primary liver cancer cases, hepatocellular carcinoma (HCC) accounts for > 90%. HCC is the sixth most common cause of cancer and third most common cause of death worldwide, accounting for nearly 10% of cancer deaths worldwide. Currently, curative treatment options for early-stage HCC detected by screening (one lesion < 5 cm or up to three lesions < 3 cm) involve resection and/or liver transplantation, with a > 50% 5-year survival. Prognosis for patients with late-stage, inoperable HCC is poor with > 90% mortality, particularly for cases associated with high serum a-fetoprotein (AFP) levels or TP53 mutation. Until 2016, sorafenib, a multi-kinase inhibitor, remained the only approved systemic agent for HCC but only prolongs patient survival by 2.8 months. In 2017, regorafenib, a multi-kinase inhibitor, and nivolumab, a PD-1 inhibitor, were approved as second-line treatments following sorafenib, bringing the total number of approved systemic agents to three, far fewer than most other cancer types. HCC shows great variation in incidence according to geographical regions. Around 80%–85% of HCCs are related to hepatitis B (HBV) and/or hepatitis C virus (HCV) chronic infection. In sub-Saharan Africa and South-East Asia, HBV chronic infection is endemic and accounts for the majority of HCCs diagnosed. In Western populations, however, the rising incidence of HCC is attributed to the increasing prevalence of chronic liver diseases associated with HCV infection, alcohol consumption, obesity and nonalcoholic fatty liver disease (NAFLD). Patients with hereditary diseases such as hemochromatosis or a-1-antitrypsin deficiency are also at increased risk of cirrhosis and fibrosis and, thus, of HCC. The male gender, exposure to aflatoxin B1 through dietary consumption and smoking are also known risk factors. In about 80% of cases, HCCs arise in cirrhotic livers or in livers with advanced fibrosis. In fact, regardless of etiologies, liver cirrhosis is considered the major clinical risk factor. The 5-year cumulative risk for HCC in patients with cirrhosis ranges between 5% and 30%, depending on the cause (with the highest risk among those infected with HCV), region or ethnic group (17% in the United States and 30% in Japan) and stage of cirrhosis (with the highest risk among patients with decompensated disease). Although liver cirrhosis itself is not considered pre-malignant per se, there is increasing evidence to support a multistep model of hepatocarcinogenesis. Following hepatic injury resulting from one or more of HBV, HCV, excessive alcohol, exposure to aflatoxin B1 and/or other agents or conditions, necrosis develops followed by hepatocyte proliferation. Repeated cycles of this destructiveregenerative process lead to chronic liver condition that culminates in liver cirrhosis. Subsequently, hyperplastic nodules arise, followed by dysplastic nodules, resulting in the development of HCC. HCCs that occur in the absence of cirrhosis may present with unique clinicopathologic features, such as the fibrolamellar subtype that is predominantly diagnosed in young female patients. Traditional pathology analyses have revealed the extraordinary histologic and phenotypic heterogeneity of HCCs. In fact, our understanding of HCC pathology is still evolving, with the last update of the WHO classification including substantial updates on the classification and diagnosis of early HCCs and lesions thought to derive from progenitor cells. From a diagnostic point

198

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65261-3

Hepatocellular Carcinoma: Pathology and Genetics

199

of view, the increased detection of small and early lesions in screening programs has highlighted the difficulties in distinguishing benign, precursor, and early HCC lesions. In addition, the advances in next-generation sequencing in recent years have also revealed the startling molecular heterogeneity of HCCs, with many deregulated pathways through an accumulation of somatic genetic and epigenetic changes in the cells.

Pathology of HCC Precursor Lesions of HCC Hepatocellular adenoma Hepatocellular adenoma (HCA) is a benign neoplasm that sometimes undergoes malignant transformation into HCC. HCA is diagnosed predominantly in young women using oral contraceptives. Less commonly, it has been described in female patients with maturity onset diabetes of the young type 3, and in men using anabolic steroids, with glycogen storage disorders or receiving androgen treatment. Recently, metabolic syndrome, obesity and alcohol have also described as likely risk factors of HCA, in particular of the inflammatory subtype. Patients with HCA have normal liver function, but may occasionally have elevated serum AFP levels. Unlike other precursor lesions of HCCs, HCA is uncommonly associated with cirrhosis. HCA is typically a large (up to 30 cm in diameter) tumor that resembles normal liver tissue microscopically but may display a pseudoglandular architecture. Large cell and fatty changes are often seen, but the cells have regular nuclei without atypia and mitoses are almost never present. In 20%–30% of cases, multiple HCAs may be present. The reticulin framework is usually present, bile ducts are absent and the sinusoids are usually compressed. In general, the risk of transformation to HCC has been reported to be between 4% and 8%. Increased risk has been reported in male patients, patients presenting with large HCA (but not the number of HCAs), with pseudogland formation. Molecular classification has divided HCA into several subgroups linked with distinct risk factors, clinical behaviors, histologic as well as imaging features: HNF1A-inactivated HCA, inflammatory HCA, CTNNB1 exon 3-mutated HCA, CTNNB1 exons 7 or 8mutated HCA, sonic hedgehog HCA and unclassified HCA. CTNNB1 exon 3-mutated HCA and sonic hedgehog HCA have been linked to high risk of malignant transformation and bleeding, respectively. Liver fatty acid-binding protein, serum amyloid A, Creactive protein, prostaglandin D2 synthetase, glutamine synthetase, and b-catenin can be used either on biopsy or surgical specimen to classify HCA into these subclasses. Transformed HCA typically appears as well differentiated HCC within the HCA with vascular invasion.

Dysplastic foci Dysplastic foci are microscopic (< 1 mm in diameter), uniform non-invasive lesions that differ from surrounding liver tissue based on morphology, cytoplasmic staining, nuclear size and cellular atypia. Dysplastic nodules are usually incidental findings and are usually found within cirrhotic nodules. Dysplastic hepatocytes within these foci may display large cell change, small cell change or iron-free foci. Large cell change refers to hepatocytes with nuclear and cytoplasmic enlargement (thus retaining the nuclear/cytoplasmic ratio), associated with nuclear pleomorphism, multinucleation, and hyperchromasia. Large cell change may encompass dysplasia as well as reactive changes, and its precise role as an HCC precursor lesion is unclear. Studies reporting low cell proliferation, increased apoptosis and association with cholestasis support the notion that large cell changes are reactive changes. On the other hand, other studies have reported increased proliferation, telomere shortening and abnormal DNA content, including chromosomal copy number aberrations, leading to the hypothesis that large cell changes are indeed precursor lesions, especially in cirrhosis associated with HBV or HCV. By contrast, small cell change refers to hepatocytes with decreased cell volume, mild nuclear pleomorphism, increased nuclear/ cytoplasmic ratio, basophilic cytoplasm and hyperchromasia. Hepatocytes with small cell change typically show increased cell proliferation, inactivation of cell cycle checkpoint and telomere shortening. Morphologically, small cell change resembles well differentiated HCC. Importantly, while expansile small cell changes foci are considered dysplastic foci, a diffuse pattern is considered regenerative changes. Compared to large cell changes, small cell changes are considered more advanced. Iron-free foci are proliferative lesions consisting of clusters of hepatocytes devoid of or low in iron content. Large or small cell change may be present. The precise clinical significance of iron-free foci is not clear, but it has been suggested that iron-free foci arising in livers with hereditary hemochromatosis are likely HCC precursor lesions.

Dysplastic nodules Dysplastic nodules are larger than dysplastic foci and are defined as those
1 mm in diameter, usually around 1 cm in diameter. Dysplastic nodules are detectable macroscopically and sometimes radiologically. Similar to dysplastic foci, dysplastic nodules are also usually found on a cirrhotic background and may occur as single or multiple nodules. Macroscopically, dysplastic nodules are distinctly nodular lesions that differ from surrounding liver tissue based on color and texture, and may bulge from the surface of the liver (Fig. 1). Dysplastic nodules are subdivided into low and high grades (LGDN and HGDN), with HGDNs associated with an increased risk of malignant transformation. LGDN shows minimal architectural atypia, mildly increased cell density, with normal to slightly elevated nuclear/cytoplasmic ratio, no mitoses and 1–2 cell wide cell plates. The borders of LGDN may be rounded

200

Hepatocellular Carcinoma: Pathology and Genetics

Fig. 1 Macroscopic and microscopic images of a dysplastic nodule. Macroscopic image of a liver dysplastic nodule taken during a routine procedure (left) and micrograph of hematoxylin and eosin stain (right).

but do not usually compress the adjacent liver tissue. Portal tracts and the reticulin network are retained. Steatosis and Mallory bodies may be present. Large cell change, but not small cell change, may also be present. However, histologic diagnosis of LGDN has not been found to be reproducible in practice, because of significant overlapping features with large regenerative nodule. By contrast, HGDN histologically and molecularly resembles HCC. In particular, HGDN displays significant architectural and cytologic atypia, including enlarged and irregular nuclei, nuclear hyperchromasia, peripheral location of the nucleus, occasional mitosis, thickening of hepatic cell plates (> 2 cells), basophilic cytoplasm, formation of pseudoglands, focal loss of the reticulin network and resistance to iron accumulation. Small cell change is frequently seen in HGDN. Some of these features may be confined to foci within the HGDN. In HGDN, portal tracts may be few and sinusoidal capillarization is increased. Neoangiogenesis may be seen in dysplastic nodules, in particular in HGDNs, including the formation of unpaired arteries with no accompanying bile ducts. Dysplastic nodules usually show an isovascular or hypovascular appearance compared to surrounding tissues (Fig. 1).

Macroscopic Features of HCC HCCs are highly heterogeneous macroscopically. HCC typically forms soft nodular masses that vary in color from gray to light brown to yellowish-green and are often punctuated by foci of hemorrhage and necrosis. HCC associated with cirrhosis typically presents as fibrous, encapsulated nodules while HCC not associated with cirrhosis is usually unencapsulated. HCC may also present as a single or as multiple nodules, with sizes from < 1 to > 20 cm. Indeed, compared to other tumor types, one of the distinguishing features of HCC is the frequent occurrence of multiple nodules. The multiple nodules of HCC may represent intrahepatic metastasis, in which malignant cells are disseminated from a single primary tumor to form additional tumor nodules, or may represent independent tumors. The distinction between these two scenarios is not trivial but has important clinical implications, given that intrahepatic metastases are likely to be more poorly differentiated and aggressive than independent, single nodules that do not disseminate. Extrahepatic metastases are most frequently identified in the lungs, the lymph nodes, bone and the adrenal glands. The WHO 2010 guidelines divide HCCs into early HCCs and progressed HCCs. Early HCC refers to single, well differentiated, vaguely nodular tumor with poorly defined margins and of < 2 cm in diameter, whereas progressed HCC refers to a single tumor > 2 cm in diameter, or small (< 2 cm), moderately differentiated HCC of distinctly nodular type or multiple tumors. Both vaguely and distinctly nodular HCCs usually occur in a cirrhotic liver, but vaguely nodular HCCs do not have well defined margins, whereas distinctly nodular HCCs are encapsulated. Compared to the distinctly nodular type, vaguely nodular type HCCs are usually smaller in size, represent a biologically earlier stage with more favorable prognosis and are rarely associated with intrahepatic metastasis. Macroscopically, progressed HCC may be nodular, massive or diffuse. Nodular HCC may involve one or more nodules. Single nodular HCC is usually encapsulated and may show extracapsular growth around the nodule. Multinodular (or expansive) HCC is defined by the presence of multiple nodules throughout the cirrhotic tissues, usually with one dominant nodule, with satellite nodules in the proximity. Multinodular HCC is the most common type and are typically seen in association with cirrhosis. Massive (or infiltrative) HCC is a large dominant, poorly circumscribed tumor with ill-defined invasive borders and may be associated with smaller satellite nodules. Massive HCCs tend to be larger, are more frequently seen in non-cirrhotic livers and are associated with a poor prognosis. HCC can also be mixed nodular and massive. The less frequent diffuse (or cirrhotomimetic) HCC consists of many small nodules in a growth pattern that resembles regenerative nodules. Diffuse HCC can involve either a liver lobe or the entire liver, may be associated with macroscopic portal vein and/or hepatic vein invasion and is associated with a poor prognosis. On rare occasions, HCCs may display a pedunculated or protruded growth pattern, referring to extrahepatic growth with and without a peduncle, respectively. Additional macroscopic features of HCC include the presence of necrotic areas, involvement of portal and hepatic veins and bile or fat production.

Hepatocellular Carcinoma: Pathology and Genetics

201

Microscopic Features of HCC The most distinctive features of HCC are that HCC cells resemble hepatocytes and attempt to mimic the growth pattern of normal liver cells. Early HCC is well differentiated and consists of tumor cells with increased nuclear/cytoplasmic ratio. Early HCC typically displays trabecular or pseudoglandular, or admixed, growth pattern. Relative to surrounding liver tissues, increased cell density is usually observed. Early HCC tends to be isovascular or hypovascular with varying numbers of portal tracts, and does not show vascular invasion. The reticulin framework is reduced but not lost. The most common presentation of progressed HCC includes the following features: one or more well vascularized tumors with vascular invasion, prominent trabecular and pseudoglandular growth pattern with small cell change, cytologic atypia with increased mitotic activity, a lack of fibrous connective tissue between tumor cords and cells, fatty change, loss of the reticulin network, and an absence of Kupffer cells. Bile canaliculi are present and bile production is frequently observed. In progressed HCC, complete neovascularization with vascular infiltration is not uncommonly seen. However, there is substantial heterogeneity between and within HCCs in terms of their architecture, cytologic features and histologic grades. HCC may display multiple architectural patterns: (1) trabecular (or sinusoidal) pattern where tumor cells grow in cords of variable thickness separated by prominent sinusoids lined by flat endothelial cells, with the trabeculae becoming thicker and contorted with dedifferentiation; (2) pseudoglandular (or acinar) pattern with a variety of abnormal, dilated bile canaliculi forming between tumor cells, often admixed with the trabecular pattern; (3) compact (or solid pattern) where tumor displays a trabecular phenotype but the tumor cells apparently grow in solid masses compressing the sinusoids and rendering the thick trabeculae inconspicuous by compression. Progressed HCC, in particular, sometimes display all three architectural patterns. Trabecular and pseudoglandular patterns are more frequently seen in well to moderately differentiated HCCs, whereas compact is more common in poorly differentiated HCCs (Fig. 2). HCC may also display the following cytologic variants: (1) pleomorphic cells that vary in size, shape, nuclei and/or staining patterns and are often associated with the presence of multinucleated or mononuclear giant cells; (2) clear cells resulting from the accumulation of glycogen; (3) spindle cells with sarcomatoid differentiation; (4) fatty change, most frequently observed in early HCCs; (5) bile production seen as plugs in pseudoglands; (6) hyaline (Mallory-Denk) bodies with aggregation of intermediate filaments within hepatocytes; (7) pale bodies which are round to ovoid cells with cystically dilated endoplasmic reticulum containing amorphous and lightly eosinophilic accumulations; (8) ground glass inclusions which are hepatocytes with granular, eosinophilic cytoplasm mainly resulting from HBsAg positivity.

Histologic grade HCC cells show variable degree of resemblance to normal hepatocytes, depending on the extent of differentiation. The most widely used system for histologic grading is the Edmondson-Steiner grading. Grade I tumors display abundant cytoplasm and minimal nuclear atypia that resembles normal liver. Grade II tumors show mild nuclear atypia with prominent nucleoli and hyperchromasia. Grade III tumors show moderate nuclear atypia with greater hyperchromasia, nuclear irregularity and granular cytoplasm. Grade IV tumors show marked nuclear pleomorphism, marked hyperchromasia, loss of trabecular pattern and are often associated with anaplastic giant cells. Most HCCs are of grades II or III. It should be noted that HCCs often vary histologically even within a single nodule. Most primary cancer nodules < 1 cm are uniformly well differentiated (grades I/II). On the other hand, around 40% of nodules between 1 and 3 cm in diameter contain tumor cells of different histologic grades, with the less differentiated tumor cells concentrated in the central regions of the tumors. Although histologic grading in liver biopsies may be subjected to tumor heterogeneity, it has been shown that tumor grade in preoperative biopsy is correlated with the final grade on resection and predicts survival of the patients. Histologic grade in preoperative biopsy tends to be lower than the grade on resection. However, compared to several other clinicopathologic factors such as tumor size, tumor stage, vascular invasion and liver function, histologic grade is a relatively weak independent prognostic indicator.

Fig. 2 Architectural patterns of HCC. Representative micrographs of the most common architectural patterns of HCC: (A) trabecular pattern (scale bar 50 mm), (B) solid pattern (scale bar 25 mm) and (C) pseudoglandular pattern (scale bar 50 mm).

202

Hepatocellular Carcinoma: Pathology and Genetics

Histologic variants of HCC Fibrolamellar HCC is a rare variant of HCC, accounting for < 1% of liver tumors. The majority of fibrolamellar HCC ( 85%) is diagnosed in children and young adults < 35 years old. Fibrolamellar HCC is characterized by well differentiated and eosinophilic tumor cells with prominent nucleoli growing between laminated collagen fibrous layers that show a lamellar pattern. This type of HCC is typically large and well circumscribed, grows with pushing borders and is often hyalinized. In some cases, fibrolamellar HCC resembles nodular hyperplasia, with a central scarred zone that is often calcified. Cytoplasmic inclusions are common, and may include Mallory bodies, pale bodies and/or cytoplasmic globules of variable positivity to PAS and immunoreactivity to anti-fibrinogen. In contrast to classical HCC, fibrolamellar HCC is reported to show abundant cytokeratin (CK) 7 and focal CK19 expression. Among the histologic variants of HCC, fibrolamellar HCC is unique as it is usually diagnosed in the absence of cirrhosis. Despite the usually large tumor size at diagnosis, fibrolamellar HCC has better prognosis than classical HCC, in part due to the fact that it is usually diagnosed in young patients not associated with primary risk factors and that the tumor can usually be resected surgically. Metastases are sometimes seen and most frequently seen in the peritoneum, the lungs and abdominal lymph nodes. Sarcomatoid HCC is seen in  2% in HCCs, alone or within classical HCC. Sarcomatoid HCC is fully or partially composed of spindle-shaped cells and shows bizarre anaplastic features that resemble those seen in leiomyosarcoma and fibrosarcoma. Giant cells may also be present. Patients with sarcomatoid HCC have been reported to show low serum AFP levels and frequently present with distant metastases. Sarcomatoid HCC has been associated with repeated chemotherapy and/or transarterial chemoembolization. Other clinical features are similar to classical HCC. Scirrhous HCC accounts for < 5% of HCCs and is characterized by the presence of diffuse fibrotic changes, alone or in conjunction with classical HCC. The tumor is usually large and unencapsulated with serrated border. Microscopically, the fibrosis is usually seen along the sinusoid-like blood spaces separating trabecular cell plates, often of > 3 cells thick. Scirrhous HCC typically does not display prominent nucleoli and has granular eosinophilic cytoplasm. Scirrhous HCC is associated with hypercalcemia but not with bone metastasis. Scirrhous HCC is often misdiagnosed as post-therapy fibrosis, cholangiocarcinoma or metastatic tumors due to the presence of diffuse fibrosis. Scirrhous HCC differs from fibrolamellar HCC in terms of its lack of lamellar fibrosis and central scar, and the presence of cirrhosis. The clear cell variant of HCC is characterized by the presence of tumor cells with clear cytoplasm due to cytoplasmic fat or glycogen. Predominant appearance of clear cell change is seen in up to 20% of HCCs and some degree of clear cell change can be found in up to 40% of HCC. The clear cell variant of HCC may display trabecular, pseudoglandular, solid or mixed architectural pattern. The clear cell variant is reported to be more frequently diagnosed in male patients. Steatohepatic-type HCC is more frequently seen in patients suffering from non-alcoholic steatohepatitis and is characterized by the steatotic appearance in > 5% of the tumor, including the presence of Mallory bodies, fibrosis, inflammation and the ballooning of hepatocytes. If present, the fibrosis usually displays a trabecular or a pericellular pattern. Inflammatory infiltrates, if present, usually consist of neutrophils, plasma cells and lymphocytes. HCC with lymphoid stroma has been described in a few case reports and is defined by the presence of prominent inflammatory infiltrates. Lymphocytes make up the majority of the stroma, with the presence of some macrophages, plasma cells, neutrophils and/ or giant cells. The T-cells are mostly CD4þ. The existence of HCC with lymphoid stroma as a distinct entity is controversial, with some suggesting that the lymphoid stroma merely represents a regression phenomenon.

Differential Diagnosis of HCC While the diagnosis of poorly to moderately differentiated HCC is usually straightforward, differentiating HCA and well differentiated HCC, or HGDN and well differentiated HCC presents significant diagnostic difficulty. The identification of surrogate markers to distinguish early HCC from pre-malignant lesions is also a major issue in clinical management. Additionally, metastasis from other organs may resemble primary HCC. Differential diagnosis of HCC involves the use of hematoxylin & eosin (H&E) stain, reticulin stain (loss of reticulin), and immunohistochemistry (IHC).

HCA versus well-differentiated HCC Differential diagnosis of well differentiated HCC from HCA is based on a combination of increased mitotic activity, nuclei atypia, trabecular growth with thickened cell plates, vascular invasion, loss of the reticulin framework and the presence of bile ducts. HCA rarely occurs on a background of cirrhosis and are negative for glypican 3 (GPC3) and AFP. Mitoses are extremely rare in HCA. If mitoses are present, a diagnosis of HCC should be considered. CD34þ arterialized sinusoids are preferentially found in HCC.

High-grade dysplastic nodule versus HCC Differential diagnosis between HGDN and well differentiated HCC is sometimes very difficult, given that they display very similar histologic features. Microscopically, stromal invasion is an important distinguishing feature of HCC but its identification is especially challenging in small biopsies. In a cirrhotic liver, stromal invasion may be difficult to distinguish from small clusters of hepatocytes within fibrous septae. Immunostains of CK7/9 that show a ductular reaction favors pseudoinvasion and therefore a diagnosis of HGDN. Elevated nuclear/cytoplasmic ratio, even when the nuclei do not show atypia, together with trabecular or pseudoglandular growth strongly suggest a diagnosis of well differentiated HCC. The complete loss of the reticulin framework is also considered one of the strongest indicators for a diagnosis of HCC.

Hepatocellular Carcinoma: Pathology and Genetics

203

Several IHC markers are used to distinguish HGDN and well differentiated HCC in cirrhotic livers. A panel of three markers GPC3, glutamine synthetase and heat shock protein 70 (HSP70) has been employed. GPC3 is an oncofetal protein that is not expressed in human livers but frequently reactivated in HCC, with cytoplasmic/membranous staining in > 5%–10% of cells. Positivity of glutamine synthetase, a target of b-catenin signaling, increases from LGDN to HGDN to HCC. In HCC, glutamine synthetase may show diffuse cytoplasmic positivity in > 50% of cells unrelated to vessels. HSP70 is negative in non-tumor hepatocytes and its positivity correlates with the grade of the HCC, in which it shows cytoplasmic or nuclear stain in > 5%–10% of tumor cells. When two of the three markers are positive, the sensitivity and specificity for detecting well differentiated HCC in resection specimens are 72% and 100%, respectively. For the detection of small (< 2 cm) early well differentiated HCC by core biopsy using a 20-21G needle, the sensitivity and specificity for 2/3 markers are 50% and 100%, respectively. Adding clathrin heavy chain to the panel of three markers increases the sensitivity and specificity for the detection of small early HCCs by core biopsy to 67% and 100%, respectively.

HCC versus metastasis In patients without underlying chronic liver disease, the vast majority of malignant tumors in the liver is from non-liver origin, particularly in Western countries. The most common sites of origin are the breasts, the lungs, colon, and pancreas. While most metastases resemble their primary tumors histologically, certain types of tumors from non-liver origin, such as clear cell adenocarcinoma, clear cell renal cell carcinoma, melanoma, neuroendocrine tumor of the gastrointestinal tract and gastrointestinal adenocarcinoma with hepatoid features, may be difficult to differentiate from primary HCC. HCC typically lacks a desmoplastic stroma, the presence of which is typical of metastatic adenocarcinomas, although HCCs of the fibrolamellar and scirrhous subtypes may contain abundant stroma. The presence of bile production, the presence of multiple types of cytologic variants usually seen in classical HCC, and/or an absence of mucin in pseudoglands would suggest predominantly hepatic phenotype. Immunostains of arginase 1, hepatocyte paraffin-1 antigen (HepPar-1, also known as hepatocyte specific antigen), polyclonal CEA (pCEA), CD10 and various cytokeratins may help determine if the tumors show hepatocellular differentiation. Arginase-1, HepPar-1, and pCEA are positive in normal liver and almost all well differentiated HCCs. HepPar-1 shows a granular, cytoplasmic (mitochondrial) pattern of staining in  90% of HCC. HepPar-1 may also be expressed by tumors of the gastrointestinal tract, lung, pancreas and biliary tract but in these tumors, staining is usually weak or focal. pCEA shows a distinctive canalicular pattern of staining in HCC. A diffuse cytoplasmic pCEA staining pattern would favor a diagnosis of tumors other than HCC. However, sensitivity and specificity for both markers decline with the dedifferentiation of the tumor. CD10 is negative in adenocarcinomas but shows similar staining patterns to pCEA in HCCs. AFP shows a patchy and cytoplasmic staining pattern in  50% of HCC and its staining increases with dedifferentiation. Cytokeratins CK7, CK19, and CK20 are usually negative in HCCs and one or more is frequently positive in adenocarcinomas. The presence of liver-specific proteins such as albumin would also suggest a liver origin. Interestingly, CK19 positivity in HCC seems to correlate with clinicopathologic features of tumor aggressiveness, more invasive characteristics, compared with CK19-negative HCCs through the upregulation of epithelial-to-mesenchymal transition-associated genes.

Combined hepatocellular-cholangiocarcinoma Combined HCC-cholangiocarcinoma (CHC) is rare and accounts for < 1% of malignant liver tumors. However, unlike HCC that is believed to have arisen from mature hepatocytes through a multistep process of hepatocarcinogenesis, CHC is thought to derive from a progenitor cell that differentiates into both hepatocytes and cholangiocytes. As its name suggests, combined HCC–CHC contains intimate mixtures of HCC and cholangiocarcinoma cells and includes tumors of transitional type in which the tumor cells display morphologic features intermediate of HCC and cholangiocarcinoma. CHC is usually positive for CK7 and CK19 although positivity is also occasionally seen in HCC. CHC must be also distinguished from collision tumors. The WHO subclassifies CHC into the classical subtype and the subtype with stem cell features. The more common classical subtype consists of tumor cells with classical features of HCC and cholangiocarcinoma, in which the HCC component may be identified by cytoplasmic staining of HepPar-1 and canalicular staining of pCEA and CD10, and the cholangiocarcinoma component with CK7/19 and epithelial membrane antigen (EMA). Additionally, the cholangiocarcinoma component may be identified using D-PAS or mucicarmine stain to show mucin production and is usually associated with abundant desmoplastic stroma. CD133 and vimentin are usually negative. Both components can be well, moderately or poorly differentiated. Foci of intermediate HCC–cholangiocarcinoma phenotype may be present. However, we have to keep in mind that the concept of stem cell features in HCC is controversial and that the definitions in the 2010 WHO classification are not precise, leading to variable interpretations. Moreover, there are no reliable immunohistochemical markers to identify stem cells. CHC with stem cell features does not produce mucin, shows high expression of CD133, EpCAM and vimentin, but less frequent expression of HepPar-1, thus recapitulating the features of progenitor cells. The subtype with stem cell features is further subclassified into typical, intermediate and cholangiolocellular subtypes, although the subclassification remains controversial. The typical subtype consists of mature hepatocytes with high nuclear/cytoplasmic ratio and hyperchromatic nuclei on the periphery. The peripheral cells stain positive for CK7/19, CD56, EMA, c-KIT, and EpCAM. The hepatocytes in the center of the tumor may show clear cell change. The intermediate subtype displays features intermediate between hepatocytes and cholangiocytes, consisting of small cells with hyperchromatic nuclei and scant cytoplasm. The cells are arranged in trabecular, solid nests or strands. CK7/19 and EMA positivity is frequent. The cholangiolocellular subtype consists of small cells with high nuclear/cytoplasmic ratio, hyperchromasia, and oval nuclei embedded in a fibrous stroma, and grows in an antler-like intersection pattern. Cellular atypia is usually mild. CK7/9 and EMA are positive and EpCAM and CD133 expression may be increased. The cholangiolocellular subtype was

204

Hepatocellular Carcinoma: Pathology and Genetics

previously considered a subtype of cholangiocarcinoma but at the moment is considered part of the spectrum of CHC with stem cell features.

Genetic Alterations in HCC HBV Integration Recent reports attribute over 50% of HCC cases worldwide to HBV infection, which is the major cause of HCC in East Asian and subSaharan countries. Compared with uninfected individuals, HBV carriers have 10- to 25-fold greater risk to develop HCC. Integration of the HBV genome into the host genomic DNA occurs randomly in infected regenerating hepatocytes, is believed to be an early event in HBV infection and is involved in the development of HBV-related HCC. Whole genome sequencing studies of HBV-related HCCs have demonstrated that HBV genome integration occurs in approximately 80% of cases with a mean of 2.5 HBV integration sites per tumor. The HBx gene (encoding for the hepatitis B viral protein) is the HBV gene most frequently integrated into the human genome and the expression of the trans-acting wild-type and truncated HBx chimeric transcript has been shown to enhance transforming potential. The first two integration sites in the human genome were described in the early 1990s when chimera sequences formed by the fusion of the HBV genome with the RARB or CCNA genes were identified. With the advancement of sequencing technologies, various additional integration sites have been described. The most recurrent HBV integration site is in the promoter region and within the gene body of the TERT gene, observed in  20% of the HBV-associated HCC. HBV has also been reported to recurrently integrate into other cancer-related genes such as SENP5, ROCK1, MLL4, CCNE1, and SOX2. Furthermore, HBV integration sites within repetitive or non-coding sequences such as long interspersed nuclear elements (LINEs), Alu and the long terminal repeats of endogenous retroviruses have also been reported. The precise mechanism of HBV genome integration in the development and the progression of HCC is under active investigation. Molecular studies have demonstrated that although HBV integration does not promote HBV replication, its integration into coding genes or their promoter regions may alter the expression of the target genes and increase genomic instability. Interestingly, HBV integration into the LINE1 region results in the generation of a chimeric transcript HBx-LINE1 detected in 21 of 90 (23%) tumors of HBV-related HCC patients and is significantly associated with poor survival of HCC patients.

Chromosomal Instability Chromosomal instability (CIN) refers to a higher than normal rate of missegregation of chromosomes or parts of chromosomes during mitosis due to defective cell cycle quality control mechanisms, resulting in copy number alterations (CNAs) or aneuploidy. CIN is frequently a consequence of genetic or epigenetic alterations in genes involved in telomere maintenance, DNA replication, chromosome segregation or cell cycle checkpoints. In the context of HCC, the most frequent alterations that result in CIN include alterations in TERT, TP53, and CDKN2A. CNAs cause dosage effects on the genes located in the affected genomic regions. The non-random distribution of CNAs in HCC suggests these CNAs, in particular focal amplifications and homozygous deletions, may lead to the overexpression of oncogenes or the loss of tumor suppressor genes. Indeed, profiling HCC using comparative genomic hybridization or single nucleotide polymorphism arrays has revealed recurrent chromosomal gains of 1q, 5p, 6p, 7q, 8q, 17q, and 20q and losses of 1p, 4q, 6q, 8p, 9p, 13q, 14q, 16p-q, 17p, 21p-q, and 22q. Analyses of focal amplifications have demonstrated recurrent amplifications targeting genomic regions of TERT (5p15.33), VEGFA (6p21.1), MYC (8q24.21), CCND1 and FGF19 (11q13.3), and CCNE1 (19q12). On the other hand, genes reported to be homozygously deleted include IRF2 (4q35.1), CDKN2A (9p21.3), PTEN (10q23.31), RB1 (13q14.2), and AXIN1 (16p13.3). Among the most commonly gained regions in HCC are copy number gains of 1q and 8q, both in around 60%–70% of HCC. The minimal amplified region 1q21 has been linked to the early development of HCC (1q21-23) and to advanced metastatic HCC cases (1q21-q22). The region has been shown to contain potential oncogenes that when overexpressed may be involved in hepatocarcinogenesis. For instance, the gene CHD1L has been shown to be amplified and overexpressed in HCC but not in matching normal parenchyma. Additionally, expression of CHD1L is positively associated with vascular invasion and has been found to be an independent predictor of decreased disease-free survival (DFS) in HCC patients after surgical resection. Chromosome arm 8q contains the well-known oncogene MYC, a central regulator of malignant transformation in early hepatocarcinogenesis and has also been found to be significantly correlated with DFS and overall survival (OS) in patients with HCC. On the other hand, 4q, 8p, and 17p are the chromosome arms most frequently lost. While how 4q loss contributes to hepatocarcinogenesis is not clear, it has been linked to high levels of serum AFP and is more frequently seen in poorly differentiated HCC. The loss of chromosome arm 8p have been extensively described in the HCC. A cluster of tumor suppressor genes on 8p have been identified, including the DLC1, CCDC25, ELP3, SH2D4A, and SORBS3 genes, and the loss of these genes has been shown to be associated with metastasis and poor prognoses for HCC patients. On chromosome 17p is the TP53 gene, one of the most frequently mutated genes in HCC. Compared to focal amplifications and homozygous deletions, the biological effects of gains or losses of whole chromosomes or large genomic regions are more difficult to pinpoint. There remain many important questions regarding the identification of the genes in these large CNAs that are involved in hepatocarcinogenesis and their biological and clinical implications.

Hepatocellular Carcinoma: Pathology and Genetics

205

Genes and Pathways Altered by Somatic Genetic Alterations Activation of the WNT/b-catenin pathway The WNT/b-catenin pathway is of great importance in physiologic embryogenesis, zonation, and metabolic control in the liver. The WNT/b-catenin pathway can be classified into canonical and non-canonical pathways (based on the dependency of b-catenin). b-catenin, encoded by the CTNNB1 gene, plays a pivotal role in the regulation of the canonical WNT signaling pathway. In the absence of WNT, cytoplasmic b-catenin protein is phosphorylated at serines 33, 37, and 45, and threonine 41 residues by the Axin complex (composed of Axin, APC, CK1, and GSK3) and then ubiquitinated and degraded by b-Trcp ubiquitin ligase. Activation of Wnt signaling induces the stabilization and translocation of b-catenin to the nucleus, where it associates with transcription factors of the TCF family and generates a functional complex able to transactivate several genes involved in the cell cycle control and proliferation, such as MYC, CCND1, MT-CO2, and MMP7. Dysregulation of the WNT pathway has been observed in 40%–70% of HCCs suggesting a major role of this pathway in HCC development. Processes WNT signaling regulates include angiogenesis, infiltration and metastasis by regulating the expression of angiogenic factors such as MMP-2, MMP-9, VEGF-A, and VEGF-C. The most common mechanism of b-catenin activation in HCC is through mutation of the CTNNB1 gene, identified in about 40% of HCC. The majority of the CTNNB1 mutations occurs at mutation “hotspots” in exon 3, specifically at the four phosphorylation sites. Mutation of serine/threonine residues results in impaired Axin/APC/GSK3b mediated degradation of b-catenin and the gain of oncogenic activity. HCCs with CTNNB1 mutation display a specific transcriptomic profile with the overexpression of glutamine synthetase (encoded by GLUL) and leucine rich repeat containing G protein-coupled receptor 5 (encoded by LGR5), as well as increased expression of glutamine synthetase on the protein level. Other alterations that lead to an upregulation of WNT signaling include mutations or homozygous deletions of AXIN1 (5%– 15%), AXIN2 (3%), and APC (1%–2%). Several studies demonstrated a critical role of the deregulation of WNT/b-catenin signaling in hepatocarcinogenesis. HCVpositive HCC tends to be associated with frequent CTNNB1 mutations. CTNNB1 mutations have been found to be associated with macrovascular and microvascular invasion and increased tumor size. At the same time, HCCs characterized by the activation of the Wnt/b-catenin pathway exhibit specific features such as high differentiation associated with a homogeneous microtrabeculoacinar pattern, low-grade cellular atypia, and cholestasis. Moreover, WNT/b-catenin signaling has been shown to play an important role in the activation of oval cells, considered to be hepatic stem cells, and HCCs with stem cell signatures have a more aggressive behavior. However, the clinical implications of CTNNB1 activating mutations in terms of overall survival and disease recurrence have been contradictory. These findings demonstrate an essential role of the Wnt signaling in hepatocarcinogenesis and suggest that targeting this pathway may be promising for therapeutic options (Figs. 3 and 4).

Alterations in TP53 and cell cycle signaling pathway TP53 is the most frequently mutated gene in human cancers. The p53 protein modulates multiple cellular functions, including transcription, DNA synthesis and repair, cell cycle arrest, senescence and apoptosis. Mutations in TP53 can abrogate these functions, leading to genomic instability and progression to cancer. In HCC, TP53 is one of the most frequent mutated genes and its

Fig. 3 Significantly mutated genes in HCC. Venn diagram illustrates the significantly mutated genes in HCC identified in three seminal studies that described in the genomic landscape of HCC. *Promoter region.

206

Hepatocellular Carcinoma: Pathology and Genetics

Fig. 4 Altered signaling pathways in HCC. Schematic representation of the major signaling pathways recurrently altered in HCC. Genes belonging to each pathway are represented and activating or inhibitory interactions between pathways are indicated with lines according to the legend.

mutational frequency is between 18% and 50%, depending on the geographic regions, etiological factors and carcinogen exposure, with more frequent TP53 mutations in HBV-associated HCCs. Inactivating mutations in the TP53 gene, or other components in the p53 pathway, may render hepatocytes susceptible to the effects of other carcinogens that activate oncogenic pathways and may also predispose to the development of HCCs with a more aggressive phenotype. For instance, TP53-mutant HCCs have been associated with features linked to poor prognosis, including high levels of serum AFP, high Edmondson grade, expression of stem-like markers, and activation of pro-oncogenic signaling pathways. Indeed, patients with TP53-mutant HCCs tend to have shorter OS and DFS. The G1 to S phase of the cell cycle (retinoblastoma pathway) is often deregulated via homozygous deletion of CDKN2A (p21, 2%–12%) or RB1 mutations (3%–8%) and is often associated with poor prognosis suggesting a role of the related pathway inactivation in tumor aggressiveness. Additionally, recurrent HBV integration into the CCNE1 locus (5%) and amplification of the CCND1/FGF19 locus (5%–14%) have been reported in HCC, both of which have implications in the regulation of cell cycle. Not all TP53 mutations in HCCs are equal. Dietary exposure to fungal aflatoxin B1 results in a specific TP53 mutation that is rarely found in other cancers, namely the R249S mutation resulting from G > T transversion; this is considered to be a driver mutation since it is also found in the normal livers of patients exposed to aflatoxin B1. There is strong epidemiologic synergism between aflatoxin B1 exposure and chronic HBV infection in the induction of HCC, and it has been shown that, in patients infected with HBV, expression of HBx is associated with an approximately twofold increase in the incidence of G/C-to-T/A transversion mutations following aflatoxin B1 exposure (Figs. 3 and 4).

Alterations in the human telomerase reverse-transcriptase The human telomerase reverse-transcriptase (TERT) gene encodes a rate-limiting catalytic subunit of telomerase, which maintains the length of telomeric DNA and chromosomal stability. Thus, TERT plays a pivotal role in cellular immortalization, cancer development and progression. Reactivation of telomerase activity allows cells to overcome replicative senescence and to escape apoptosis, both of which are fundamental steps in the initiation of malignant transformation. The interests in the role of telomerase maintenance in the development of HCC have grown since the discovery of TERT promoter hotspot mutations in HCC and other cancers. While telomere shortening is a feature of chronic liver diseases and cirrhosis, telomerase reactivation through TERT promoter mutations (30%–60%), amplifications (5%–6%) and HBV integration into the TERT promoter or the gene body of TERT (10%–15%) is associated with hepatocarcinogenesis. Two hotspot mutations c.-124C > T and c.-150C > T have been reported in HCC, and in particular the c.-124C > T mutation has now been shown to be the most common mutation in HCC, observed in

Hepatocellular Carcinoma: Pathology and Genetics

207

30%–60% of HCC. This mutation is in fact the most frequent mechanism of telomerase activation. The c.-124C > T and c.-150C > T mutations result in the formation of novel ETS transcription factor binding sites upstream of the TERT transcription start site, which leads to increased TERT transcript expression. TERT promoter mutations have been reported to be more frequent in HCV-associated HCC. On the other hand, compared to HCCs of other etiologies, HBV-induced HCCs are less likely to harbor TERT promoter mutations but may be deregulated by integration of HBV sequences into the TERT gene locus, which serves as a complementary mechanism for telomerase activation. Deregulation of TERT has been suggested to be an early driver event in hepatocarcinogenesis. TERT mRNA has been reported to be detectable in the serum of patients with HCC with higher sensitivity and specificity than serum AFP, AFP-L3, and des-gammacarboxy prothrombin (DCP) in the diagnosis of early stage HCC. Thus, measuring serum TERT mRNA levels may potentially be used as diagnostic tool for HCC detection at early stage. In fact, TERT promoter mutations have been suggested to be early driver mutations, given that TERT promoter mutations have been found in 30% of cirrhotic preneoplastic lesions. These findings suggest that TERT promoter mutations are among the earliest genetic alterations in hepatocarcinogenesis, occurring at preneoplastic stages and behaving as a “gatekeeper” during the malignant transformation sequence. It has also been suggested that telomerase inhibition may be a potential therapeutic target in treating HCC (Figs. 3 and 4).

Alterations in chromatin modifiers Chromatin remodelers are epigenetic modifiers that play important roles in maintaining nucleosome positioning and thus in transcriptional regulation. Alterations in ATP-dependent chromatin remodeling complexes have been implicated in carcinogenesis, tumor heterogeneity and cellular response to therapy. Among others, the most frequently altered chromatin remodelers in human cancer are part of the SWI/SNF complex (20%). In fact, this complex has been associated with epigenetic modification including roles in maintaining nucleosome positioning and interacting with other chromatin modifiers. The SWI/SNF complex contains two main subunits, namely the BAF and the PBAF complexes. In HCC, inactivating mutations are frequently found in the ARID1A and ARID1B genes (4%–17%), encoding the AT-rich interacting domain containing protein 1A-B which are components of the BAF complex, and in ARID2 (3%–18%), encoding the AT-rich interacting domain 2, a component of the PBAF complex. The frequent occurrence of inactivating mutations in the SWI/SNF complex suggests that the complex acts as a tumor suppressor. Although it has been suggested that alterations in the SWI/SNF complex modify chromatin structure and nucleosome position, the functional role of the mutations and the molecular mechanisms that result in the initiation and progression of HCC are not yet fully understood. In addition to the SWI/SNF complex, recurrent somatic genetic alterations in the histone methylation family members MLL (3%–4%), MLL2 (2%–3%), MLL3 (3%–6%), and MLL4 (2%–3%) or HBV insertions into the MLL4 gene locus (10%) are also frequent in HCC. The MLL gene family encodes for proteins that modify histone methylation by adding and removing H3K4 methyl groups. As with alterations in the SWI/SNF complex, the functional significance of alterations in the MLL genes in hepatocarcinogenesis remains to be further elucidated (Figs. 3 and 4).

Alterations in genes involved in oxidative stress response Oxidative stress pathway dysfunctions have been linked to carcinogenesis and the progression of cancer via the production of reactive oxygen species (ROS) that cause damage to proteins, DNA and lipids. One of the major oxidative stress pathways is the NRF2KEAP1 signaling pathway, which is a regulator of cytoprotective responses to endogenous and exogenous stresses caused by ROS and electrophiles. In this pathway, NRF2 is the mediator of the oxidative stress response while KEAP1 is the negative regulator of NRF2 activity. Deregulation of the NRF2-KEAP1 signaling pathway prevents proteasome degradation of NRF2, which is physiologically induced by KEAP1/CUL3 complex ubiquitinylation. This results in epigenetic instability or altered chromatin status, leading to abnormal methylation of tumor suppressor genes. In HCC, the NRF2-KEAP1 signaling pathway is altered by activating mutations of NRF2 (encoded by NFE2L2) or inactivating KEAP1 mutations in 5%–15% of the cases. The role of the somatic mutations in these genes in cancer formation/development is, however, controversial. Studies carried out in rats have reported the presence of these somatic alterations in preneoplastic lesions and early HCCs. By contrast, mutations in NFE2L2 and KEAP1 in humans have only been observed in advanced HCC and not in premalignant nodules or early HCC, suggesting that these mutations are late genetic events in hepatocarcinogenesis in humans. Furthermore, mutations in NFE2L2 or KEAP1 are significantly correlated with the deregulation of the WNT/b-catenin pathway via CTNNB1 or AXIN1 mutations, suggesting that the NRF2/KEAP1 pathway may interact with WNT/b-catenin signaling to promote hepatocarcinogenesis (Figs. 3 and 4).

Deregulation of PI3K-AKT-mTOR and RAS/RAF/MAPK signaling pathways The PI3K/AKT/mTOR pathway is an intracellular signaling pathway that plays a pivotal role in the cell cycle control. Phosphorylation of PI3K activates AKT which regulates several downstream molecules, including mTOR. Several factors may constitutively activate the PI3K/AKT/mTOR signaling pathway, such as EGF, shh, IGF-1, insulin, and CaM. On the other hand, PTEN, GSK3B and HB9 are inhibitors of the pathway. In many cancers including HCC, the PI3K/AKT/mTOR pathway is hyperactive, thus reducing apoptosis and promoting cellular proliferation. Activation of the PI3K/AKT/mTOR pathway is a common feature of HCC and has been reported in 40%–50% of HCC. Pathway activation may be achieved through several distinct mechanisms. Activating mutations of PIK3CA are observed in 0%–2% of patients and inactivating mutations of TSC1 or TSC2 are observed in 3%–8% of patients. Furthermore, homozygous deletion of PTEN has been identified in 1%–3% of the HCCs. However, some HCCs with activation of PI3K/AKT/mTOR signaling have no

208

Hepatocellular Carcinoma: Pathology and Genetics

genetic alterations in the pathway. Indirect activation of the upstream insulin growth factor pathway has been proposed as an alternative mechanism to active the PI3K/AKT/mTOR pathway. Genetic alterations in the RAS/RAF/MAPK pathway have also been implicated in the development of HCC, albeit rarer. Activating mutations in the RAS family genes (e.g., KRAS, NRAS) are rare (< 2%). More common are inactivating mutations in RP6SKA3, encoding for the RAS inhibitor RSK2, identified in 2%–9% of HCCs. RSK2 is downstream of MAPK and is a negative regulator of RAS signaling. Inactivating RSK2 disrupts the negative feedback loop and results in the constitutive activation of the RAS pathway (Figs. 3 and 4).

Alterations in liver metabolic pathways Liver is unique organ and has very different gene expression patterns compared to other organs. A number of genes highly expressed and/or only expressed in the liver have been found to be frequently mutated in HCC. Of these, ALB (encoding for albumin) and APOB (encoding for apolipoprotein B) have been found to be affected by mutations in 10%–15% of HCC. Mutations in ALB and APOB are dominated by inactivating mutations (i.e., nonsense, splice site and insertions/deletions), strongly suggesting a tumor suppressive role for these genes in liver. Indeed, ALB and APOB-mutant HCCs have been found to have reduced expression of the respective gene. It has been suggested that ALB and APOB mutations may be under positive selection to divert energy for other cancer-associated metabolic requirements. HNF1A and HNF4A (encoding for hepatocyte nuclear factors 1-alpha and 4-alpha) are each mutated in  2% of HCC. Both encode for transcription factors essential for the transcription of a number of genes required for hepatocyte differentiation, with HNF4A regulating the transcription of HNF1A. Alterations in HNF1A and HNF4A are also preferentially loss-of-function alterations (Figs. 3 and 4).

Alterations in non-coding elements Considering that protein-coding regions only account for 1%–2% of the human genome, it is not surprising that recent wholegenome sequencing studies of HCC and other cancers have revealed that the number of somatic mutations in non-coding elements far outnumber that in coding regions. The significance of the vast majority of the non-coding mutations is still unknown but several non-coding regions/genes have been identified to harbor elevated number of mutations. Besides the hotspot mutations in the promoter regions of TERT, an additional 10 promoter regions, including those of TFPI2, MED16 and WDR74, and nine untranslated regions, including those of BCL6 and AFF4, have also been identified as regions with significantly elevated number of mutations. Among these, mutations in the TFPI2 (encoding for tissue factor pathway inhibitor 2, a serine proteinase inhibitor) promoter have been linked to lower expression of TFPI2, a gene reported to have tumor suppressive roles in several cancers. Finally, the long noncoding genes NEAT1 ( 20%) and MALAT1 ( 5%) are recurrently mutated in HCC. Both NEAT1 and MALAT1 encode for transcripts that are subnuclear structural components and regulate the transcription of cancer genes. In particular, MALAT1 is upstream of p53 and is required for G1/S cell cycle progression. On the other hand, NEAT1 expression is reported to be influenced by p53. Our understanding of the precise effect of mutations in non-coding elements is very preliminary and the functional impact of these alterations is currently under active investigation.

DNAJB1-PRKACA fusion gene in fibrolamellar HCC Fibrolamellar HCC is a clinically and morphologically distinctive subtype that accounts for < 1% of HCC. Unlike classical HCC, fibrolamellar HCC is usually diagnosed in young adults and is rarely associated with pre-existing liver disease or cirrhosis. Whole genome and transcriptome sequencing studies have identified a recurrent fusion gene resulting from a 400 kb deletion on chromosome 19 in 100% of fibrolamellar HCC. This in-frame chimeric transcript fuses exon 1 of the DNAJB1 gene, which encodes for a member of the heat shock 40 protein family, to exons 2–10 of the PRKACA gene, which encodes for the adenosine 30 ,50 -monophosphate (cAMP)-dependent protein kinase A (PKA) catalytic subunit alpha. Additional sequencing studies revealed no other highly recurrent genetic lesions in fibrolamellar HCC and the absence of the fusion in non-fibrolamellar HCC. These observations suggest that the fusion is pathognomonic of fibrolamellar HCC. The gene fusion leads to the formation of a chimeric protein that differs from wild-type PKA in size. The functional catalytic domain and kinase activity are retained but the fusion protein lacks the domain that binds to the regulatory subunit of PKA. The functional role of this fusion transcript is still under investigation (Fig. 5).

HCC Susceptibility Loci Germline polymorphisms module HCC risk. Genome-wide association studies have identified susceptibility loci in genes associated with signaling pathways known to be involved in carcinogenesis. These include SOD2 and MPO involved in oxidative stress response, TNFA, IL1B, and TGFB1 in inflammatory response, MTHFR and XRCC3 in DNA repair, MDM2 and TP53 in cell cycle regulation, and EGF (epidermal growth factor). The majority of the susceptibility loci associated with HCC development are linked to specific risk factors. For instance, single nucleotide polymorphisms (SNPs) of GTSM1 and GSTT1 are linked to aflatoxin B1 exposure and HBV infection. Also in HBVpositive patients, SNPs in STAT4, TPTE2, DCL1 are associated with increased risk of HCC. In HCV-positive patients, however, polymorphisms in MICA (a gene involved in regulating immune response), DEPDC5 (a gene in the mTOR pathway) and in the 50 UTR of EGF have been reported as susceptibility loci. A PNPLA3 polymorphism has been linked to HCC associated with fatty and alcoholic liver diseases. These associations suggest interaction between these SNPs, environmental factors and chronic infection.

Hepatocellular Carcinoma: Pathology and Genetics

209

Fig. 5 Recurrent molecular alteration in fibrolamellar HCC. High power and low power micrographs of a fibrolamellar HCC (left) and a schematic representation of the major recurrent fusion gene DNAJB1-PRKACA present in nearly 100% of fibrolamellar HCCs (right).

Genetic Syndromes Associated With Increased Risk of HCC Several rare monogenic syndromes have been linked to increased risks of hepatocarcinogenesis. The autosomal recessive a-1-antitrypsin deficiency results from mutations in the SERPINA1 gene, a serine protease synthesized at high levels in the liver and responsible for the inhibition of neutrophil elastase. Glycogen storage disease type I, caused by the deficiency of glucose-6-phosphatase (G6Pase) activity results in excess glycogen storage in the liver and causes hepatomegaly, fasting hypoglycemia, lactic acidosis, hyperlipidemia, hyperuricemia, and growth retardation. Hemochromatosis is caused by a mutation in the HFE gene (most commonly single base mutation C282Y) and leads to the development of fibrous tissue at the site of iron deposition, causing liver dysfunction. Patients with hemochromatosis are at increased risk of developing hepatic cirrhosis followed by HCC. Acute intermittent porphyria (AIP) is a low penetrant autosomal dominant disease caused by mutation in enzymatic activity of hydroxymethylbilane synthase (HMBS), one of the enzymes in the heme biosynthesis pathway. AIP patients have an > 30-fold increase in the risk of HCC compared to the general population. HCC in AIP patients are rarely associated with cirrhosis. Tyrosinemia type I is an autosomal recessive disorder caused by biallelic pathogenic variants in the FAH gene. The deficiency of the fumarylacetoacetate hydrolase enzyme leads to an accumulation of tyrosine catabolic intermediates, acute hepatic failure in infancy or chronic liver disease associated with cirrhosis and HCC development. Finally, HCC has also been reported in patients with familial adenomatous polyposis associated with APC germline mutation.

Conclusions With sorafenib, regorafenib, and nivolumab still the only approved systemic therapies for HCC, the identification of additional therapeutic targets is crucial to improve the prognosis of HCC patients. On the one hand, the consensus on the classification between early and late HCCs is having an impact in the clinic to optimize clinical management. On the other hand, our evolving understanding of the genetic heterogeneity of HCC is beginning to reveal additional pathways and potential vulnerabilities that may be exploited in future studies. Integration of histopathologic and molecular analyses will provide further insight into the pathogenesis and genotype–phenotype interactions important in HCC.

See also: Cholangiocarcinoma: Diagnosis and Treatment.

210

Hepatocellular Carcinoma: Pathology and Genetics

Further Reading Atlas, C.G., 2017. Comprehensive and integrative genomic characterization of hepatocellular carcinoma. Cell 169, 1327–1341. Di Tommaso, L., Destro, A., Seok, J.Y., et al., 2009. The application of markers (HSP70 GPC3 and GS) in liver biopsies is useful for detection of hepatocellular carcinoma. Journal of Hepatology 50, 746–754. Fujimoto, A., Furuta, M., Totoki, Y., et al., 2016. Whole-genome mutational landscape and characterization of noncoding and structural mutations in liver cancer. Nature Genetics 48, 500–509. Schlageter, M., Terracciano, L.M., D’Angelo, S., et al., 2014. Histopathology of hepatocellular carcinoma. World Journal of Gastroenterology 20, 15955–15964. Schulze, K., Imbeaud, S., Letouzé, E., et al., 2015. Exome sequencing of hepatocellular carcinomas identifies new mutational signatures and potential therapeutic targets. Nature Genetics 47, 505–511.

Hereditary Cancer Syndromes: Identification and Management Roberta Pastorino and Alessia Tognetto, Institute of Public Health, Catholic University of the Sacred Heart, Rome, Italy Stefania Boccia, Institute of Public Health, Catholic University of the Sacred Heart, Fondazione Policlinico “Agostino Gemelli”, Rome, Italy © 2019 Elsevier Inc. All rights reserved.

Cancer Control From a Public Health Perspective Incidence and prevalence of cancer have been increasing for most neoplastic diseases in the past decades, due to different causes, including the progressive aging of the population and the continuous improvement in diagnostic and treatment strategies, that, if not always able to cure, at least make people live longer and with an acceptable quality of life. In this context, the healthcare assistance related to cancer, from screening to diagnosis, to cure or treatment, has become a very crucial, but challenging, public health topic. Cancer is a multifactorial disease, whose pathogenesis is complex, but strongly driven by genetic modifications in the genome DNA. This makes cancer field of study of Public Health genomics, the field in epidemiology wherein molecular data at population scale are integrated into new strategies both from a personalized medicine and a public health perspective. The completion of the human genome-sequencing project in 2003, and the further advances in numerous sectors of biotechnology, created great expectations about health benefits for several diseases, including cancer. Moreover, the decrease of the costs related to the DNA sequencing techniques has contributed to the spread of genetic testing in healthcare. The genomic approach to cancer means a consistent reduction in terms of mortality, morbidity, and, therefore, a possible reduction of costs for the healthcare systems. Genetic testing may contribute to cancer control by identifying individuals at high risk of disease. Some monogenic subtypes of common cancers such as BRCA-related breast cancer and Lynch syndrome (LS) colorectal cancer (CRC) are autosomal dominant disorders with a risk of 50% for first-degree relatives of the index case to be affected. The identification of these hereditary forms of cancer that show high incidence, mortality rate, and that are potentially preventable or efficiently treatable is of critical public health importance because it changes cancer risk management options and enables patients and their at-risk family relatives to benefit from targeted intervention measures.

Hereditary Breast Cancer Breast cancer is the most common malignancy in women, and one of the three most common cancers worldwide, with around 1,671,000 new cases diagnosed in 2012. The majority of breast cancers show a multifactorial pathogenesis; however, approximately 30% of breast cancer cases occur in family clusters and up to 7% have a strong hereditary component. Among affected women, the genetic variants of BRCA1 and BRCA2 (breast cancer susceptibility genes) account for up to the 30% of the genetic predisposition to breast cancer and show a high-penetrance pattern of hereditary predisposition. The prevalence of pathogenic variants of BRCA genes has been estimated as 1/800; however, a number of founder effects with higher prevalence have been observed in different populations as Ashkenazi Jewish, Swedish, Dutch, Hungarian, and Icelandic. Mutations in these genes predispose women to develop cancer at a younger age than general population. In women, BRCA1 or BRCA2 mutations result in a 57%–65% or 45%–55% risk of developing breast cancer by age 70 years, and a lifetime risk of 39%– 44% or 11%–18% of developing ovarian cancer (includes fallopian tube and primary peritoneal cancers), respectively. In men, the lifetime risk of breast cancer is 1.2% for BRCA1 mutation carriers and 6.8% for BRCA2 mutation carriers. Besides, carriers have a considerable risk to develop a contralateral breast cancer, after the first one. In 2003 the European Council recommended the implementation of cancer screening programs for women aged 50–69 years. Also the U.S. Preventive Services Task Force and the American Cancer Society recommend that women aged over 50 years undergo regular screening mammography for the early detection of breast cancer. However, women with a BRCA1 or BRCA2 mutation may develop cancer at a younger age. Identifying BRCA mutated women is important not only for cancer-affected women because of the prognosis and the related treatment choices but also for healthy carriers because of the different options in terms of primary and secondary prevention. According to international guidelines, surveillance of breast cancer in BRCA carriers includes monthly self-examinations, clinical breast examinations twice a year, and yearly mammograms and magnetic resonance imaging of breasts starting at age 25 or younger based on the earliest age at which breast cancer was diagnosed in the family. Screening for ovarian cancer is also recommended and preventive surgery (mastectomy and salpingo-oophorectomy) can be considered. Considering the relevance of breast cancer, international communities are strengthening the efforts in order to implement adequate integrated diagnostic-therapeutic pathways for the affected women. In this context, breast units have a key role, coordinating high quality and appropriate assistance, taking care of patients in all the stages of their disease.

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65126-7

211

212

Hereditary Cancer Syndromes: Identification and Management

To date, according to the most recent international guidelines it is not indicated to perform a screening program for genetic testing of BRCA genes, neither in the general population nor in all women with breast cancer. In fact, the U.S. Preventive Services Task Force recommends to perform the genetic testing only in the presence of specific anamnestic criteria that identify the high-risk woman that might carry the BRCA mutations. Women with positive screening results should receive genetic counseling and, if indicated after counseling, BRCA testing. Considering economic evaluations, there are good evidences in support of the implementation of dedicated screening pathways to identify high-risk women. According to a recent systematic review of the economical evaluations of BRCA screening strategies, family history-based genetic testing is potentially very cost-effective, but there was no evidence of cost-effectiveness for testing all the newly diagnosed cases of cancers. Hereditary breast cancer is addressed as one of the main topics in cancer control policies and the implementation of prevention programs would reduce the burden of the inherited disease. In some countries public health genomics policies have already been formulated; however, the implementation of integrated pathways for the identification and follow-up of BRCA carriers is still at the beginning.

Hereditary Colorectal Cancer CRC is the third leading cause of cancer death worldwide, accounting for 774,000 deaths in 2015, and with an estimated annual worldwide incidence of 1.4 million new cases. CRC ranked third in prevalence, after lung and prostate cancers, for men, and second, after breast cancer, for women. Due to its high incidence rate, its long preclinical phase, its recognizable and tractable precursor, and the association between the tumor stage and mortality rate, CRC is a condition suitable for screening according to the criteria established by Wilson and Jungner as the “gold standard of screening assessment.” The recommended screening can be performed through an annual or biennial fecal immunochemical test, or sigmoidoscopy every 5 years, or colonoscopy every 10 years. The burden of this disease could be dramatically reduced by screening, but CRC screening programs in many countries start at an age when the neoplastic process has already gone in hereditary forms of CRC, leading to a late-stage diagnosis for those patients. Between 5% and 10% of CRC are hereditary forms, of which the most common is LS, that is associated with approximately 3%– 5% of all CRCs. LS, previously known as hereditary nonpolyposis CRC, is an autosomal dominant disorder caused by mutations in DNA mismatch repair (MMR) genes that have the role to maintain genomic integrity during DNA replication. Inactivation of MMR genes causes an increased mutation rate (genomic instability) and a microsatellite instability. In individuals with LS, current CRC screening recommendations include colonoscopy every 1–2 years beginning at age 20–25 years. Annual transvaginal ultrasound of the uterus and ovaries and endometrial sampling are also recommended although efficacy is still debated as well as the utility of screening for other LS cancers. Furthermore, prophylactic surgery can be considered as an effective option for women with LS. Additionally, regular, long-term aspirin use is reported as a preventive measure to reduce incidence and mortality due to CRC and the most relevant impact of chemopreventive strategies is expected in patients with an hereditary predisposition as LS. Hence, the identification of LS-related CRC can lead to beneficial outcomes in terms of survival for the patients, and in terms of prevention for their affected family members. Several international guidelines recommend to screen for LS all newly diagnosed CRC patients, regardless the age or the clinical criteria. The first step of the screening is performed on tumor tissue through immunohistochemistry staining or the microsatellite instability molecular testing, two laboratory procedures with comparable sensitivity and specificity. In case of positive tumor test, the patient is referred to genetic services for counseling and germline testing for MMR mutations to establish the diagnosis of LS. Moreover, economic evaluations documented the cost-effectiveness of the universal tumor screening strategy versus no screening because it reduces the costs related to morbidity and mortality from CRC from the early identification of LS-carriers among family members. However, in resource-limited settings, a universal tumor screening could be hardly achievable. This strategy requires a multidisciplinary team (involving pathologists, surgeons, gastroenterologists, oncologists, geneticists, genetic counselors) that might be not available in all clinical contexts. Furthermore, genetic services are not always available in all clinical settings, and some laboratories could lack in resources, expertise, and quality assessment to perform all the required tests for tumor screening.

Implementation of Genetic Testing Programs in Health Care: Barriers and Future Challenges The increasing evidence of genomic contribution to cancer onset and progression makes it a relevant part of research and public health which can no longer be overlooked. The awareness of the health benefits of using genetic tests and family health history in clinical practice has been increasing, as underscored in the agenda items in Healthy People 2020: “Increase the proportion of women with a family history of breast and/or ovarian cancer who receive genetic counseling” and “Increase the proportion of persons with newly diagnosed colorectal cancer who receive genetic testing to identify Lynch syndrome.”

Hereditary Cancer Syndromes: Identification and Management

213

In this framework, to make genetic testing for hereditary forms of cancer accessible to all the patients and their family members, when appropriate, as part of the routine diagnostic and therapeutic pathways, many barriers still need to be overcome. The health professionals’ lack of knowledge about genetic testing and their lack of confidence in interpreting familial patterns of disease are some of the most significant barriers to integrate genetic testing into cancer prevention. The lack of consensus and providers’ awareness as well as specific guidelines for every country, that take into account the feasibility of different approaches in different contexts, have been associated with divergent referral and testing practices. The miscommunication between patient and physicians and the refusal of genetic counseling by the patients, due to a perceived lack of benefit, have to be also considered. Therefore, further efforts are necessary to overcome the gap between expectations on genetics and clinical results. A promotion of genetic education and training for healthcare professionals is crucial to assure awareness of emerging issues and appropriate utilization of new genetic technologies, as well as measures to foster the public understanding of scientific developments in human genetics and associated ethical, legal, and social issues. Moreover, a reduction in financial, geographic, and cultural barriers to access genetic services, an optimal management of test-positive individuals, and a family support through integrated services will be crucial. A multidisciplinary team is usually involved in the screening pathways; therefore, a standardized plan should be created to drive their efforts in order to provide effective and equitable cancer assistance to all patients. In conclusion, genetic research is rapidly increasing the opportunities for the detection of inherited cancer risk, but coordinated strategies in educating clinicians and policy makers and addressing public concerns are necessary to ensure that public health needs are satisfied through appropriate and sustainable healthcare services.

Acknowledgment This work was supported by the PRECeDI project (Marie Skłodowska-Curie Research and Innovation Staff ExchangedRISE No. 645740) http://www. precedi.eu/site/.

See also: Genetic and Epigenetic Deregulation of Enhancers in Cancer. Hereditary Risk of Breast and Ovarian Cancer: BRCA1 and BRCA2. Li–Fraumeni Syndrome. Lynch Syndrome. TP53.

Further Reading Boccia, S., Michelazzo, M.B., Tognetto, A., Del Sole, A.M., 2016. The prevention of hereditary breast cancer in Italy: towards the implementation of the national prevention plan in the Italian regions. Epidemiology Biostatistics and Public Health 13 (2), 119431–119432. Editorial. Burke, W., Khoury, M., Stewart, A., Zimmern, R.L., for the Bellagio Group, 2006. The path from genome-based research to population health: development of an international public health genomics network. Genetics in Medicine 8 (7), 451–458. D’Andrea, E., Marzuillo, C., De Vito, C., et al., 2016. Which BRCA genetic testing programs are ready for implementation in health care? A systematic review of economic evaluations. Genetics in Medicine 18 (12), 1171–1180. Evaluation of Genomic Applications in Practice and Prevention (EGAPP) Working Group, 2009. Recommendations from the EGAPP Working Group: Genetic testing strategies in newly diagnosed individuals with colorectal cancer aimed at reducing morbidity and mortality from Lynch syndrome in relatives. Genetics in Medicine 11, 35–41. Giardiello, F.M., Allen, J.I., Axilbund, J.E., et al., 2014. Guidelines on genetic evaluation and management of Lynch syndrome: a consensus statement by the US Multi-Society Task Force on colorectal cancer. The American Journal of Gastroenterology 109 (8), 1159–1179. Hampel, H., Frankel, W.L., Martin, E., et al., 2008. Feasibility of screening for Lynch syndrome among patients with colorectal cancer. Journal of Clinical Oncology 26, 5783–5788. Hampel, H., Bennett, R.L., Buchanan, A., Guideline Development Group, American College of Medical Genetics and Genomics Professional Practice and Guidelines Committee and National Society of Genetic Counselors Practice Guidelines Committee, 2015. A practice guideline from the American College of Medical Genetics and Genomics and the National Society of Genetic Counselors: referral indications for cancer predisposition assessment. Genetics in Medicine 17, 70–87. Lerner-Ellis, J., Khalouei, S., Sopik, V., Narod, S.A., 2015. Genetic risk assessment and prevention: the role of genetic testing panels in breast cancer. Expert Review of Anticancer Therapy 15 (11). Moyer, V.A., 2014. Risk assessment, genetic counseling, and genetic testing for BRCA-related cancer in women: U.S. Preventive Services Task Force recommendation statement. Annals of Internal Medicine 160 (4), 271–281. Pastorino, R., Tognetto, A., Boccia, S., 2017. Screening programs for Lynch syndrome in Italy: state of the art and future challenges. Epidemiology Biostatistics and Public Health 14 (2), 126151–126153. Editorial. Rubenstein, J.H., Enns, R., Heidelbaugh, J., et al., 2015. Clinical Guidelines C. American Gastroenterological Association Institute. Guideline on the diagnosis and management of Lynch syndrome. Gastroenterology 149 (3), 777–782. Tognetto, A., Michelazzo, M.B., Calabrò, G.E., et al., 2017. A systematic review on the existing screening pathways for Lynch syndrome identification. Front. Public Health 5, 243.

Relevant Websites https://www.cancer.org/health-care-professionals/american-cancer-society-prevention-early-detection-guidelines/breast-cancer-screening-guidelines.html. http://globocan.iarc.frdInternational Agency for Research on Cancer; 2013. GLOBOCAN. https://www.healthypeople.gov/2020/topics-objectives/topic/genomics/objectivesdHealthy People 2020. https://www.nice.org.uk/guidance/DG27dNICE guidelines. Molecular testing strategies for Lynch syndrome in people with colorectal cancer. https://www.uspreventiveservicestaskforce.org/.

Hereditary Risk of Breast and Ovarian Cancer: BRCA1 and BRCA2q Claus Storgaard Sørensen, University of Copenhagen, Copenhagen, Denmark © 2019 Elsevier Inc. All rights reserved.

Glossary BRCA1 gene A breast cancer susceptibility gene located on chromosome 17q21, known to be mutated in families prone to a high incidence of early onset breast cancers and also ovarian cancers. BRCA2 gene A gene mapped to chromosome 13q12-13; known to be mutated in families prone to a high incidence of early onset breast cancers and also ovarian cancers. HBOC Hereditary breast and ovarian cancer (HBOC). Homologous recombination repair (HRR) A DNA repair pathway acting on DNA double-strand breaks that uses the undamaged sister chromatid as template for error-free repair. Locus heterogeneity A genetic term describing how variations in different genes cause the same disorder. The genes are not linked physically; examples are hereditary breast and ovarian cancer predisposition by BRCA1, BRCA2, partner and localizer of BRCA2 (PALB2) and TP53. Variant of unknown significance (VUS) Variants in genes are classified according to their impact on the protein function, for example, BRCA1 function. A variant with an unknown clinical function owing to lack of functional or clinical data is classified as a VUS.

Characterization of BRCA1 and BRCA2 Fifteen years of analysis of large multigenerational families with a strong history of breast cancer led in 1990 to the identification of a gene on chromosome 17q12–21 that conferred a greatly increased risk of breast cancer in an autosomal-dominant manner. The gene itself, BRCA1, was cloned in 1994. The discovery of the complete sequence of a second such gene, BRCA2 on 13q12-13, was reported in 1995. The protein-coding region of BRCA1 consists of 5592 bp in 22 exons that encode a protein of 1863 amino acids, and the proteincoding region of BRCA2 consists of 10,254 bp in 26 exons that encode a protein of 3418 amino acids. Studies of the interactions of these proteins led to the conclusion that they are involved in the regulation of genomic stability and DNA repair. The BRCA1 and BRCA2 proteins interact through a third breast cancer tumor suppressor, PALB2 (partner and localizer of BRCA2) that serves as a major link. A key function of the BRCA1–PALB2–BRCA2 pathway is to secure proper and stable loading of the RAD51 protein at sites of DNA lesions such as DNA double-strand breaks. RAD51 is crucial for homologous recombination repair as well as for the protection of exposed DNA from undergoing excessive degradation during DNA replication. Mice lacking Brca1 and Brca2 (the murine homologues of BRCA1 and BRCA2) undergo developmental arrest during embryogenesis. Furthermore, the modeling of mice with mutations in Brca1 and Brca2 revealed that such genetic changes display enhanced tumorigenesis. This can be modeled to take place in the murine breast tissue, particularly when combined with the TP53 loss. Mice homozygous for a truncating mutation of Brca2 that survive to adulthood have a wide range of defects, including improper tissue differentiation, absence of germ cells, and the development of lethal thymomas, and cultured embryonic fibroblasts from these mice are unable to repair radiation-induced DNA damage. Other studies suggest that mouse embryonic stem cells deficient in Brca1 are unable to carry out a transcription-coupled repair of oxidative DNA damage and are hypersensitive to ionizing radiation and hydrogen peroxide. Human BRCA2-defective cancer cells are also deficient in the repair of double-strand DNA breaks induced by ionizing radiation, although individuals with germline mutations in BRCA1 and BRCA2 do not appear to be hypersensitive to ionizing radiation.

Cancer Risk Due to Mutations in BRCA1 and BRCA2 Mutations in BRCA1 and BRCA2 are responsible for increasing the risk of breast cancer to between 56% and 87% for carriers by the age 70. The lower estimate of risk was derived from an analysis of a general population with little or no family history of cancer, and the higher estimate was derived from families with multiple women affected by breast cancer. The increased risk of breast cancer is observed not only over a woman’s lifetime, but particularly at a young age, to 33%–50% before age 50 compared to the general population risk of 2%.

q

Change History: August 2017. Claus Sørensen has updated the text and references. This is an update of Thomas S. Frank, Hereditary Risk of Breast and Ovarian Cancer: BRCA1 and BRCA2 in Encyclopedia of Cancer (Second Edition), edited by Joseph R. Bertino, Academic Press, 2002, Pages 381–385.

214

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65227-3

Hereditary Risk of Breast and Ovarian Cancer: BRCA1 and BRCA2

215

Mutations in BRCA1 also confer a greatly increased risk of ovarian cancer of up to 44%, by age 70, whereas mutations in BRCA2 increase this risk to approximately 27% compared to 1% for the general population of women. The age of onset of hereditary ovarian cancer is typically later than that of hereditary breast cancer, especially when associated with a BRCA2 mutation. Women already diagnosed with breast cancer are at a greatly increased risk of a second cancer of breast and ovary if they carry a mutation in BRCA1 or BRCA2. Mutations in BRCA1 confer a 64% risk of contralateral breast cancer by age 70, or 20% within 5 years of the initial diagnosis, whereas mutations in BRCA2 increase these risks to about 50% by age 70, or 12% within 5 years of the first breast cancer. The risk of developing ovarian cancer after breast cancer is increased 10-fold to at least 16% compared to women with early onset breast cancer without such mutations. Mutations in both BRCA1 and BRCA2 also confer increased (albeit low) risks of some cancers of men, particularly male breast cancer. Mutations in BRCA2 are responsible for an increased risk of prostate cancer (8% by age 70 and 20% by age 80), as well as an elevated but still low (2%–3%) risk of pancreatic cancer. There is also a familial component in “triple-negative” breast cancer where cancer cells do not express the receptors estrogen, progesterone, and HER2/neu. About 10%–30% of women under the age of 60 diagnosed with “triple-negative” breast cancer will have a gene mutation in BRCA1. Triple-negative breast cancer has relatively poor prognosis, hence these women are also included in clinical genetic programs.

Assessment of Hereditary Risk of Breast and Ovarian Cancer Genetic testing for mutations in BRCA1 and BRCA2 genes is no longer confined to research protocols, but is now clinically available to healthcare professionals. Mutations in the BRCA1 or BRCA2 genes can be identified through a blood or saliva test. DNA sequence analysis is generally acknowledged as the most sensitive method of analyzing these genes because more than thousand mutations have been described throughout the lengths of each. Sequence analysis detects the majority of clinically significant abnormalities in BRCA1 and BRCA2. Large duplications and deletions in the respective genomic regions have also been reported for which clinical testing has become available. Prior to genetic testing, a patient is generally provided pretest education and counseling regarding hereditary risk and genetic testing, including consideration of which relatives with whom the patient would share test results. Individuals who choose to be tested are asked to sign an “informed consent” form indicating that they understand the benefits and limitations of the test that they have chosen. An identified mutation in either BRCA1 or BRCA2 can usually be assumed to have been inherited from one parent or the other, as spontaneous germline mutations in either gene are rare. An individual who does not have a germline mutation in BRCA1 or BRCA2 at the time of birth cannot acquire one later in life so tests for mutations in these genes are normally performed only once in a person’s lifetime. If a mutation is identified in HBOC predisposing genes, a tailored surveillance plan is developed based on the pattern of cancers associated with the specific genes and family history of cancer. This plan must be developed by a team of healthcare professionals and must include expertise in clinical cancer genetics. Testing for mutations in the BRCA1 or BRCA2 genes may not be beneficial for the average woman since most breast and ovarian cancers are sporadic. Genetic testing for hereditary cancer risk is generally preceded by thorough assessment of an individual’s family history in order to determine whether she (or he) is likely to carry a mutation responsible for hereditary cancer risk. The hallmarks of mutations in BRCA1 and BRCA2 include two or more family members on the same side of the family with breast cancer at an early age of onset (usually before age 50) or ovarian cancer at any age. Breast and ovarian cancer in the same individual or male breast cancer at any age also indicates the possibility of HBOC. In evaluating a family history for the possibility of hereditary cancer risk, it is important to equally assess the father’s side of the family as well as the mother’s, as half of women with a hereditary risk of breast and ovarian cancer inherited it from their father. Mutations in BRCA1and BRCA2 have been described throughout the world but are more prevalent in some populations than others, such as individuals of Ashkenazi Jewish descent (i.e., central and eastern European origin). Evaluation of hereditary cancer risk may be warranted for any Ashkenazi Jewish woman with early onset breast cancer or ovarian cancer at any age regardless of family history. Notably, identifying a BRCA1 and BRCA2 variant does not immediately suggest that the individual is a carrier of a pathogenic variant. Indeed, many variants of BRCA1 and BRCA2 have no clear disease linkage, which can be due to a documented lack of cancer predisposition, or, the individual may carry a rare or newly discovered variant. These variants of unknown significance (VUS) can constitute a major issue for the genetic counseling, and such result generally indicate that the individual should be monitored more carefully. BRCA1 and BRCA2 are not the only HBOC predisposing genes. Mutations in other genes may be associated with an increased risk of developing breast cancers, including mutations in the PALB2, ATM, and CHEK2 genes. Another set of genes are involved in syndromes with more varied pattern of cancer types, which is including the Li–Fraumeni syndrome (TP53 gene), Cowden syndrome (PTEN gene), and others. Thus, the pattern of cancers in a family can indicate the specific gene predisposing to hereditary cancer for that family. To aid in optimal genetic analysis, new panels of multiple genes have been developed for testing in patients with a suggestive personal and family history. These panel tests include BRCA1 and BRCA2 and other genes that increase the risk of breast, ovarian, and other cancers. Panels may include 6, 20, 40, or more genes depending on the personal and family history. Thus, even if an individual has a negative test result for BRCA1 and BRCA2, mutations in other genes may still be elucidated. These mutations can be relevant as BRCA1 and BRCA2 operate in complex pathways where locus heterogeneity can be observed. Thus, several gene products collaborate to maintain genome stability, and mutations in several such factors can lead to a BRCA1/2-like phenotype. Because these non-BRCA1/2 mutations are rare, accurate risk assessments may not be possible. However, if a genetic analysis indicates a potential pathogenic variant, prevention and screening strategies may still be initiated.

216

Hereditary Risk of Breast and Ovarian Cancer: BRCA1 and BRCA2

It is in principle possible to use massive in-depth sequencing (next-generation sequencing) of blood samples from individuals with suspected family history. However, this leads to the inclusion of vast number of genes with no clear link to HBOC, which makes it virtually impossible to use such massive data in the clinical genetic setting.

Hereditary Risk of Breast and Ovarian Cancer: Implications for Medical Care Evaluation for hereditary cancer risk should occur in the context of the medical interventions available to address that risk. Management options generally include increased surveillance, prophylactic surgery, and chemoprevention (medications that reduce the risk of cancer). The risks and benefits of these options should each be considered not only for women who do not themselves have cancer, but also women with breast cancer who are at the risk of subsequent malignancy of the breast and ovary. Elaborate guidelines have been developed to establish standardized and adequate clinical practices. These are based on expert opinions and are from organizations such as the National Comprehensive Cancer Network (NCCN), the National Institutes of Health and Clinical Excellence (NICE), and the European Society for Molecular Oncology (ESMO). The reader is referred to these (see references) for comprehensive and recently updated guidelines on initial risk assessment, genetic counseling, and testing. Specifically, it is recommended that women with mutations in BRCA1or BRCA2 perform monthly breast self-examinations, beginning at age 18. The female carrier then undergoes annual or semiannual clinician breast examinations from the age of 25. Annual screening MRI should be started from age 25 as well as annual mammography to commence from age 30. Increased surveillance may be implemented in association with chemoprevention. Tamoxifen, a selective estrogen receptor modulator, may reduce the risk of breast cancer in high-risk women and specifically reduces the risk of contralateral breast cancer in women with BRCA1or BRCA2 mutations. It is thus likely, although not yet specifically demonstrated, that tamoxifen will reduce the occurrence of primary cancer in such women. Several risk-reducing breast surgical techniques exist ranging from total to nipple-sparing mastectomy (NSM). Prophylactic mastectomy has been shown to reduce the risk of breast cancer in “high-risk” women by at least 90% with mutations in BRCA1 or BRCA2. Bilateral risk-reducing total mastectomy is an option, however, this procedure is not the sole approach because of the efficacy of surveillance in detecting most early stage breast cancer and the high cure rate of breast cancers detected at an early stage. Prophylactic total mastectomy may be considered by women whose mammographic assessment is compromised by extensive fibrocystic change or by women whose perception of breast cancer has been affected by relatives or friends with the disease. Following NSM, there might be somewhat elevated residual risk due to breast tissue being left behind after surgery, however, such elevated risk is not clearly established. In any case, all aspects of surgical procedures should be carefully discussed with the individual carrier or patient. Oral contraceptives reduce the risk of ovarian carcinoma by 50% in the general population and have also been associated with a 60% reduction in the risk of ovarian cancer in women with mutations in BRCA1 and BRCA2. It should be noted that oral contraceptives may increase the risk of breast cancer for carriers. Unfortunately, surveillance for ovarian cancer is often ineffective. Measurement of serum CA-125 is regarded as a quite unreliable screen for ovarian cancer and transvaginal ultrasound lacks specificity. Prophylactic oophorectomy should therefore be considered by women around age 35–40 with mutations in BRCA1or BRCA2 as recommended by the clinical practice guidelines. Prophylactic oophorectomy is believed to reduce the risk of ovarian cancer in women with mutations in BRCA1 and BRCA2 by as much as 95% and also reduces their risk of breast cancer by nearly 50%. A negative as well as a positive test result can have significant implications for patient care. If a mutation in BRCA1 or BRCA2 has previously been characterized in a relative, a negative result indicates that a woman is at the population risk of breast and ovarian cancer, despite having a strong family history of either or both diseases. It has also been shown that individuals who learn through genetic testing that they do not carry the mutation identified in their family have significant reductions in depressive symptoms compared with untested individuals. Options exist for couples interested in having a child when they know that one of them carries a gene mutation that increases the risk for this or any other hereditary cancer syndrome. Preimplantation genetic diagnosis done in conjunction with in vitro fertilization is used to avoid the carrier chromosome. Thus, it allows both women and men who carry a specific known genetic mutation to have children who do not carry the mutation.

Summary Inherited mutations in the genes BRCA1 and BRCA2 are associated with a significantly increased risk of breast cancer, particularly before age 50, as well as an increased risk of ovarian cancer. Laboratory analysis of these genes can determine whether a woman has inherited increased risks of breast and ovarian cancer. Identification of a hereditary risk of breast and ovarian cancer can facilitate the medical care of healthy mutation carriers, as well as those already diagnosed with cancer who are at risk of a second malignancy. Once a mutation has been characterized in a family, a family member whose test indicates that he or she did not inherit the mutation has no elevated risk of cancer despite the strong family history, and can therefore avoid unnecessary interventions that might have previously been considered appropriate. Clinical laboratory analysis of BRCA1 and BRCA2 can be used to assist in the identification and management of individuals with a hereditary risk of breast and ovarian cancer.

Hereditary Risk of Breast and Ovarian Cancer: BRCA1 and BRCA2

217

Prospective Vision The identification of BRCA1 and BRCA2 as major HBOC factors has been crucial in the development of tailored clinical approaches to improve counseling and treatment of affected individuals. However, a number of unresolved issues still remain, including the detailed analysis and validation of VUS as well as the elucidation of non-BRCA1/2 genes contributing to a substantial number of HBOCs. The latter may even operate in the same functional pathways as BRCA1/2. Thus further understanding of genetic risk factors as well as nongenetic risk contributors are important to further advance the field. In this regard, ongoing international large-scale efforts aim at tackling these issues, and it is likely that comprehensive whole exome/genome studies will contribute in the near future. When coupled with functional assays to characterize and validate potential variant dysfunction, it is conceivable that HBOC clinical practices will be markedly improved within the coming decade.

See also: Breast Cancer: Pathology and Genetics. Hereditary Cancer Syndromes: Identification and Management. Ovarian Cancer: Diagnosis and Treatment. Ovarian Cancer: Pathology and Genetics. Role of DNA Repair in Carcinogenesis and Cancer Therapeutics.

Further Reading COMPLEXO, Southey, M.C., Park, D.J., et al., 2013. COMPLEXO: identifying the missing heritability of breast cancer via next generation collaboration. Breast Cancer Research 15, 402. Easton, D.F., Pharoah, P.D., Antoniou, A.C., et al., 2015. Gene-panel sequencing and the prediction of breast-cancer risk. New England Journal of Medicine 372, 2243–2257. Hartmann, L.C., Lindor, N.M., 2016. The role of risk reducing surgery in hereditary breast and ovarian cancer. New England Journal of Medicine 374, 454–468. Kuchenbaecker, K.B., Hopper, J.L., Barnes, D.R., et al., 2017. Risks of breast, ovarian, and contralateral breast cancer for BRCA1 and BRCA2 mutation carriers. Journal of the American Medical Association 317 (23), 2402–2416. Lord, C.J., Ashworth, A., 2016. BRCAness revisited. Nature Reviews Cancer 16, 110–120. Millot, G.A., Carvalho, M.A., Caputo, S.M., et al., 2012. A guide for functional analysis of BRCA1 variants of uncertain significance. Human Mutation 33, 1526–1537. National Collaborating Centre for Cancer (UK), 2013. Familial breast cancer: classification and care of people at risk of familial breast cancer and management of breast cancer and related risks in people with a family history of breast cancer. NICE Clinical Guidelines, No. 164. NICE, Cardiff. NCCN, 2016. Clinical practice guidelines in oncology: genetic/familial high risk assessment: breast & ovarian. Version 2.2016. nccn.org/professionals/physician_gls/pdf/genetics_ screening.pdf (Version 2.2017 – December 7, 2016). Nielsen, F.C., van Overeem Hansen, T., Sørensen, C.S., 2016. Hereditary breast and ovarian cancer: new genes in confined pathways. Nature Reviews Cancer 16 (9), 599–612. Nik-Zainal, S., Davies, H., Staaf, J., et al., 2016. Landscape of somatic mutations in 560 breast cancer whole-genome sequences. Nature 534, 47–54. Paluch-Shimon, S., Cardoso, F., Sessa, C., et al., 2016. Prevention and screening in BRCA mutation carriers and other breast/ovarian hereditary cancer syndromes: ESMO Clinical Practice Guidelines for cancer prevention and screening. Annals of Oncology 27 (Suppl. 5), v103–v110. Spurdle, A.B., Healey, S., Devereau, A., et al., 2012. ENIGMAdevidence-based network for the interpretation of germline mutant alleles: an international initiative to evaluate risk and clinical significance associated with sequence variation in BRCA1 and BRCA2 genes. Human Mutation 33, 2–7.

HIV (Human Immunodeficiency Virus)q Jay A Levy and JoAnn C Castelli, University of California San Francisco, San Francisco, CA, United States © 2019 Elsevier Inc. All rights reserved.

Glossary Acquired immunodeficiency syndrome (AIDS) A disease of the immune system caused by HIV infection and characterized by increased susceptibility to opportunistic infections, certain cancers, and neurological disorders. CD4 A cellular surface protein found on T cells as well as other cells. It functions as a receptor for HIV binding and facilitates recognition of the T-cell receptor to antigens bound to MHC class II complexes. Cellular immunity The immune system response by a number of cellular components such as CD8 þ T cells, CD4 þ T cells, NK cells, neutrophils, macrophages, and dendritic cells. The effector function involves contact of the immune cells with infected cells or cancer cells. Chemokines A class of proinflammatory cytokines that have the ability to attract and activate leukocytes. They can be divided into at least three structural branchesdC, CC, and CXCdaccording to variations in a shared cysteine motif. Cytokines Proteins secreted by leukocytes and some nonleukocytic cells that act as intercellular mediators. They are also produced by a number of tissue or cell types and generally act locally in a paracrine or autocrine manner. Dendritic cells Immune cells, derived from bone marrow, with long, tentacle-like branches called dendrites. Among dendritic cells are the Langerhans cells of the skin and follicular dendritic cells in the lymph nodes. Most dendritic cells function as antigen-presenting cells that digest extracellular pathogens and present segments of protein (antigen) on the cell surface to induce a primary immune response. Humoral immunity An antibody-mediated immunity involving secreted products of B cells that interact and neutralize extracellular pathogens. Innate immunity A nonantigen specific immune system that is the first line of host defense against incoming pathogens and foreign substances. Unlike the adaptive immune system, the innate immune system response is not to a specific antigen nor does it usually have memory of previous exposure to an antigen. This immune system acts within hours to days whereas the adaptive immune system, reflected by humoral and cellular immunity, responds after days to weeks. The innate immunity plays a role in activating the adaptive immune system. Lentivirus A genus of the family Retroviridae consisting of nononcogenic retroviruses that produce multiorgan diseases characterized by long incubation periods and persistent infection. Lentiviruses are unique in that they contain open reading frames between pol and env genes and in the 30 env region. Long terminal repeats (LTR) Identical DNA sequences, several hundred nucleotides long, found at either end of transposons and proviral DNA. LTRs are formed by reverse transcription of retroviral RNA. In proviruses, the upstream LTR acts as a promoter and enhancer, whereas the downstream LTR acts as a polyadenylation site. Long-term survivors Individuals infected by HIV for many years who, without therapy, do not show any signs of the infection. They have a normal immune system reflected by normal CD4 þ cell counts and low virus levels in the blood. Some are elite controllers who have had undetectable HIV in their plasma for several years. Macrophages Large white blood cells found mainly in connective tissue and in the bloodstream that ingest foreign particles and infectious microorganisms by phagocytosis. Retroviridae A family of viruses with a single-stranded RNA that, upon infection, generates a DNA copy via a viral reverse transcriptase; lentivirus is a genus in a subfamily. Simian immunodeficiency virus (SIV) A member of the genus lentivirus that induces acquired immunodeficiency syndrome in monkeys and apes. SIV and HIV-2 exhibit close structural and immunologic properties and are 75% homologous. T lymphocyte A subset of lymphocytes that develop in the thymus and circulate in the blood and lymphoid tissue. They orchestrate the response of the immune system to infected or malignant cells, either by lymphokine secretions or by direct contact. Helper T cells recognize foreign antigen on the surfaces of other cells, thus causing the stimulation of B cells to produce antibody and to cytotoxic T cells to destroy antigen-displaying cells.

q

Change History: November 2017. Jay A. Levy added 3 new sections: Intrinsic Anti-HIV Cellular Factors; Cancers associated with HIV infection; Anti-HIV Approaches. Table 2 and Figure 3. Bibliography has been updated. This article is an update of JoAnn C. Castelli, and Jay A. Levy, HIV (Human Immunodeficiency Virus), In Encyclopedia of Cancer (Second Edition), edited by Joseph R. Bertino, Academic Press, 2002, Pages 407-416.

218

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.11085-2

HIV (Human Immunodeficiency Virus)

219

The human immunodeficiency virus (HIV), is a retrovirus that gradually destroys the immune system and has caused the worldwide epidemic of acquired immunodeficiency syndrome (AIDS). In 1983, HIV was first isolated from AIDS patients and was later identified as the infectious agent responsible for the disease. Other reports indicated the presence of this agent in many populations, including healthy individuals. HIV is cytotoxic to CD4þ cells and does not transform them into malignant cells characteristic of HTLV infection. This infection of CD4 þ cells appeared to contribute to the dramatic decrease in CD4þ cell numbers observed in AIDS patients. HIV can be transmitted to individuals through sexual contact, blood products, or via mother-to-child transmission before or shortly after birth. Consistent with this observation, the levels of infectious HIV virions are substantially higher in the peripheral blood and genital fluids of individuals with the greatest chance of transmitting the virus.

Stages of HIV Infection The typical course of HIV infection begins with an acute infection characterized by the presence of increasing levels of infectious viral RNA in plasma and a transient decrease in CD4 þ T lymphocytes. Within weeks to a few months, the amount of virus detected in the peripheral blood decreases as a result of immune responses, while the CD4 þ T lymphocyte count returns to a near-normal level. The initial infection is often associated with a flu-like syndrome that occurs within weeks and is followed by a quiescent period characterized by a healthy clinical state. The length of this asymptomatic latent period is influenced by environmental, genetic, and immunologic factors as well as by the predominant virus type. In the final stages of the infection, which takes place after an average of 10 years, the viral load increases in the peripheral blood and lymphoid tissues. This phenomenon is accompanied by a major decrease in CD4þ cell numbers and the occurrence of opportunistic infection or cancer leading to a fatal outcome.

Genetic Structure and Products HIV-1 and HIV-2 are grouped within the lentivirus genus as members of the Retroviridae family. The two types of AIDS viruses, HIV-1 and HIV-2, share up to 50% overall nucleotide homology and are similar in structure and cellular tropism. Based on virus genetic similarities, four major groups of HIV-1 (M, N, O, P) have been identified. Each group differs genetically by up to 30%. Within the HIV-1 group M there are at least 9 subtypes or clades that vary in sequence by about 15%. Most HIV infections worldwide involve HIV-1 clade C. For HIV-2, there are 2 groups and no clades. Most evidence suggests that the difference in pathogenicity of these various HIV isolates is determined by host specific factors (intrinsic or immune) rather than the particularly virus isolate. The genomes of HIV-1 and HIV-2 consist of two identical single RNA strands enclosed within a cone-shaped core, surrounded by a glycosylated outer virion surface (Fig. 1). The 9.8-kb RNA genome contains a 50 cap (Gppp), a 30 poly(A) tail, and several open reading frames (ORFs). This genome codes for 15 major molecular products (Table 1). The longest ORF encodes for viral structural proteins, whereas smaller ORFs encode regulators of viral binding, fusion, replication, and assembly.

Structural Proteins Env The Env structural polyprotein, gp160, is produced from a long ORF transcript. Gp160 is cleaved into gp120 and gp41 in the endoplasmic reticulum. These proteins are highly glycosylated and can be expressed on either the cell or the virion surface. Interaction of the viral gp120 with the cellular receptor CD4 causes a conformational change that facilitates virus binding and entry. The gp120 protein forms trimeric structures that are anchored to the membrane of the virion by the viral transmembrane protein gp41. The

Fig. 1 Structure of the HIV virion. The HIV RNA genome is encapsulated in a capsid (CA) composed of p24 and is stabilized by a matrix protein, MA. The cellular lipid bilayer containing envelope glycoproteins completes the outer structure. Several proteins located inside the capsid are required for maturation and infectivity of HIV.

220

HIV (Human Immunodeficiency Virus) Table 1

HIV proteins and their functions

Protein

Size (kDa)

Function

Gag

Tat Rev Nef Vif

p24 p17 p6 p7 p66, p51 p10 p32 gp120 gp41 p14 p19 p27 p23

Vpr Vpu Vpx

p15 p16 p15

Capsid (CA) structural protein Matrix (MA) protein-myristoylated Role in budding Nucleocapsid (NC) protein; helps in reverse transcription Reverse transcriptase (RT); RNase H–inside core Posttranslational processing of viral proteins Viral cDNA integration Envelope surface (SU) protein Envelope transmembrane (TM) protein Transactivation Regulation of viral mRNA expression Pleiotropic, can increase or decrease virus replication Increases virus infectivity and cell-to-cell transmission; helps in proviral DNA synthesis and/or in virion assembly Helps in virus replication; transactivation Helps in virus release; disrupts gp160–CD4 complexes Helps in infectivity

Polymerase (Pol) Protease (PR) Integrase (IN) Envelope

Adapted from Levy, J. A. (2007). HIV and the pathogenesis of AIDS (3rd ed.). Washington, DC: ASM Press.

gp120 surface protein has several loops highly variable in amino acid sequence (V1–V5) among viral strains. These regions primarily regulate the efficiency of virus entry and cellular tropism. The envelope also contains five constant regions (C1–C5) interspersed within its structure. Mutations in the V3 loop can decrease or alter infectivity and the fusion capabilities of HIV. One of the conserved regions of gp120, C4, has a complex folded structure that, if mutated, can affect conformational changes and thus the efficiency of virus binding to CD4. The C2 region of gp120 is important for interactions with CD4 and with gp41, to which gp120 binds by hydrophobic interaction. The gp41 protein mediates fusion and, along with gp120, cytotoxicity (see Section IVB).

PR, RT, and IN Another long viral ORF generates precursor Gag and Gag-Pol transcripts, which can be transported to the cytoplasm for translation by free ribosomes into polyproteins. A ribosomal frameshifting event produces the Gag-Pol polyprotein Pr160Gag-Pol that can be proteolytically processed by the viral protease, PR. The autocatalysis of PR, a product of Pol, is activated in the presence of high Pr160Gag-Pol protein concentrations. The PR-mediated cleavage of Pr55gag and Pr160Gag-Pol, during assembly of the virion, is necessary for the generation of infectious particles. In addition, PR processes the reverse transcriptase (RT) and integrase (IN) proteins from Pol and matrix (MA), p24, p9, and p6 proteins from Gag. Reverse transcriptase generates the complementary strand from the RNA genome producing the provirus cDNA form. Because RT is colocalized with the RNA genome in the capsid, some reverse transcription can occur in the viral particle before infection of the target cell. However, the completion of provirus production requires infection of the host cell and uncoating of the virion. Within the capsid, viral IN proteins are found along with MA and the accessory viral protein, Vpr. IN is required for integration of the provirus into the host genome.

MA, p24, p9, and p6 The Gag precursor protein Pr55gag is responsible for generating structural proteins required for virus assembly. Due to myristoylation of Pr55gag, the protein can be targeted to the plasma membrane for the regulation of viral assembly at the cell surface. Proteolytic processing of Pr55gag by PR generates MA, p24, p9, and p6. MA is a myristoylated protein (p17) that forms a matrix assuring the structural integrity of the virion. The p24 protein is the major structural component of the cone-shaped core or capsid (CA) and can influence infectivity of the virion through interaction with cyclophilin A. The p9 protein is an RNA-binding protein in the core that protects viral RNA from nucleases and transports full-length viral RNA to the assembly complex. The p6 protein can assist in budding of the virion and affect infectivity.

Regulatory Proteins Tat and Rev The smaller ORFs generate viral regulatory and accessory proteins. These viral gene products, Tat, Rev, Nef, Vpu, Vpr, and Vif, can be spliced in the nucleus by cellular spliceosomes and transported to the cytoplasm for translation by membrane-bound ribosomes. Tat regulates viral transcription through interactions with specific cellular factors and TAR, the transactivation response element in the 50 stem–loop region of the HIV RNA genome. Tat can also enhance, but is not required for, virus replication. The Rev protein regulates nuclear export of transcripts by interacting with a Rev responsive element (RRE) located in the viral env-coding region. Rev controls the ratio of unspliced to spliced transcripts, which is important for initiating virus assembly.

HIV (Human Immunodeficiency Virus)

221

Accessory Proteins Vif, Vpu, Vpr, Nef In addition to regulatory proteins, several accessory proteins have important functions in the HIV replicative cycle. Vif increases virus infectivity, and Vpu, found in HIV-1 only, mediates downregulation of the CD4 molecule and enhances infectious HIV virion release from the cell surface. Vpr, a nuclear protein found in most HIV subtypes, regulates transport of the preintegration complex. Both Vpr and Vpx, a cytoplasmic protein found only in HIV-2 and SIV, affect assembly and budding, as well as enhance viral replication. Nef is a pleiotropic protein that can positively regulate virus replication in quiescent cells and mediate pathogenicity. Nef appears to affect T-cell signaling activity, resulting in altered or lower levels of cytokines, particularly IL-2. The role for Nef in the regulation of viral infectivity and pathogenesis is supported by the recovery of Nef mutants of HIV-1 from infected individuals with no evidence of disease progression. However, the full mechanism of Nef activity remains unclear.

HIV Replication Cycle The infection cycle of HIV can be divided into four major stages: binding and entry, reverse transcription and integration, transcription and translation, and assembly and budding (Fig. 2). (Step 1) In the binding and entry stage, the HIV envelope protein, gp120, has a high-affinity interaction with the cellular surface protein, CD4, and specific cellular coreceptors, which facilitate fusion with the cellular membrane. Once inside the cell, viral CA protein p24 and cyclophilins mediate the uncoating of the HIV core and release of the RNA genome. (Step 2) In association with several structural proteins, the RNA genome is primed by viral tRNA-lysine and is reverse transcribed into a double-stranded cDNA. The MA and Vpr viral proteins target the preintegration complex, composed of viral cDNA and MA, Vpr, and IN, to the nucleus. Integration of the provirus occurs randomly in the cellular genome and is regulated by IN, which is capable of cleaving chromosomal DNA and mediating ligation of the provirus.

Fig. 2 Steps in the HIV replication cycle. HIV infection proceeds through several stages involving binding, entry, reverse transcription, integration, transcription, translation, assembly, and budding. (A) The HIV gp120 envelope protein binds CD4 on the cellular surface and specific chemokine receptors facilitate fusion with the cellular membrane. (B) Entry involves uncoating of the virion, releasing the capsid containing the viral RNA genome and viral proteins into the cytoplasm. (C) Virally encoded reverse transcriptase generates the RNA genome into double-stranded cDNA. (D) MA, IN, and Vpr mediate nuclear transport of the cDNA preintegration complex, and IN completes integration of the HIV cDNA into random sites of the host genome. (E) Transcription of viral genes is initiated at the 50 LTR and requires Tat for optimal transcription. Splicing occurs in the nucleus, and the export of long viral transcripts is mediated by Rev. (F) Translation of viral transcripts in the cytoplasm produces viral products that assist in (G) assembly of the virion at the cell surface and the subsequent (H) budding from the cellular surface.

222

HIV (Human Immunodeficiency Virus)

The HIV proviral cDNA is flanked by a 50 LTR and 30 LTR that regulate initiation of transcription and RNA termination, respectively, of regulatory enzymatic and structural genes. (Step 3) The 50 LTR is composed of several regulatory sites for the transcription of viral genes by RNA polymerase II. NF-kB, Sp1, and TATA binding protein (TBP) are among some of the cellular factors that bind upstream of the 50 LTR and assist in initiating viral transcription. The viral regulatory protein Tat is essential for increasing the rate of transcription. The regulation of splicing events, directed by the Rev protein, allows for the proper translation of many of the viral RNAs. Unspliced viral RNA can incorporate, at the cell membrane, into virus capsids that subsequently bud from the cell surface, incorporating the virus envelope expressed on the cell surface. (Step 4) During the budding process, proteolytic processing (mediated by the viral protease) brings about the maturation of the viral particle into an infectious virion.

Cellular Binding and Tropism Cellular determinants of the host range include receptors for HIV binding and entry, such as the CD4 surface molecule, chemokine coreceptors, galactosyl ceramide, complement, and Fc receptors used when virus is complexed with antibody. However, high levels of productive infection are predominantly observed in CD4 þ hematopoietic cells. The CD4 surface protein expressed by mononuclear phagocytes and helper T lymphocytes contains region D1 important for viral entry and cellular tropism. This CD4 molecule provides a docking site for, and induces a conformational change in, the envelope protein gp120, which exposes a site(s) for virus binding to coreceptors on the cell and fusion with the cell.

HIV: Cell Interactions Coreceptors are necessary for efficient HIV virion attachment that leads to viral fusion and entry. Two such receptors are the G protein-coupled seven transmembrane chemokine receptors CXCR-4 and CCR5. Chemokine receptors have important roles in the trafficking of T lymphocytes and macrophages. In HIV infection, chemokine receptors are essential for HIV entry after initial binding to the CD4 surface molecule and are determinants of cellular tropism. A morphologic hallmark of productive HIV infection in cultured cells is the presence of syncytia, a fusion of infected cells resulting in multinucleated cellular complexes, and subsequent cell death. All HIV isolates can replicate in CD4þ T lymphocytes and some are more cytopathic than others, as observed by the rapid formation of syncytia. Those HIV strains that replicate in established T-cell lines, as well as CD4þ lymphocytes, are the syncytia-inducing (SI) strains and utilize the CXCR-4 coreceptor, expressed on T lymphocytes. In contrast, non-syncytia-inducing (NSI) strains that also infect CD4þ lymphocytes grow well in macrophages using the CCR5 coreceptor. This receptor is expressed on monocytes, dendritic cells, T cells, and microglia. Additionally, some isolates are dual-tropic and can infect both cell types. Therefore, the diversity of HIV isolates and coreceptor expression can affect cellular tropism. Evidence has shown that SDF-1, the natural ligand to CXCR-4, and the b-chemokines that bind to CCR-5 can suppress HIV replication in cell culture by blocking entry. These findings further support the importance of chemokine receptors in mediating HIV infection. Moreover, at the genetic level, homozygosity for a CCR5 allele with a 32-bp deletion, which affects CCR5 surface expression, confers resistance in some HIV-infected individuals to the NSI virus strains. HIV can infect CD4 negative cells, although with less replication capacity, through alternate receptors. In brain cells and bowel epithelium, gp120 interacts with surface galactosyl ceramide at a site distinct from its CD4 receptor-binding region. Another mode of entry may involve Fc and complement receptors. These receptors can interact with the Fc portion of nonneutralizing antibodies that attach to HIV and can transport the virus into the cell.

Role of HIV Envelope Protein HIV envelope proteins act as viral determinants of cellular binding and host range. Both gp120 and gp41 appear important in regulating the rate and efficiency of virus attachment and entry through defined conformational changes or glycosylation patterns. Specific regions in the V3 loop also influence the pathogenicity or syncytia-inducing capability of the virion. The V3 loop is particularly important for determining the host range. Additionally, cleavage of gp120 by cellular proteases may be an essential step for viral entry and facilitate gp41 fusion with the cellular membrane.

Virus Replication and Cytopathicity The difference in virus production and cytopathogenicity in T cells and macrophages may correlate with the state of activation. Quiescent T lymphocytes expressing CD4 on the surface are very susceptible to HIV infection. However, until activated, they cannot elicit a productive infection or permit integration of viral DNA. In nondividing CD4 þ macrophages, HIV infection can be productive with integration of viral DNA. Levels of virus replication are generally lower than in CD4 þ lymphocytes but some NSI strains can be produced to high levels in macrophages. In some CD4 negative cells, HIV can enter but generally replicates to low levels. Differences in the intracellular factors influencing transcription, reverse transcription and virus production may account for the incapacity of some cell types to produce substantial amounts of virus (see below). In addition, virus entry or envelope processing can vary among cell types and affect the extent of viral replication.

HIV (Human Immunodeficiency Virus)

223

Cytopathic Effects The loss of CD4þ cells could be due to a variety of mechanisms, including syncytia formation, antibody-dependent cellular cytotoxicity, gp120-induced apoptosis, cytotoxic T lymphocyte (CTL) lysis of infected cells, or death of uninfected bystander cells by apoptosis, or necrosis. Syncytia formation (or cell:cell fusion) is followed by membrane damage, leading to “balloon cell” degeneration, or necrosis. As noted earlier, SI viruses are generally more cytopathic than NSI viruses, and the appearance of SI isolates in infected individuals strongly correlates with progression to AIDS. Regions on CD4 and gp120 can regulate syncytia formation and, therefore, toxicity to CD4þ cells. Programmed cell death, or apoptosis, in response to virus infection, appears to account in part for the decline in CD4þ cell numbers in individuals as they progress to disease. This process is increased with inflammation and cell activation. Direct infection of the cells is not needed to observe this effect. Many reports suggest that alterations in cytokine production can influence the rate of cell death by inhibiting survival cytokines (e.g., IL-2) or by directly inducing the death of uninfected bystander lymphocytes as well as infected cells. However, the pathway(s) leading to apoptosis in vivo is still unclear.

Cellular Latency Cellular latency occurs following HIV provirus integration and is characterized by minimal transcriptional or translational activity of viral genes. Virus activation from a latent state is often the result of stimulation by mitogens, cytokines, or DNA-damaging agents. The regulation of viral latency remains elusive. Some of the cellular factors recruited to the LTR for the active transcription of viral genes include NF-kB, Sp-1, and TBP However, Rev, Tat, Vpu, and Nef have been implicated in mediating latency through their interactions with the LTR and replication cycle. Further understanding of the factors preventing induction of and reactivation from viral latency would be helpful for developing approaches to inhibit progression to disease. Activation of HIV from a latent state is a recent approach at bringing about a cure (see below).

Host Immune Responses to HIV Infection The effect of HIV on the immune system heavily influences the pathogenesis of the infection. Some of the differences observed in immune responses to HIV infection can be associated with small genetic changes in the virus or differences in the host genome. Certain HLA genotypes (e.g., HLA-B57 B27 and B51) have been associated with long term survival and slow disease progression. Also HLA-KIR interactions can influence the natural killer (NK) cell anti-HIV activity. A humoral response to the initial acute infection is reflected by the production by B cells of anti-HIV antibodies. As a host response to HIV infection, some of these antibodies are able to neutralize the virus, generally blocking the epitopes responsible for binding and/or fusion of gp120 or gp41 to the target cell. Antibody-dependent cellular cytotoxicity (ADCC) is another host humoral antiviral response that targets specific HIV epitopes for destruction by interactions of antibody-coated infected cells with effector cells such as NK cells, neutrophils, or macrophages. In contrast, antibody-dependent enhancement (ADE) can potentiate HIV infection through complement or Fc receptors. A decreased concentration of inhibitors of complement in infected individuals suggests that the antibody-independent complement cascade pathway can be activated to lyse either the virus or the virus-infected cell. This and other innate immune responses may play a role in controlling HIV infection.

Natural Killer Cells A component of the innate immune system, NK cells, can directly eliminate virus-infected cells. The cytotoxic mechanism is not MHC-dependent and potentially involves ADCC. In individuals progressing to disease, a loss in NK cell number and function, partly reflecting direct infection of these cells by HIV, alters the production of proinflammatory cytokines such as interferon-g. A decrease in interferon-g production can also interfere with CD8þ cell function.

CD4D T Lymphocytes The CD4 þ cell number and proliferative responses are dramatically reduced during disease progression. Their loss is caused by several processes including direct infection, apoptosis, and immune cell killing. Dysfunctional T helper cells and a loss of CD4þ cells are characteristically observed during infection. Pronounced deficits in CD4þ cell function are manifested by a decreased proliferative response first to recall antigens, then alloantigens, and, in later stages, mitogenic stimulation. Lower levels of IL-2 production in addition to decreased IL-2 receptor expression are observed with progression to disease. Generally, CD4 þ T helper cells produce higher levels of type 2 cytokines (IL-4, IL-5, IL-13) than type 1 cytokines (e.g., IL-2) during the course of symptomatic infection.

CD8D T Lymphocytes Unlike NK cells, CD8 þ cells with cytotoxic function require MHC restriction and are antigen-specific. These cytotoxic T lymphocytes (CTL) eliminate virus-infected cells through perforin or Fas/FasL-mediated pathways. CD8 þ T lymphocytes are critical for

224

HIV (Human Immunodeficiency Virus)

controlling the dissemination of HIV in stages early in infection and before the appearance of antibodies. Importantly, an increase of CD8 þ cells is observed during the asymptomatic stage of the infection, reflecting a response to infection as well as a compensatory reaction to the loss of CD4 þ T helper cells. Both transient and long-lived CTL populations, specific for viral proteins, such as Env, Nef, RT, and Gag gene products, are found in vivo. However, during the progression to disease, a decrease in the frequency of HIV-specific CTLs has been observed along with a decline in cytotoxic responses. Proposed mechanisms for the loss of function include the generation of viral mutants that escape CTL recognition, downregulation of MHC I molecules (potentially through viral tat or nef gene products), and decreased T helper cell function. CD8 þ cells can also exhibit noncytotoxic, non-MHC-dependent anti-HIV activity. This innate CD8 þ cell noncytotoxic antiviral response (CNAR), which is not HIV isolate-specific, acts on all strains of HIV-1 and HIV-2, and is mediated, at least in part, by a factor secreted by CD8 þ cells. This CD8 þ cell antiviral factor (CAF) decreases HIV replication in CD4 þ cells and in macrophages by acting on the HIV LTR to block virus transcription. Lower levels of viral RNA and protein are observed in infected cells exposed to CAF. The activity of CAF is distinct from chemokines and other known cytokines; its identity remains unknown. CNAR activity correlates directly with a healthy clinical state; CD8 þ cells from asymptomatic individuals exhibit a greater capacity to suppress HIV replication than individuals progressing to disease. People infected for more than 10 years who are healthy and without therapy have a strong CNAR. Thus, this host innate immune response appears to be very important in controlling HIV infection.

Dendritic Cells Dendritic cells and other antigen-presenting cells (APC) migrate in the peripheral blood and line mucosal surfaces. They are necessary, after interacting with infectious virus, for triggering subsets of antiviral effector cells. The replication of virus in these cell types is relatively low and stable and therefore they can remain for long periods of time as reservoirs releasing virus in the body. Because certain dendritic cells are believed to harbor virus, they can readily transfer HIV to T lymphocytes and contribute to the loss of CD4 þ cells. Many of the functions of APCs, including presentation capabilities and cytokine regulation for the differentiation, survival, and maintenance of memory and naive T cells, are compromised in HIV-infected individuals. Because dendritic cells are among the first cell subset to encounter infectious HIV and are capable of modulating a broad range of lymphocyte functions, a further understanding of their role in HIV pathogenesis is of particular importance.

Intrinsic Anti-HIV Cellular Factors When HIV infects a cell there can be differences in the extent of virus replication depending on intracellular intrinsic factors that have been identified. These include APOBEC3 G/F, Trim 5a, LD2, tetherin, MURR-1, and SAMHD1. APOBEC, a cytidine deaminase, causes mutations in the HIV genome affecting virus replication. The viral protein Vif can counter this anti-viral response by inducing APOBEC degradation. Trim 5a, a cytoplasmic protein, prevents the uncoating of the virus by binding to the capsid. LD2 has a similar effect in blocking HIV-2 replication after entry. Tetherin, a membrane protein, prevents budding of virus particles brought to the cell surface and can be countered by HIV Vpu. MURR-1 can affect viral transcription. Finally SAMHD1 blocks proviral synthesis during reverse transcription, particularly in monocytes and dendritic cells. This antiviral intrinsic effect can be countered by HIV Vpx which induces proteolytic degradation.

Cancers Associated with HIV Infection HIV-infected patients and recipients of organ transplants show an increased prevalence of cancer particularly those associated with a viral infection (Table 2). As seen in transplant patients, HIV infection leads to an immune deficiency that can decrease immune responses against cancer or increase immune responses associated with inflammation that are associated with cancer development (Fig. 3). It is estimated that about 40% of cancers in HIV-infected subjects can be attributed to viral infections. Herpes and papilloma viruses are responsible for cancers that were initially linked to the diagnosis of AIDS. The mechanism for the induction of malignancy involves initiation by the virus, promotion of cell proliferation, and cytokine production that leads to cellular karyotypic and genomic mutational events. Obviously the current vaccines against known viruses such as Hepatitis B and HPV should reduce the prevalence of these cancers.

Anti-HIV Approaches Current therapeutic approaches to block HIV replication involve targeting the three viral enzymes reverse transcriptase, integrase and protease. These treatments have decreased plasma viral loads dramatically and increased the life expectancy for symptomatic individuals infected with HIV. Recent combination therapy approaches (using three drugs) have given infected people the possibility of living a near normal lifetime, although some treated subjects develop resistance to the antiviral drugs or toxic side effects

HIV (Human Immunodeficiency Virus) Table 2

Cancers associated with HIV infection

Cancer AIDS-defining cancers Kaposi’s sarcoma Non-Hodgkin’s lymphoma Invasive cervical cancer Non-AIDS-defining cancers Anal Cancer Hodgkin’s disease (Hodgkin’s lymphoma) Adult T cell leukemia/lymphoma Melanoma skin cancer Liver cancer Lung cancer Mouth and throat cancers Testicular cancer Squamous cell and basal cell skin cancers

Fig. 3

225

Virus KSHV (HHV8) EBV HPV HPV EBV HTLV Hep B; Hep C HPV HPV?

HIV Infection and Cancer Development.

affecting tissues such as the kidney, liver, and pancreas. Most recently, combination HIV therapy can be delivered as one pill taken daily. That has reduced resistance by helping increase adherence. Also, because HIV infection often leads to the development of Bcell lymphomas and several central nervous system disorders, the extended life expectancies as a result of therapeutics could increase the prevalence of HIV-associated cancers and neuropathies. With the success of antiviral drugs, many in the HIV field are advocating therapy for all infected individuals. Reducing the viral load by using anti-viral drugs can also markedly decrease the transmission rate. Thus the aim in some communities is to establish universal anti-HIV drug administration that could reduce transmission rates and overall prevalence of the infection. In addition, the success of antiretroviral therapy (ART) has led to strategies to prevent HIV infection by preexposure therapy and to effect a cure. In the latter case, the drugs are used in attempts to rid the body of HIV completely or reduce HIV presence to a state in which there are no signs of the infection. Attempts at establishing a cure involve activation of HIV from latent cellular reservoirs so that the immune system can eliminate them, gene therapy in which the CCR 5 gene and other host genes required for HIV infection are deleted, as well as approaches to boost the immune system to control virus-infected cells and prevent HIV spread in the host. Short-term advances toward a cure have been reported but elimination of the virus that integrates into several different cells in the body presents a notable challenge. Many researchers are suggesting that combination antiviral therapies and immune enhancing drugs could lead to a “functional” cure in which ART will not be needed for a lifetime. The host immune system would control the virus as seen in healthy long-term infected individuals who have no detectable circulating virus or very low viraemia. Innovative approaches at enhancing antiviral immunologic responses can offer advantages to individuals infected with HIV, as the immune system can recognize a variety of viral subtypes and control infection in many different tissues of the body (e.g., bowel, brain). Finally, the development of vaccines using antigenic components of HIV delivered through a variety of different modalities and with different adjuvants will be necessary for the prevention of infection. HIV is very variable and can be transmitted by infected cells as well as by free virus. The approaches needed for inhibiting all HIV isolates and for blocking an incoming infected cell, therefore, raise major challenges to the development of an effective vaccine. It is hoped that by studying the cellular, molecular, and biochemical pathways involved in HIV replication and the effective immune responses that can control the virus, particularly in long-term survivors, optimal therapies and strategies for the prevention of HIV and AIDS can be developed.

226

HIV (Human Immunodeficiency Virus)

See also: Non-Hodgkin Lymphoma: Diagnosis and Treatment.

Further Reading Allers, K., Hütter, G., Hofmann, J., Loddenkemper, C., Rieger, K., Thiel, E., Schneider, T., 2011. Evidence for the cure of HIV infection by CCR5D32/D32 stem cell transplantation. Blood 117 (10), 2791–2799. Altfeld, M., Fadda, L., Frleta, D., Bhardwaj, N., 2011. DCs and NK cells: Critical effectors in the immune response to HIV-1. Nature Reviews. Immunology 11 (3), 176–186. Barré-Sinoussi, F., Chermann, J.C., Rey, F., Nugeyre, M.T., Chamaret, S., Gruest, J., Dauguet, C., Axler-Blin, C., Vézinet-Brun, F., Rouzioux, C., Rozenbaum, W., Montagnier, L., 1983. Isolation of a T-lymphotropic retrovirus from a patient at risk for acquired immune deficiency syndrome (AIDS). Science 220 (4599), 868–871. Berger, E.A., Murphy, P.M., Farber, J.M., 1999. Chemokine receptors as HIV-1 coreceptors: Roles in viral entry, tropism, and disease. Annual Review of Immunology 17, 657–700. Cheng-Mayer, C., Seto, D., Tateno, M., Levy, J.A., 1988. Biologic features of HIV-1 that correlate with virulence in the host. Science 240 (4848), 80–82. Clerici, M., Stocks, N.I., Zajac, R.A., Boswell, R.N., Lucey, D.R., Via, C.S., Shearer, G.M., 1989. Detection of three distinct patterns of T helper cell dysfunction in asymptomatic, human immunodeficiency virus-seropositive patients. Independence of CD4þ cell numbers and clinical staging. The Journal of Clinical Investigation 84 (6), 1892–1899. Cohen, M.S., Smith, M.K., Muessig, K.E., Hallett, T.B., Powers, K.A., Kashuba, A.D., 2013. Antiretroviral treatment of HIV-1 prevents transmission of HIV-1: Where do we go from here? Lancet 382 (9903), 1515–1524. Deeks, S.G., Tracy, R., Douek, D.C., 2013. Systemic effects of inflammation on health during chronic HIV infection. Immunity 39 (4), 633–645. Frankel, A.D., Young, J.A.T., 1998. HIV-1: Fifteen proteins and an RNA. Annual Review of Biochemistry 67, 1–25. Review. Grant, R.M., Lama, J.R., Anderson, P.L., McMahan, V., Liu, A.Y., Vargas, L., Goicochea, P., Casapía, M., Guanira-Carranza, J.V., Ramirez-Cardich, M.E., Montoya-Herrera, O., Fernández, T., Veloso, V.G., Buchbinder, S.P., Chariyalertsak, S., Schechter, M., Bekker, L.G., Mayer, K.H., Kallás, E.G., Amico, K.R., Mulligan, K., Bushman, L.R., Hance, R.J., Ganoza, C., Defechereux, P., Postle, B., Wang, F., McConnell, J.J., Zheng, J.H., Lee, J., Rooney, J.F., Jaffe, H.S., Martinez, A.I., Burns, D.N., Glidden, D.V., iPrEx Study Team, 2010. Preexposure chemoprophylaxis for HIV prevention in men who have sex with men. The New England Journal of Medicine 363 (27), 2587–2599. Homsy, J., Meyer, M., Tateno, M., Clarkson, S., Levy, J.A., 1989. The Fc and not CD4 receptor mediates antibody enhancement of HIV infection in human cells. Science 244 (4910), 1357–1360. Jones, K.A., Peterlin, B.M., 1994. Control of RNA initiation and elongation at the HIV-1 promoter. Annual Review of Biochemistry 63, 717–743. Killian, M.S., Levy, J.A., 2011. HIV/AIDS: 30 years of progress and future challenges. European Journal of Immunology 41 (12), 3401–3411. Levy, J.A., 2015. Dispelling myths and focusing on notable concepts in HIV pathogenesis. Trends in Molecular Medicine 21 (6), 341–353. Levy, J.A., 2007. HIV and the pathogenesis of AIDS, 3rd ed. ASM Press, Washington, DC. Levy, J.A., Scott, I., Mackewicz, C., 2003. Protection from HIV/AIDS: The importance of innate immunity. Clinical Immunology 108 (3), 167–174. Levy, J.A., Hoffman, A.D., Kramer, S.M., Landis, J.A., Shimabukuro, J.M., Oshiro, L.S., 1984. Isolation of lymphocytopathic retroviruses from San Francisco patients with AIDS. Science 225 (4664), 840–842. Maartens, G., Celum, C., Lewin, S.R., 2014. HIV infection: Epidemiology, pathogenesis, treatment, and prevention. Lancet 384 (9939), 258–271. Margolis, D.M., 2010. Mechanisms of HIV latency: An emerging picture of complexity. Current HIV/AIDS Reports 7 (1), 37–43. Neil, S., Bieniasz, P., 2009. Human immunodeficiency virus, restriction factors, and interferon. Journal of Interferon & Cytokine Research 29 (9), 569–580. Silverberg, M.J., Chao, C., Leyden, W.A., Xu, L., Horberg, M.A., Klein, D., Towner, W.J., Dubrow, R., Quesenberry Jr., C.P., Neugebauer, R.S., Abrams, D.I., 2011. HIV infection, immunodeficiency, viral replication, and the risk of cancer. Cancer Epidemiology, Biomarkers & Prevention 20 (12), 2551–2559.

Hodgkin Lymphoma: Pathology and Genetics Stefano A Pileri, European Institute of Oncology, Milan, Italy; and Bologna University School of Medicine, Bologna, Italy Stefano Fiori, European Institute of Oncology, Milan, Italy Vincenzo De Iuliis, Gabriele D’Annunzio University of Chieti, Chieti, Italy Valentina Tabanelli, European Institute of Oncology, Milan, Italy © 2019 Elsevier Inc. All rights reserved.

Glossary The HUGO (Human Genome Organization) Committee nomenclature is used throughout: genes are quoted in italics and capital letters, except for those encoding for immunoglobulins (IG) or the variable region of their heavy chains (IGVH) and Tcell receptor (TR), which are reported in ordinary capital letters. The gene products are quoted in ordinary capital letters. The cluster of differentiation (CD) nomenclature of human leukocyte antigens is used through the text. The terminology of the Revised 4th Edition of the World Health Organization: WHO Classification of tumors of hematopoietic and lymphoid tissues is used throughout.

Abbreviations ABVD Adriamycin, bleomycin, vinblastine, dacarbazine ALK Anaplastic large cell lymphoma kinase CD Cluster of differentiation EBER Epstein–Barr encoding region HIV Human immunodeficiency virus MW Molecular weight OS Overall survival PCR Polymerase chain reaction PET Positron emission tomography (R-)CHOP (Rituximab-)cyclophosphamide, hydroxydaunorubicin, oncovin, prednisone WHO World Health Organization

Definition Hodgkin lymphoma (HL) is usually characterized by the presence of a low number of neoplastic cells, either multinucleated (Reed– Sternberg cells, RSCs) or mononucleated (Hodgkin cells and variants, HCs), which are comprised in an overwhelming inflammatory background with variable cytological composition. RSCs, which measure about 60 mm in diameter, display a large rim of cytoplasm and two or more huge nuclei with central, inclusion-like nucleoli, often acidophilic at hematoxylin and eosin (H&E) (Fig. 1). HCs differ from RSCs by the presence of a single nucleus.

Historical Annotations The tumor was first described by Sir Thomas Hodgkin in 1832 and subsequently named “Hodgkin’s Disease” by Samuel Wilks. The histogenesis of the condition was a matter of debate for several decades, until the use of single cell micro-dissection allowed establishment of the monoclonal B-cell nature of the neoplastic cells. As a result of this finding the term “Hodgkin’s disease” was abandoned in favor of “Hodgkin lymphoma,” with as implication its inclusion among lymphoid neoplasms. On histological grounds, the Lukes and Butler classification became the reference in the mid-1960s, by distinguishing four varieties of HL (lymphocyte predominant, nodular sclerosing, mixed cellularity, and lymphocyte depleted), mainly based on the composition of the microenvironment. This classification still represents the basis for the diagnosis of HL, although some refinements have been introduced meanwhile. Nowadays, a major distinction is made between nodular lymphocyte predominant HL (NLPHL) and classical HL (CHL), each characterized by particular clinical, morphologic, immunophenotypic and molecular features. Accordingly, the Revised Edition of the WHO Classification has abandoned the unifying term “Hodgkin lymphoma” in favor of “Hodgkin lymphomas.” CHL is further subdivided into four types: lymphocyte-rich (LR), nodular sclerosing (NS), mixed cellularity (MC), and lymphocyte depleted (LD). LRCHL was in the past lumped with NLPHL. However, although it shares architectural and microenvironmental features with the latter, it shows the morphologic and immunophenotypic profile of CHL.

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65258-3

227

228

Fig. 1

Hodgkin Lymphoma: Pathology and Genetics

Example of a binucleated Reed–Sternberg cell (center) surrounded by some mononucleated Hodgkin cells.

Epidemiological Notes HL usually affects lymph nodes, except for NSCHL that can develop in the thymus. About 95% of HL cases belong to the classical category, the remaining 5% corresponding to NLPHL. In resource-rich countries, the latter more often occurs in males in the fourth decade of life. It has a rather indolent course with possible late recurrences and a spread that is closer to that of non-Hodgkin lymphomas. NLPHL can be concomitant with or preceded or followed by a peculiar modality of lymph node regression, known as progressively transformed germinal centers (PTGC). The latter, however, is not necessarily associated with NLPHL. CHL shows a bimodal age distribution with peaks in the second and seventh decades of life. It is more common in males, except for NSCHL that is more predominant among females. It has a distinct pattern of extension in the body, starting from a single node or group of adjacent nodes; this pattern represents the rationale for the staging procedures.

Lymphocyte Predominant HL This form of HL has unique morphologic and immunophenotypic features. In 80% of cases it is characterized by the presence of nodules mostly composed of small B-lymphocytes (CD20 þ, CD79aþ, PAX5þ) with some epithelioid elements (CD68 þ/ CD163 þ). Inside the nodules, which tend to expand and coalesce through time, producing a nodular and diffuse growth pattern, there are mononucleated cells measuring about 60 mm in diameter, which represent 1%–2% of the cell population (Fig. 2A–C). They show polylobated nuclei, with dispersed chromatin and multiple small nucleoli at times adjacent to the nuclear membrane, and a narrow rim of slightly basophilic cytoplasm. Because of the nuclear morphology, they were called “pop-corn” cells, a term that has been almost completely abandoned in favor of “lymphocyte predominant” cells (LPCs). RSCs are exceedingly rare and are usually detected only when serial sections are examined. LPCs express a complete B-cell phenotype (CD20 þ, CD75 þ, CD79a þ, PAX5 þ, BOB.1 þ, OCT2 þ) as well as the leukocyte common antigen CD45 anddin most casesdepithelial membrane-antigen (EMA) (Fig. 3A–G). They also stain positive for BCL6 and IRF4 (Fig. 2J and K), which supports their derivation from a cell still residing in the germinal centers along with the presence of a high mutational load of the variable region of the immunoglobulin heavy chain genes (IGVH). In 10%–20% of cases, they express IgD and CD38, but not IgM or CD27 (Fig. 3G). This peculiar pattern is usually observed in young males and is associated with a more favorable course. A meshwork of follicular dendritic cells (FDCs) (CD21 þ, CD23 þ) is detected within the nodules. Importantly, LPCs lack expression of CD30 and CD15, which are characteristic of HRSCs of CHL. Furthermore, LPCs are typically surrounded by rosettes of follicular T-helper (FTH) lymphocytes (CD3 þ, CD279/ PD1 þ, BCL6 þ, CD57 þ) (Fig. 3H and I). A few scattered CD30 þ elements may be observed, which usually correspond to activated B-immunoblasts and only exceptionally have the typical LP morphology. The regular CD20 positivity of LPCs represents the rationale for treatment with Rituximab, either as monotherapy in low stage disease, which may be followed by maintenance therapy, or in combination with CHOP or ABVD in more advanced stages. Interestingly, the administration of Rituximab might prevent disease transformation. Besides this classic nodular, B-cell rich appearance, LPHL can display some variant patterns: serpiginous/interconnected nodular, nodular with prominent extra-nodular LPCs, nodular with T-cell rich background, diffuse (T-cell rich B-cell lymphoma-like), and diffuse “moth-eaten” with B-cell rich background. The latter two variants correspond to the 20% of cases in which no nodular pattern is observed and may be difficult to distinguish from T-cell/histiocyte-rich large B-cell lymphoma (THRLBCL), also due to the absence of FDCs and the similarity in terms of gene expression profile between the two neoplasms. The following elements can help under these circumstances to reach a diagnosis: (1) the clinical presentation, which is aggressive with widespread disease

Hodgkin Lymphoma: Pathology and Genetics

229

Fig. 2 So-called “lymphocyte predominant cells” (LPCs), scattered in a background rich in small lymphocytes (A, inset: nuclear detail of a LPC). Nodular lymphocyte predominant Hodgkin lymphoma may show a nodular (B) or diffuse growth pattern (C). These patterns are frequently coexistent in the same lymph node.

and systemic symptoms in THRLBCL; (2) the presence of CD279/PD1 þ T-cell rosettes around otherwise canonical LPCs; (3) the expression of PU.1 and the negativity for LSP-1 in LPCs, the opposite pattern being observed in the neoplastic elements of THRLBCL; and (4) the presence of a residual nodule with the typical characteristics of NLPHL. Molecular studies have shown that LPCs harbor clonally rearranged IGVH. As mentioned above, the clonal rearrangements can be easily detected in the DNA of single microdissected neoplastic cells, and the variable regions of IGVH carry a high load of somatic mutations indicative of ongoing mutational activity. Epstein-Barr virus (EBV) infection detected by EBER 1/2 in situ hybridization may be found in LPCs in 3%–5% of cases in both children and adults. EBV may be also present in bystander lymphocytes. BCL6 rearrangements (involving IG, IKAROS, ABR and other partner genes) are present in about half NLPHL cases. Aberrant somatic hypermutations are found in 80% of cases, most frequently in PAX5, but also in PIM1, RHOH/TTF, and MYC. Mutations of the genes SGK1, DUSP2, and JUNB are also reported in about half of NLPHL cases by targeted next-generation sequencing (NGS). An increased risk of NLPHL has been noted in some families. PHL has also been identified in patients affected with HermanskyPudlak type 2 syndrome or autoimmune lymphoproliferative syndrome (ALPS) with mutations in FAS. As previously mentioned, nodular and nodular and diffuse forms of NLPHL develop slowly, with frequent late relapses. It usually remains responsive to therapy and thus is rarely fatal, although in an advanced stage and higher age group it may behave more aggressively, as do histopathologic variants characterized by the presence of LPCs outside B-cell nodules or B-cell depletion of the microenvironment or THRLBCL-like transformation. Therefore, it is useful to mention these variant features in the pathology report. Progression to diffuse large B-cell lymphoma (DLBCL) has been observed in approximately 3%–5% of cases. In such cases the neoplastic cells maintain their typical immunophenotype. If localized, DLBCL associated with NLPHL generally has a good prognosis. NLPHL and the associated DLBCL are clonally related. Bone marrow involvement is rare in NLPHL and raises the possibility of THRLBCL, when there are only PD1/CD57-negative T-cells and no small B-cells in the background. Cases of NLPHL with bone-marrow involvement show aggressive clinical behavior. Advanced stage NLPHL responds poorly to the chemotherapy regimens traditionally used for CHL treatment but responds better to R-CHOP or regimens used to treat aggressive B-cell lymphomas.

230

Hodgkin Lymphoma: Pathology and Genetics

Fig. 3 Immunophenotypic features of nodular lymphocyte predominant Hodgkin lymphoma. LPCs are positive for CD20 (A), BSAP/PAX5 (B), CD75 (C), epithelial membrane antigen, EMA (D), BCL6 (E) IRF4/MUM1 (F), and, in a subset of cases, IgD (3G). LPCs are, as a rule, surrounded by small T-lymphocytes, which express CD3 (H) and PD-1 (I), with formation of rosette-like structures.

Classical HL Morphology and Histotypes Classical HL is characterized by the presence of typical HRSCs in a cellular milieu that varies depending on its histotype. Neoplastic cells can at times become smaller due to apoptotic changes: they are accordingly termed “dwarfs.”

Lymphocyte-rich classical Hodgkin lymphoma Lymphocyte-rich classical Hodgkin lymphoma (LRCHL) may proliferate in a nodular or, more rarely, diffuse pattern. The reactive component consists of small lymphocytes (mostly of B-cell origin) and some histiocytes with or without the formation of microgranulomas. In this context, HRSCs are scattered through the nodules or more rarely located at their periphery (Fig. 4A). At times, residual germinal centers are seen within the nodules in an eccentric location. Neutrophils and eosinophils are absent. Based on morphological characteristics, in the past LRCHL was included in the setting of LPHL. With the advent of immunohistochemistry, it became clear that neoplastic cells had the typical phenotype of HRSCs (see below) and not that of LPCs. With the currently

Hodgkin Lymphoma: Pathology and Genetics

231

Fig. 4 Morphologic features of classical Hodgkin lymphoma (CHL). Lymphocyte rich CHL (A) displays scattered Hodgkin and Reed–Sternberg (HRS) cells in a background rich in small lymphocytes; eosinophils and sclerosis are not observed. Nodular sclerosis CHL is characterized by nodules surrounded by fibrous bands departing from a thickened lymph node capsule (B–D). The nodules contain variable amounts of small lymphocytes, eosinophils, macrophages, and plasma cells. HRS cells have a lacunar appearance and may be dispersed (BNLI type I, E) or form cohesive sheets (BNLI type II, F). Mixed cellularity CHL is distinguished by lack of sclerosis, frequent interfollicular growth and scattered classic HRS cells (G). Lymphocyte depletion CHL can show two different morphologic patterns: the fibro-histiocytic variant (H), that is characterized by a predominance of fibroblasts and histiocytes, with scattered HRS cells, and the sarcomatous variant (I) that is extremely rich in HRS cells. In both subtypes, the small lymphocytic component is extremely sparse.

available therapeutic approaches, LRCHL has a more favorable course than the other types of CHL, similar to that of LPHL. On the contrary, however, relapses are exceedingly rare.

Nodular sclerosing classical Hodgkin lymphoma Nodular sclerosing classical Hodgkin lymphoma (NSCHL) represents about 70% of CHL cases in Western Countries, although its prevalence varies in different parts of the World. In contrast to the other types of CHL, it affects females more often than males and presents with a mediastinal mass (with possible features of bulky disease) in 80% of patients. At microscopic examination (Fig. 4B– D), NSCHL is characterized by nodules surrounded by collagen bands (birefringent on polarized light) that depart from a thickened lymph node capsule. The cellular composition of the nodules can significantly vary by mimicking the one of the other types of CHL

232

Hodgkin Lymphoma: Pathology and Genetics

(LR, MC, and LD). Occasionally, only nodule formation is seen with minimal or absent fibrosis: this pattern is called the cellular phase of NSCHL. Neoplastic cells are mostly mononucleated and can acquire a “lacunar” appearance in improperly fixed material, consisting of the retraction of cytoplasm close to the nucleus with some thin connections to the cytoplasmic membrane, forming what look like small lacunae (Fig. 4E). The British National Lymphoma Investigation (BNLI) distinguishes two grades of NSCHL, which are potentially relevant for prognosis. Grade I corresponds to the presence of a small amount of scattered HRSCs in a composite cellular milieu, consisting of variable amounts of small lymphocytes, histiocytes, neutrophils, eosinophils, mastcells, and plasmacells. Grade II is assigned (Fig. 4F) when (1) > 25% of the nodules show sarcomatous lymphocyte depletion, (2) > 80% of the nodules show features of the fibro-histiocytic variant, or (3) > 25% of the nodules show numerous bizarre, anaplasticappearing Hodgkin cells without lymphocyte depletion. Following such criteria, approximately 15%–25% of NSCHL cases are classified as grade II. Grading is not mandatory for clinical purposes but has been applied in some clinical trials. However, it has become less relevant, due to advances in therapy which obscure differences seen in less effectively treated patients. Within the morphologic spectrum of NSCHL, the so-called “syncytial” variant merits to be mentioned. It is characterized by large aggregates of neoplastic cells within the nodules, resembling metastatic involvement by undifferentiated carcinoma. Immunohistochemistry is pivotal for the differential diagnosis as it is in the distinction between grade II NSCHL and ALK-negative anaplastic large cell lymphoma (ALCL) (see below).

Mixed cellularity classical Hodgkin lymphoma In this variant, HRSCs can be easily found comprised in a mixed reactive population composed of lymphocytes (mainly T), histiocytes, plasma cells, neutrophils, and eosinophils (Fig. 4G). The normal lymph node structure is complete effaced, at times with sparing of some follicles with a germinal center. It should be differentiated from nodal peripheral T-cell lymphoma (PTCL), especially of the FTH-cell type, that can contain EBV-positive HRS-like cells. However, T-lymphocytes in mixed cellularity classical Hodgkin lymphoma (MCCHL) show a complete T-cell phenotype and do not homogeneously express FTH-related markers in contrast to those of FTH-PTCL. In addition, they do not display the large rim of clear cytoplasm that is characteristic of nodal FTH-PTCL.

Lymphocyte depleted classical Hodgkin lymphoma Lymphocyte depleted classical Hodgkin lymphoma (LDCHL) includes two morphologic subtypes (Fig. 4H and I). The fibrohistiocytic subtype has a cellular milieu mostly consisting of fibroblasts and histiocytes. The sarcomatous subtype is extremely rich in neoplastic cells, which frequently form a palisade at the periphery of large necrotic areas but can also occur diffusely throughout sinuses. Immunophenotyping is pivotal to differentiate them from ALK-negative ALCL (see below).

Phenotype HRSCs of CHL have a distinctive immunophenotypic profile. They strongly express CD30 in a characteristic dot-like and membrane-bound pattern, corresponding to the synthesis of the protein backbone (90 kDa MW) of the molecule, which is glycosylated in the Golgi apparatus (120 kDa MW) and then translocates to the cell membrane, where it acquires a transmembrane location (Fig. 5A). Ber-H2 is the reference antibody for immunohistochemical identification of CD30 and SGN30 is a humanized monoclonal antibody used for the construction of brentuximab-vedotin by conjugation with monomethyl auristatin E. The epitopes detected by these antibodies are located on the external domain of CD30, but are not shed in the circulation, in contrast to the CD30 epitopes that can be detected in peripheral blood. This is quite important, since CD30 is an important diagnostic and therapeutic target. CD30 is the target of specific immunoconjugates, administered either as monotherapy or in combination with chemotherapy with promising results, both in refractory cases or as first line treatment. In about 60% of cases HRSCs express CD15, also known as X-hapten, with a staining pattern similar to that of CD30 (Fig. 5B). This is one of the many molecules aberrantly expressed in at least 20% of HRSC cases, which include CD15, GATA3, TRAF1, ID2, ABF1, JUN, JUNB, AP-1, FLIP (CFLAR), JAK/STAT, STAT5, and the cytotoxic markers TIA-1, Granzyme B and perforin. It is assumed that these aberrant attributes facilitate the immune escape of neoplastic cells. A main feature of HRSCs is downregulation of the B-cell program and loss of most B-cell markers (see also section “Molecular Characteristics”), in spite of their B-cell derivation. In > 90% of cases, they retain only the expression of the PAX5 gene product, also known as BSAP (B-cell specific activator protein), although at a significantly lower intensity than normal B-cells (Fig. 5C). CD20 is expressed in about 30% of cases by a proportion of neoplastic cells with variable intensity (Fig. 5D). CD79a is detected in about 10% of cases with a staining pattern like CD20. The transcription factors BCL6, BOB.1, OCT2 and PU.1 are usually absent. At times, either BOB.1 or OCT2 may be expressed, which impairs the transcription of IG genes (Fig. 5E). CD45 and EMA are also defective (Fig. 5F and G). IRF4 and Ki-67 are expressed in most if not all cases of HRSC (Fig. 5H and I). Even though they express Ki-67, HRSC enter the cell cycle but do not proceed through due to abortive cytokinesis. BCL2 and TP53 are expressed by > 50% and 25% of neoplastic cells in one third and 12% of cases respectively, which has been found of prognostic relevance (Fig. 6A and B). In 87% of cases, HRSCs express CD274/PD-L1 and CD273/PD-L2 a rationale for treatment with immune check-point inhibitors (e.g., nivolumab and pembrolizumab) (Fig. 6C). Copy number alterations of 9p24.1/PDCD1/LG contribute to robust PD-L1 and PD-L2 expression by HRSCs, although PD-L1 is also strongly expressed by reactive tissue macrophages. The combination of brentuximab-vedotin and nivolumab has recently been tested in relapsed or refractory CHL with promising results (82% overall response rate). In a small percentage of cases HRSCs may express one or more T-cell associated molecules. Intracytoplasmic and globular staining suggests passive absorption, but detection at the plasma

Hodgkin Lymphoma: Pathology and Genetics

233

Fig. 5 Immunophenotypic features of classical Hodgkin lymphoma. HRS cells are positive for CD30 (A), CD15 (B) and, weakly, for BSAP/PAX5 (C). CD20 is negative or weakly expressed by a minority of cells (D). Furthermore, HRS cells are negative for CD45/LCA (E), OCT2 (F), BOB.1 (F inset) and EMA (G). Most cells are positive for IRF4/MUM1 (H) and for the proliferation marker KI-67 (I).

membrane level has been associated with IGVH gene rearrangements with polyclonal T-cell receptor (TR) (see section “Molecular Characteristics”). When a tumor is rich in neoplastic cells (NSCHL grade II or sarcomatous LDCHL), a differential diagnosis of ALK-negative ALCL should be considered. Expression of PAX5, IGVH clonal rearrangement, lack of TR clonality, and possible EBV infection will then favor a diagnosis of CHL. Reactive cells in the microenvironment are important for cross-talk with neoplastic cells through cytokines and cytokinereceptors, and also convey prognostic information. The presence of macrophages (with cut-off values ranging from 5% to 25%), of T-lymphocyte subsets and of myeloid suppressor cells has been found of potential prognostic value, although conflicting data have been reported in the literature. Numerous markers of HRSC or cells in the microenvironment have been proposed as prognostic indicators, and these have been compared with interim-PET for prediction of response to two cycles of ABVD treatment. In patients with a negative interim PET, expression of CD68 (> 25%) and CD279/PD1 (diffuse or rosetting pattern) in microenvironmental cells, and STAT1 negativity in HRSC identify a subset of patients with a significantly lower 3-year progression-free survival (Fig. 6E and F).

234

Hodgkin Lymphoma: Pathology and Genetics

Fig. 6 Biologic prognostic markers in CHL. HRS cells can express BCL2 (A) and p53 (B). Expression above 50% and 25% of neoplastic cells represents an adverse prognostic factor. Positivity for PD1 ligand is found in HRS cells of most cases of CHL (C), and this can have impact on therapy choice. An increased macrophagic population above 25% of cellular background (D), as well as an increase of PD1-positive T-lymphocytes, with rosette formation (E) and down-regulation of STAT1 in HRS cells (F), are considered adverse prognostic factors.

EBV Infection By in situ hybridization for EBER 1/2 and immunohistochemical detection of LMP1 and EBNA1 (latency II pattern) it has been convincingly shown that HRSC are EBV infected. LMP-1 expression is regarded as potentially implicated in the pathogenesis of the tumor because of its in vitro transforming capacity. Conceivably, EBV infection of a B-cell replaces one of the genetic alterations necessary for the development of CHL. The prevalence of EBV in HRSCs varies according to the histological subtype and epidemiologic factors. The highest frequency ( 75%) is found in MCCHL, and the lowest (10%–40%) in NSCHL. In resource-poor regions and in patients infected with HIV, the rate of EBV infection is close to 100%. The type of EBV strain also varies between geographical areas. In resource-rich countries strain 1 prevails, and in resource-poor countries strain 2.

Cell of Origin As previously mentioned, IGH gene rearrangement studies carried out by single cell microdissection provided convincing evidence that HRSCs of CHL are clonal and derived from germinal center B-cells, even if they lack morphologic or phenotypic similarities other than their PAX5/BSAP expression. In rare cases, HRSCs were reported to harbor clonally rearranged TR, indicating that cases with morphological features of CHL might be derived from T-cells. This issue remains unsolved, since differentiating between these exceptional cases and PTCLs, especially ALK-negative ALCL, can be very challenging.

Molecular Characteristics HRSCs reveal clonal IGVH rearrangements in > 98% of cases. As mentioned above, single cell microdissection is required to detect clonal rearrangements, as DNA extracted from the whole tissue specimens provides poor results. The rearranged IGVH of tumor cells

Hodgkin Lymphoma: Pathology and Genetics

235

harbor a high load of somatic mutations (hypermutated), usually without signs of ongoing mutational activity. These findings support the view that in most if not all instances HRSCs are derived from a germinal center B-cell. Nonetheless, HRSCs have lost much of the B-cell specific expression program and have acquired B-cell inappropriate gene products, as described earlier. In addition, deregulated transcription factors in CHL promote proliferation and abrogate apoptosis of the neoplastic cells. The transcription factor NF-kB is constitutively activated in HRSCs, and there is altered activity of the NF-kB target genes, which regulate proliferation and survival, the AP-1 complex and the Janus kinase/signal transducers and activators of transcription (JAK/STAT) signaling pathway. Mutations of the JAK regulator SOC-1 are associated with nuclear STAT5 accumulation in HRSCs, indicating blocking of the negative feedback loop of the JAK/STAT5 pathway. Even though TP53 may be overexpressed, mutations of TP53 are rare or absent in primary CHL. Classical cytogenetic and fluorescence in situ hybridization (FISH) studies show aneuploidy and hypertetraploidy, consistent with the multinucleation of the neoplastic cells; however, these techniques fail to demonstrate recurrent and specific chromosomal changes in CHL. Comparative genomic hybridization, however, shows recurrent gains of sub-regions on chromosomal arms 2p, 9p, and 12q and distinct high-level amplifications on chromosomal bands 4p16, 4q23-q24, and 9p23-p24. Special attention deserves the alterations at 9p24.1, since this region includes the genes encoding for CD274/PD-L1, CD273/PD-L2, and JAK2, which herald sensitivity to specific targeted therapies as highlighted above in the “Phenotype” section. Gene expression profiling (GEP) studies have confirmed all these findings, including global down-regulation of the B-cell program, and highlighted similarities between CHL and primary mediastinal large B-cell lymphoma (PMBL). PBML show overexpression of genes encoding for CD274/PD-L1, CD273/PD-L2 and JAK2, due to 9p24.1 alterations. The similarities between the signature of CHL and that of PMBL are not surprising, which is a reflection of the gray-zone between the two neoplasms. In the Revised WHO Classification, the latter condition for which the term “B-cell lymphoma unclassifiable with features intermediate between CHL and DLBCL” is used, corresponds to two different conditions: a tumor showing CHL morphology but complete B-cell immunophenotype or a tumor with the PMBL morphology but a CHL immunophenotype. By profiling HL cell lines and micro-dissected HRSCs two molecular subgroups of CHL were identified, associated with differences in activity of transcription factors of NOTCH1, MYC, and IRF4. Moreover, HRSCs displayed deregulated expression of several genes potentially highly relevant to lymphoma pathogenesis, including silencing of the apoptosis-inducer BIK and of INPP5D, an inhibitor of the PI3K-driven oncogenic pathway. In further GEP studies, using mRNA extracted from HRSCs micro-dissected from formalin-fixed, paraffin-embedded tissues, a macrophagelike signature in HRSCs significantly correlated with treatment failure. CSF1R is a representative of this signature, and its expression by mRNA in situ hybridization was significantly associated with progression-free and overall survival. A combined score of CSF1R in in situ hybridization and CD68 in immunohistochemistry was found to be independently predictive for progression-free survival in multivariate analysis. In one study, a 23-gene outcome predictor was generated, which identified a 29% difference in in 5-year overall survival between the high- and low-risk groups population in a validation cohort. The predictor was claimed to be superior to the International Prognostic Score and CD68 immunohistochemistry, but this was not confirmed in two later studies. Whole exome sequencing on purified HRS cells has revealed inactivating B2M mutation as the most frequently detected gene mutation in CH. This leads to loss of major histocompatibility complex class I (MHC-I) expression. The absence of B2M protein in HRSCs has been associated with lower stage of disease, younger age at diagnosis, and better overall and progression-free survival. In a further study, fusions involving CIIITA (MHC II trans-activator) in 15% of CHL cases caused down-regulation of HLA class II expression and overexpression of CD274/PD-L1 and CD273/PD-L2. Some novel gene loci have been identified as being associated with increased risk for the development of CHL. An association between rs6903608 and EBV-negative cHL was confined to the nodular sclerosis histological subtype. Other associations involving HLA Class I have been identified in EBV-positive CHL, mainly mixed cellularity subtype. These observations confirm the relevance of histological subtyping of CHL, and differences between EBV-positive and EBV-negative cases. It has recently been reported that sequencing of circulating cell free (cf)DNA in CHL patients provides a surrogate for the mutational landscape of HRSCs obtained by micro-dissection. Genes recurrently mutated in > 20% of CHL cases include STAT6 (37.5%), TNFAIP3 (35%), and ITPKB (27.5%), and are clustered in major pathways, including NF-kB, PI3K-AKT, cytokine and NOTCH signaling, and immune evasion. The mutational landscape of newly diagnosed and chemorefractory CHL largely overlaps. Refractory CHL does not show an increased rate of TP53 mutation (see above). In contrast, TET2 mutations, including newly acquired mutations, are more frequent in refractory CHL, which is potentially important for therapy choice. Longitudinal studies in patients relapsing under/after chemotherapy or brentuximab vedotin, of paired pre-treatment/relapse tumor tissue samples, have shown that recurrent tumor cell clones branch from ancestral clones through acquisition of specific mutations, while preserving the mutational landscape of the ancestral clone. Conversely, in patients under nivolumab maintenance therapy, ancestral clones were replaced by completely novel clones.

Conclusions Only once the goal of curing all HL patients has been reached, the long journey begun by Sir Thomas Hodgkin can be concluded.

Acknowledgment This manuscript was supported by the grant AIRC 5x1000 number 10007 to SAP.

236

Hodgkin Lymphoma: Pathology and Genetics

Further Reading Agostinelli, C., Gallamini, A., Stracqualursi, L., et al., 2016. The combined role of biomarkers and interim PET scan in prediction of treatment outcome in classical Hodgkin’s lymphoma: A retrospective, European, multicentre cohort study. The Lancet Haematology 3, e467–e479. Carey, C.D., Gusenleitner, D., Lipschitz, M., et al., 2017. Topological analysis reveals a PD-L1-associated microenvironmental niche for Reed–Sternberg cells in Hodgkin lymphoma. Blood 130, 2420–2430. Cencini, E., Fabbri, A., Bocchia, M., 2016. Rituximab plus ABVD in newly diagnosed nodular lymphocyte-predominant Hodgkin lymphoma. British Journal of Haematology 176, 831–833. Connors, J.M., Ansell, S.M., Fanale, M., et al., 2017. Five-year follow-up of brentuximab Vedotin combined with ABVD or AVD for advanced-stage classical Hodgkin lymphoma. Blood 130, 1375–1377. Della Pepa, R., Picardi, M., Giordano, C., et al., 2017. Rituximab in a risk-adapted treatment strategy gives excellent therapeutic results in nodular lymphocyte-predominant Hodgkin lymphoma. British Journal of Haematology. https://doi.org/10.1111/bjh.14856. Fan, Z., Natkunam, Y., Bair, E., et al., 2003. Characterization of variant patterns of nodular lymphocyte predominant Hodgkin lymphoma with immunohistologic and clinical correlation. The American Journal of Surgical Pathology 27, 1346–1356. Fanale, M.A., Cheah, C.Y., Rich, A., et al., 2017. Encouraging activity for R-CHOP in advanced stage nodular lymphocyte-predominant Hodgkin lymphoma. Blood 130, 472–477. Herrera, A.F., Moskowitz, A.J., Bartlett, N.L., et al., 2018. Interim results of brentuximab vedotin in combination with nivolumab in patients with relapsed or refractory Hodgkin lymphoma. Blood 131 (11), 1183–1194. Hummel, M., Ziemann, K., Lammert, H., et al., 1995. Hodgkin’s disease with monoclonal and polyclonal populations of Reed–Sternberg cells. New England Journal of Medicine 333, 901–905. Küppers, R., Rajewsky, K., Zhao, M., et al., 1995. Hodgkin’s disease: Clonal Ig gene rearrangements in Hodgkin and Reed–Sternberg cells picked from histological sections. Annals of the New York Academy of Sciences 764, 523–524. Marafioti, T., Hummel, M., Anagnostopoulos, I., et al., 1997. Origin of nodular lymphocyte-predominant Hodgkin’s disease from a clonal expansion of highly mutated germinalcenter B cells. The New England Journal of Medicine 337, 453–458. Shimabukuro-Vornhagen, A., Haverkamp, H., Engert, A., et al., 2005. Lymphocyte-rich classical Hodgkin’s lymphoma: Clinical presentation and treatment outcome in 100 patients treated within German Hodgkin’s Study Group trials. Journal of Clinical Oncology 23, 5739–5745. Spina, V., Bruscaggin, A., Cuccaro, A., et al., 2017. Genotyping of classical Hodgkin lymphoma on the liquid biopsy. Blood 130, 307. Swerdlow, S.H., Campo, E., Harris, N.L., Jaffe, E.S., Pileri, S.A., Stein, H., Thiele, J., 2017. WHO Classification of tumours of haematopoietic and lymphoid tissues. Revised 4th edn. IARC Press, Lyon. Tiacci, E., Döring, C., Brune, V., et al., 2012. Analyzing primary Hodgkin and Reed–Sternberg cells to capture the molecular and cellular pathogenesis of classical Hodgkin lymphoma. Blood 120, 4609–4620.

Hormones and Cancer Kristen D Brantley, Harvard T.H. Chan School of Public Health, Boston, MA, United States Susan E Hankinson, University of Massachusetts Amherst, Amherst, MA, United States A Heather Eliassen, Harvard T.H. Chan School of Public Health, Boston, MA, United States; Brigham and Women’s Hospital; and Harvard Medical School, Boston, MA, United States © 2019 Elsevier Inc. All rights reserved.

Introduction A range of biologic and epidemiologic evidence documents a key role of the hormonal milieu in the etiology of several cancers. Hormones regulate major physiological systems, playing key roles in metabolism, growth and development, reproduction, and neurological function. Produced in and secreted by endocrine glands, hormones enter circulation and target specific organs, where they bind receptors, producing downstream effects. In some cases, secretion may also occur locally, resulting in paracrine or autocrine effects. Mechanistically, hormones may contribute to carcinogenesis through disruption of normal regulatory action. For example, estrogens and other sex-steroid hormones promote cell proliferation, which increases the opportunity for mutations to occur and can lead to increased tumor growth. Some hormones, including insulin-like growth factor I (IGF-I) and prolactin, have been shown to reduce apoptosis, promoting tumor development. In addition, hormones can affect one another to influence cancer progression; for example, gonadotropins may influence tumorigenesis by stimulating sex-steroid production. Antioxidant properties of certain hormones, such as melatonin, may prevent carcinogenesis. Dysregulation of broader functional processes including circadian rhythm and metabolic control, which is heavily influenced by insulin activity, may increase susceptibility to carcinogenesis through additional mechanistic pathways. Hormonally related risk factors have long been associated with several cancer types, particularly breast cancer. These established breast cancer risk factors include parity and ages at menarche and menopause. Adulthood obesity is linked to as many as 13 cancer types. This association is mediated, at least in part, through hormonal pathways; for instance, most estrogen production among postmenopausal women takes place in adipose tissue, and obesity is also closely linked to insulin function. Diabetes, defined by insulin resistance, is also a moderate risk factor for several cancers, including breast and endometrial cancers.

Methodologic Considerations To evaluate the role of hormones in cancer, measurement considerations in epidemiologic studies must be addressed. First, collection of blood or urine (or other tissue) must occur well before the cancer is clinically evident to establish temporality of the association between hormonal exposure and cancer development, given that the tumor and treatments may affect hormone levels. It is rare for more than one blood or urine sample to be collected per participant in prospective studies, due to logistic and financial constraints. This increases the potential for misclassification of the participant’s true long-term hormone levels (generally the exposure of interest), and precludes the ability to evaluate hormonal levels over time. Several hormone measurements show temporal variability, requiring careful consideration in study design and analysis (Fig. 1). However, single blood sample collection has been shown in multiple studies to reasonably reflect long-term hormone levels, with correlations over a 2-to-3-year period ranging between 0.5 and 0.9 for IGF-I, IGFBPs, gonadotropins, and sex-steroid hormones. Phase of the menstrual cycle influences some hormones, and correlations over time for estrogens and progesterone are substantially lower among premenopausal compared to postmenopausal women. Thus, use of a single blood measurement results in some attenuation of relative risk (RR) estimates. Overall, however, for many hormones of interest, reproducibility of hormone measures in blood are similar to reproducibility of serum cholesterol measures, and therefore can likewise be considered consistent predictors of disease. Whether blood, urine, or both are collected varies based on the study goals and funding limitations. In either case, fasting samples are generally preferred to limit influence of metabolic effects on hormone levels, although exceptions exist, including insulin, where both fasting and nonfasting levels are potentially of scientific interest. Because serum is more proximal to tissues of interest, hormones measured in serum may be a more accurate surrogate marker for tissue exposure compared to urinary levels. Hormone levels tend to remain stable in serum over time, and storage at
80 C or colder is acceptable to prevent degradation or dissociation over time. Urine collection is less invasive, and some urinary hormones are well correlated with serum levels. Urinary analysis is advantageous when breakdown products of hormones are of interest; evaluation of such products may provide insight into mechanistic pathways. However, the concentration of urine varies substantially within and among individuals, and correction for concentration must be considered when measuring urinary products. Moreover, it is less clear whether urine levels are representative of high tissue exposure, or if they simply represent enhanced excretion efficiency, making connections between urine measurements and cancer risk more difficult. The ability to accurately represent tissue hormone levels via blood or urine measurements is a major consideration, although limited data exist because of challenges in obtaining tissues of interest and measuring hormone levels in these tissues. In a recent study, relatively low correlations were observed between circulating sex hormones and prostate tissue specific levels, while strong correlations were observed between plasma estradiol and expression of estrogen dependent genes in ER þ breast tumors, supporting use of blood biomarkers to evaluate the breast tissue milieu. In addition, external factors, such as exercise, alcohol intake, and

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65108-5

237

238

Hormones and Cancer

Fig. 1 Temporal variability of selected biomarkers. (A) Melatonin excretion by hour (aMT6s) for adults 19–40 years (young) or 58–84 years (elder). (B) Estradiol concentration by days centered to ovulation for women 19–42 years of age. (C) IGF-1 and IGF-II concentrations by age and sex. (D) IGFBP concentrations by age and sex. (A) Adapted from Kripke, D. F., Youngstedt, S. D., Elliot, J. A., et al. (2005). Circadian phase in adults of contrasting ages. Chronobiology International22 (4), 695–709. (B) Adapted from Welt, C. K., McNicholl, D. J., Taylor, A. E., et al. (1999). Female reproductive aging is marked by decreased secretion of dimeric inhibin. The Journal of Clinical Endocrinology & Metabolism84 (1), 105–111, Fig. 1. (C and D) Reprinted from Yu, H., Mistry, J., Nicar, M. J., et al. (1999). Insulin-like growth factors (IGF-I, free IGF-I, and IGF-II) and insulin-like growth factor binding proteins (IGFBP-2, IGFBP-3, IGFBP-6, and ALS) in blood circulation. Journal of Clinical Laboratory Analysis13 (4), 166–172, Fig. 1 (C), Fig. 3 (D).

smoking, among others, can influence endogenous hormone production, and thus should be considered in epidemiologic analyses. Despite these concerns, consistency of associations between hormones measured in blood or urine and risk of specific cancers provides evidence for a connection between these measures and actual tissue exposure.

Overview This article reviews endogenous hormones in relation to incident cancer risk. Data from biologic studies inform mechanistic actions, while data from prospective cohort studies (i.e., “nested” case–control studies) provide the strongest epidemiologic evidence for associations with cancer. The hormones included are sex steroids (estrogen and estrogen metabolites, progesterone, androgens), anti-Müllerian hormone (AMH), gonadotropins, prolactin, insulin/C-peptide, insulin-like growth factors (IGF), adiponectin, and melatonin.

Sex Steroids Derivatives of cholesterol formed via steroidogenesis, sex-steroid hormones include estrogens, progesterone, and androgens. These hormones bind to their respective receptors, triggering multiple downstream gene pathways. The aromatase enzyme converts androgens to estrogens, representing the primary mode of estrogen production in males and postmenopausal females. Sex steroids contribute to cancer risk primarily by promoting cellular proliferation, thereby increasing the chance of cellular mutations. Estrogens are thought to influence tumorigenesis via estrogen receptor-mediated transcriptional activation of estrogen-responsive genes, including proto-oncogenes (Fig. 2B). Estrogen also may be tumorigenic through genotoxic mechanisms. Androgens may increase cancer risk by directly influencing proliferation or via conversion to estrogens. Progesterone promotes

Hormones and Cancer

239

apoptotic signaling, and thus may balance the proliferative impact of estrogens; the absence of progesterone underlies the “unopposed estrogen” hypothesis.

Estrogens Estrogens play an essential role in the development of female secondary sexual characteristics, and contribute to several biological systems, including reproductive, cardiovascular, and neuroendocrine. Major estrogens include estrone (E1), 17-b-estradiol (E2), the most biologically active estrogen, and estriol (E3). In premenopausal women, estrogen is primarily produced in the granulosa cells of ovarian follicles with assistance from follicle-stimulating hormone (FSH), and is produced in smaller amounts in the liver, adrenal glands, breasts, and adipose tissue. Estrone is a primary estrogen in postmenopausal women, produced from conversion of adrenal androgens. In males, estrogen production occurs in the Leydig cells of the testes, through conversion of testosterone to estradiol via the cytochrome p450 aromatase enzyme, and through conversion of androstenedione to estrone. Estrogens affect signaling pathways by binding to and activating one of two estrogen receptors, ER-a or ER-b, in target organs. Parent estrogens, estradiol and estrone, can be metabolized via 2-, 4-, and 16a hydroxylation pathways. These oxidative metabolites of estrogen have been implicated in carcinogenesis (Fig. 2A). Catechol estrogens are formed following irreversible oxidation of estrone and estradiol at the C-2 or C-4 positions, while oxidation at the C-16 position yields 16a-hydroxyestrone. Catechol estrogens can be further methylated to 2-methoxyestrone and 4-methoxyestradiol, 2-hydroxyestrone-3-methyl ether, 4-methoxyestrone, and 4-methoxyestradiol, while 16a-hydroxyestrone can be metabolized to 17-epiestriol, 16-ketoestradiol, and 16-epistriol. The estrogenic and proposed genotoxic activity of metabolites differs across pathways. The 4- and 16a- pathway metabolites have higher estrogenic activity than estrogen, and as such are thought to increase proliferation in the same manner as estradiol; in contrast, 2-catechol estrogen metabolites may inhibit or mitigate proliferation. These differences may be explained in part by each metabolites’ binding affinity to ER: 2- and 4-catechol estrogen metabolites (EMs) bind with a similar affinity as estradiol, while 16a-hydroxyestrone binds with lower affinity, though covalently. Thus, bound 16a-hydroxyestrone results in a constitutively activated ER, whereas 2- and 4-catechol EM can dissociate. 4-Catechol EMs have lower dissociation rates than estradiol, so they tend to upregulate ER-dependent processes. In addition, the metabolism of 4-hydroxy catechols to reactive quinones leads to generation of reactive oxygen species and consequent oxidative damage which may contribute to tumorigenesis. While 2-catechol estrogens are similarly metabolized to quinones, they instead form stable, reversible, DNA adducts. In either case, the methylation of an adjacent hydroxyl group will block formation of quinones. Women who have traits associated with high cumulative estrogen exposure, such early age at menarche or late age at natural menopause, and long menopausal hormone therapy use, are more likely to develop several cancers, including postmenopausal breast, ovarian, and endometrial cancers, supporting a role for estrogens in carcinogenesis. More recently, evidence has also emerged for estrogen pathways in premenopausal breast, prostate, and colorectal cancers.

Postmenopausal breast cancer Increasing levels of circulating estrogens are consistently associated with increased risk of postmenopausal breast cancer. A 2002 pooled analysis of nine prospective studies by the Endogenous Hormones and Breast Cancer Collaborative Group included 663 women who developed breast cancer after blood collection and 1765 controls. Positive associations were observed between all measured estrogens and postmenopausal breast cancer (Fig. 3), with evident dose–response trends (5th vs. 1st quintile total estradiol relative risk (RR) ¼ 2.00, 95% confidence interval [CI]: 1.47–2.71, p-trend < 0.001; free estradiol RR ¼ 2.58, 95% CI: 1.76– 3.78, p-trend < 0.001). Large prospective studies published subsequently confirmed these pooled findings. In a 20-year followup within the Nurses’ Health Study (NHS), positive associations between estradiol and breast cancer risk were stronger among women with ER þ/PR þ tumors (5th vs. 1st quintile RR ¼ 2.8, 95% CI: 2.0–4.0), and for women with more aggressive disease (i.e., recurrent or fatal disease) (RR ¼ 2.6, 95% CI: 1.3–5.1). In an analysis investigating serum or urinary EMs in postmenopausal breast cancer, including 1298 cases from four studies, an inverse association was observed between 2-hydroxylation pathway metabolites and postmenopausal breast cancer, as well as an inverse association between the ratio of 2-hydroxylation: 16-hydroxylation pathway metabolites.

Catechol estrogen (A) Estrogen metabolites

Semiquinone

Quinone Oxidative damage

Estrogens

Point mutation(s)

CANCER

(B) Estrogen receptor Estrogen receptor activation

Cell division/ proliferation

Fig. 2 Mechanisms of estrogens in carcinogenesis. (A) Catechol estrogens form semiquinones and quinones, that may lead to oxidative damage and genetic mutations. (B) Estrogen directly causes increased cell proliferation following ER activation, increasing likelihood of point mutations. Mutations via either mechanism promote carcinogenesis. Adapted from Fig. 1 in Bohra, A. and Bhateja, S. (2015). Carcinogenesis and sex hormones: A review. Endocrinology & Metabolic Syndrome4, 156. https://doi.org/10.4172/2161-1017.1000156.

240

Hormones and Cancer

Exposure

Cases/Controls

P-trend

OR (95% CI)

Premenopausal Estradiol Q1 Q2 Q3 Q4 Q5

104/280 108/277 120/271 128/277 140/270

1.00 (ref) 1.07 (0.77, 1.48) 1.19 (0.87, 1.63) 1.26 (0.92, 1.73) 1.41 (1.02, 1.95)

Doubling

600/1375

1.19 (1.06, 1.35)

Q1 Q2 Q3 Q4 Q5

144/350 130/331 158/330 152/329 165/329

1.00 (ref) 0.99 (0.75, 1.32) 1.21 (0.92, 1.61) 1.16 (0.87, 1.54) 1.32 (0.98, 1.76)

Doubling

750/1662

1.18 (1.03, 1.35)

Q1 Q2 Q3 Q4 Q5

111/405 115/307 113/351 152/317 434/502

1.00 (ref) 1.42 (1.04, 1.95) 1.21 (0.89, 1.66) 1.80 (1.33, 2.43) 2.00 (1.47, 2.71)

Doubling

583/1555

1.31 (1.17, 1.48)

Q1 Q2 Q3 Q4 Q5

86/337 101/312 123/312 120/307 155/306

1.00 (ref) 1.34 (0.96, 1.87) 1.61 (1.16, 2.24) 1.59 (1.13, 2.23) 2.22 (1.59, 3.10)

Doubling

583/1555

1.42 (1.25, 1.61)

0.0042

Testosterone

0.018

Postmenopausal Estradiol

75% of IGF-1 and IGF-2. In the bound state, IGFs are not available for IGF-I receptor activation. IGF bioactivity has thus been proposed to be a function of circulating levels of IGF, expression of IGF genes on tissues of interest, and protease action to cleave IGBP complexes to their functional form, though additional complexities also must be considered. For instance, most evidence supports an anticarcinogenic role for the binding proteins, particularly IGFBP-3 and IGFBP-1. However, certain binding proteins are also able to enhance the delivery of IGFs to certain tissues. Overexpression of both IGFBP-2 and IGFBP-5 have been associated with increased IGF action, and IGFBP-1, which is closely associated with insulin function, also increases IGF-1 induced DNA synthesis in its dephosphorylated state. Overall, an imbalance of IGFs and IGFBPs may indicate issues in cellular turnover and kinetics, characteristic of carcinogenesis. Given the biologic plausibility that IGFs are important in carcinogenesis, these hormones, along with IGFBP-3 and IGFBP-1, have been investigated in several cancers to date, including breast, endometrial, ovarian, prostate, colorectal, pancreatic, and lung cancers. Serum measured IGFs and IGFBPs, used in epidemiologic studies, correlate well with tissue levels in murine models, supporting the use of blood-based measures as surrogates for tissue expression.

Postmenopausal Breast Cancer In a recent pooled analysis combining data from 17 prospective studies in Europe, Australia, and the United States, 15 of which included women who were postmenopausal at blood collection (n ¼ 2853 cases), IGF-1 was positively associated with breast cancer risk in postmenopausal women (5th vs. 1st quintile RR ¼ 1.33, 95% CI: 1.14–1.55). Contrary to the idea that IGFBP-3 is inversely related to the bioactivity of IGF, IGFBP-3 levels were also positively associated with breast cancer risk (RR ¼ 1.23, 95% CI: 1.04–1.45). Consistent with evidence from laboratory studies that estrogen and IGF signaling pathways interact in a synergistic manner, risk was only apparent among women with ER þ tumors (for 80% increase in IGF-1 RR ¼ 1.38, 95% CI: 1.14–1.68), with no association noted among ER
tumors (RR ¼ 0.80, 95% CI: 0.57–1.13). However, risk also increased consistently with increasing tertile of estrogen concentration, among women with either low, medium, or high levels of IGF-1, suggesting no clear interaction of effects, though these results were not confined to ER þ tumors.

Premenopausal Breast Cancer IGF-1 has been associated with mammographic density in premenopausal women, a known risk factor for breast cancer, in a few recent studies. For instance, one study of 783 premenopausal healthy women found high IGF-1 and low IGFBP-3 to be independently correlated with high breast density. In the above-mentioned pooled analysis, 11 of the 17 studies included information on women who were premenopausal at blood collection (n ¼ 1937 cases). The association between IGF-1 and breast cancer risk was

248

Hormones and Cancer

similarly positive (5th vs. 1st quintile RR ¼ 1.21, 95% CI: 1.00–1.45). An analysis by age at diagnosis (< 50 years or  50 years) similarly observed a small, suggestive positive association between IGF-1 and breast cancer among younger women, though circulating IGFBP-3 levels were not significantly associated with breast cancer risk among women premenopausal at blood collection (RR ¼ 1.00, 95% CI: 0.82–1.22).

Endometrial Cancer A large nested case–control study within EPIC (n ¼ 286 cases) examined prediagnostic concentrations of IGFBP-1 and IGFBP-2 and risk of endometrial cancer (the IGF hormones were not examined). No association between IGFBP-1 and endometrial cancer risk was observed (4th vs. 1st quartile RR ¼ 0.76, 95% CI: 0.47–1.21), though there was an inverse association between IGFBP-2 and endometrial cancer risk (RR ¼ 0.56, 95% CI: 0.35–0.90). This finding corresponds with a potential role for the IGF axis in the carcinogenesis of endometrial cancer; though, the null finding for IGFBP-1 suggests that the primary mode through which insulin resistance increases endometrial cancer risk is not through IGF signaling.

Prostate Cancer A large pooled analysis of 17 prospective studies (n ¼ 7682 cases) examined circulating IGF-1, IGF-2, IGFBP-2, and IGFBP-3 with prostate cancer risk (Fig. 4). IGF-1 was associated with an increased risk of prostate cancer (5th vs. 1st quintile RR ¼ 1.29, 95% CI: 1.16–1.43), as was IGFBP-3 (RR ¼ 1.25, 95% CI: 1.12–1.40). IGFBP-1 was inversely associated with risk (RR ¼ 0.81, 95% CI: 0.68– 0.96). After mutual adjustment for IGFs and IGFBPs, only IGF-1 remained associated with prostate cancer risk. While experimental evidence suggests that the activation of the IGF pathway is associated with progression of prostate cancer to lethal disease, the association between IGFs and prostate cancer risk did not appear to differ by stage or grade of disease in this pooled analysis.

Colorectal Cancer Markers of hyperinsulinemia, such as obesity and central adiposity and type II diabetes have been associated with risk of colorectal cancer. Thus, insulin and corresponding IGF signaling pathways have been implicated in colorectal carcinogenesis. IGF-1 receptor is overexpressed in colon cancer cells, and overexpression of IGF-1 in a human CRC cell line resulted in a highly invasive tumor, further supporting a carcinogenic role. IGF-2 is also highly overexpressed in a subset of colorectal cancers. Moreover, CRC patients with low IGFBP-3 or high IGFBP-2 have worse prognosis, suggesting that these binding proteins negatively regulate IGF activity, or act independently, to influence tumor growth. Given the evidence for IGFs and IGFBPs in CRC progression, prospective epidemiologic studies have investigated this association. A study within the Physicians’ Health Study (PHS) examined IGF-1 and IGFBP-3 serum levels and colorectal cancer risk (n ¼ 193 male cases). IGF-1 was positively associated with risk of colorectal cancer (5th vs. 1st quintile RR ¼ 2.51, 95% CI: 1.15–5.46), and, correspondingly, IGFBP-3 was associated with decreased risk of colorectal cancer (RR ¼ 0.28, 95% CI: 0.12–0.66). This association was attenuated among obese men, suggesting that obesity also contributes to colorectal cancer through non-IGF related pathways. These findings were supported by an investigation within the Northern Sweden Cohort. In contrast, no association was observed in the EPIC cohort (n ¼ 1121 cases) between IGF-1 and CRC, or between IGFBP-3 (total and intact) and CRC. A meta-analysis combining these results with those of nine other prospective studies conducted between 1999 and 2008 (n ¼ 2862 cases), found a small positive association between IGF-1 and CRC risk (per standard deviation increase RR ¼ 1.07, 95% CI: 1.01–1.14).

Pancreatic Cancer Experimental studies have shown the presence of IGF-1 and IGF-1R in pancreatic cell lines, and have demonstrated increased cellular proliferation and decreased apoptosis via signaling through IGF-1R. In the PLCO study (n ¼ 187 cases) no statistically significant associations were observed for IGF-1, IGF-2, or IGFBP-3, although associations with IGF-1 (and the ratio of IGF-1/ IGFBP-3, which would generally represent the proportion of bioavailable IGF-1) were suggestively positively associated with pancreatic cancer risk (e.g., IGF-1/IGFBP-3 4th vs. 1st quartile RR men ¼ 1.39, 95% CI: 0.69–2.80; RR women ¼ 1.47, 95% CI: 0.58–3.75). Another pooled study from the NHS, HPFS, PHS, and the WHI (n ¼ 212 cases), observed no significant association between plasma IGF-1, IGF-2, IGFBP-3, or the IGF-1:IGFBP-3 M ratio and pancreatic cancer risk. Overall, the relatively limited data available to date suggests no substantial association of circulating IGFs and IGFBPs with pancreatic cancer risk.

Lung Cancer IGF-1 exhibits mitogenic and antiapoptotic effects in lung cancer cell lines, while IGFBP-3 has been shown to inhibit IGF-1R signaling through interference with MAPK and Akt signaling pathways. A recent meta-analysis examining of circulating IGF-1 and IGFBP-3 and lung cancer risk included six nested case–control studies (n ¼ 1043 cases). Neither IGF-1 nor IGFBP-3 were associated with increased lung cancer risk. One of the larger studies included found a positive association between IGF-1 and lung cancer risk, though this increased risk was only reported after controlling for IGFBP-3 levels. Overall, current evidence suggests no association between IGF-1 and IGFBP-3 and risk of lung cancer.

Hormones and Cancer

Cases/Controls

OR (95% CI)

Q1 Q2 Q3 Q4 Q5

1402/2037 1452/2018 1523/2024 1607/2011 1698/2006

1.00 (ref) 1.05 (0.95, 1.16) 1.11 (1.01, 1.23) 1.19 (1.07, 1.31) 1.29 (1.16, 1.43)

Per 80% increase

7682/10096

1.29 (1.17, 1.41)

484/754 511/744 596/745 546/745 557/740

1.00 (ref) 1.06 (0.90, 1.25) 1.27 (1.08, 1.50) 1.15 (0.97, 1.36) 1.20 (1.00, 1.43)

2694/3728

1.19 (1.01, 1.39)

581/633 462/602 503/606 502/605 442/602

1.00 (ref) 0.85 (0.72, 1.01) 0.89 (0.75, 1.06) 0.89 (0.75, 1.06) 0.81 (0.68, 0.96)

2490/3048

0.85 (0.73, 1.00)

349/506 396/502 429/504 474/502 434/502

1.00 (ref) 1.15 (0.95, 1.40) 1.25 (1.02, 1.52) 1.40 (1.15, 1.70) 1.26 (1.03, 1.54)

2082/2516

1.29 (1.08, 1.55)

Q1 Q2 Q3 Q4 Q5

1158/1769 1326/1780 1353/1729 1337/1755 1357/1742

1.00 (ref) 1.16 (1.04, 1.29) 1.21 (1.09, 1.35) 1.20 (1.08, 1.34) 1.25 (1.12, 1.40)

Per 80% increase

6531/8775

1.21 (1.09, 1.34)

Exposure

249

P-trend

IGF-1

85% of cases) that might be therapeutically exploitable by the use of PARP inhibitors. Up to now, genetic analyses of osteosarcoma have been limited to tumors of the peripheral skeleton and have not included gnathic OS. The molecular basis for the differences in biologic behavior and prognosis is therefore still unknown.

Microenvironment Including Immune Response There are no established data on the microenvironment of gnathic OS. On pure morphologic basis, tumor infiltrating inflammatory cells are not a common feature in these tumors.

Staging and Grading There are no commonly accepted guidelines for staging procedures in gnathic OS but besides the assessment of local tumor extent, pulmonary metastases should be ruled out by a CT scan of the chest. Grading is generally performed according to tumors of the peripheral skeleton.

Fig. 3

Osteosarcoma: Neoplastic matrix formation permeating through preexisting bone and encircling the inferior alveolar nerve (arrowheads).

318

Fig. 4

Jaws Cancer: Pathology and Genetics

Osteosarcoma: Pleomorphic and hyperchromatic spindle cells with immature and neoplastic new bone formation.

Prognostic and Predictive Biomarkers For yet unknown reasons, gnathic OS metastasize far less frequently (6%–21% of cases) and later in the course of the disease compared to OS of the peripheral skeleton, rendering complete resection with clear margins the single most important prognostic factor for affected patients. If achievable, 10-year survival rates exceed 80% and the prognosis remains excellent even if clear margins can only be accomplished after repeated surgery. The role of (neo-)adjuvant chemotherapy is thus still controversial and difficult to investigate systematically due to the rarity of the disease. There are, however, tumors that follow an aggressive course which might not be obvious morphologically. Notably, the favorable outcome of most OS is restricted to tumors developing in the jaws, whereas tumors of the skull or facial bones behave similar to extragnathic tumors.

Craniofacial Fibrous Dysplasia Definition Fibrous dysplasia (FD) is a skeletal anomaly caused by a post-zygotic missense mutations of the GNAS gene encoding the activating alpha-subunit of the stimulatory G-protein. The mutation results in the proliferation of undifferentiated bone marrow related stem cells that transform into functionally impaired bone forming progenitors. Marrow spaces are replaced by a fibroblastic spindle cell stroma containing immature trabeculae of woven bone. Fibrous dysplasia might involve single (monostotic) or multiple bones (polyostotic) and can occur together with a range of endocrinopathies and skin lesions (Café-au-lait spots) as McCune Albright syndrome or along with intramuscular myxoma as Mazabraud’s syndrome. In the craniofacial bones, multiple adjacent bones are characteristically affected which is still considered as a monostotic involvement and designated as craniofacial fibrous dysplasia (CFD). Symptomatic patients generally notice painless diffuse swellings of the affected region which might cause significant cosmetic deformity and compression/obstruction of vital structures (e.g., the optic nerve or foramina of the skull base). Radiographically, early FD is primarily lytic but develops the characteristic homogeneous ground-glass appearance over time. Lesions are not well defined and gradually blend into the adjacent normal bone. The radiologic and clinical picture is highly characteristic so the diagnosis can often be made without requiring histologic proof (Fig. 5).

Burden Exact figures on the incidence of FD or CFD are unknown because many patients are asymptomatic. Monostotic involvement is 6– 10 times more common than polyostotic disease, craniofacial involvement is frequent (> 50% of cases). CFD most frequently involves the maxilla and adjacent bones but can develop in virtually any (craniofacial) bone.

Risk Factors There are no established risk factors for developing crandiofacial dysplasia.

Pathology Morphologically, CFD is similar to FD in the peripheral skeleton and reveals a monomorphic and cytologically bland spindle cell stroma containing islands or trabeculae of woven bone. In early stages, the stroma appears immature and richly cellular although mitotic activity is low. Directly evolving from the spindle cells, varying amounts of metaplastic woven formation are recognizable that typically lack osteoblastic rimming. The architecture of the trabeculae can be curvilinear and has been described to resemble

Jaws Cancer: Pathology and Genetics

319

Fig. 5 Fibrous dysplasia: Fibrous dysplasia presenting as an ill-defined expansion of the left maxilla with homogenous ground-glass appearance (CT scan, coronal plane).

Chinese script letters (Fig. 6). Although nonspecific, the kind of matrix formation is almost unique and diagnostic for FD, especially in the hand of an experienced pathologist. Using special stains (e.g., van Gieson), a typical pattern of Sharpey’s fibers radiating perpendicularly from the trabeculae into the surrounding stroma can furthermore be seen. Over time, the stromal cellularity decreases and the woven bone matures slowly into lamellar bone that can also show some degree of osteoblastic rimming due to an ongoing remodeling. Occasionally, FD can undergo infarction which can further blur the diagnostic changes and further complicate the histologic diagnosis. Like in other tumors and tumor-like lesions of bone, diagnosis should thus always include the correlation with clinical and radiologic findings, particularly in long-standing disease.

Molecular Pathology and Genetics Point mutations in the GNAS gene can be detected in the vast majority (> 90%) of cases. Over time, however, the lesional cells progressively decrease in number and the amount of mutant gene copies can fall below the detection threshold of the diagnostic method used. Particularly in older or matured lesions, lack of GNAS mutations therefore does not necessarily argue against the diagnosis.

Microenvironment Including Immune Response There are no established data on the microenvironment of craniofacial fibrous dysplasia. On pure morphologic basis, tumor infiltrating inflammatory cells are not a common feature in these tumors.

Staging and Grading Fibrous dysplasia is a benign condition and does not require staging procedures besides the assessment of local extent and individual symptoms. There is no grading system either.

Fig. 6

Fibrous dysplasia: Irregularly shaped woven bone formation embedded in a monomorphous fibroblastic spindle cell stroma lacking atypia.

320

Jaws Cancer: Pathology and Genetics

Prognostic and Predictive Biomarkers In most patients, CFD stabilizes or even undergoes some degree of remission after skeletal maturity. Surgical intervention should only be considered in severe cases to preserve function and optimize esthetics. Although there is no unified consensus on the exact timing or even the indication for surgery, it is usually postponed until the patient has reached the end of puberty. There are no prognostic or predictive biomarkers to anticipate the clinical course of individual patients. Malignant transformation, most commonly into high-grade OS, is exceedingly rare.

Ossifying Fibroma Definition Ossifying fibroma together with (craniofacial) fibrous dysplasia and cemento-osseous dysplasia belong to the descriptively defined group of fibro-osseous lesions of the jaws. Among three variants that can be distinguished, the most common type is the cementoossifying fibroma (COF), a tumor of odontogenic origin that exclusively develops in the tooth-bearing areas of the jaws. Juvenile trabecular ossifying fibromas (JTOF) and juvenile psammomatoid ossifying fibromas (JPOF) are rare bone tumors that can develop also in extragnathic sites. All variants generally present as painless intraosseous swellings that can reach considerable and disfiguring dimensions if left untreated. Early lesions are radiologically lytic but over time matrix mineralization increases, resulting in a mixed lytic and sclerotic appearance; usually, a lytic rim remains encompassing the lesion. JTOF and JPOF can grow rapidly and are therefore also designated as juvenile aggressive subtypes, but in general, ossifying fibroma are slowly growing, well circumscribed and expansile lesions that lead to cortical thinning. Aggressive periosteal reactions or wide destruction of preexisting bone are generally not present and are important aspects in the differential diagnosis of osteosarcoma. Occasionally, the lesions are associated with secondary aneurysmal bone cysts.

Burden Cemento-ossifying fibroma preferentially develops in women in the third to fourth decade of life (female to male ratio ¼ 4:1). Although any tooth-bearing area of the jaws might be affected, the mandible and primarily the molar/premolar region is the most common site of origin. The juvenile variants occur at younger age (8.5–12 years mean age for JTOF and 16–33 years for JPOF) but the age distribution is broad and ranges up the seventh decade. JTOF are most common in the maxilla but can rarely occur also at extragnathic sites, JPOF only infrequently affect the jaws and are more prevalent in extragnathic locations, preferentially in the periorbital frontal and ethmoid bones. All subtypes generally develop as solitary lesions with the exception of multifocal manifestations in the exceedingly rare hyperparathyroidism jaw tumor syndrome. Multifocal involvement of lesions with a similar morphology have also been reported as (familial) gigantiform cementoma.

Risk Factors There are no established risk factors for developing ossifying fibroma.

Pathology Ossifying fibromas are characterized by varying amounts and kinds of hard tissue formation embedded in a moderately cellular and fibroblast-like spindle cell stroma. The latter usually appears monomorphic and without significant atypia although hyperchromatic nuclei can be present. Mitotic activity is usually low. In COF, the matrix component consists of woven bone trabeculae with prominent osteoblastic rimming and different stages of maturation (Fig. 7). Additionally, hypo- or acellular cementum-like material can usually be encountered. Over time, the mineralized matrix increases and can coalesce to form more complex structures, that are also visible radiologically as enlarging areas of sclerosis. JTOF can infiltrate into the surrounding marrow spaces and typically demonstrates delicate strands of woven bone that lack evidence of maturation. The matrix and stroma component blend into each other resulting in very immature appearing bone (Fig. 8). In JPOF, spherical and paucicellular matrix deposits reminiscent of psammoma bodies/cementum-like material predominate that can coalesce over time and secondarily induce new bone formation (Fig. 9). Fibrous dysplasia and particularly cemento-osseous dysplasia can mimic ossifying fibroma histologically, underlining the need for a thorough radiologic and clinical correlation of findings. Cemento-osseous dysplasia is usually not expansile and often occurs multifocally, fibrous dysplasia is less well defined, can affect several adjacent craniofacial bones and typically lacks osteoblastic rimming.

Molecular Pathology and Genetics Due to the rarity of the disease and the intense mineralization requiring decalcification procedures, molecular and genetic analyses are restricted to few and smaller case series so far. Mutations in the CDC73 gene causes the hyperparathyroidism jaw tumor syndrome and has been detected also in sporadic cases. Recent studies have suggested juvenile ossifying fibromas to represent

Jaws Cancer: Pathology and Genetics

321

Fig. 7 Cemento-ossifying fibroma: Bony particles of varying size and shape embedded in a monomorphous fibroblastic spindle cell stroma lacking atypia.

Fig. 8 Juvenile trabecular ossifying fibroma: Immature and slender strands of woven bone directly emerging from and blending into a fibroblastlike spindle cell stroma without atypia.

Fig. 9 Juvenile psammomatoid ossifying fibroma: Multiple psammoma-body like ossicles embedded in a cellular spindle cell stroma without significant atypia.

potential precursor lesions of osteosarcoma of the jaws due to an increased amount of MDM2 and RASAL1 templates detected by qPCR, both located on the long arm of chromosome 12. Whereas MDM2 and RASAL1 flank nearly the complete chromosomal arm, the amplification typical for parosteal osteosarcoma and a smaller subset of low-grade central osteosarcoma (approximately 30% of cases) is confined to a very small chromosomal region encompassing the MDM2 and CDK4 genes that are located in close proximity to each other. Although interesting, the data published is not sufficient to conclude a relationship between ossifying fibroma and osteogenic sarcoma or even to consider juvenile ossifying fibroma as a potential precursor of osteosarcoma. GNAS mutations typical for fibrous dysplasia do not occur in ossifying fibroma.

322

Jaws Cancer: Pathology and Genetics

Microenvironment Including Immune Response There are no established data on the microenvironment of ossifying fibroma. On pure morphologic basis, tumor infiltrating inflammatory cells are not a common feature in these lesions.

Staging and Grading As ossifying fibromas are uniformly benign, staging is confined to the assessment of local tumor extent. There is no grading system applicable to these lesions.

Prognostic and Predictive Biomarkers There are no established prognostic or predictive biomarkers in ossifying fibroma. If excised completely, recurrences are rare. Malignant transformation has not been convincingly described.

Central Giant Cell Granuloma Definition Central giant cell granulomas (CGCG) are localized and benign but locally aggressive tumors occurring exclusively in the jaws. The majority of lesions grow slowly and present as asymptomatic expansions of bone but one third of patients experience rapid growth, pain, resorption and displacement of teeth, bone errosion (or destruction), and/or soft tissue infiltration. Generally, CGCG are solitary lesions but can occur multifocally, particularly when associated with neurofibromatosis type 1 or Noonan /LEOPARD syndrome (so-called RASopathies). Radiographically, CGCG appear lytic and polylobulated. They can reach considerable dimensions with cortical thinning and sometimes show central septae of woven bone resulting in a typical honeycomb pattern. Aggressive permeative growth or periosteal reactions are usually absent.

Burden CGCG represent about 10% of all benign tumors in the jawbones. They occur more commonly in women and usually develop under the age of 20. The anterior mandible is the most frequently affected site.

Risk Factors There are no established risk factors for developing CGCG.

Pathology Histologically, CGCG show a lobular proliferation of mononuclear cells with a storiform pattern of growth. The cells appear spindle-shaped or polygonal and lack cytologic atypia (Fig. 10). Mitoses are frequently seen but are not atypical. Intermingled and partly clustered around vascular spaces and hemosiderin deposits, a varying number of rather small multinucleated giant cells

Fig. 10

Central giant cell granuloma: Monomorphic spindle cells arranged in a storiform pattern with intermingled multinucleated giant cells.

Jaws Cancer: Pathology and Genetics

323

(< 15 nuclei per section) can be observed. In some cases, they can be present only focally but they generally do not predominate the histology like in classic giant cell tumor of bone. The lobules of tumor cells are typically incompletely separated by fibrous septae that can contain metaplastic new bone formation. Brown tumors related to hyperparathyroidism can appear morphologically identical to CGCG so laboratory tests should be complemented to rule out such a condition. The solid variant of aneurysmal bone cyst is another differential diagnosis that can histologically closely mimic CGCG and molecular tests can be useful to exclude these lesions (see below). Conventional giant cell tumors of bone generally do not occur in the jaws with very rare convincing exceptions affecting the condylar process. Syndrome-related giant cell lesions are morphologically identical to sporadic cases but can show a striking predominance of spindle-shaped mononuclear cells. Another syndrome with a similar histologic appearance is cherubism that typically affects all quadrants of the jaws and leads to a marked and sometimes disfiguring expansion of bone.

Molecular Pathology and Genetics Although the genetic background of sporadic CGCG is still unknown, molecular tests can help to rule out histological mimics. Giant cell tumors of bone typically harbor point mutations in the H3F3A gene (p.G34W) which have not been detected in CGCG. These lesions are therefore not only histologically but also molecularly distinct. Primary aneurysmal bone cysts show rearrangements of the USP6 gene in about 70% of cases that can be detected by FISH analyses. Although a negative FISH result does not exclude the diagnosis, aneurysmal bone cysts usually have larger pseudocystic areas that should make a distinction from CGCG easily possible in the majority of cases.

Microenvironment Including Immune Response There are no established data on the microenvironment of CGCG. On pure morphologic basis, tumor infiltrating inflammatory cells are not a common feature in these tumors.

Staging and Grading As CGCG is considered a benign tumor, staging is confined to local tumor extent and there is no grading system.

Prognostic and Predictive Biomarkers Most CGCG can be cured by local curettage but recurrences occur, particularly in clinically aggressive and syndrome-related lesions. The role of adjuvants including antiresorptive drugs (e.g., RANK ligand inhibitors and bisphosphonates) is currently still unclear and should be restricted to patients in which a complete removal of the lesion is surgically difficult or unfeasible.

Malignant Odontogenic Tumors The odontogenic tissues occasionally are the source of malignant neoplasms occurring within the jaw bones. They may be epithelial, mesenchymal, or mixed. The following malignant epithelial types (odontogenic carcinomas) are recognized: ameloblastic carcinoma, primary intraosseous carcinoma, sclerosing odontogenic carcinoma, clear cell odontogenic carcinoma and ghost cell odontogenic carcinoma. Mesenchymal malignancies are labeled as odontogenic sarcoma and those of a mixed nature are known as odontogenic carcinosarcoma. Their extreme rarity precludes more extensive discussion; for more information the reader is referred to specialized texts on odontogenic tumors.

Prospective Vision The field of genetics of odontogenic tumors is yet largely unexplored. It is to be expected that within the near future, elucidation of the pathways involved in these tumors will offer new tools for diagnosis and treatment.

Further Reading Baumhoer, D., Brunner, P., Eppenberger-Castori, S., et al., 2014. Osteosarcomas of the jaws differ from their peripheral counterparts and require a distinct treatment approach. Experiences from the DOESAK Registry. Oral Oncology 50, 147–153. Behjati, S., Tarpey, P.S., Haase, K., et al., 2017. Recurrent mutation of IGF signalling genes and distinct patterns of genomic rearrangement in osteosarcoma. Nature Communications 8, 15936. Heikinheimo, K., Kurppa, K.J., Elenius, K., 2015. Novel targets for the treatment of ameloblastoma. Journal of Dental Research 94, 237–240. Maiorano, E., Slootweg, P.J., 2016. Maxillofacial skeleton and teeth. In: Cardesa, A., Slootweg, P.J., Gale, N., Franchi, A. (Eds.), Pathology of the head and neck, 2nd edn. Springer, Heidelberg, pp. 179–227.

324

Jaws Cancer: Pathology and Genetics

Odell, E.W., Muller, S., Richardson, M., et al., 2017. Odontogenic and maxillofacial bone tumours. In: El-Naggar, A.K., Chan, J.K.C., Grandis, J.R., Takata, T., Slootweg, P.J. (Eds.), WHO classification of head and neck tumours, 4th edn. Lyon, IARC, pp. 204–260. Presneau, N., Baumhoer, D., Behjati, S., et al., 2015. Diagnostic value of H3F3A mutations in giant cell tumour of bone compared to osteoclast-rich mimics. Journal of Pathology: Clinical Research 1, 113–123. Salinas-Souza, C., De Andrea, C., Bihl, M., et al., 2015. GNAS mutations are not detected in parosteal and low-grade central osteosarcomas. Modern Pathology 28, 1336–1342. Santos, J.N., Sousa Neto, E.S., Franca, J.A., et al., 2017. Next-generation sequencing of oncogenes and tumor suppressor genes in odontogenic myxomas. Journal of Oral Pathology and Medicine (Epub ahead of print). Slootweg, P.J., 2009. Lesions of the jaws. Histopathology 54, 401–418. Tabareau-Delalande, F., Collin, C., Gomez-Brouchet, A., et al., 2015. Chromosome 12 long arm rearrangement covering MDM2 and RASAL1 is associated with aggressive craniofacial juvenile ossifying fibroma and extracranial psammomatoid fibro-osseous lesions. Modern Pathology 28, 48–56. Vered, M., Wright, J.M., 2017. Update from the 4th edition of the World Health Organization classification of head and neck tumours: odontogenic and maxillofacial bone tumors. Head and Neck Pathology 11, 68–77.

Kidney Cancer: Diagnosis and Treatment ska, Science to the Point, Archamps Technopole, Archamps, France Katarzyna Szyman © 2019 Elsevier Inc. All rights reserved.

Abbreviations ccRCC Clear cell renal cell carcinoma CN Cytoreductive nephrectomy CT Computed tomography FISH Fluorescence in situ hybridization HRCC Hereditary renal cell carcinoma IARC International Agency for Research on Cancer (WHO) ICD-O International Classification of Diseases for Oncology ISUP International Society of Urological Pathology LND Lymph node dissection MRI Magnetic resonance imaging PN Partial nephrectomy PRCC Papillary renal cell carcinoma RCC Renal cell carcinoma TCGA The Cancer Genome Atlas TNSA Total number of aberrations VEGF Vascular endothelial growth factor VEGFR Vascular endothelial growth factor receptor VHL von Hippel-Lindau (gene or syndrome) WHO World Health Organization

Definition and Classification Kidney tumors include tumors of the renal cortex and renal pelvis. Approximately 90% of renal tumors are renal cell carcinomas (RCC). Of those, about 80% are clear cell tumors (ccRCC; ICD-O code: 8310/3) which are a heterogeneous group of malignant neoplasms composed of cells with clear or eosinophilic cytoplasm. They have a typical vessel formation and a characteristic molecular background with VHL inactivation and the upregulation of HIF (see the section “Pathology and Genetics”). Other, less common, variants include papillary (ICD-O code: 8260/3), chromophobe (8317/3), MiT family translocation (8311/3), and Bellini duct (collecting duct) renal cell carcinoma (8319/3). Medullary renal carcinoma (8510/3) is a variant of collecting duct renal carcinoma and was initially described as occurring in sickle-cell trait-positive patients. Renal cancer associated with end-stage acquired cystic kidney disease is regarded as an independent histological subtype.

Presentation and Diagnosis The two most common renal cell carcinomas are clear cell and papillary type. Clear cell renal cell carcinomas (ccRCCs) are typically sporadic solitary globular tumors protruding from the renal cortex in either kidney. The border with the kidney is usually sharp, with a pseudocapsule. Multifocality and/or bilaterality are rare, anddtogether with an early age onsetdare characteristic for familial cancer syndromes, such as von Hippel-Lindau (VHL) syndrome. Papillary renal cell carcinoma (PRCC) occurs in the renal cortex and may be multifocal. Multiple tumors may occur in association with renal scarring. Multiple and/or bilateral tumors are also characteristic for hereditary disease. PRCC is often well circumvented with a pronounced pseudocapsule and varies in color, depending on the degree of intratumoral hemorrhage. It usually has a friable consistency. Larger tumors may contain fibrosis as well as foci of necrosis and/or cystic degeneration. The most common RCC symptoms are hematuria and flank pain. However, the classic triad of flank pain, visible hematuria, and palpable abdominal mass is rare (6%–10%) in earlier disease stages. About 60%–80% of kidney tumors are detected incidentally on ultrasound, computed tomography (CT), or magnetic resonance imaging (MRI), many of them at an asymptomatic stage. Laboratory evaluation includes a complete blood count and a comprehensive metabolic panel which may include serum corrected calcium, serum creatinine, liver function studies, and urinalysis. CT of the abdomen, with or without pelvic CT, and chest x-ray are essential studies in the initial workup. For metastatic evaluation, chest radiography must be performed at the very least,

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65311-4

325

326

Kidney Cancer: Diagnosis and Treatment

although chest CT is more accurate than radiograph for chest staging. Abdominal MRI is used to evaluate the inferior vena cava if tumor involvement is suspected, or it can be used instead of CT for detecting renal masses and for staging when contrast material cannot be administered. As the recommended abdominal imaging studies provide high diagnostic accuracy, a needle biopsy is not always necessary before surgery. Histological diagnosis includes, besides difining the RCC type, an evaluation of the nuclear grade, sarcomatoid features, vascular invasion, tumor necrosis, and invasion of the collecting system and peri-renal fat as well as defining pT and pN categories. A variety of grading systems for renal cell neoplasia have been proposed, the most widely used being the Fuhrman grading system. However, none of these systems has been validated for many of the most recently described renal cell carcinoma morphotypes and using the four-tiered Wolrd Health Organization/International Society of Urological Pathology (WHO/ISUP) grading system is therefore recommended.

Epidemiology and Risk Factors Kidney Cancer Burden Kidney cancer is the 13th most common cancer in the world and the 9th most frequent in men. In 2012, 214,000 new cases were reported in men and 124,000 in women. Approximately 70% of cases occurred in countries with high and very high levels of socioeconomic development. Renal cell carcinoma is more frequent in men than in women and is rare in children. With 143,000 deaths from kidney cancer (91,000 in men and 52,000 in women) in 2012, it was the 16th cause of cancer death worldwide (Fig. 1). Both incidence and mortality rates have been increasing in many countries, across different levels of socioeconomic development. The number of new kidney cancer cases is predicted to rise by 184,064 by the year 2030, and the number of associated deaths by 86,754.

Etiology and Risk Factors Renal cell carcinoma is more common in men than in women (approximately 2:1), with the peak incidence between 60 and 70 years of age. Approximately 2%–4% of renal carcinomas are hereditary and having a first-degree relative with RCC roughly doubles the risk of developing the disease. Hereditary clear cell RCCs are rare. Chromosome 3p14.2 was suggested as a genomic region containing a kidney cancer susceptibility gene as chromosomal translocations involving this region have been found in

Fig. 1 Incidence and mortality of kidney cancer worldwide. (A) Age-standardized incidence and mortality rates (ASR) by gender and geographical area. (B) Incidence distribution worldwide. From Ferlay, J., et al. (2013). GLOBOCAN 2012 v1.0, cancer incidence and mortality worldwide: IARC CancerBase No. 11 [Internet]. Lyon, France: International Agency for Research on Cancer. Available from http://globocan.iarc.fr (accessed February 20, 2018).

Kidney Cancer: Diagnosis and Treatment Table 1

327

Risk factors for renal carcinoma Carcinogenic agents classified by IARC Monographs on the Evaluation of Carcinogenic Risks to Humans (volumes 1–120) Agents of sufficient evidence for kidney carcinogenicity in humans Tobacco smoking X-radiation and gamma-radiation Trichloroethylene Plants containing aristolochic acid Phenacetin and phenacetin-containing analgesic mixtures Agents of limited evidence for kidney carcinogenicity in humans Arsenic and inorganic arsenic compounds Cadmium and cadmium compounds Perfluorooctanoic acid Printing processes Welding fumes Aristolochic acid Other carcinogenic agents and lifestyle risk factors Obesity Hypertension and the use of antihypertensive drugs

a small proportion of families with hereditary renal cell carcinoma (HRCC) and a high prevalence of somatic loss of 3p heterozygosity was identified by tumor studies. However, no causative gene has been identified so far. Several familial syndromes are associated with an increased risk of renal tumors. The most common is the von Hippel-Lindau (VHL) syndrome associated with germline mutations in the VHL gene (3p25.3), which predisposes to multiple bilateral renal cell carcinomas and renal cysts. The gene for the hereditary nonpapillary renal cell carcinoma, however, is thought to be different form the VHL gene. Smoking, obesity (body fatness) and hypertension are established risk factors for the development of RCC. The proportion of all renal cancer cases attributable to overweight and obesity in the United States and in European countries has been estimated to be approximately 40%. Hypertension and the use of antihypertensive medication are associated with an increased risk of RCC independently of obesity. Additionally, a number of occupational, lifestyle and health-related exposures have been suggested as potential risk factors for RCC development (Table 1). Finally, the incidence of renal cancer among patients with end-stage acquired cystic kidney disease is reported to be markedly increased.

Pathology and Genetics The gene which is by far most frequently altered in kidney cancer is the von Hippel-Lindau (VHL) gene which is a recessive tumor suppressor gene controlling cellular oxygen sensing. Clear cell renal cell carcinoma (ccRCC), the most common type of RCC, is closely associated with VHL mutations which lead to stabilization of hypoxia-inducible factors (HIF-1a and HIF-2a) in both sporadic and familial forms. In sporadic ccRCC, this gene has also been found to be inactivated by epigenetic silencing and loss of the 3p chromosome. Overall, this gene is biallelilcally inactivated in a majority of ccRCCs and the loss of the VHL protein function appears to be a critical event in renal carcinogenesis, contributing to tumor initiation, progression and metastasis. Alterations in genes controlling the maintenance of chromatin states are also frequent in ccRCC. PBRM1, a subunit of the PBAF SWI/SNF chromatin remodeling complex has been reported to be mutated in 40%–50% of ccRCCs, whereas histone deubiquitinase BAP1 and histone methyltransferase SETD2 in 15% and 12% of cases, respectively. BAP1-mutant tumors are typically high-grade and have poor survival. Whole-exome sequencing of ccRCC tumors from 417 patients has identified 19 significantly mutated genes, with VHL, PBRM1, SETD2, KDM5C, PTEN, BAP1, MTOR, and TP53 being the eight most commonly altered genes. At the level of chromosome aberrations, deletions of chromosome 3p have been shown to be very frequent in ccRCC (91% of samples in a recent study by the Cancer Genome Atlas (TCGA) Research Network). This locus encompasses all the four most commonly mutated genes: VHL, PBRM1, BAP1 and SETD2. Arm level losses on chromosome 14q associated with loss of HIF1A are also frequent (45% of ccRCCs samples in the recent TCGA study) and have been associated with a more aggressive disease and a worse prognosis. Gains of 5q copy numbers are found in approximately 70% of ccRCC cases. Additional focal amplifications refined the region of interest to 60 genes in 5q35. Focally deleted regions included among others the tumor suppressor genes CDKN2A at 9p21 and PTEN at 10q23. The PI3K/AKT pathway was also shown to be altered by specific DNA methylation events. Overall, the most recent data have highlighted the importance of the well-known VHL/HIF pathway, the newly emerging chromatin remodeling/histone methylation pathway, and the PI3K/AKT pathway in ccRCC. The latter presents a strong therapeutic target, supporting the potential value of MTOR and/or related pathway inhibitor drugs for this type of cancer. Of note, despite a proven high intratumoral genetic heterogeneity of renal tumors, the most common gene-level alterations seem to be shared

328

Kidney Cancer: Diagnosis and Treatment

(A)

(B)

VEGF/VEGFR inhibitors

Oxygen Sensing Angiogenesis

mTOR inhibitors

VHL HIF1a

PTEN MTOR

Bioenergecs Metabolism

PRMB1 BAP1 SETD2 KDM5C Chroman Remodelling Histone methylaon Ectopic gene expression Epidrugs Immunotherapies? Fig. 2 Genetic alterations in renal cell carcinoma (RCC). The tumor develops by accumulating alterations in three major categories of genes, each of them containing potential therapeutic targets (A). Some alterations (e.g., VHL inactivation) occur early in the tumorigenesis and are common to all cells within a tumor. However, further progression results in the development of multiple clones with different alterations and an overall high intratumoral heterogeneity of the tumor (B).

between different areas of the same tumor (Fig. 2). However, studies evaluating the heterogeneity of tumors with regard to specific alterations will be needed to validate their utility as biomarkers or therapeutic targets. Compared to ccRCC, the genetic background of other, less common, subtypes of renal cell carcinoma is less known. PRCC is typically associated with trisomy and tetrasomy of chromosome 7, trisomy of chromosome 17, and loss of the Y chromosome. Chromosome 7 contains the locus of the MET tyrosine kinase gene (7q32) whose heterozygous germline mutations are associated with the hereditary form of PRCC. The MET gene may also be altered by other mechanismsdsomatic MET mutations have been reported in a proportion of sporadic PRCC cases. Moreover, a number of chromosomal aberrations other than the most typical ones have also been shown in PRCCs, with the type 1 and type 2 tumors showing different alterations (Table 2).

Biomarkers The range of histologies and tumor phenotypes a renal mass can represent is a challenge for the diagnostics and treatment. Renal tumors are genetically diverse with variable prognoses and treatment response rates even within one histological subtype. Therefore, precise molecular differentiation of renal carcinomas would be key to proper treatment. However, while imaging biomarkers (see the section “Presentation and diagnosis”) are reliable tools for early detection of kidney tumors, identifying prognostic and predictive biomarkers which would characterize the tumor aggressiveness and metastatic potential on an individual level remains an unmet need. Until now, risk stratification in localized tumors is based on T-category and grading but this does not reliably predict metastases. Many putative prognostic and predictive biomarkers have been published for ccRCC, including in particular molecules involved in angiogenesis, inflammatory response and the regulation of cell cycle. Baseline serum levels of vascular endothelial growth factor (VEGF) which is key to the angiogenesis of RCC and is targeted by many anticancer drugs have been shown to be prognostic of overall survival in some studies. It has also been suggested it might be a predictive biomarker for sorafenib treatment benefit in RCC patients. High baseline levels of plasma carbonic anhydrase IX (CA-IX), tissue inhibitor of metalloproteinase 1 (TIMP-1), and Ras p21 were shown to be prognostic of reduced overall survival in metastatic RCC patients but not predictive of the sorafenib benefit. Low levels of IL-8, hepatocyte growth factor (HGF), osteopontin and TIMP-1 were all shown to be associated with significantly longer progression-free survival in patients treated with pazopanib. In patients receiving placebo, IL-6, IL-8 and osteopontin were all strong prognostic markers, while IL-6 levels appeared to be predictive of pazopanib treatment benefit. Moreover, Hypoxiainducible factor (HIF), Ki67 (proliferation), p53, phosphatase and tensin homolog (PTEN; cell cycle), E-cadherin, and other molecules detectable in serum, urine, or circulating free DNA obtained from peripheral blood have also been investigated. However, none of all these biomarkers has been shown to clearly improve the predictive accuracy of current prognostic systems, none has been externally validated, and as of now their routine use in clinical practice is not recommended. It becomes clear that we need to use molecular signatures rather than single-molecule biomarkers. A 16-gene signature based on the analysis of the expression of genes involved in angiogenesis (APOLD1, EDNRB, NOS3, PPAP2B), cell growth/division

Kidney Cancer: Diagnosis and Treatment Table 2

329

Gene loci that are most frequently altered in renal carcinoma

Pathway

Gene (locus)

– – –

Loss of chromosome 3p Gains of 5q copy numbers Loss of chromosome 14q, associated with HIF1A deletion VHL (3p25.3)

HIF-1 signaling (response to hypoxia) Chromatin remodeling/histone methylation

P13K/PTEN/Akt: control of cell metabolism C-Met signaling

PBRM1 (3p21.1) BRCA1-associated protein BAP1 (3p21.1) SETD2 (3p21.31) Lysine-specific demethylase KDM5C (Xp11.22) PTEN (10q23) MTOR (1p36.22) MET protooncogene (7q31.2)

Alteration type

Estimate Up to 91% of ccRCCs Approximately 70% of ccRCCs Up to 45% of ccRCCs

Biallelic inactivation by germline or somatic mutations, loss of chromosome, and epigenetic silencing by promoter methylation Somatic mutations Somatic mutations

Somatic mutations in nearly 40% of all kidney tumors and in 83% of RCCs

Somatic mutations Somatic mutations

40%–50% of ccRCCs 15% of ccRCCs; associated with higher grade and poor prognosis 12% of ccRCCs 4%–9% of ccRCCs

Somatic mutations, deletions Somatic mutations Hyperactivation by point mutations and/or chromosome amplifications

6%–25% of ccRCCs 7% of ccRCCs Somatic mutations in approximately 13% of sporadic PRCC cases

ccRCC, Clear cell renal cell carcinoma; PRCC, Papillary renal cell carcinoma; RCC, Renal cell carcinoma.

(EIF4EBP1, TUBB2A, LMNB1), immune response (CEACAM1, CX3CL1, CCL5), and inflammation (IL-6), plus reference genes (AAMP, ARF1, ATP5E, GPX1, RPLP1) elaborated by Rini and collaborators has been reported to predict renal carcinoma recurrence and reliably stratify patients within disease subgroups by recurrence risk. This assay is particularly promising as there seem to be little intratumoral heterogeneity in the expression of these 16 genes at the RNA level. Brannon and collaborators have described a gene signature which allows to individually classify ccRCC tumors into two distinct molecular subtypes: ccA and ccB, based on their gene expression profiles which are associated with different clinical outcomes and could potentially be used to guide treatment decisions. The ccA group was shown to relatively overexpress genes associated with hypoxia, angiogenesis, fatty acid metabolism, and organic acid metabolism, whereas ccB tumors overexpressed a more aggressive panel of genes that regulate the epithelial-mesenchymal transition (EMT), the cell cycle, and wound healing. A recent study has validated these signatures as being independently associated with survival, with ccA patients having a markedly better prognosis. Based on these results, a 34-gene classifier (ClearCode34) for assigning ccRCC tumors to the two subtypes has been developed and validated as having an added prognostic value in patients with localized tumors. Combining this molecular profiling with standard clinical tumor evaluation criteria may allow to stratify patients according to the recurrence risk. However, prospective studies with large cohorts of patients will be needed to fully refine the integrated prognostic algorithm. Moreover, a substantial intratumoral heterogeneity so as to these two molecular subtypes has been shown, which implies that multiple biopsies from the same tumor will have to be used to identify clones of poor prognosis. Therefore, the accuracy of the putative prognostic signatures depending on the tumor size and the usefulness of biomarkers based on biopsy material should be evaluated. Finally, the clinical relevance of this model to metastatic tumors has yet to be investigated. Another group has suggested using the so called total number of aberrations (TNSA) score, a signature of four copy number alterations evaluated using a fluorescence in situ hybridization (FISH) test to identify primary tumors with a high metastatic potential. This putative biomarker has been validated in two independent cohorts and has been shown to independently predict metastasis and correlate with survival. A prospective analysis of a larger multicenter cohort is necessary to define the degree of influence of each specific aberration on the different survival end-points and to develop a multivariate prognostic model for ccRCC, integrating each genomic aberration and well-established prognostic parameters such as T-stages, tumor size, nodal status, and histological grade. Our knowledge on putative molecular biomarkers in RCC subtypes other than clear cell type is very limited. In papillary RCC, we have to distinguish two subtypes: type 1 with good prognosis and type 2 which is much more aggressive and associated with a shorter survival. Recently, it has been suggested that the aggressive subtype 2 of the PRCC is associated with hypermethylation, 9p alterations, specific miRNA changes anddto some extentdFH mutations. However, the development of corresponding prognostic biomarkers remains at an early investigational stage. Overall, it seems possible to identify aggressive renal cell carcinomas by analyzing the molecular signatures in tumor samples and so stratify patients into groups with different prognosis based on a specific molecular background. However, more studies are needed to validate the putative biomarkers suggested so far and develop clinically relevant algorithms for integrating them into prognostic patient stratification models. In addition, the issue of intratumoral heterogeneity needs to be meticulously addressed. Liquid biopsies containing free nucleic acids as well as extracellular vesicles may help overcome this problem but their utility as prognostic biomarkers in addition to their diagnostic potential also needs to be evaluated.

330

Kidney Cancer: Diagnosis and Treatment

Management and Therapy For patients with surgically resectable RCC, the standard of care is surgical excision with a curative intent. For early stage (T1) localized RCC, the treatment of choice is partial nephrectomy (PN). Compared with radical nephrectomy, this nephron-sparing surgery better preserves kidney function and limits the development of metabolic and cardiovascular disorders in a long term. If there is clinical evidence of the adrenal gland invasion, adrenalectomy should simultaneously be performed. In patients with adverse clinical features (a large diameter of the primary tumor or sarcomatoid histological features), extended lymph node dissection (LND) should be considered. Patients with T2 tumors and localized masses not treatable by partial nephrectomy as well as patients not fit for PN for other reasons (e.g., due to the use of anticoagulants) are offered laparoscopic radical nephrectomy. Elderly and comorbid patients with incidental small renal masses, who have a low risk of RCC-specific mortality and a significant competing-cause mortality risk, may be managed by alternative (nonsurgical) approaches. Active surveillance defined as the initial monitoring of tumor size by serial abdominal imaging (ultrasound, CT, or MRI) should be considered at first, with delayed intervention using ablative therapies (cryoablation or radiofrequency ablation performed laparoscopically or percutaneously) reserved for tumors showing clinical progression during follow-up. In patients with locally advanced nonresectable tumors, embolization can control symptoms, such as visible hematuria or flank pain. The use of neoadjuvant targeted therapy to downsize tumors is still experimental and cannot be recommended outside clinical trials. In case of clinically positive lymph nodes, lymph node dissection is always justified and can add staging information. LND is also recommended for staging purposes in case of enlarged lymph nodes. However, the extent of lymph node dissection remains controversial. Tumor resection is curative only if all tumor deposits are excised. For most patients with metastatic disease, cytoreductive nephrectomy (CN) is palliative and systemic treatments are necessary. However, metastatic RCC is often resistant to conventional chemotherapy. Therefore, inoperable or metastatic RCCs are typically managed by systemic treatment with targeted agents and/or immune checkpoint inhibitors. Over the past decade, a number of targeted therapies have been shown to provide a benefit in the treatment of advanced RCC. These therapies include small molecules or monoclonal antibodies inhibiting signaling through the vascular endothelial growth factor (VEGF) and its receptor (VGFR): sunitinib, sorafenib, bevacizumab, pazopanib, axitinib, cabozantinib and lenvatinib. These treatments, as mono-agent or combined regimens, are the current standard of care as first or second-line systemic therapy for advanced RCC. Single-agent treatment with inhibitors of mTOR signaling, such as everolimus and temsirolimus, is an approved second-line treatment. Current clinical trials are investigating combinations of anti-VEGF therapy with immunotherapy using checkpoint inhibitors, such as monoclonal antibodies against the programmed cell death protein 1 (PD1; e.g., nivolumab, pembrolizumab) and its ligand (PDL1; avelumab, atezolizumab). Another combination under investigation is nivolumab with ipilimumab, an antibody inhibiting the cytotoxic T lymphocyte-associated protein 4 (CTL4). Blocking T-cell inhibitory signals by these antibodies promotes T-cell activation and antitumor response, providing an overall survival benefit in patients who have failed to respond to first- or second-line therapy with small drug inhibitors.

Prospective Vision In the early 2000s, there were very limited options for the treatment of metastatic kidney cancer and the median patient survival was about 15 months. Introducing small drug inhibitors of VEGF-VEGFR and mTOR pathways in the past decade has caused the median survival to double. Using combinations of these drugs with checkpoint immunotherapies which are being developed is expected to further improve these significant benefits, achieving durable remission in a high number of patients. Progressing towards this objective will, however, require further advances in diagnosing and managing high-risk patients as well as development of systemic therapies. A major gap in the current knowledge is the lack of suitable biomarkers for predicting treatment responses and informing therapeutic decisions. One of the main areas for research to this effect is exploiting the data generated by integrative approaches combining genomics, transcriptomics and epigenomics in order to identify reliable predictive biomarkers that take into account the dynamics of intratumoural heterogeneity. This would enable a better management of off-target effects causing severe toxicities as well as provide biomarker-based rationales for combination regimens allowing to overcome drug resistance.

See also: Renal Cell Cancer: Pathology and Genetics.

Further Reading Brannon, A.R., et al., 2010. Molecular stratification of clear cell renal cell carcinoma by consensus clustering reveals distinct subtypes and survival patterns. Genes & Cancer 1, 152–163. Brooks, S.A., et al., 2014. ClearCode34: A prognostic risk predictor for localized clear cell renal cell carcinoma. European Urology 66 (1), 77–84. Cancer Genome Atlas Research Network, 2013. Comprehensive molecular characterization of clear cell renal cell carcinoma. Nature 499 (7456), 43–49. European Association of Urology. Guidelines on Renal Cell Carcinoma, 201x: http://uroweb.org/guideline/renal-cell-carcinoma/ (accessed on February 5, 2018).

Kidney Cancer: Diagnosis and Treatment

331

Farber, N.J., 2017. Renal cell carcinoma: The search for a reliable biomarker. Translational Cancer Research 6 (3), 620–632. Moch, H., Humphrey, P.A., Ulbright, T.M., Reuter, V.E. (Eds.), 2016WHO classification of tumors of the urinary system and male genital organs, 4th edn, vol. 8. International Agency for Research on Cancer, Lyon, France. National Comprehensive Cancer Guidelines (NCCN). NCCN Clinical Practice Guidelines in Oncology. Kidney Cancer. Version 2.2018 – November 30, 2017: https://www.nccn.org/ professionals/physician_gls/default.aspx#site. Pastore, A.L., et al., 2015. Serum and urine biomarkers for human renal cell carcinoma. Disease Markers 2015, 251403. Rni, B., et al., 2015. A 16-gene assay to predict recurrence after surgery in localized renal cell carcinoma: Development and validation studies. The Lancet Oncology 16 (6), 676–685. Sanjmyatav, J., et al., 2014. Identification of high-risk patients with clear cell renal cell carcinoma based on interphase-FISH. British Journal of Cancer 110, 2537–2543. Serie, D.J., et al., 2017. Clear cell Type A and B molecular subtypes in metastatic clear cell renal cell carcinoma: Tumor heterogeneity and aggressiveness. European Urology 71 (6), 979–985.

Relevant Websites http://www.auanet.org/dAmerican Urological Association. www.euroweb.orgdEuropean Association of Urology.

Laryngeal Cancer: Diagnosis and Treatment Fernando Lo´pez A´lvarez and Juan Pablo Rodrigo, Asturias Central University Hospital, University of Oviedo, University Institute of Oncology of Asturias (IUOPA), ISPA, CIBERONC, Oviedo, Spain © 2019 Elsevier Inc. All rights reserved.

Glossary Glottis The glottis extends from an artificial horizontal plane extending bilaterally across the apex of the laryngeal vestibule to approximately 1 cm below the true vocal folds. Laryngectomy Partial (partial laryngectomy) or complete (total laryngectomy) surgical removal of the larynx, usually as a treatment for cancer of the larynx. Laryngoscopy A procedure to look at the larynx and pharynx for abnormal areas. A mirror or a rigid or flexible laryngoscope (a thin, tube-like instrument with a light and a lens for viewing) is inserted through the mouth or the nose (fibroscope) to see the larynx. A special tool on the laryngoscope may be used to take samples of tissue. Supraglottis The supraglottis extends from the tip of the epiglottis to an artificial horizontal plane extending bilaterally across the apex of the laryngeal vestibule, including the epiglottis, the false vocal cords (ventricular bands), the ventricles, the aryepiglottic folds and the arytenoids. Subglottis The lower part of the larynx between 1 cm below the true vocal folds and the trachea (windpipe). Tracheostomy The surgical formation of an opening into the trachea through the neck to allow the passage of air. The term “tracheotomy” refers to the incision into the trachea that forms a temporary or permanent opening, which is called a “tracheostomy,” however; the terms are sometimes used interchangeably.

Introduction Cancers of the larynx represents approximately 0.8% of the total cancer risk and is the second most frequent head and neck cancer. The number of new cases of cancer (cancer incidence) is 3.1 per 100,000 men and women per year (based on 2010–14 cases). The American Cancer Society’s most recent estimates for laryngeal cancer in the United States for 2017 are that about 13,360 new cases of laryngeal cancer (10,570 in men and 2790 in women) will be diagnosed and about 3660 people (2940 men and 720 women) will die from laryngeal cancer, which is 0.6% of all cancer deaths. Cancer of larynx is strongly related to cigarette smoking. The role of alcohol in inducing laryngeal cancer remains unclear, although it may be play a role specially in supraglottic cancer. More than 95% of laryngeal tumors are squamous cell carcinomas (SCC). Other laryngeal malignancies include: verrucous SCC, papillary SCC, nonkeratinizing HPV-positive SCC, spindle cell carcinomas, basaloid SCC, lymphoepithelial carcinomas, acantholytic SCC, adenosquamous carcinomas, salivary tumors and sarcomas, among others. Glottic, supraglottic, and subglottic cancers represent approximately two-thirds, one-third, and 2% of laryngeal cancers, respectively. Tumors that extend past the ventricle from either a glottic or supraglottic primary are considered transglottic tumors. Nearly 75% of patients with glottic carcinoma present with hoarseness and often they are diagnosed at an early stage. However, nearly 70% of patients with supraglottic and subglottic laryngeal cancers usually present with advanced disease due to a paucity of symptoms, propensity for local extension, and rich lymphatics resulting in a high incidence of lymph node metastases (mainly supraglottic tumors). Early diagnosis can lead to a reduction in mortality.

Diagnosis The initial evaluation of patients with suspected laryngeal carcinoma needs to include a detailed medical history to assess clinical risk factors (history of tobacco and alcohol use and environmental exposures) and symptom severity. Moreover, a complete head and neck physical examination should undergo, including examination of all mucosal sites of the upper aerodigestive tract, due to the frequent occurrence of multiple primary tumors in patients with a head and neck tumor. It is advisable an assessment of speech and swallowing function, communication needs, nutrition, health behaviors and availability of social support for what is typically prolonged period of treatment and rehabilitation. Finally, pretreatment dental evaluation is also recommended for patients who will undergo radiation, given the risk of dental infection, damage, and treatment-induced osteoradionecrosis. Then, if a malignant tumor is suspected, it must be confirmed by obtaining a biopsy of the lesion for histopathological study. Finally, radiologic imaging is performed to obtain precise anatomical details regarding the tumor localization and extension. Fig. 1 represent an algorithm for diagnostic evaluation of a laryngeal cancer.

332

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65318-7

Laryngeal Cancer: Diagnosis and Treatment

333

Fig. 1 Algorithm for diagnostic evaluation of a laryngeal lesion. CT, computed tomography; MRI, magnetic resonance imaging; PET, positron emission tomography.

Clinical Evaluation Patients with primary early-stage tumors arising on the true vocal cords usually present with complaints of dysphonia and hoarseness; sore throat, odynophagia, dysphagia, referred otalgia, pain localized to the thyroid cartilage, hemoptysis, airway obstruction, stridor and neck adenopathy are features of advanced-stage glottic lesions. Early diagnosis is crucial for both improved survival and laryngeal preservation. In general, persistent hoarseness (> 15 days) in an adult patient requires endoscopic visualization of the larynx to rule out a tumor. Supraglottic tumors are usually more advanced than glottic cancer at presentation and patients may experience the following symptoms or signs: discomfort or sensation of a lump in the throat (the most frequent initial symptom), dysphagia, odynophagia, the sensation of something stuck in the throat, occasional respiratory obstruction, hemoptysis, referred pain to the ipsilateral ear (by way of the vagus and auricular nerves) or a mass in the neck; however, hoarseness is not a prominent symptom of supraglottic cancer until the lesion becomes quite extensive. Primary subglottic tumors are usually asymptomatic at an early stage and they do not present until more advanced in size where dyspnea ad stridor are frequent. Dysphonia is relatively common in advanced subglottic tumors due to the involvement of the true vocal cord or direct extension to the recurrent laryngeal nerve. An emergent tracheotomy is not uncommon in these patients. Late symptoms of laryngeal cancer include weight loss, dysphagia, foul breath and aspiration.

Physical Examination After the medical history, a thorough clinical examination using a rigid or flexible fiber-optic endoscope is performed (Fig. 2A). It allows an adequate assessment of the surface extent of the primary tumor, the mobility of the vocal cords and performing an initial tumor staging. Accurate evaluation of the primary tumor prior to initiating treatment is vitally important, especially if any voicesparing surgical procedure is a consideration. It is advisable to document the lesions of the larynx by a photograph or to depict them on a drawing to describe the site of origin of the primary tumor with its local extension to adjacent sites within the same region of the larynx or from one region to another region. Bulky lesions may extend beyond the larynx into the adjacent base of the tongue, pyriform sinus, or retrocricoid region and this should be assessed. The relation to the anterior commissure is very important in glottic tumors and in lesions involving the laryngeal surface of the epiglottis. Determination of the mobility of the vocal cords is critical because there are subtle distinctions between mobile, partially fixed and fixed cords. Fixation of the vocal cord from laryngeal cancer is caused by invasion or destruction of the vocal cord musculature, invasion of the cricoarytenoid muscle or joint or invasion of the recurrent laryngeal nerve. Ulceration of the infrahyoid epiglottis, fullness of the vallecula or palpation of a diffuse, firm fullness above the thyroid notch with widening of the space between on the hyoid and the thyroid cartilages are indirect signs of preepiglottic space invasion. Retrocricoid invasion may be suspect when the laryngeal click disappears on examination or when the thyroid cartilage protrudes interiorly, producing a fullness of the neck. Pain or tenderness to palpation or a small bulge over the thyroid cartilage is suggestive of thyroid invasion.

334

Laryngeal Cancer: Diagnosis and Treatment

Fig. 2 (A) Endoscopic view of the larynx showing a tumor of the supraglottic. (B) Axial contrast enhanced CT showing an advanced supraglottic carcinoma. Moderately thick lesion arising from laryngeal surface of epiglottis with involvement of preepiglottic space.

Nowadays, apart from the endoscopy with conventional white light, narrow band imaging (NBI) endoscopy should be use in the routine pharyngo-laryngeal examination because it helps to make an early diagnosis of tumors of the larynx. Moreover, it is a useful tool in the postoperative surveillance to detect recurrences. NBI may be used in both flexible fiberoptic or rigid laryngoscopes. NBI is an imaging technique for the depiction of tumor-specific neoangiogenesis. A higher contrast between the mucosal epithelium and blood vessels is achieved in NBI endoscopy using filtered light comparing to white light observations. This allows detection of small suspicious mucosal changes or small tumors, few millimeters in diameter, which are not observable using white light. NBI also allows to define better the extension of tumors, which is crucial for perform targeted biopsy and for determination of resection margins in cancer surgery. NBI is increasingly used for follow-up of patients after treatment for head and neck malignant tumors, when early detection of possible recurrence is crucial. Other imaging techniques such as auto fluorescence, contact endoscopy and optical coherence tomography (OCT) are increasingly used in ENT practice. However, these techniques have not been reproduced and is consequently not a standardized method. Sometimes it is necessary to carry out a direct laryngoscopy under general anesthesia. The ventricles, subglottic area, apex of the pyriform sinus and retrocricoid area must be carefully examined when these areas are not well examined by the above techniques and they are at risk of being affected by the tumor. Through the laryngoscope may be introduced rigid telescopes (0, 30,45 or 70 degrees) or microscopic examination can be performed. A detailed examination of the neck is mandatory whenever a laryngeal tumor is suspected. Regional metastases are the most important negative prognostic factor, decreasing the survival rate by 50%. The subdigastric basin (level II) is the most frequently involved. The submandibular area (level I) is rarely affected and the risk of spinal accessory lymph node (level V) involvement is small. Anterior commissure and anterior subglottic invasion are associated with involvement of the Delphian node. In tumors of the vocal cord, the incidence of clinically positive lymph nodes at diagnosis approaches 0% for T1 lesions and 2% for T2 lesions. This incidence increases from 20% to 65% for T3 and T4 lesions. The supraglottic tumors tend to have a higher rate of lymphatic metastasis and the incidence of clinically positive nodes is up to 60% at the time of diagnosis and 15% are bilateral. Spread to the pyriform sinus, vallecula and the base of tongue increases the risk of lymph node metastases. The incidence of distant metastases is generally thought to be 10% or less. Although this increases with locoregional extent of disease and is more common with supraglottic (and subglottic) tumors, distant metastases are still found in only a small number of patients diagnosed with laryngeal cancer (15% or less). If spread through the bloodstream does occur, the lungs are the most common site of metastasis, followed by the bones.

Histopathological Diagnosis Biopsy should be performed at the time of diagnosis, ideally after imaging, but it should not delay the treatment. A correct histologic diagnosis is critical due to the impact on therapeutic approach and prognosis and highlights the need for both a high index of clinical suspicion and adequate representative biopsies. A generous biopsy specimen is taken from the suspicious areas, looking to take tissue samples in depth and avoiding areas of necrosis. The biopsy should be performed through a flexible fiberoptic endoscope or, if this is not possible, by direct laryngoscopy.

Imaging When malignancy is suspected, computed tomography (CT) scan with contrast enhancement (Fig. 2B) and/or magnetic resonance imaging (MRI) must be performed to define the submucosal extent and deeper margins of the tumor. CT is preferred because it is

Laryngeal Cancer: Diagnosis and Treatment

Fig. 3

335

Axial contrast enhanced CT performed in a patient with a tumor invading the anterior commissure and penetrating through the cartilage.

much faster than MRI. Small and superficial mucosal tumors may not be appreciated at CT or MRI and hence, it is mandatory that an endoscopy is done prior to any imaging study. Integration of cross-sectional imaging with endoscopy findings significantly improves the accuracy of T staging. The accuracy of clinical T staging alone for laryngeal cancer to be 57.5%, but as high as 80% when combined with contrast-enhanced CT. CT and/or MRI are valuable to obtain precise anatomical details regarding the tumor localization and extension (preepiglottic and paraglottic space involvement, subglottic extension, cartilage destruction and submucosal and extralaryngeal extension, which may not be obvious on clinical examination) (Fig. 3). CT and/or MRI of the entire neck provide information on nodal disease that may not be clinically evident. A minimum axial diameter more than 10 mm, round or spherical shape, a necrotic node of any size and a node with indistinct spiculated margins (suggesting extra nodal disease spread) are the generally accepted radiological criteria to diagnose malignant nodes at CT and MRI. The sensitivity and specificity of CT to detect nodal disease using these criteria are 90% and 75% respectively. It is important to note that imaging should be performed before biopsy so that abnormalities that may caused by the biopsy are not confused with tumor. The general radiological criteria used for tumor involvement include asymmetric soft tissue prominence or thickening, abnormal contrast enhancement, a bulky mass, obliteration of the normal fat planes and spaces, or a combination of these. A high-resolution CT scan with contrast enhancement (1–2 mm thick sections through the larynx) is generally preferred. Contrast enhancement helps to outline the blood vessels and the tumor. CT delineates the extent of disease and the presence and extent of lymphatic involvement. CT offers high spatial resolution; discriminates among fat, muscle, bone, and other soft tissues; and surpasses MRI in the detection of bony erosion. Nevertheless, MRI is probably better to define subtle extralaryngeal extension and to assess early thyroid cartilage destruction. MRI has a high sensitivity (89%–95%) but lower specificity (74%– 84%) as compared to CT for the detection of cartilage invasion. The negative predictive value of MRI to exclude cartilage invasion is also very high, at around 94%–96%. T2-weighted or postcontrast T1-weighted cartilage signal intensity greater than that of the adjacent tumor is considered to indicate inflammation, and signal intensity similar to that of the adjacent tumor was considered to indicate neoplastic invasion. Supraglottic tumors may arise in the epiglottis or in the aryepiglottic folds and false cords. Epiglottic tumors primarily invade into the preepiglottic space. While the tumors arising from the mobile portion of the epiglottis may spread from the preepiglottic space further into the base of tongue and laterally into the paraglottic space, those arising from the petiole often invade the low preepiglottic space and via the anterior commissure, reach the glottis or subglottis. The primary sign of preepiglottic space invasion at imaging is replacement of the normal fat by abnormal enhancing soft tissue. The sensitivity of CT and MRI to detect invasion of the preepiglottic space is 100% and the corresponding specificities are 93% and 84–90%. Tumors arising in the aryepiglottic folds present as exophytic or infiltrative masses. They expand the aryepiglottic fold and spread into the paraglottic space. They may spread further anteriorly into the preepiglottic space or posteriorly to invade the piriform sinus. Tumors originating in the false vocal cords are lateral masses with a strong predilection for submucosal spread into the paraglottic space. More extensive tumor may destroy the thyroid cartilage and spread transglottically into the glottis and subglottis. Tumor spread to the paraglottic space on CT or MRI is seen as replacement of the normal paraglottic fat by the enhancing tumor tissue. Both CT and MRI have a high sensitivity of about 95% to detect paraglottic tumor spread, the specificity, however, ranges between 50 and 75% as peritumoral inflammation may mimic tumor resulting in false positive assessments. Glottic tumors commonly arise from the anterior half of the vocal cord and spread into the anterior commissure. Anterior commissural disease is seen on CT or MRI as soft tissue thickening of more than 1–2 mm. The accuracy of CT in predicting anterior commissure involvement is about 75%. From the anterior commissure, the tumor may spread further anteriorly into the contralateral cord and the thyroid cartilage or posteriorly into the posterior commissure, the arytenoids, cricoarytenoid joint and the cricoid. While vocal cord mobility is best assessed at endoscopy, disease in the cricoarytenoid joint and interarytenoid region have been described as imaging correlates for vocal cord fixation. The tumor may spread superiorly to access the preepiglottic space and

336

Laryngeal Cancer: Diagnosis and Treatment

Fig. 4 (A) Axial contrast enhanced CT depicting a supraglottic tumor invading the preepiglottic space. (B) PET-CT showing a hypermetabolic focus of active tumor occupying the preepiglottic space that corresponds to the CT abnormality.

the paraglottic space, or inferiorly to reach the subglottis. Subglottic spread below the anterior commissure is seen as an irregular thickening of the cricothyroid membrane. Tumor may gain access into the extralaryngeal tissues through the cricothyroid membrane. Subglottic cancer is diagnosed if any tissue thickening is noted between the airway and the cricoids ring. Due to their late presentation, invasion of the cricoids cartilage, trachea and the cervical esophagus with extralaryngeal spread are common findings in these patients at imaging. Laryngeal tumors encroaching on both, the glottis and supraglottis, with or without subglottic component and when the site of origin is unclear, is termed as transglottic tumor. This tumor spread is frequently through the paraglottic space and is readily identified on CT or MRI. Transglottic carcinoma is frequently accompanied by metastatic lymphadenopathy. Coronal images are particularly helpful in assessing transglottic extension of tumor. Nowadays, to assess the possibility of diagnosing distant disease and second primary tumors, a chest CT should be routine. Second primary malignancies have been reported in up to 20% of patients with laryngeal cancer. Moreover, the use of positron emission tomography (PET) scanning with 18F-fluorodeoxyglucose (FDG) and combined CT with PET imaging allow to combine functional and anatomical studies (Fig. 4). FDG imaging has the potential to distinguish between benign and malignant processes, grade tumors, identify metastases, and diagnose tumor recurrence. Although standardized uptake values (SUV) of FDG uptake has been used by some authors to identify tumoral tissue, to date, no universally accepted SUV threshold has been determined to differentiate benign from malignant nodal disease. In head and neck cancer, FDG imaging is useful for detecting clinically occult recurrences and for determining residual disease in the neck following definitive radiotherapy. CT-PET can detect occult disease and use alter the treatment plan in up to 30% of patients. CT-PET has a high negative predictive value but the specificity is not very high. The overall accuracy of in identifying nodal disease is higher than that of CT alone, by almost 20%. However, CT-PET is not useful to exclude the presence of metastases in the clinically N0 neck because PET cannot detect very small tumors (< 5–7 mm).

Staging Staging is based on TNM criteria developed by the AJCC/UICC (8th edition) (Tables 1–4). This classification incorporates all information available prior to treatment, including the clinical examination, endoscopy, endoscopic biopsy and cross-sectional imaging. The guidelines rely heavily on the use of cross-sectional imaging for the T staging; however, no recommendation is made regarding the preference of one technique over the other. Laryngeal cancer staging is dependent on tumor location and subsite involvement. For supraglottic tumors, the subsites that are important for staging include the suprahyoid and infrahyoid epiglottis, aryepiglottic folds, arytenoids and ventricular bands (false vocal cords). The glottic subsites include the true vocal cords including the anterior and posterior commissure. Other important factors involved in laryngeal cancer staging include the presence of impaired true vocal cord mobility or frank paralysis, base of the tongue involvement, preepiglottic space involvement, paraglottic space involvement and thyroid or cricoid cartilage invasion. It is important to note the fact that a tumor that erodes only the internal cortex of the thyroid cartilage is classified as T3, while if the tumor passes through the cartilage is considered as a T4 tumor. Those tumors that cause vocal cord fixation, involve base of the tongue preepiglottic or paraglottic spaces, show thyroid or cricoid cartilage invasion, invade tissues beyond the larynx (trachea, soft tissues of the neck, esophagus thyroid gland, strap muscles or extrinsic muscle of the tongue) or are associated with regional lymphatic metastases, are considered as advanced disease. Cancers that invade the prevertebral space, encase the internal carotid artery or extend into the mediastinum are considered incurable with surgery.

Laryngeal Cancer: Diagnosis and Treatment Table 1

Primary tumor (T) according to TMN 8th edition

TX

Primary tumor cannot be assessed

T0 Tis Supraglottis T1 T2

No evidence of primary tumor Carcinoma in situ

T3 T4a T4b Glottis T1 T1a T1b T2 T3

Tumor limited to one subsite of supraglottis with normal vocal cord mobility Tumor invades mucosa of more than one adjacent subsite of supraglottis or glottis or region outside the supraglottis without fixation of the larynx. Tumor limited to the larynx with vocal cord fixation and/or invades any of the following: postcricoid area, preepiglottic space, paraglottic space, and/or inner cortex of thyroid cartilage Moderately advanced local disease. Tumor invades through the thyroid cartilage and/or invades tissues beyond the larynx Very advanced local disease. Tumor invades prevertebral space, encases carotid artery, or invades mediastinal structures Tumor limited to the vocal cord(s) with normal mobility Tumor limited to one vocal cord Tumor involves both vocal cords Tumor extends to supraglottis and/or glottis, and/or with impaired vocal cord mobility Tumor limited to the larynx with vocal cord fixation and/or invasion of the paraglottics space, and/or inner cortex of the thyroid cartilage Moderately advanced local disease. Tumor invades through the outer cortex of the thyroid cartilage and/or invades tissues beyond the larynx Very advanced local disease. Tumor invades prevertebral space, encases carotid artery, or invades mediastinal structures

T4a T4b Subglottis T1 T2 T3 T4a T4b

Table 2

Tumor limited to the subglottis Tumor extends to vocal cord(s) with normal or impaired mobility Tumor limited to the larynx with vocal cord fixation Moderately advanced local disease. Tumor invades cricoid or thyroid cartilage and/or invades tissues beyond the larynx Very advanced local disease. Tumor invades prevertebral space, encases carotid artery, or invades mediastinal structures

Regional lymph nodes (N) according to TMN 8th edition Clinical (cN)

Nx N0 N1 N2a N2b N2c N3a N3b

337

Pathological (pN)

Regional lymph nodes cannot be assessed No regional lymph nodes metastasis Metastasis in a single ipsilateral lymph node, 3 cm or less in greatest dimension Metastasis in a single ipsilateral lymph node, more than 3 cm but not more than 6 cm in greatest dimension Metastasis in multiple ipsilateral lymph nodes, none more than 6 cm in greatest dimension Metastasis in bilateral or contralateral lymph nodes, none more than 6 cm in greatest dimension Metastasis in a lymph node more than 6 cm in greatest Metastasis in a lymph node more than 6 cm in greatest dimension without extranodal extension dimension without extranodal extension Metastasis in a single or multiple lymph nodes with clinical Metastasis in a lymph node more than 3 cm in greatest extranodal extension (skin involvement or soft tissue dimension with extranodal extension or, multiple invasion with deep fixation/tethering to underlying ipsilateral, or any contralateral or bilateral node(s) with muscle or adjacent structures or clinical signs of extranodal extension nerve involvement)

Table 3 M0 M1

Distant metastasis (M) according to TMN 8th edition No distant metastasis Distant metastasis

338

Laryngeal Cancer: Diagnosis and Treatment Table 4 Stage 0 Stage I Stage II Stage III Stage IVA Stage IVB Stage IVC

Stage groups according to TMN 8th edition Tis N0 M0 T1 N0 M0 T2 N0 M0 T3 N0 M0 T1–3 N1 M0 T4a N0–1 M0 T1-4a N2 M0 T4b Any N M0 Any T N3 M0 Any T Any N M1

Treatment The primary goal of treatment in head and neck cancer is to achieve optimal oncological outcomes while preserving function and quality of life as much as possible. This is particularly important in the treatment of laryngeal cancer. The larynx has important functionsdincluding breathing, voice making and swallowing, and therefore both the disease and its treatment may significantly affect quality of life. As previously mentioned, 95%–98% of malignant neoplasms of the larynx are squamous cell carcinomas (SCC), so we will only refer to this histological type. In the treatment of these cancers we have several options available (surgery, radiation therapy, chemotherapy and molecular targeted therapies), which can be used alone or in combination. In general, patients in early stages (I–II) are treated with a single modality therapy, either surgery of radiotherapy. But approximately 40% of laryngeal cancers occur at advanced stages (III–IV) at diagnosis, requiring aggressive treatment that usually involves combining several treatment modalities. Primary treatment decisions may depend on patient desires, tumor extent and location, experience of the treatment team, the adequacy of follow up surveillance, medical comorbidities and long-term voice and swallowing expectations. Since most cases of laryngeal cancer are associated with high consumption of tobacco and alcohol, co-morbidities are common, which not only determine the choice of treatment and compliance, but are an important determinant of overall patient survival.

Treatment of Early Stage (T1–T2) Tumors Treatment options for T1–T2N0 laryngeal SCC include radiotherapy (RT), transoral laser surgery (TLS), and open partial laryngectomy (see below). Functional results after open partial laryngectomy are usually worse compared to RT and TLS so that this alternative is rarely employed. Given the high probability (> 20%) of occult nodal metastasis in supraglottic cancers, even in these initial stages, the neck must be included in the treatment plan of the tumors on this location, and it is usually treated with the same modality therapy (surgery or RT) than the primary tumor. The decision whether to select RT or TLS (with neck dissection in supraglottic cancers) depends on a number of factors, including the location and extent of the tumor, anatomical circumstances (probability of exposure of the tumor in TLS), the medical condition of the patient, the likelihood of tumor control after treatment, anticipated functional outcome (voice quality), the expertise of the attending physicians and logistical considerations. In this complex decision-making process, we should also include patient preference, after an informed discussion of the pros and cons of each treatment modality. The likelihood of local control after RT or TLS is equivalent and is approximately 85%–95%. An advantage of RT is that it is applicable to all patients with T1–T2N0 SCCs, whereas TLS could not be used in patients with inadequate exposure of the tumor (due to the location of the tumor or anatomical circumstances). Patients with significant medical comorbidities who are poor candidates for anesthesia may be better treated with RT. On the other hand, in glottic tumors, we must keep in mind that the more of the glottis that is involved with SCC, requiring a wider resection, the poorer the voice quality after TLS. Therefore, RT may be preferred treatment option for patients with more demanding requirements for voice quality. When RT is used, the dose of radiation is from 63 Gy (2.25 Gy/fraction) to 66 Gy (2.0 Gy/fraction) for T1 and 65.25 Gy (2.25 Gy/fraction) to 70 Gy (2.0 Gy/fraction) for T2 disease. Patients with anterior commissure involvement will provide technical challenges and, even in experienced hands, may have local control rates that are somewhat lower compared with T1–T2 SCCs without anterior commissure invasion. Efficacy of RT is not affected by involvement of the anterior commissure. In addition, voice quality is likely to be worse after TLS in these cases. One advantage of TLS is that it can be repeated several times in contrast to RT. The ability to repeat TLS may contribute to the fact that the likelihood of laryngeal preservation may be higher when TLS can be offered as initial treatment. Many patients with recurrences after RT will undergo total laryngectomy, although laryngeal preservation may be feasible with salvage open partial laryngectomy or TLS in selected patients after radiation failure. Another advantage of TLS is the shorter length of treatment. In cases where the neck is not dissected, TLS can be done on an outpatient basis whereas RT is delivered once daily on weekdays over 5–7 weeks. In relation to this, an important point to be taken

Laryngeal Cancer: Diagnosis and Treatment

339

into consideration is that TLS is the most cost-effective treatment of early laryngeal SCC, radiation therapy being 2–4-fold more expensive. The presence and extent of a cost differential will vary with the medical system. Regardless of the modality chosen, physicians should track their own patient’s functional and survival data rather than rely on the best reported results from the most experienced institutions. Analysis of outcomes should include tumor control, survival, functional outcomes (quality of voice) and larynx preservation rates.

Treatment of Advanced Stage (T3–T4) Tumors For years, total laryngectomy (with postoperative radiotherapy in high-risk cases) was the only treatment option for patients with intermediate to advanced laryngeal cancer (T3–T4). In addition, either ipsilateral or bilateral neck dissections should be performed in these cases, based on tumor location and extent of nodal metastasis (if present). Over the past two decades, great progress has been made in the management of this disease, with multimodality approaches aimed at laryngeal preservation reshaping the treatment landscape. In response to the common use of total laryngectomy, the nonsurgical approaches have often been referred to as “organ preservation” strategies. In many institutions, it appears that there are organ preservation strategies and then there is surgery. In contrast, we think there are both nonsurgical organ preservation strategies and surgical organ preservation strategies (Fig. 5). The key is that in both approaches the goal is to spare the functions of the larynx. A new paradigm has emerged in which both the surgical and nonsurgical approaches have equal value in functional laryngeal preservation. However, in T4 tumors laryngeal preservation rates are much lower and often present complications which compromise laryngeal function. In addition, patients with T4 tumors treated with up-front surgery had superior overall survival than those treated with CRT, and therefore in these cases total laryngectomy continues to be recommended.

Surgical options for functional laryngeal preservation In addition to the time-honored approaches of vertical partial laryngectomy and horizontal or supraglottic laryngectomy, the options for conservation laryngeal surgery have significantly improved over the past two decades. Transoral minimally invasive surgery and supracricoid partial laryngectomy (SCPL) have emerged as important function-preserving approaches for patients with laryngeal cancer. After either partial laryngectomy or open resection, absence of adverse features, including extra nodal spread, positive margins, perineural invasion, or lymphovascular invasion, allows for observation of the patient with no need for adjuvant therapy. When these high-risk features are present, adjuvant RT is recommended. Vertical partial laryngectomy (VPL) VPL (or vertical hemylaryngectomy) encompasses a spectrum of procedures ranging from laryngofissure with cordectomy to extended hemi-laryngectomy. Common to all these procedures is vertical transection of thyroid cartilage and glottic resection extending into the paraglottic space (Fig. 6). In VPL, vertical incisions are made through the thyroid cartilage near the anterior commissure and just anterior to the posterior edge of the thyroid cartilage. The resulting resection includes the true vocal cord and immediate sub glottis, ventricle, false vocal cord, and arytenoepiglottic fold, and usually crosses just in front of the vocal process of the arytenoid posteriorly. This area can extend around the anterior commissure to involve the anterior one-third of the opposite vocal cord if required. When the anterior commissure is removed, the procedure is termed a frontolateral hemylaryngectomy.

Fig. 5

Algorithm for treatment of advanced laryngeal cancer.

340

Laryngeal Cancer: Diagnosis and Treatment

Fig. 6 Vertical partial laryngectomy. (A) Incisions in laryngeal framework. (B) and (C) Sagittal and coronal views showing the resected portions of the larynx. (D) Endoscopic picture of the larynx after a right hemilaryngectomy.

It is a well-established procedure for T1 and T2 glottic cancers. Some authors believe that patients with fixation of the true vocal cord (T3) caused by direct invasion of the cancer into the thyroarytenoid muscle are still candidates for a vertical hemylaryngectomy. However, in patients with vocal cord fixation caused by cricoarytenoid joint invasion a hemilaryngectomy should not be considered. Other contraindications are involvement of the posterior commissure or the thyroid cartilage, and extension superiorly to the aryepiglottic fold. With this technique overall local control and laryngeal preservation rates, between 82 and 95% have been reported for T1–T2 cases, and 5-year survival rates were greater of 90%. The local control and 5-year survival rates were lower for T3 cases, with reported local control rates between 73% and 85%. These results reflect the continuing value of the VPL in selected cases. However, currently, with the advancement of laser surgery, the role of the VPL is questionable. For most patients with lesions amenable to VPL, laser surgery provides equal local control rates, with superior voice and swallowing function and less complications. And because the relatively high recurrence rates in T3 cases, VPL was replaced by supracricoid laryngectomy for these cases in many centers. Supraglottic laryngectomy Supraglottic laryngectomy involves resection of the epiglottis, the false vocal cords, the aryepiglottic folds, the hyoid bone (in most cases), the upper aspect of the thyroid cartilage, and the contents of the preepiglottic space (Fig. 7). The resection can be extended to include one arytenoid, the tongue base or pyriform sinus. The result is that the patient has a nearly normal voice but a significant challenge in developing normal swallowing caused by the loss of the protective mechanisms of the epiglottis and false cords. Rehabilitation, which involves temporary tracheostomy in all patients and a temporary feeding tube (usually nasogastric), is achieved in most patients within 1 month of surgery with removal of the feeding tube and tracheostomy. The rehabilitative process is complicated by either preoperative or postoperative radiation therapy as well as extension of the surgical resection to include the tongue base, arytenoid cartilage, or pyriform sinus. Supraglottic laryngectomy is indicated in all T1–T2 supraglottic cancers, but also in patients with T3 and T4 supraglottic tumors that involve preepiglotic space or one arytenoid, or that extend into the pyriform sinus or the tongue base. Massive tumors with cartilage erosion, subglottic extension, or involvement of the lateral wall of the pyriform sinus remain subject to total laryngectomy. In addition, the patient must have adequate pulmonary function to be a candidate for a supraglottic laryngectomy. Local control was better for those with tumor confined to the endolarynx (> 90%), but was over 80% for all sites, and laryngeal preservation rates described are also over 80% cases. Overall five-year survival rates are comparable to that obtained with total

Laryngeal Cancer: Diagnosis and Treatment

341

Fig. 7 Supraglottic laryngectomy. (A) Incisions in laryngeal framework. (B–D) Coronal, sagittal, and axial views showing the resected portions of the larynx. (E) Endoscopic picture of the larynx after a supraglottic laryngectomy.

laryngectomy, and range from 67% to 90%. These rates are over 85% for stage I–II tumors, between 75% and 80% for stage III, and between 55% and 70% for stage IV patients. However, in these tumors survival rates depends more from the presence and extension of nodal metastases than the size of the primary tumor. The laryngeal preservation rates with this technique are very good, with overall reported rates over 85% of patients. However, these rates were lower (60%–80%) in T3–T4 tumors. In addition, functional results were good to fair, with more than 90% of patients achieving decannulation and oral diet. Nowadays, conventional supraglottic laryngectomy is being replaced by laser supraglottic laryngectomy because the oncological results of transoral laser surgery for early and moderately advanced laryngeal cancer appear to be comparable to those of classic supraglottic laryngectomy, and the endoscopic approach offers functional advantages. Supracricoid laryngectomy (SCPL) The SCPL is an alternative to (chemo) radiation therapy, supraglottic laryngectomy, and near total and total laryngectomy in selected cases of supraglottic and transglottic carcinoma. This procedure is a true functional preservation technique and should be considered as a conservative laryngeal technique as it preserves physiological rehabilitation of speech, swallowing, and respiration without a permanent tracheostomy. SCPL involves resection of the following structures: the true vocal cords, the false vocal cords, the aryepiglottic folds, the epiglottis (to a variable degree), the subglottic to the superior aspect of the cricoid cartilage, the thyroid cartilage, and the contents of the pre- and paraglottic spaces. The resection can include one arytenoid but must preserve the hyoid bone (Fig. 8). Two different reconstructions are possible depending on the extent of disease involving the epiglottis. In cases in which only the inferior portion of the epiglottis is involved, the suprahyoid epiglottis can be preserved and used in the reconstruction (cricohyoidoepiglottopexy, CHEP). In cases in which it is not oncologically feasible to preserve the epiglottis, the reconstruction will involve the impaction of the base of tongue/hyoid complex to the cricoid cartilage (cricohyoidopexy, CHP). Supracricoid laryngectomy with cricohyoidoepiglottopexy (SCPL-CHEP) is indicated in glottic tumors: T2 (especially with anterior commissure involvement), T3 and selected T4 (limited thyroid cartilage invasion). Is contraindicated in cases with fixation of the cricoarytenoid joint, invasion of the posterior commissure, cricoid invasion, extralaryngeal spread of tumor or poor pulmonary function. Supracricoid laryngectomy with cricohyoidopexy (SCPL-CHP) is also indicated in T2–T4 laryngeal tumors: supraglottic tumors extended onto the vocal cord or anterior commissure and in transglottic tumors. Limitations are the same than with SCPL-CHEP, and also invasion of the hyoid bone.

342

Laryngeal Cancer: Diagnosis and Treatment

Fig. 8 Supracricoid laryngectomy (with crico-hyoidopexy). (A) Incisions in laryngeal framework. (B–D) Sagittal, coronal, and axial views showing the resected portions of the larynx. (E) Endoscopic picture of the larynx after a supracricoid laryngectomy.

Local control and organ-preservation rates with SCPL as primary therapy in patients with T2 and selected T3 lesions exceed 90% and are comparable withdif not better thandrates seen with chemotherapy and radiation and with total laryngectomy. With respect to functional outcomes, although speech and swallowing are restored following SCPL, voice quality is substantially different postoperatively, although subjective voice analysis rank voice as globally “acceptable” by the patient and the physician. Restoration of normal swallowing may take several weeks and requires intensive rehabilitation; nonetheless, more than 80%– 90% of patients can be expected to have swallowing function restored within the first year.

Transoral laser surgery Transoral laser surgery (TLS) is minimally invasive and is performed under suspension-direct laryngoscopy, with an operating microscope, microsurgical instruments, and the surgical CO2 laser. In conjunction with SCPL, this has been one of the two areas of greatest development in conservation laryngeal surgery in recent years. The approach challenges a basic surgical tenet, as the tumor is transected and removed piecemeal through a laryngoscope. However, transection reveals the depth of tumor penetration and allows for clear visualization of tumor margins during the procedure. In contrast to open laryngeal surgery, the cartilaginous laryngeal framework and the infrahyoid muscles are preserved during endoscopic resections, which is believed to improve postoperative function. Additionally, the concept of adequate margins is viewed differently for endoscopic resections: the goal is preservation of as much adjacent normal tissue as possible while ensuring a clear margin. With respect to patient selection, exposure through the laryngoscope dictates which tumors can be managed by TLS. It is a wellestablished procedure for T1–T2 glottic or supraglottic carcinomas, as previously indicated. Additionally, some authors indicated this technique in selected T3 glottic tumors (fixation of the true vocal cord caused by direct invasion of the cancer into the thyroarytenoid muscle), T3 supraglottic tumors (with limited preepiglotic space invasion) and also in some T4 cases (limited base of tongue invasion). Several reports have shown good oncologic results in intermediate and advanced laryngeal cancer. The 5-year local control with laser alone and laryngeal preservation rates are approximately 95% and 98% for T1 tumors, 85% and 95% for T2 tumors, and 70% and 75% for T3 tumors. These results compare well to those of standard supraglottic laryngectomy. Regardless of the surgical technique employed, negative margins are essential in limiting local recurrence. Tumor involvement of the surgical margin after TLS has been associated with higher rates of local recurrence and distant metastasis, lower specific survival rate, and the necessity of salvage surgery.

Laryngeal Cancer: Diagnosis and Treatment

343

Functional outcomes following TLS for laryngeal cancers are excellent. By maintaining one valve of the larynx, the airway is protected and voice and swallowing can be resumed with appropriate rehabilitation. Aspiration occurred in most patients soon after surgery, but recovered within 1–6 months, with recovery being faster in partial resections. Functional results depend on the extent of the resection. The return of voice quality also depends on the depth and extent of resection. The functional results of TLS are generally superior to those of the conventional open approach, in terms of the time required to restore swallowing, tracheotomy rate, incidence of pharyngocutaneous fistulae, and shorter hospital stay. These functional advantages can be attributed to the more conservative nature of the endoscopic procedure, because normal tissues are not disrupted during the procedure. With open procedures, the thyroid cartilage, soft tissues and infrahyoid and suprahyoid muscles are divided, and the hyoid bone is frequently resected. There is invariably airway compromise and a need for a temporary tracheotomy. With endoscopic resection, tracheotomy is almost never indicated. Avoidance of tracheotomy and preservation of the strap muscles may facilitate faster return and ensure improved long-term swallowing function. Transoral robotic surgery The concept of robot-assisted surgery is gaining popularity for multiple different specialties, and more recently in minimally invasive head and neck surgery. The overriding advantages from the proponents of robot-assisted surgery are the excellent 3-dimensional visualization and 2- or 3-handed surgery through minimally invasive approaches that are afforded by the instrument. The wider angle of vision and angled lenses increases the range of the endoscopic visual surgical field compared with the “line of sight” visual field gained by microscopes. The 2-dimensional visualization provided by single channel optical systems in current endoscopes lacks the depth perception of 3-D vision provided by the binocular optical systems used in standard microsurgery. The 5-mm robotic endoscope commonly used has a dual-channel optical system coupled with a dual charge-coupled device, which allows for 3-D visualization of the surgical field at the surgeon’s console. Another advantage of the technology used in the da Vinci robotic instrumentation is its ability to provide movement at the instrument tip with 7
of freedom and 90
of articulation and motion scaling. This allows the surgeon, who sits at the console with an adjustable arm, support to perform precise tremor-free movement in a deep and confined space, with working angles usually not achievable with nonrobotic instruments. Weinstein and O’Malley have previously reported on the development and refinement of a novel procedure called transoral robotic surgery (TORS) in preclinical experimental models. These foundational studies established the technical feasibility of TORS to gain access to the oral cavity, oropharynx, hypopharynx, supraglottis, and glottis, and they also introduced basic concepts of patient safety and methods for controlling active bleeding. Although the present literature reports early findings, without long-term oncologic outcomes, the results are consistently encouraging. Indeed, some institutions have shown that transoral robotic surgery programs can be successfully established yielding excellent clinical results.

Nonsurgical organ preservation protocols In the 1990s, organ preservation treatment protocols combining chemotherapy and radiotherapy were introduced as an alternative to total laryngectomy with the objective to preserve a functional larynx without compromising oncological outcome. The effectiveness of concomitant chemoradiation (CRT) as an effective organ preservation strategy was initially established by The Department of Veterans Affairs (VA) Laryngeal Cancer Study Group in 1991. Their conclusions were subsequently confirmed by the GETTEC and RTOG 91–11 trials and two individual data meta-analyses. According to the different treatment guidelines (European Society of Medical Oncology -ESMO-, National Comprehensive Cancer Network -NCCN-, American Society of Clinical Oncology -ASCO-) the current standard of treatment for patients with T3 laryngeal carcinoma who desire a nonsurgical organ preservation treatment is concomitant radiotherapy with cisplatin, with salvage surgery in cases of persistent disease, or induction chemotherapy followed by definitive radiation/chemoradiation or surgery depending on clinical response. The dose considered standard for cisplatin by most researchers is 100 mg/m2 administered on days 1,22, and 43 of radiotherapy, and radiation therapy is administered with a conventional fractionation (2 Gy/day to give 70 Gy in 7 weeks). With these treatments, the larynx is preserved in approximately 2/3 of patients. Although initially not addressed by the VA protocol, it was recognized later that instead of organ preservation, functional preservation is a more relevant outcome. This functional preservation is defined as an “in situ” larynx without need for permanent tracheostomy and permanent gastrostomy at 2 years after finishing the treatment. Despite the various randomized trials that have all confirmed that the CRT approach achieves survival rates similar to treatment with total laryngectomy, none have shown improvement in survival rates with an organ preservation approach. Furthermore, some investigators have been concerned about the long term toxic effects of CRT treatment on laryngeal function and the decreases in overall survival rates with nonsurgical treatment reported more recently from large tumor registries and long-term follow-up of original trials. This raises a critical question of whether the results of a complex multidisciplinary treatment approach developed in controlled clinical trials by skilled investigators can be effectively generalized to standard practice. Patients specifically included in the pivotal trials were predominantly patients with T3 tumors located at supraglottis, half of them without vocal cord fixation and a small proportion of patients with T4 tumors with minimal invasion to the cartilage, which do not represent the entire spectrum of “advanced larynx cancer.” When the results of CRTs were extrapolated and applied in clinical practice and low volume nonteaching centers, selection criteria may not have been applied as strictly as suggested, and many patients who would not have been candidates to receive CRT (most T4 cancers) were included in this type of treatment. There

344

Fig. 9

Laryngeal Cancer: Diagnosis and Treatment

Role of the salvage surgery in the context of nonsurgical treatment protocols of laryngeal cancer.

are few data about the rate of cartilage invasion or vocal cord fixation in these large cohort studies, but it could be expected to be higher than those reported in pivotal trials. Therefore, despite state of the art treatment facilities, survival may be decreased due to an incomplete treatment dose, lack of supportive care and treatment of adverse effects of these treatments, loss of follow up, etc. Moreover, if salvage laryngectomy, which is part of an organ preserving strategy (Fig. 9) to obtain comparable survival rates between nonsurgical and surgical treatments, is not offered to patients in case of recurrence, survival is hampered in this patient category. It is highly probable that salvage surgery is not offered due to the lack of organized strategies to assess primary response to CRT, to the delay in the detection of early recurrences, to the lack of surgical experience in resection and reconstruction and for the lack of support from health systems to manage postoperative complications and rehabilitation. For advanced laryngeal cancers, as laryngectomy is seen as a “mutilating” treatment and organ preservation is an important achievement in treatment, organ preservation protocols will be used more frequently and laryngectomy will be avoided as primary treatment or even when salvage is necessary. In the VA trial, salvage laryngectomy was more frequently done in patients with glottic tumors, vocal cord fixation, and gross invasion of the cartilage and T4 tumors, which indicates early evidence of worse therapeutic response to CRT in these patients. However, CRT was still offered to all patients under the assumption that it will offer better quality of life in comparison with laryngectomy. There is enough evidence to say that treatment of T4 advanced larynx cancer should consider total laryngectomy since survival outcomes appear better than with CRT in most reports. For patients with T3 tumors, definitive CRT strategies are acceptable on the condition that all resources for the administration of the treatment, follow-up and surgical salvage are available. The entire factors mentioned gives some light to the contradictory results but also offer information for physicians who treat patients with advanced larynx cancer, in order to consider not only the results derived from CRT, but to consider geographical, cultural and socioeconomic conditions before offering the treatment. The introduction of new systemic therapies should be analyzed carefully before applying them in patients with advanced larynx cancer in order to avoid indiscriminate administration to patients that are not going to get clinical advantages from its use. Customization of treatment is also critically important if overall survival rates are to be improved upon. Future developments of more precise functional imaging capabilities to monitor the response to nonsurgical therapies and possible persistence/recurrence of cancers could allow earlier intervention leading to changes of therapy or salvage surgery which would be expected to enhance survival results. The development of useful biomarkers reflecting disease characteristics such as chemo- or radiosensitivity could be useful in more precise decision making, thereby reducing redundancy of treatments and toxicities. In summary, decision making for treatment of patients with advanced laryngeal cancer has never been more complex. Primary treatment decisions may depend on patient desires, tumor extent, experience of the treatment team, the adequacy of follow up surveillance, medical comorbidities and long-term voice and swallowing expectations.

Prognosis Some authors have explored prognostic factors that affect survival in larynx cancer and identified that age, sex, functional status, comorbidities, tumor stage and subsite, type of therapy and treatment of recurrences are closely related with this outcome.

Laryngeal Cancer: Diagnosis and Treatment

345

Five-year disease-specific survival is excellent (> 95%) in early stage (I–II) patients, and can be considered good (70%–80%) in intermediate stage (III) patients. In advanced disease stages (stage IV) 5-year survival rates drop to 50%–60%. The main cause of death is local-regional recurrence, with distant metastasis being infrequent (less than 10%).

See also: Larynx Cancer: Pathology and Genetics.

Further Reading American Society of Clinical Oncology, Pfister, D.G., Laurie, S.A., Weinstein, G.S., et al., 2006. American Society of Clinical Oncology clinical practice guideline for the use of larynxpreservation strategies in the treatment of laryngeal cancer. Journal of Clinical Oncology 24, 3693–3704. Ferlito, A., Rinaldo, A., Silver, C.E., et al., 2008. Neck dissection for laryngeal cancer. Journal of the American College of Surgeons 207, 587–593. Haigentz Jr, M., Silver, C.E., Hartl, D.M., et al., 2010. Chemotherapy regimens and treatment protocols for laryngeal cancer. Expert Opinion in Pharmacotherapy 11, 1305–1316. Holsinger, F.C., Nussenbaum, B., Nakayama, M., et al., 2010. Current concepts and new horizons in conservation laryngeal surgery: An important part of multidisciplinary care. Head & Neck 32, 656–665. Lefebvre, J.L., 2006. Laryngeal preservation in head and neck cancer: Multidisciplinary approach. Lancet Oncology 7, 747–755. López, F., Williams, M.D., Skálová, A., et al., 2017. How phenotype guides the treatment of the most common malignant salivary neoplasms of the larynx? Advances in Therapy 34, 813–825. López, F., Williams, M.D., Cardesa, A., et al., 2017. How phenotype guides management of non-conventional squamous cell carcinomas of the larynx? European Archives of Otorhinolaryngology 274, 2709–2726. Mendenhall, W.M., Takes, R.P., Shah, J.P., et al., 2015. Current treatment of T1N0 squamous cell carcinoma of the glottic larynx. European Archives of Otorhinolaryngology 272, 1821–1824. Rodrigo, J.P., Coca-Pelaz, A., Suárez, C., 2011. The current role of partial surgery as a strategy for functional preservation in laryngeal carcinoma. Acta Otorrinolaringologica Española 62, 231–238. Rodrigo, J.P., Suárez, C., Silver, C.E., et al., 2008. Transoral laser surgery for supraglottic cancer. Head & Neck 30, 658–666. Sanabria, A., Chaves, A.L.F., Kowalski, L.P., et al., 2017. Organ preservation with chemoradiation in advanced laryngeal cancer: The problem of generalizing results from randomized controlled trials. Auris Nasus Larynx 44, 18–25. Suárez, C., Rodrigo, J.P., Silver, C.E., et al., 2012. Laser surgery for early to moderately advanced glottic, supraglottic, and hypopharyngeal cancers. Head & Neck 34, 1028–1035. Yesensky, J., Agrawal, N., Bayan, S., et al., 2017. AHNS series-do you know your guidelines? Principles of treatment for glottic cancer: A review of the National Comprehensive Cancer Network guidelines. Head & Neck 39, 1729–1732.

Relevant Websites https://www.asco.org/practice-guidelines/quality-guidelines/guidelines/head-and-neck-cancerdAmerican Society of Clinical Oncology (Head and neck cancer). http://www.esmo.org/Guidelines/Head-and-Neck-CancersdEuropean Society of Medical Oncology (Head and neck cancer). https://www.nccn.org/store/login/login.aspx?ReturnURL¼https://www.nccn.org/professionals/physician_gls/pdf/head-and-neck.pdfdNational Comprehensive Cancer Network. https://www.cancer.gov/types/head-and-neck/patient/laryngeal-treatment-pdqdNational Cancer Institute. http://www.cancerresearchuk.org/about-cancer/laryngeal-cancerdCancer research UK (laryngeal cancer). https://www.ncbi.nlm.nih.gov/pubmedhealth/PMH0032515/dNCBI (laryngeal Cancer treatment).

Larynx Cancer: Pathology and Genetics Nina Zidar and Nina Gale, Faculty of Medicine, University of Ljubljana, Ljubljana, Slovenia © 2019 Elsevier Inc. All rights reserved.

Glossary Epithelial-mesenchymal transition Process in which epithelial cells change to mesenchymal cells. It is essential in normal embryonal development; in later life, it contributes to wound healing, development of fibrosis and progression of cancer. Invasion Direct extension and penetration of cancer cells through the epithelial basement membrane into the underlying stroma/lamina propria. Invasive front Deepest part of the tumor, at tumor-host interface, the most important part of the tumor for invasion and progression. Precancerosis Lesions with morphologic atypia, associated with an increased risk of cancer development.

Cancer of the Larynx; Pathology and Genetics Cancer of the larynx is the second most common respiratory cancer, after lung cancer. The vast majority belongs to squamous cell carcinoma (SCC) and its variants. Other carcinoma types are rare and include neuroendocrine carcinomas and adenocarcinomas, particularly adenoid cystic carcinoma and mucoepidermoid carcinoma. They are diagnosed, graded and classified similarly to their counterparts in other organs. Sarcomas are also uncommon, the most important being chondrosarcoma. Malignant lymphomas and malignant melanomas can rarely arise in the larynx, as well as metastases from various primary tumors. Only SCC and its variants will be described in details.

Squamous Cell Carcinoma of the Larynx Definition Squamous cell carcinoma (SCC) is a malignant tumor showing evidence of squamous differentiation, e.g., intercellular bridges and/ or keratinization. It arises from squamous epithelium or from respiratory epithelium that has undergone squamous metaplasia.

Burden SCC is the most common malignant tumor of the larynx, accounting for approximately 95% of all malignant tumors at this location. The majority are conventional type SCC. Laryngeal cancer accounts for 1.6%–2% of all malignant tumors in men, and for 0.2%–0.4% of malignant tumors in women, and for approximately 1.0% of all cancer deaths. Its incidence is on the rise in some countries, particularly in women, presumably related to increased incidence of smoking. It occurs most frequently in the 6th and 7th decades, with a strong male predominance. It rarely occurs in children.

Risk Factors Cigarette smoking and alcohol consumption, especially in combination, are the most important risk factors in laryngeal cancer. Among them, smoking is more important. Avoiding of smoking and alcohol consumption could prevent about 90% of laryngeal carcinomas. Some other factors, such as gastroesophageal reflux, radiation exposure, diet and nutritional factors have been also related to an increased risk of laryngeal cancer, particularly in patients who lack the major risk factors. Much attention has been recently paid to the possible role of infection with human papillomavirus (HPV) in the pathogenesis of laryngeal cancer. The results of recent studies, using modern techniques, such as in situ hybridization for HPV E6/E7 mRNA, which enable to detect transcriptionally active, integrated virus, suggest that HPV, particularly type 16, plays a role in the development of 5%–10% of laryngeal SCCs.

Pathology Macroscopic features SCC of the larynx may present as a flat lesion with raised edges, as an exophytic lesion frequently with a central ulceration, or an ulcerative endophytic lesion. Most of them occur in the supraglottic and glottic region, while subglottic SCC is very rare.

346

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65092-4

Larynx Cancer: Pathology and Genetics

• • •

347

Supraglottic SCC arises most commonly in the epiglottis, followed by the false vocal cords, aryepiglottic folds, ventricles and the arytenoids. It tends to spread to oropharynx and pyriform sinus, but it rarely invades the glottis and thyroid cartilage. Glottic SCC arises mostly from the anterior half of the vocal cord or from the anterior commissure. Because of poor lymphatic supply, glottic SCC tends to remain localized for a long period. As SCC progresses, it invades the vocal muscle resulting in the fixed vocal cord which is an ominous clinical sign. In late stages of the disease, it may extend to the opposite true vocal cord, to the supraglottis and subglottis; it may also extend through the thyroid cartilage and invade the soft tissue of the neck. Subglottic carcinoma involves the region extending 1 cm below the true vocal cord, between the lower edge of the true vocal cord and the first tracheal cartilage (Fig. 1).

Microscopic features SCC is characterized by invasive growth and variable degrees of squamous differentiation. Invasive growth is manifested by interruption of the basement membrane and the growth of islands, cords or single (dyscohesive) tumor cells in the subepithelial stroma. Large tumors may extend into deeper structures, i.e., muscle, cartilage or bone. Perineural invasion and invasion of lymphatic and blood vessels may be present, which is a reliable proof of an invasive cancer. Squamous differentiation is manifested by intercellular bridges (desmosomes) and/or keratinization, with keratin pearl formation. SCC is traditionally graded into well-, moderately- and poorly differentiated SCC. The criteria for grading are: the degree of differentiation, nuclear pleomorphism and mitotic activity. Well differentiated SCC resembles closely normal squamous epithelium and contains varying proportions of large, differentiated keratinocyte-like cells and small basal-type cells, which are usually located at the periphery of the tumor islands. There are intercellular bridges and usually full keratinization; mitoses are scanty. Moderately differentiated SCC exhibits more nuclear pleomorphism and more mitoses, including abnormal mitoses; there is usually less keratinization. In poorly differentiated SCC, basal-type cells predominate, with a high mitotic rate, including abnormal mitoses, barely discernible intercellular bridges and minimal, if any keratinization. Keratinization should not be considered an important histological criterion in grading SCC (Fig. 2). Tumor growth at the invasive front can show an expansive pattern, an infiltrative pattern or both. An expansive growth pattern is characterized by large tumor islands with well defined pushing margins, whereas an infiltrative pattern is characterized by scattered small cords or single tumor cells, with poorly defined, diffusely spreading, infiltrating margins.

Differential diagnosis The diagnosis of SCC does not usually present a diagnostic problem for pathologists. Nevertheless, well differentiated SCC must be distinguished from pseudoepitheliomatous hyperplasia, which is a benign condition that consists of deep, irregular tongues of epithelium that lack atypia and abnormal mitoses, and is associated with granular cell tumor or infections. Other differential diagnoses include verrucous carcinoma and papillary carcinoma, discussed later. Poorly differentiated SCC must be differentiated from malignant melanoma, lymphoma, neuroendocrine carcinoma and adenocarcinoma, using immunohistochemistry and special stains for demonstration of mucin production. Melanoma is distinguished

Fig. 1 Squamous cell carcinoma of the larynx. (A) Large exophytic tumor of the glottis extending to the supraglottic and subglottic regions. (B) Small flat tumor of the epiglottis with slightly raised edges.

348

Larynx Cancer: Pathology and Genetics

Fig. 2 (A) Moderately differentiated keratinizing squamous cell carcinoma: moderate nuclear and cellular pleomorphism, rare mitotic figures. (B) Poorly differentiated nonkeratinizing squamous cell carcinoma: marked nuclear and cellular pleomorphism, high mitotic rate.

from SCC by expression of S100, HMB45, melan A, MITF. Neuroendocrine carcinoma expresses neuroendocrine markers (synaptophysin, chromogranin, CD56). Lymphoma is distinguished from SCC by the presence of CD45 and markers of B-cell or T-cell differentiation.

Molecular Pathology and Genetics SCCs of the head and neck including laryngeal SCCs are believed to be among the most highly mutated malignant tumors. Loss of heterozygosity (LOH) and comparative genomic hybridization studies have shown gains of 3q, 5p, 8q, 11q13, and 18p with losses at 3p, 5q, 8p, 9p, 11q23–24, 13q, and 18q. Cyclin-dependent kinase inhibitor 2A (CDKN2A) and TP53 are among the most commonly affected tumor suppressor genes. Epidermal growth factor receptor (EGFR), vascular endothelial growth factor A (VEGFA), prostaglandin-endoperoxide synthase 2 (PTGS2), phosphatidylinositol-4,5-bisphosphate 3-kinase, catalytic subunit alpha (PIK3CA) and matrix metalloproteinases are among the most commonly amplified and mutated oncogenes. EGFR, for example, regulates cell growth, migration and survival, and is overexpressed in 80–90% of cases, increasing progressively from precancerous lesions to SCC. Anti-EGFR treatment is now available but has not proven effective in the majority of patients with laryngeal SCC. P53 is involved in apoptosis and cell cycle regulation. Loss-of-function mutations have been found in laryngeal SILs and SCC. In SCC, the presence of mutations is associated with poor survival and may predict the response to treatment. PIK3CA plays a role in signaling pathway affecting tumor cell growth, metabolism and survival. Mutations are frequent in SCC of the larynx and is a potential target for treatment with PIK3CA inhibitors. Much progress has been achieved in our understanding of genetic abnormalities in precancerosis and SCC of the larynx. Hopefully, this will be used in the future as diagnostic criteria for assessing risk of carcinoma development in precancerous lesions and in surgical margins, and the risk of unfavorable course in SCC. Importantly, they are also potential targets for new treatment modalities. Some modern drugs have already been introduced, but patients with head and neck SCC, including laryngeal SCC, respond poorly, and we are still waiting for new targeted drugs to increase survival of patients with laryngeal SCC.

Microenvironment Including Immune Response Invasive SCC of the larynx is associated with desmoplastic stromal reaction, which consists of proliferation of myofibroblasts, excessive deposition of extracellular matrix and neovascularisation. Together with inflammatory and immune cells, cytokines and other signaling molecules, they constitute tumor microenvironment. It resembles in many ways the processes of inflammation and wound healing and has been compared to a wound which does not heal. The tumor cells and microenvironment are closely related, and the constant interaction between them is complex, playing an important role in the final outcome. It seems that desmoplastic stromal reaction is present only in invasive SCC and never in precancerosis, regardless of the grade, and may be considered an additional marker of invasion. The desmoplastic stromal reaction tends to be pronounced in well- and moderately differentiated SCC and weak or absent in poorly differentiated SCC, as well as in carcinomas, related to infection with HPV and Epstein–Barr virus (EBV). Inflammatory cells comprise cells of innate and adaptive immune system. Some of these cells possess antitumoral properties, while others suppress antitumor immunity and are manipulated by the tumor cells to evade the immune system. Recent studies

Larynx Cancer: Pathology and Genetics

349

suggest an important role of regulatory T cells, tumor-associated macrophages and myeloid-derived suppressor cells. There is merging evidence that immune cells mediating host response contribute significantly to the highly variable outcome, providing a clue why some precancerous lesions progress to cancer and others do not, and why some SCCs progress, metastasize and kill the patient and others do not. Targeting tumor microenvironment to reverse tumor immune escape and induce inhibitory immune cells could represent basis for the development of new treatment modalities in the future.

Staging and Grading The TNM system (T-tumor, N-node, M-metastasis), established by the International Union Against Cancer (UICC) is widely used for staging laryngeal cancer. The TNM status is based on the extent of the tumor and infiltration of the surrounding structures and organs, and on the presence of regional and distant metastases. The nodal status is assessed on the basis of the number of metastatic nodes, their location and size (expressed as the greatest dimension of the node with metastasis, with cut-off values of 3 and 6 cm) and the presence of extranodal extension. The most common sites for distant metastases are the lungs and bones. Stage remains the most significant predictor of survival. Consistently, the prognosis varies from excellent for patients with early disease (stage I and II), with more than 90% 5-year survival rate, to poor for patients with advanced disease (stage III and IV), with less than 60% 5-year survival. Laryngeal SCCs are usually graded as G1 (well differentiated), G2 (moderately differentiated), G3 (poorly differentiated) or G4 (undifferentiated). The criteria and prognostic significance of grading are described elsewhere.

Prognostic and Predictive Biomarkers Clinical prognostic factors

• • •

Stage is the most significant predictor of survival. Extent of the tumor, depth of invasion and the presence of regional and distant metastases are independent predictors of survival. Localization is an important prognostic factor for laryngeal SCC. The most favorable course has been reported for glottic SCC; glottis has a poor lymphatic supply and glottic SCC rarely spreads to the regional lymph nodes. The supraglottic and subglottic regions have a rich lymphatic network, resulting in common metastases in the regional lymph nodes. Other factors that may have a significant impact on the outcome of SCC laryngeal include patient age at presentation, comorbidity and performance status.

Histopathological prognostic factors

• • •

• • •

Differentiation: the prognostic significance of traditional grading into well-, moderately- and poorly differentiated SCC is controversial and mostly correlates poorly with outcome. The main criticism of this widely used system is related to its subjectivity and lack of objective criteria. Invasive front: the growth pattern at the invasive front has prognostic implication: an infiltrative pattern is associated with a more aggressive course and poorer prognosis than an expansive pattern. Lymphovascular invasion: tumor cells often invade the thin-walled lymphatic vessels, capillaries and veins. The finding of the penetration of tumor cells in the lymphatic and/or blood vessels is associated with a high probability of lymph node and/or distant metastases, with recurrence and poor survival. However, the presence of lymphovascular invasion should not be considered synonymous with metastasis, because many tumor cells that enter in the lymphatic system and circulation are destroyed (Fig. 3). Perineural invasion is defined as tumor cells within any of the three layers of the nerve sheath or tumor foci outside of the nerve with involvement of at least 33% of the nerve’s circumference. Perineural invasion enables tumor to spread well beyond the extent of local invasion. It is a marker of increased locoregional recurrence rates, a shorter time to recurrence and decreased survival. Extracapsular spread (ECS), also termed extranodal extension, in lymph node metastases is defined as penetration of the lymph node capsule by tumor cells, with infiltration of the perinodal fibro-adipose tissue. It can be visible by the naked eye (macroscopic/gross ECS) or on histopathologic examination (microscopic ECS). ECS has been recognized to worsen the adverse outcome associated with nodal metastasis. Resection margins: the aim of cancer surgery is to remove as much neoplastic tissue and retain as much healthy tissue as possible. Total surgical removal of the tumor can be proved by clear surgical/resection margins. Margins are considered clear if there is no invasive SCC or carcinoma in situ. Although the concept of clear margins seems straightforward, the definition of adequate margins, i.e., how much healthy tissue around the tumor must be removed, is highly controversial. It is generally believed that a margin of 5 mm is adequate but some believe that even margins of 1–2 mm are adequate, particularly in glottic cancer.

Resection margins clear of tumor are associated with a lower recurrence rate and better survival. However, up to 22% of patients with histologically negative margins undergo treatment failure. In these cases, relapse may be due to a minimal residual disease or to field cancerisation, the latter being related to the presence of genetically damaged cells in microscopically normal mucosa. It has

350

Fig. 3

Larynx Cancer: Pathology and Genetics

(A) Lymphatic invasion: carcinoma cells in a thin-walled lymphatic vessel. (B) Vascular invasion: carcinoma cells within a small vein.

therefore been proposed to introduce novel molecular assays to detect these genetically damaged, phenotypically normal mucosal cells in resection margins. This form of margins has been termed “molecular margins.” Some studies reported promising results using various genetic alterations including TP53 mutations, loss of heterozygosity, promoter hypermethylation, eIF4E protooncogene overexpression, mitochondrial DNA mutations etc. Nevertheless, there is no clinical experience with these methods and no consensus on how to use them in routine work.

Precancerous Lesions of the Larynx Transition from normal laryngeal epithelium to precancerous lesions and SCC is a stepwise process, in which genetic changes progressively accumulate, followed by architectural and cytologic abnormalities. The entire spectrum of morphologic changes in this process has been referred to as dysplasia or squamous intraepithelial lesions (SILs). Several classification systems are used worldwide. Aiming to harmonize the various concepts of different classifications, the new edition of the WHO classification of head and neck tumors proposed a two-tiered system, classifying these lesions as low-grade or high-grade dysplasia (SILs). Some authors believe that carcinoma in situ must be defined as a separate category. Dysplasia (SILs) is associated with an increased risk of progression to SCC, from 1.6% for low-grade dysplasia to 12.5% for high-grade dysplasia, and up to 40% for carcinoma in situ (Fig. 4).

Fig. 4 (A) Low-grade dysplasia (squamous intraepithelial lesion): hyperplastic squamous epithelium with augmented basal-type cells, occupying lower half of epithelial thickness, with rare mitotic figures and no atypia. (B) High-grade dysplasia (squamous intraepithelial lesion): hyperplastic squamous epithelium with augmented basal-type cells, occupying almost the entire epithelial thickness, with marked atypia and high mitotic rate. (C) Carcinoma in situ: epithelial cells show severe atypia, with frequent mitoses, architectural disorder and preserved basement membrane.

Larynx Cancer: Pathology and Genetics

351

Subtypes of Squamous Cell Carcinoma of the Larynx Several variants of SCC in the larnyx have been described in the WHO Classification of the head and neck tumors, including verrucous carcinoma, basaloid SCC, papillary SCC, spindle cell carcinoma, adenosquamous carcinoma and lymphoepithelial carcinoma. Most of them are true clinico-pathologic entities with important prognostic implications. Their recognition also helps to distinguish them from other tumors. Variants of SCC are rare and together account for 5%–10% of laryngeal malignant tumors. Similarly to conventional SCC, they mostly occur in the 6th and 7th decades, more frequently in males, and are often associated with alcohol abuse and tobacco.

Verrucous Carcinoma Verrucous carcinoma (VC) is a variant of well differentiated SCC which grows slowly but can cause extensive destruction. Pure VC is believed to be unable to metastasize. The possible relation of VC to HPV infection has been controversial. Recent studies using highly sensitive and specific molecular methods suggest that VC is not related to infection with HPV. Macroscopically, VC presents as an exophytic, broad based tumor with a warty surface. Microscopically, it consists of thickened, club-shaped filiform projections with prominent surface keratinization and blunt stromal invaginations lined by thick, well differentiated squamous epithelium, composed of a few layers of basal cells and multiplied, voluminous spinous cells lacking cytologic criteria of malignancy. Mitoses are present only in the basal layer. The tumor invades the subjacent stroma with a well defined, pushing margin. Approximately 10% of VCs of the larynx contain foci of conventional SCC. They are referred to as hybrid (mixed) tumors and have the potential for metastasis (Fig. 5). VC is characterized by a high frequency of initial misdiagnosis. An adequate, full-thickness biopsy specimen must be taken, when a clinician suspects VC; moreover, multiple biopsies may be needed to rule out a conventional SCC component in a VC. Differential diagnosis includes verrucous hyperplasia, well differentiated SCC, papillary SCC and squamous papilloma. Invasion below the level of the basal cell layer of the neighboring normal squamous epithelium distinguishes VC from verrucous hyperplasia. Lack of atypia helps to rule out conventional SCC and papillary SCC. VC also lacks the well formed, wide papillary fronds of a squamous cell papilloma. Molecular pathology and genetics of VC are largely unknown. Pure VC has a significantly better prognosis than conventional SCC and doesn’t affect overall survival of the patients when properly treated. Patients with hybrid carcinoma must be treated aggressively as if they had conventional SCC.

Basaloid Squamous Cell Carcinoma Basaloid squamous cell carcinoma (BSCC) is a biphasic tumor composed of neoplastic basaloid cells and foci of SCC. In the larynx, BSCC is not associated with HPV infection. Macroscopically, BSCC does not show any characteristic features, distinguishing it from conventional SCC. It usually present as slightly elevated tumor with a central ulceration and elevated edges. Microscopically, it consists of a SCC component and basaloid cells. The SCC component can present as foci of conventional SCC or as dysplastic changes in the surface epithelium. Basaloid cells are small, with hyperchromatic nuclei without nucleoli and scant cytoplasm. They grow in a solid nests which are closely packed, with a jigsaw-puzzle pattern. Stromal hyalinization between and within the tumor nests is a characteristic feature. Large central necroses of comedo type and prominent peripheral palisading are

Fig. 5 Verrucous carcinoma. (A) Filiform projections of well differentiated squamous epithelium with prominent surface keratinization, with no atypia. (B) Tumor invades the subjacent stroma with well defined, pushing margins.

352

Larynx Cancer: Pathology and Genetics

Fig. 6 Basaloid squamous cell carcinoma. (A) Islands of closely packed small cells with hyperchromatic nuclei without nucleoli and scant cytoplasm. (B) Stromal hyalinization between and within the tumor islands.

usually present. Distinctive features of basaloid SCC are small cystic spaces mimicking gland formation, containing PAS-positive material. BSCC may occasionally lack the SCC component (Fig. 6). Immunohistochemically, BSCC expresses high-molecular-weight cytokeratins, p63 and p40, but it does not express neuroendocrine markers. The most important differential diagnosis includes adenoid cystic and neuroendocrine carcinoma. The former lacks squamous differentiation and usually expresses S100, vimentin and focally p63. Neuroendocrine carcinoma is distinguished from BSCC by positivity for neuroendocrine markers, e.g., synaptophysin, chromogranine and/or CD56. In large tumors extending to the oropharynx, HPV-positive SCC must be considered; it is characterized by immunohistochemical overexpression of p16 and the presence of HPV16/18. Molecular pathology and genetics of BSCC of the larynx are largely unknown. There is some controversy regarding the prognosis of basaloid SCC. Some studies suggested that it has a worse outcome than conventional SCC, whereas others suggested that prognosis in both tumor types is similar when matched for site and stage.

Papillary Squamous Cell Carcinoma Papillary squamous cell carcinoma (PSCC) is characterized by an exophytic, papillary growth pattern and a good prognosis. In the larynx, it is rarely associated with HPV infection. Macroscopically, PSCC presents as exophytic, friable, soft tumors. Microscopically, they are composed of papillary projections, which consist of a central delicate fibrovascular core covered by neoplastic squamous epithelium. The covering epithelium is nonkeratinizing or minimally keratinizing, it may be composed of immature basaloid cells or may be more pleomorphic resembling carcinoma in situ. Stromal invasion is often difficult to demonstrate in biopsy specimens (Fig. 7). Differential diagnosis includes squamous papilloma and verrucous carcinoma. Papilloma and VC share similar architecture with PSCC, but PSCC is differentiated from both VC and papilloma by the presence of atypia of the squamous epithelium covering the papillae. Molecular pathology and genetics of PSCC of the larynx are largely unknown. PSCC is generally believed to have a better prognosis than conventional SCC, though recurrences are frequent.

Spindle Cell Carcinoma Spindle cell carcinoma (SpCC), also referred to as sarcomatoid carcinoma, is a biphasic tumor composed of conventional SCC and malignant spindle cells. The characteristic spindle cell phenotype of the neoplastic cells in SpCC is the result of epithelialmesenchymal transition. Similar to conventional SCC, SpCC has been etiologically related to cigarette smoking and alcohol consumption. It has been suggested that SpCC may develop after radiation exposure; however, some authors believe that this is not a major etiologic factor. Macroscopically, SpCC can present either as an exophytic, polypoid lesion or as a flat, ulcerated tumor. Microscopically, the SCC component may be well-, moderately- or poorly differentiated, keratinizing or nonkeratinizing, and transition between the two components may be abrupt or gradual. The spindle cell component usually forms the bulk of the tumor. Spindle cells are often pleomorphic, with large hyperchromatic nuclei, prominent nucleoli, and numerous mitoses. Sometimes,

Larynx Cancer: Pathology and Genetics

353

Fig. 7 Papillary squamous cell carcinoma. (A) Exophytic tumor, consisting of papillary projections, covered by nonkeratinizing squamous epithelium and fibrovascular cores. (B) The epithelium covering papillae shows nuclear and cellular atypias and increases mitotic activity.

only spindle cells are present. If such tumors are less cellular, they may mimic benign reactive lesions. Foci of osteosarcomatous, chondrosarcomatous or rhabdosarcomatous differentiation may be present, particularly in patients who had previously been treated by radiotherapy (Fig. 8). Immunohistochemically, tumor cells in SpCC often express epithelial and mesenchymal markers. Cytokeratin and vimentin coexpression has been observed in individual tumor cells. Cytokeratin expression can be demonstrated in spindle cells in 40%– 85% of cases. The most sensitive/reliable epithelial marker for SpCC seems to be keratin (AE1/AE3, K1) K1, K18 and EMA. Spindle cells always express vimentin and often other mesenchymal filaments, such as myogenic markers (smooth muscle actin, muscle specific actin, desmin). A diagnosis of a SpCC is based on demonstration of an invasive or in situ SCC and a malignant spindle cell component. However, when a SCC component cannot be demonstrated, the diagnosis is more difficult and SpCC must be distinguished from a number of benign and malignant processes, such as spindle cell sarcomas, nodular fasciitis, inflammatory myofibroblastic tumor and malignant melanoma. In the larynx, true sarcomas (with the exception of chondrosarcoma) and benign mesenchymal tumors are very rare. It is therefore the general view, that a malignant spindle cell tumor in the mucosa of the larynx tract is probably a SpCC and not a sarcoma. A negative reaction for S100 protein and HMB45 helps to distinguish SpCC from malignant melanoma. Molecular pathology and genetics of SpCC is complex and is similar to poorly differentiated SCC. SpCC also shows features of epithelial-mesenchymal transition, e.g., up-regulation of transcription repressors (Snail, Slug, SIP and Twist etc.) and downregulation of microRNAs implicated in the induction of epithelial-mesenchymal transition. Prognosis of the laryngeal SpCC is controversial, but it seems to be favorable for patients with no history of radiation, glottic location and polypoid appearance of the tumor.

Fig. 8 Spindle cell carcinoma. (A) Small islands and cords of squamous cell carcinoma and dense proliferation of neoplastic spindle cells. (B) Spindle cell carcinoma with abundant edematous, myxoid stroma.

354

Larynx Cancer: Pathology and Genetics

Adenosquamous Carcinoma Adenosquamous carcinoma (ASC) is a malignant tumor that arises from the surface epithelium and shows both squamous and glandular differentiation. Macroscopically, ASC cannot be distinguished from conventional SCC. Microscopically, the two components occur in close proximity but are generally distinct and separate and are not closely intermingled as in mucoepidermoid carcinoma. The SCC component can present either as in situ or as invasive SCC, and the adenocarcinoma component is usually located in the deeper parts of the tumor, consisting of tubular, alveolar or ductal structures (Fig. 9). Immunohistochemistry shows distinctive staining patterns in the two components: positive staining for CEA and lowmolecular-weight cytokeratins, such as K7 and CAM5.2 in the adenocarcinomatous component and positive staining for p63, K5/6 and K7 in the SCC component. Both components express high-molecular-weight cytokeratins. Molecular pathology and genetics of ASC of the larynx are largely unknown. Differential diagnosis includes mucoepidermoid carcinoma, adenoid SCC and conventional SCC invading the normal salivary glands. It is important to differentiate ASC from mucoepidermoid carcinoma because ASC has a worse prognosis. Features favoring a diagnosis of ASC are separate and distinct areas of SCC and adenocarcinomatous components and the involvement of the surface epithelium, exhibiting either atypical hyperplasia, carcinoma in situ or invasive SCC. The presence of mucin in the true glandular spaces helps to distinguish ASC from adenoid SCC. Conventional SCC invading or entrapping the normal salivary or mucoserous glands can be confused with ASC, especially in small biopsy specimens. In such cases, preservation of the lobular gland architecture and lack of significant atypia are observed, helping to distinguish conventional SCC from ASC. In general, ASC has a more aggressive course than conventional SCC, with a tendency for early lymph node metastases, frequent local recurrences and dissemination.

Lymphoepithelial Carcinoma Lymphoepithelial carcinoma (LC) is a poorly differentiated SCC or undifferentiated carcinoma, associated with a dense lymphocytic stromal infiltration. It is morphologically similar to nonkeratinizing nasopharyngeal carcinoma, undifferentiated subtype. It is extremely rare in the larynx. In contrast to nasopharyngeal LC, laryngeal LC is rarely associated with EBV infection. Macroscopically, LC cannot be distinguished from conventional SCC. Microscopically, LC is composed of small clusters or large syncytial masses of cells with oval or round vesicular nuclei and prominent nucleoli, and poorly defined, sparse cytoplasm. The stroma is densely infiltrated by T lymphocytes, occasionally admixed with plasma cells, follicular dendritic cells and eosinophils. Immunohistochemically, LC expresses cytokeratins distinguishing it from other malignant neoplasms, e.g., malignant melanoma and lymphoma (Fig. 10). Molecular pathology and genetics of LC of the larynx are largely unknown. LC is more aggressive than conventional SCC, with a higher incidence of cervical lymph node metastases and a propensity for distant metastases.

Prospective Vision A step towards a unique classification of precancerous lesions has been made in the last WHO edition of Classification of head and neck tumors.

Fig. 9 Adenosquamous carcinoma. Two components, clearly separated: squamous cell carcinoma component on the left, and adenocarcinoma component on the right, with mucin production within the cytoplasm of tumor cells.

Larynx Cancer: Pathology and Genetics

355

Fig. 10 Lymphoepithelial carcinoma. (A) and (B) Islands of poorly differentiated carcinoma beneath the surface epithelium, with stroma densely infiltrated by lymphocytes and occasional plasma cells.

In contrast to oropharyngeal carcinoma, no significant progress has been achieved in the understanding of etiology, pathology and treatment of laryngeal carcinoma. Early diagnosis of precancerous lesions, proper classification and optimal choice of treatment is therefore essential for better outcomes of patients at risk for laryngeal cancer. It is now clear that a small proportion of laryngeal SCC is related to high-risk HPV, but clinical significance of this finding, if any, is yet to be determined. Discovery of new targets and treatment modalities are needed, possibly targeting not only tumor cells, but also other constituents of the tumor microenvironment.

See also: Laryngeal Cancer: Diagnosis and Treatment.

Further Reading Alos, L., Castillo, M., Nadal, A., et al., 2004. Adenosquamous carcinoma of the head and neck: Criteria for diagnosis in a study of 12 cases. Histopathology 44, 570–579. Christofori, G., 2006. New signals from the invasive front. Nature 441, 444–450. de Miguel-Luken, M.J., Chaves-Conde, M., Carnero, A., 2016. A genetic view of laryngeal cancer heterogeneity. Cell Cycle 15, 1202–1212. El-Naggar, A.K., Chan, J.K.C., Grandis, J.R., Takata, T., Slootweg, P.J. (Eds.), 2017. WHO classification of head and neck tumors, 4th edn.,. IARC, Lyon. Gale, N., Gnepp, D.R., Poljak, M., et al., 2016. Laryngeal squamous intraepithelial lesions: An updated review on etiology, classification, molecular changes, and treatment. Advances in Anatomic Pathology 23, 84–91. Hinni, M.L., Ferlito, A., Brandwein-Gensler, M.S., et al., 2013. Surgical margins in head and neck cancer: A contemporary review. Head and Neck 35, 1362–1370. Liebig, C., Ayala, G., Wilks, J.A., Berger, D.H., Albo, D., 2009. Perineural invasion in cancer: A review of the literature. Cancer 115, 3379–3391. Mehrad, M., Carpenter, D.H., Chernock, R.D., et al., 2013. Papillary squamous cell carcinoma of the head and neck: Clinicopathologic and molecular features with special reference to human papillomavirus. American Journal of Surgical Pathology 37, 1349–1356. Odar, K., Kocjan, B.J., Hosnjak, L., et al., 2014. Verrucous carcinoma of the head and neckdnot a human papillomavirus-related tumour? Journal of Cellular and Molecular Medicine 18, 635–645. Puri, S.K., Fan, C.Y., Hanna, E., 2003. Significance of extracapsular lymph node metastases in patients with head and neck squamous cell carcinoma. Current Opinion in Otolaryngology & Head and Neck Surgery 11, 119–123. Trivedi, S., Rosen, C.A., Ferris, R.L., 2016. Current understanding of the tumor microenvironment of laryngeal dysplasia and progression to invasive cancer. Current Opinion in Otolaryngology & Head and Neck Surgery 24, 121–127. Zidar, N., Bostjancic, E., Gale, N., et al., 2011. Down-regulation of microRNAs of the miR-200 family and miR-205, and an altered expression of classic and desmosomal cadherins in spindle cell carcinoma of the head and neckdhallmark of epithelial-mesenchymal transition. Human Pathology 42, 482–488. Zidar, N., Gale, N., Cardesa, A., Ortega, L., 2016. Larynx and hypopharynx. In: Cardesa, A., Slootweg, P.J., Gale, N., Franchi, A. (Eds.), Pathology of the head and neck, 2nd edn. Springer-Verlag, Berlin, pp. 338–386.

Li–Fraumeni Syndromeq Orli Michaeli and David Malkin, The Hospital for Sick Children, Toronto, ON, Canada; and University of Toronto, Toronto, ON, Canada © 2019 Elsevier Inc. All rights reserved.

Glossary Familial cancer A form of cancer that occurs in various members of the same family at a far higher rate than would be expected to occur by chance or that would be attributed to shared environmental conditions, for example, Li–Fraumeni syndrome. Germline mutation A mutation which occurs in the germ cell and is therefore propagated to every cell in the body and can be inherited, in contrast to somatic mutations which are acquired. Li–Fraumeni syndrome (LFS) A familial cancer syndrome associated with occurrence of a wide spectrum of cancers that typically develop at significantly earlier ages than their sporadic counterparts. These include breast carcinomas, soft tissue and bone sarcomas, leukemia, and brain tumors among others. For families at risk, the probability of developing some invasive form of cancer by age 30 is close to 50%. p53 A tumor suppressor protein encoded by the TP53 tumor suppressor gene whose primary function is to activate maintenance of genome integrity. Somatic TP53 mutations occur in over 50% of human cancers; germline TP53 mutations occur in
80% of LFS patients. Sporadic cancer Nonhereditary malignancy. Trp53 (transgenic) mice Mice that have been genetically engineered either to be devoid of functional p53 or to express mutant alleles that compromise wild-type p53 function. Tumor suppressor gene A gene whose protein product, when inactivated, can contribute to the malignant transformation of a normal cell to a tumor cell.

Nomenclature LFS Li–Fraumeni syndrome MRI Magnetic resonance imaging yr Years

Introduction Inherited mutations in “cancer genes” pose the most potent and penetrant oncogenic influence in human carcinogenesis, exceeding the effects of such environmental factors as ionizing radiation, tobacco, and occupational carcinogens. Hereditary cancers have been reported to occur in autosomal dominant, recessive or X-linked patterns of inheritance, and for some (most notably breast and colon) at least two different hereditary forms exist, distinguished by their clinical features. Different genetic events are thought to influence the respective phenotypic associations. Some neoplasms such as acute myelogenous leukemia occur only rarely in a familial form, whereas others such as retinoblastoma occur in the heritable form almost as frequently as in the sporadic form. There tends to be a wide degree of penetrance of the susceptibility to tumors arising among family members for each particular gene. Furthermore, the number of different types of cancer varies from family to family. For example, the only well-documented second tumor in the context of hereditary retinoblastoma is osteosarcoma. However, members of a Li–Fraumeni syndrome (LFS) family are at risk for a wide spectrum of tumors, including, but not limited to, carcinomas of the breast, lung and adrenal cortex, soft tissue sarcomas, osteosarcoma, leukemia, and a spectrum of brain tumor subtypes. Studies of hereditary cancers have identified a class of genes that is critical to both carcinogenesis and normal development. The identification of these genes facilitates the identification of increasing numbers of individuals (with or without cancer) who harbor them, as well as opportunities for development of novel therapeutic options. These advances have generated both rays of hope as well as significant challenges both for families and clinicians.

History and Definition In 1969, Li and Fraumeni reported the results of a survey of 280 medical records and 418 death certificates of childhood rhabdomyosarcoma (RMS) patients diagnosed in the United States. In five families in whom siblings or cousins had a childhood sarcoma, q

Change History: May 2018. O Michaeli and D Malkin updated all text and added new text. This article is an update of David Malkin, Li–Fraumeni Syndrome, in Encyclopedia of Cancer (Second Edition), edited by Joseph R. Bertino, Academic Press, 2002, Pages 21–29.

356

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.96122-1

Li–Fraumeni Syndrome

357

a high concentration of cancers of diverse types was noted on the ancestral line of one parent. Soft tissue sarcomas and breast carcinomas, as well as acute leukemias, brain tumors, carcinomas of the lung, pancreas, and skin frequently occurred in first- and seconddegree relatives, and adrenocortical carcinomas were often seen in siblings. This family cancer syndrome came to be known as the Li–Fraumeni Syndrome (LFS). A subsequent prospective analysis of more than 50 similarly affected families defined the list of cancer phenotypes in affected members to include osteosarcoma, brain tumors, leukemias, and adrenocortical carcinomas, in addition to the originally described component tumors, namely soft tissue sarcomas and breast cancer (Fig. 1). It was suggested in the original reports that the familial occurrence of neoplasms originating at discordant sites might represent a counterpart of the tendency for a single individual to develop multiple primary tumors. Indeed, neoplasms in LFS tend to develop in children and young adults, often as multiple primary cancers in affected individuals. Although LFS shares with other hereditary cancer syndromes the tendency for tumors to develop at unusually early ages at multiple sites, the constellation of tumors is quite distinct. “Classic” LFS families include one member diagnosed with a sarcoma before age 45, a first-degree relative with cancer before age 45, and another first- or second-degree relative in the same ancestral line with any cancer diagnosed under 45, or a sarcoma at any age. It was recognized that the identification of a defective gene or genes conferring a predisposition in carriers within these families would assist in clarification of the definition of the syndrome. Although LFS had been characterized in both a statistical and a classical genetic manner, identification of the gene that yielded this striking cancer predisposition remained elusive until 1990. Recognition of the limitations placed on standard linkage analysis suggested that the candidate approach to isolating the responsible gene might be productive. The class of genes most strongly associated with familial neoplasms has been the tumor suppressor genes. Of those genes known at the time, TP53 was most likely to be linked to LFS, and indeed was found to be its cause. Subsequent to the initial report of TP53 mutations in classic LFS families, numerous cases were reported of individuals who did not fulfill the classical LFS definition but who either had an extensive familial history or who carried a germline TP53 mutation. These families were termed “LFS variants” or “LFS like” with several somewhat different definitions that, in general, broadened the classic criteria. For example, the Eeles (1995) definition of LFS like families include individuals with any childhood cancer or sarcoma, brain tumor, or adrenocortical carcinoma diagnosed under 45 years of age, with one first or second degree relative, with typical LFS cancer (sarcoma, breast cancer, brain tumor, leukemia, or adrenocortical carcinoma) diagnosed at any age, plus one first or second degree relative in the same lineage with any cancer diagnosed under age 60. In 2001, a wider set of criteria were proposed by the French LFS consortium that expanded referral opportunities for individuals for genetic testing for TP53 mutations. The original “Chompret criteria” included the following (Chompret et al., 2001): 1. Proband affected by one of a narrow spectrum cancer (see below) before 36 years and at least one first or second degree relative affected by a narrow spectrum tumor (other than breast cancer if the proband is affected by breast cancer) before 46 years or multiple primary tumors.

Colon CA Dx 59y

Leukemia Dx 33y

Breast CA Dx 40y

Breast CA Dx 35y Brain CA Dx 32y

ACC Dx 4y Fig. 1

Li–Fraumeni syndrome “classic” pedigree.

Sarcoma Dx 7y

358 Table 1

Li–Fraumeni Syndrome Criteria for referral for TP53 mutation analysis (based on “revised Chompret criteria”)

Familial presentation Multiple primitive tumors Rare tumors

Proband with tumor belonging to LFS tumor spectrum before age 46 years, AND at least one first/second-degree relative with LFS tumor (except breast cancer if proband has breast cancer) before age 56 years or with multiple tumors Proband with multiple tumors (except multiple breast tumors), two of which belong to LFS tumor spectrum and first of which occurred before age 46 years Proband with one of: adrenocortical carcinoma, choroid plexus carcinoma, anaplastic rhabdomyosarcoma, very early-onset breast cancer (< 31 years), hypodiploid acute lymphoblastic leukemia, medulloblastoma (SHH subtype), irrespective of family history (at any age unless specified)

LFS tumor, premenopausal breast cancer, soft tissue sarcoma, osteosarcoma; CNS tumor, adrenocortical carcinoma. CNS, central nervous system; SHH, Sonic Hedgehog; LFS, Li–Fraumeni syndrome. Based on “2015 Version of Chompret Criteria”, in: Bougeard, G. et al. (2015). Revisiting Li–Fraumeni syndrome from TP53 mutation carriers. Journal of Clinical Oncology 33(21), 2345–2352.

2. Proband with multiple primary tumors, two of which belong to the narrow spectrum and the first of which occurred before 36 years, whatever the family history. 3. Proband with adrenocortical carcinoma whatever the age of onset and family history. The “narrow spectrum of tumors” includes soft tissue sarcoma, osteosarcoma, brain tumor, breast cancer, and adrenocortical carcinoma. These criteria were further revised in 2009 (Tinat et al., 2009) and again in 2015 (Bougeard et al., 2015) (Table 1). These criteria can now be expanded even more, as recently several other cancers have been added to the list for whom gene testing is highly suggested (see “Diagnosis and TP53 Germline Prevalence in Selected Cohorts” section). The current criteria that should be generally used to guide TP53 gene testing is depicted in Table 1.

The TP53 Tumor Suppressor Gene Alterations of the TP53 gene or its encoded protein are the most frequently observed genetic events in human cancer; more than 50% of sporadic cancers harbor somatic TP53 mutations and p53 is functionally dysregulated in more than 80%. The human gene, located on chromosome 17p13, encodes a 53-kDa nuclear phosphoprotein. p53 binds specific DNA consensus sequences and is a transcription factor that regulates the expression of other growth regulatory genes. Structural analysis suggests that several highly conserved amino acids contact the minor or major grooves of the DNA helix. p53 function may be inactivated through sequence changes at these or other codons that alter the conformation of the protein and prevent it from forming tetramers or activating transcription. Activation of p53 is the consequence of various stress signals to the cell: oncogene expression, DNA damage, hypoxia, nutrient deprivation, oxidative stress, replication defects, and more. Activation of p53 induces several downstream effects, including cell cycle arrest, senescence, apoptosis or DNA repair, through modulation of numerous genes. The exact effect will be determined according to the type of stress, the specific cell type, and other factors. Inhibition of growth is achieved by blocking progression at a checkpoint control site prior to G1/S and at the G2/M restriction point. p53, commonly referred to as the “guardian of the genome” is important in maintaining the fidelity of DNA repair, particularly in response to double-stranded (ds) DNA breaks induced by ionizing radiation or certain chemotherapeutic drugs. Wild-type p53 mediates apoptosis when overexpressed in cultured cells in the absence of appropriate differentiation or proliferation signals. Transcriptional targets of p53 are many and include Mdm-2 which is involved in a negative regulatory system that terminates the response of a cell to DNA damage; GADD45, a DNA repair protein; CDKN1a, which is critical for G1 cell-cycle arrest and cell senescence; PUMA and NOXA, mediators of apoptosis; and miR34, a microRNA which regulates genes that promote apoptosis, cell-cycle progression and differentiation. In a cell with normal p53, levels of the protein rise in response to DNA damage, and the cell arrests prior to the G1/S transition, where genomic repair or apoptosis ensues with the mechanism being determined by the transforming oncoprotein(s). In a cell in which p53 is inactivated, G1 arrest does not occur and damaged DNA is replicated. During mitosis, the presence of damaged DNA results in mutation, aneuploidy, mitotic failure, and cell death. As “the guardian of the genome” the functions of p53 are therefore very complex. Indeed, p53 has additional roles that involve autophagy, modulation of reactive oxygen species levels, ferroptosis and metabolism, as preventing tumorgenesis can be achieved by controlling aberrant cellular metabolism. Moreover, TP53 alterations are necessary but not sufficient, for the ultimate cancer that arises from these malignant clones, as other genetic events are clearly required. While the scope of this chapter focuses on germline TP53 mutations and LFS, we suggest further reading of TP53 tumor suppressor gene itself.

Germline TP53 Mutations In 1990 (Malkin et al., 1990), five LFS families were studied to determine whether TP53 played any role in the occurrence of cancer in affected family members. Base pair mutations were identified in the germline of affected members in each of the five families

Li–Fraumeni Syndrome

359

studied. Although the missense mutations were initially observed between codons 252 and 258 in exon 7, within one highly conserved region of the gene, further analysis of one family revealed a 2-bp deletion at codon 184 in exon 5 instead of the codon 252 mutation. The wild-type allele was deleted in tumors of affected individuals. Several unaffected relatives were mutant gene carriers, suggesting that they might be at risk to develop cancer at a later date. Since that original report, germline TP53 mutations were reported in additional LFS families, with other constitutional mutations in the same region of the gene, as well in different ones. However, while most LFS families harbor a TP53 alternation, there is a lack of 100% concordance with the “classic” LFS phenotype. In fact, only 60%–80% of “classic” LFS families harbor germline TP53 mutations. The other clinical criteria as mentioned above are broader and less specific; therefore, only approximately 40% of “LFS-Like” families and 30% of individuals meeting the 2009 version of the Chompret criteria, carry mutations. This genotype–phenotype discordance, in which not even all “classic” LFS exhibit TP53 mutation, may be explained in several ways. These include posttranslational p53 alterations, endogenous undetected promoter defects leading to aberrant expression of the TP53 message, complete TP53 deletion, effects of modifier genes, or alterations of other genes that may influence the phenotype generated by the presence of a specific germline TP53 mutation. Nevertheless, the high frequency of germline TP53 mutations in LFS families together with the tight association of tumor formation in Trp53-deficient or Trp53-mutant mice (see “Animal Models of Li–Fraumeni Syndrome” section) do confirm a causal association of germline TP53 alterations and cancer predisposition. Currently, hundreds of germline TP53 mutations have been reported. For the most part, these correspond to the somatic mutations and can be found across the gene. However, the most common mutations include those found in the original hotspots along with several additional ones, located in the highly conserved DNA binding domain of the gene, in exons 5–8 (Fig. 2). According to the recent version of International Association for Research on Cancer (IARC. R18, April 2016), the most common mutation hotspots, excluding a unique one at codon 337 (see below), are at codons 248, 273, 175, 245, 282, 213, 125, and 158, in decreasing order of frequency. Point mutations are the most prevalent (70%–96%), specifically missense mutations. Other, less common alternations are splice site and nonsense mutations that encode a truncated p53 protein, as well as rare frameshift mutations, intragenic deletions, in-frame insertion/deletions, and also intronic mutations.

Animal Models of Li–Fraumeni Syndrome Transgenic animals carrying distinct deregulated oncogenes develop tumors that appear to be cell-type specific. To better study p53 in vivo, mice have been created that either lack functional p53 or express dominant-negative mutant alleles that inhibit wild-type p53 function. In one of the early studies by Donehower et al. (1992), homozygous knockout mice with a germline Trp53 null mutation (p53/) were created. These mice were highly susceptible to malignant lymphomas and sarcomas, which occurred in > 75% before 6 months of age. Tumor development, primarily sarcomas, was delayed in heterozygotes mice (p53þ/); however, by 18 months of age, 50% of them develop tumors. Multiple primary tumors were noted in
30% of tumor-bearing p53/ mice, while virtually none of the p53þ/þ mice had developed tumors by 18 months. The heterozygous mice are a closer genetic model for LFS, since affected LFS individuals are heterozygous for mutant p53 rather than homozygous, and a substantial fraction of the germline mutations are functionally null for p53. Developmentally, the mice were normal, implying lack of crucial function of p53 in the evolving embryo. In a different study, the p53/ state was again compatible with normal development. However, the yield of p53/ offspring varied from 16.6% to 23%. The apparent increase in fetal loss is related to presence of fetal exencephaly, although increased fetal loss is not seen in human LFS families. The genetic background of mice (C57BL/6 mice vs. 129/Sv), determined by strain-associated modifier genes, was shown in the aforementioned study to influence the tumor spectrum in the setting of p53 functional deficiency. The importance of genetic background in influencing the tumor type is as well exemplified by the occurrence of unusual cancers, including pineoblastomas and islet cell tumors in p53/ mice crossed with mice heterozygous for an RB1 mutation. In a study by Kuperwasser et al. (2000), which looked at female mice and breast carcinomas, high rate of mammary tumors developed in p53-deficient mice in a BALB/c background, but not in other genetic backgrounds (C57Bl/6 or 129/Sv). Notably, BALB/c-p53/ mice did not develop high rates of mammary tumors, but this was likely due to early death from other tumors, especially lymphomas. Indeed, when mammary glands of BALB/c-p53/ female mice were transplanted into wild-type (BALB/c-p53þ/þ) hosts, 75% developed mammary tumors.

Fig. 2

Structural organization of p53.

360

Li–Fraumeni Syndrome

Mice deficient for p53 are more sensitive to the effects of certain carcinogenic agents. p53þ/ mice exposed to dimethylnitrosamine developed liver hemangiosarcomas more rapidly than treated p53þ/þ animals. p53/ mice treated with an initiator, dimethylbenzanthracene, and a promoter, 12-O-tetradecanoyl-phorbol-13-acetate, show a more rapid rate of malignant progression of skin papillomas to carcinomas. These studies help distinguish whether p53 mutations play rate-limiting or tissue-specific roles in the tumor progression pathway. p53-deficient and p53-Tg mice (mice carrying mutant transgene) exposed to sublethal doses of girradiation develop tumors, usually sarcomas, earlier than in untreated animals. This susceptibility is associated with a twofold increase in the accumulation of radiation-induced dsDNA breaks compared to that seen in p53þ/þ animals. These studies confirm that p53 prevents the accumulation of cells sustaining radiation-or chemically induced DNA damage. More than 80% of the TP53 germline mutations in humans are missense mutations in the conserved regions of the p53 DNA binding domain, whereas only few are nonsense mutations that result in no p53 or a truncated p53 protein. Therefore, it was crucial to test missense mutations in mice, and not only null p53 models. A model by Liu et al. (2000) looked at heterozygous mice with TP53 R172H mutation (p53R172H/), which corresponds to the TP53 R175H hotspot mutation in human cancers, and demonstrated different spectrum of tumors. Although mice with this mutation expressed low levels of mutant p53, there was high prevalence of metastatic disease compared to heterozygous null mice (p53þ/). Notably, most of the tumors from these mice had an intact TP53 allele, suggesting that loss of wild-type TP53 was not important to tumor development. Actually, in LFS patients, approximately 40% of tumors retain an intact wild-type TP53 allele, in contrast to sporadic tumors, in which mutations or loss of both TP53 alleles occur more frequently. In humans with germline missense TP53 mutation, when p53 is functionally null, all the tumors will show LOH (loss of heterozygosity), meaning loss of the wild-type allele. In mouse models, however, tumors from mice with one null TP53 allele (p53þ/) were analyzed, and over half were found with an intact, wildtype TP53 allele, which was functionally active. Provocative studies have tested whether the tumorigenic activity of a mutant p53 is altered by the presence or absence of wildtype p53 in vivo. Mice carrying the 135Ala > Val mutant transgene were crossed with p53-deficient mice. The mutant p53-Tg accelerated tumor formation in p53þ/, but not in p53/ mice, suggesting that this loss-of-function mutation had a dominant-negative effect with respect to tumor incidence and cell growth rates. Although the tumor spectrum was similar in transgenic and nontransgenic mice, the transgenics showed a predisposition to lung adenocarcinomas. Thus, a given p53 alteration may have distinct tissue specificity with respect to tumorigenic potential. The dominant-negative effect, by which mutant p53 functionally inactivates the wild-type p53 allele, was demonstrated by Chène (1998) in a mouse model with two types of mutations. The first, Arg175His, is a structural mutation, which affects the global configuration of the p53 protein, hence interfering with its ability to bind DNA. The p53 protein is active as a tetramer. When wildtype p53 forms a tetramer with such a mutant protein to form heterotetramers, that heterotetramer does not bind DNA due to its defective configuration. The second, Arg248Trp (and similarly-Arg 273Leu), is an active site mutation, which affects the actual site of DNA binding on the p53 protein, and not its configuration. This is also a dominant negative mutation: the complete lack of ability to bind DNA causes the wild type subunit of the heterotetramer to act improperly so that all parts of the heterotetramer cannot bind DNA. However, a different kind of mutation, which does not allow any tetramerization with the wild type protein, (e.g., Arg175His; Leu330Ala) will not cause such a dominant negative effect. Likewise, an active site mutation which causes low affinity of DNA binding but does not completely preclude it (e.g., Arg273His) also do not cause such dominant negative effects, since some binding is possible. The role of those two different types of point mutations were examined in study by Olive et al. (2004). Trp53 R270H is a missense mutation which corresponds to the TP53 R273H mutation in humans and is an active site mutation. A different mutation, Trp53 R172H, corresponds to the TP53 R175H in humans. Heterozygous mice (p53R270H/þ and p53R172H/þ) developed different tumor spectra when compared to p53þ/ mice, to each other (in the same genetic background) and to p53/ mice. Tumors arising in p53/ mice were primarily carcinomas (specifically lung adenocarcinomas), most of which were metastatic or invasive, as well as B cell lymphomas. In contrast, p53R172H/þ mice showed a higher incidence of sarcomasdparticularly osteosarcomas, with frequent metastatic disease. Thus, at least in mice, different genotypes produce different phenotypes. Lang et al. (2004) also created p53R172H/þ mice. These mice developed osteosarcomas and carcinomas but with a much higher frequency of metastases than p53/ mice. Importantly, both studies demonstrated the “gain-of-function” effect common to many p53 missense mutations, meaning that the mutant p53 exhibits additional inherent tumorigenic activity. Both also highlighted possible tumor suppressor roles of p63 and p73. These other p53 “family members” were functionally inactivated by the mutant p53R172H/þ in mouse embryonic fibroblasts, providing an explanation for their role in the “gain of function” effect. Owing to the fact that p53 has several functions, one can speculate that different mutations will cause distinct altered functions. An interesting study evaluated mice harboring a R172P mutation, which corresponds to the human R175P. This mutant retains cell cycle arrest while abrogating initiation of apoptosis. Consequently, cells of mice with a homozygous mutation (p53R172P/R172P) were able to induce cell-cycle arrest and maintain a stable genome. Notably, the mice had delayed tumor onset compared with homozygous null mice (p53/). This finding suggests that delayed cell-cycle arrest contributes to tumor suppression. These mouse models and many others shed light on the critical functions of p53 and the phenotypic manifestations of different TP53 mutations. The continuing expansion of our knowledge through these models is still crucial to further understand the phenotype: genotype correlations in LFS. However, although mice are powerful models with which to study various aspects of p53 biology, they are still somewhat imperfect representations of LFS and cannot be entirely relied upon to explain the clinical and genotype associated phentoypes of the human disease.

361

Li–Fraumeni Syndrome

Clinical Aspects and Epidemiology of Li–Fraumeni Syndrome Prevalence Currently, more than 700 families have been reported in the database of the International Association for Research on Cancer (IARC). However, estimations of the actual prevalence of the syndrome are much higher, until recently estimated at approximately 1:5000, as awareness of cancer predispositions is increasing and many cases are not reported. In addition, newer ways of determining the prevalence are emerging (i.e., by surveying genetically larger cohorts and through the use of next generation sequencing technologies) which suggest that the TP53 mutation population carrier rate may be as high as 1:500.

Penetrance Penetrance is thought to be very high, although it is challenging to determine as cancer development in an individual is not entirely dependent on the specific mutation. As genetic tests are becoming more prevalent, more and more people are identified with TP53 germline mutations without having a rich familial history or even having cancer. Some are identified by the occurrence of cancer in a family relative. Moreover, others are discovered incidentally, for instance by genetic testing which was performed for reasons other than a cancer diagnosis. Studies have demonstrated a high penetrance rate with estimates of a lifetime cancer risk of 93% in females and 75% in males. Some reports attribute 100% penetrance by the age of 70 years. This cancer risk is determined by various variables including gender and age. Hwang et al. (2003) looked at 56 carriers of germline TP53 mutations and found that 12%, 35%, 52%, and 80% developed cancer by ages 20, 30, 40, and 50 years, respectively. In comparison, Bougeard et al., 2015 found an earlier age of onset, with 22% of carriers diagnosed with a cancer by age 5 years and 41% by age 18 years. While these differences are attributed to different cohorts with different inclusion criteria, it is notable that cancer risk is substantially higher than the general population. Notably, risk of cancer begins from infancy, as one of these studies reports that 4% of participants developed a malignancy during the first year of life.

Tumor Spectrum LFS is a cancer predisposition syndrome with no other phenotypic features. The cancer spectrum is vast and virtually any cancer can occur. However, the “core cancers” that were included in Li and Fraumeni’s original report are indeed more prevalent. In the report by Bougeard et al., 2015, 322 affected carriers were described. The five core cancers were the most frequent, with breast cancer more prevalent than any other, occurring in 60% of females. Soft tissue sarcoma and osteosarcoma followed, affecting 27% and 16% of patients, respectively. ACC and brain tumors affected 13% of individuals each. IARC reports 1644 tumors in TP53 carriers, of which 27% are breast cancers, 13% STS, 12% brain cancers, 11% are adrenal and 10% involve the bones (Fig. 3). Other cancers that occur in TP53 carriers with some frequency are leukemia, lymphoma, embryonal cancers, various carcinomas, other kinds of sarcomas, neuroblastomas, and more. Interestingly, several tumor types that frequently harbor a somatic TP53 mutation are less common among individuals with LFS, though when they do occur, they tend to happen at a lower age of onset than

27.31% (449)

BREAST 13.14% (216)

SOFT TISSUES

12.35% (203)

BRAIN

11.56% (190)

ADRENAL GLAND BONES

10.16% (167) 8.64% (142)

OTHER 3.47% (51)

HEMATOLOGICAL COLORECTUM

3.1% (51)

LUNG

2.49% (41) 2.49% (41)

SKIN

1.64% (27)

OVARY STOMACH

1.22% (20)

KIDNEY

1.03% (17) 0.43% (7)

TESTIS LIVER PROSTATE

0.24% (4)

LARYNX

0.18% (3)

0.24% (4)

HEAD&NECK

0.18% (3)

ESOPHAGUS

0.06% (1)

BLADDER

0.06% (1)

0.0

5.0

10.0

15.0

20.0 %

Fig. 3

Tumors associated with TP53 germline mutations (n ¼ 1644).

25.0

30.0

35

362

Li–Fraumeni Syndrome

their sporadic counterparts. For example, colorectal cancer occurred in only 3% of TP53 carriers, though on average at least two decades earlier than the median age of onset for sporadic CRC. Type of cancer naturally changes in relation to age. Although adults with LFS are more commonly affected by breast cancer and soft tissue sarcomas (STS), in adolescents the most common tumors are osteosarcomas and young children are noted for adrenocortical carcinoma (ACC), brain tumors, and STS. In regards to gender differences, since LFS is an autosomal dominant disorder, males and females have an equal chance of carrying the affected gene. However, penetrance differs and women appear to have a higher chance of developing cancer. This difference is attributed in part to the relative abundance of breast cancer, although several studies report that additional factors are involved. In a study by Olivier et al., 2003, among 494 tumors in individuals with germline TP53 mutations, there was an excess in males of brain tumors, hematopoietic cancers, and stomach cancer. Excess of females was observed for skin cancer and adrenocortical carcinoma. This distribution is in concordance with gender predilection in sporadic tumors, though the excess of females with ACC is exceptional. Males and females were equally affected by soft tissue and bone sarcoma, lung cancer, and colorectal cancer. Mai et al. (2016) report that among females, the cumulative incidence rates by age 70 years were 54% for breast cancer, 15% for STS, 6% for brain tumors, and 5% for osteosarcoma. Among males, the incidence rates were 22% for STS, 19% for brain tumors and 11% for osteosarcoma. Females not only have a greater chance for cancer, but also tend to be affected at an earlier age. A difference of 11 years between median age of cancer onset was reported among females (28.0 years) and males (17.0 years). This difference was still present after excluding breast cancers, though it was narrowed to 4 yearsdtumor onset in females was 13 years, compared to 17 years in males.

Multiple Cancers A striking feature in LFS kindreds is the high frequency of affected members who develop multiple primary neoplasms. One multicenter study demonstrated germline TP53 mutations in leukocyte DNA from 4 of 59 patients (6.8%) who had survived second cancers but did not have family histories compatible with the LFS. The increased risk for multiple cancers is pronounced in patients who had their first cancer at a young age or were previously treated with radiation. While early studies state a relatively low incidence of second primaries, more recent studies report a higher incidence, perhaps due to different inclusion criteria, differences in treatment, and survival. For example, in a study by Hisada et al. (1998) among patients diagnosed between 1968 and 1986, the incidence of second, third and fourth primary cancers were 15%, 4%, and 2%, respectively. Notably, 71% of the multiple cancers developed were the LFS core tumors. In contrast, Mai et al. (2016) report in 2016 that almost 50% of LFS patients develop second primary tumors after a median of 10 years following their initial cancer. Breast cancer was the most commonly occurring second malignancy, followed by STS and brain and lung cancer. Interestingly, females had a higher risk of second cancer when compared with males, but lower risk of death. Similarly, Bougeard et al., 2015 reported multiple primary tumors in 43% of mutation carriers, metachronous presentation being more common than synchronous. This higher chance of multiple primary cancers cannot be attributed solely to therapy, as the incidence is also increased when compared to patients with a history of sporadic cancers. According to a study by Hwang et al. (2003), there is 12-fold higher risk to develop a second primary cancer among carriers, when compared to the general population. In this report, among LFS patients the second cancer occurred at a median of 9.3 years after the first malignancy, which is longer than the median time for treatmentrelated secondary malignancy in noncarriers in this cohort. As survivorship increases and surveillance is implemented (see “Management” section), more instances of patients surviving multiple cancers are reported, several of which are quite remarkable with documentation of up to 17 primary cancers in a single patient. As mentioned, the estimates of second cancer risk will continue to evolve as more data from cases of TP53 carriers becomes available, mostly of individuals that do not fulfill the classical LFS criteria, or have any familial history of cancers.

Germline TP53 Aberrations and Phenotype: Genotype Correlations Understanding genotype: phenotype correlations in individuals with LFS can aid clinicians in determining more precise estimates of the likelihood of occurrence of specific tumors according to age group, gender, and genotype. Ideally, surveillance and management can be tailored according to those characteristics. Although not fully understood, several important correlative associations have been described. One important observation is that dominant-negative TP53 missense mutations within the DNA-binding domain are associated with the highest cancer risk and young age of onset. Such mutations were detected most commonly in pediatric cases of LFS with brain tumors, osteosarcoma and rhabdomyosarcoma (RMS). In contrast, the majority of pediatric cases with adrenocortical carcinoma (ACC) harbor other types of alterations. Other studies have noted that gain of function mutations were associated with younger age of onset when compared to loss of function mutations. A unique association between genotype and phenotype is described for the germline TP53 mutation R337H. This mutation is abundant in Brazil, where this founder mutation results in a high prevalence of LFS. The specific mutation (c.1010G > A) is located at codon 337 in exon 10, within the oligomerization domain of the p53 protein. Interestingly, the pH level is a determinant of the mutant protein’s function, so that only in the physiological pH range (up to 7.5), does the protein form normal oligomers. At higher pH, and higher temperatures, the mutant protein is unable to dimerize and therefore does not assemble into a functional protein.

Li–Fraumeni Syndrome

363

The TP53 R337H mutation is estimated to be present in 0.3% of residents of the South/Southeastern regions of Brazil, resulting in estimate of more than 300,000 carriers. The spectrum of cancers is similar to that observed with other mutations; however, there is a greater than 15-fold increase in childhood ACC in the Brazilian population. In addition, there is high occurrence of papillary thyroid cancer, renal cancer, and lung adenocarcinoma. The age of tumor onset is somewhat older so that the penetrance in the younger age group is lower: cancer occurs before age 30 in only 15%–20%, compared with 50% in carriers of other TP53 mutations. The lifetime penetrance is, however, as high as in “classic” LFS. The phenotype: genotype correlations are complicated by various alterations that modify the function of p53. One example are specific TP53 polymorphisms, such as a duplication within intron 3 (PIN3). When this duplication is found in a TP53 R337H mutation carrier, the malignancies are more likely to appear at a later age, compared to carriers of the same TP53 mutation that do not harbor the PIN3 duplication. PIN3 also confers a higher susceptibility to breast and colorectal cancer. Another genetic modifier is a specific polymorphism of MDM2, an important regulator of p53. MDM2 ubiquinates p53 so that it is targeted for degradation. Bond et al. (2004) found that a T> G substitution at nucleotide 309 (SNP309) in the promoter of the first intron of MDM2 leads to increased levels of MDM2 and hence accelerated degradation of p53. As a result, it is associated with a higher rate of tumor formation in TP53 mutation carriers and was shown to correlate with younger age of tumor onset. This higher risk is augmented when, in addition to the MDM2 SNP309 polymorphism, the TP53 mutation carrier also harbors a TP53 polymorphism at codon 72Pro (Arg72Pro). If a 72Arg variant of the p53 protein exists instead of 72Pro, a higher affinity of p53 to MDM2 will be achieved, causing greater degradation of p53 and younger age of cancer. A different mechanism of functional modification of TP53 is through differential methylation of other targets. One example are the microRNAs (miRs), short noncoding RNA molecules that modulate protein expression by promoting RNA degradation and inhibiting mRNA translation. miR-34a is a regulatory microRNA of p53. It was found that miR-34a promotor is hypomethylated in TP53 mutation carriers, thus making it more active as a “cover” mechanism for the absent p53 tumor suppressor activity. On the other hand, hypermethylation of miR-34a, which decreases its function, is associated with decreased overall survival. This finding was demonstrated in choroid plexus carcinomas (CPC), a typical LFS tumor. Investigators found increased hypermethylation across three CpG sites of the miR-34a promoter, when compared with wild type TP53 CPCs. Moreover, patients with CPCs that harbored miR-34a promoter hypermethylation had a lower median survival, a finding consistent with the fact that TP53 mutant CPCs are known to have a more clinically aggressive course, with poorer survival. An additional microRNA that was found to modify the LFS phenotype is miR-605, which regulates the p53–MDM2 loop. MDM2 as mentioned is a cellular inhibitor of p53, and is correlated with earlier age of tumor onset when increased. A variant of miR-605 gene, the G allele, was shown to be associated as well with an earlier age of tumor onset in LFS patients, in almost all tumor types. In vitro, when nucleofected as pre-miR 605 plasmids to RMS cell lines, the G allele showed decreased processing from its precursor form to its mature form, as compared to the “normal” A allele. Furthermore, in the context of mutant p53, overexpression of miR605 resulted in a significant reduction in cell viability and oncogenic properties, including cellular proliferation and migration potential. An intriguing genotype: phenotype phenomenon in LFS is anticipation, wherein successive generations manifest with earlier onset of malignancies. While the exact mechanisms remain unknown, there is probably a role for accelerated telomere attrition. Tabori et al. (2007) compared the telomere length in peripheral blood leukocytes of TP53 mutation carriers and noncarriers. Carriers who were affected with cancer had shorter telomeres than unaffected carriers or healthy controls. The same was true when comparing affected children with cancer in comparison to their unaffected family members. Excess DNA copy number variation (CNV) was also found to be one of the characteristics of the LFS phenotype. In a study by Shlien et al., 2008, of individuals with LFS compared the general population, TP53 mutation carriers had a significant increase in CNVs across their genome, particularly if they had a family history of cancers. This is postulated to be the consequence of the underlying genomic instability caused by the mutant p53. In addition, some LFS individuals had only a few CNVs but of exceptionally large deletions or duplications. Subsequently, large deletions that encompass the entire TP53 gene, appear to be associated with a distinctive phenotype of congenital anomalies without a predisposition to cancers. While such genetic modifiers continue to be explored, it is now acknowledged that phenotype: genotype correlations are the result of complex genetic interactions including genetic modifiers within and outside of TP53 itself.

Diagnosis and TP53 Germline Prevalence in Selected Cohorts LFS can be defined as an inherited form of cancers caused by germline mutations of the TP53 gene; therefore, its diagnosis is best confirmed by the identification of a pathogenic variant in the gene. However, prior to the genetic test, the diagnosis in most cases has to stem from a high index of suspicion. The revised Chompret criteria described above should aid in the decision of which cancer patient should be referred for genetic testing. As mentioned, these criteria are broad and more sensitive for identifying individuals with LFS who do not have familial cancer. In fact, estimations of de novo mutation frequency in LFS patients range between 7% and 20%, though the number might be even higher as more individuals are now being referred for genetic testing. In particular, the revised criteria focus also on specific cancer diagnoses which confer high risk of TP53 germline mutation, regardless of the familial history. For instance, 9% of children diagnosed with RMS harbor germline TP53 alterations. This fraction increases to 23% for patients younger than 4y and up to 73% of children with RMS with diffuse anaplasia. The incidence of TP53 germline alternations in other sarcomas vary according to specific

364

Li–Fraumeni Syndrome

type; for example, up to 10% of children with osteosarcoma harbor germline TP53 mutations. Among children with ACC, about 50% have TP53 germline alterations, and newer estimates range as high as 70%. However, in adult ACC patients, germline TP53 mutations are exceedingly rare. Two studies of patients with CPC found that about 40% carry TP53 mutations. This rate is increased when cohorts focus on patients with somatic TP53 mutation or on patients with a familial history of cancer. More than 10% of children with the Sonic Hedgehog (SHH) subgroup of medulloblastoma (SHH-MB) harbor TP53 germline mutations, though the number increases to 30% when children between 8 and 17 are studied. Furthermore, 56% of children with the subgroup SHH/TP53-MB (SHH-MB with somatic TP53 mutation) carry such a mutation. While most patients with childhood acute lymphoblastic leukemia (ALL) do not have cancer predisposition, an uncommon subset of them with hypodiploid ALL, have substantial risk to harbor TP53 germline mutation. Studies show that 19%–43% of such children are in fact carriers. The prevalence of TP53 germline carriers among breast cancer patients substantially changes with age. At older age they represent a minority. However, up to 10% of very early onset breast cancer (< 30 years of age) have LFS, similar to the rate of BRCA 1/2 mutations in this population. While all of these types of malignancies do confer high risk of underlying constitutional TP53 mutation, it is important to bear in mind that for all of these cohorts, ascertainment bias is likely to lead to an overestimation of tumor risk in individuals with LFS. Regardless, germline testing is still recommended in these clinical contexts. Suspicion of an underlying predisposition for cancer should also be raised with other features of a given malignancy. For instance, detection of specific tumor signatures should promote germline testing. In a report by Rausch et al., 2012, the phenomenon of chromothripsisdmassive genomic rearrangementsdwas associated with TP53 germline mutations in SHH MB. When looking at medulloblastomas, the vast majority displayed low numbers of single nucleotide variant (SNV) per chromosome. However, 13 of 98 tumors showed high number of SNVs, consistent with chromothripsis. Eleven of these thirteen were SHHMBs, of whom 10 harbored TP53 mutations, whereas none of the wild-type SHH-MBs showed this phenomenon. Therefore, the finding of chromothripsis in medulloblastoma, should raise the suspicion of an underlying TP53 germline mutation. Likewise, comparing SNV rate in a given tumor to the expected SNV rate in the same “reference” sporadic tumor, can point to the need of germline testing. All these features of a specific cancer diagnosis should advance the suspicion for an underlying predisposition. Other “universal criteria” that should raise suspicion of LFS include multiple primary tumors (synchronous or metachronous) and familial history of cancer, though not necessarily as depicted in the “classic” LFS criteria. In an era of increasing genetic tests, many individuals are being diagnosed without even being affected. Individuals can be identified following a family member’s diagnosis of LFS or incidentally. Once a referral for genetic testing has been made, the type of genetic testing is determined according to the indication. A known familial variant usually leads to testing for this variant only. Where the mutation is not known, single gene testing by Sanger sequencing is often the initial test of choice with the addition of copy number analysis for intragenic deletions/duplications as well as examination of the intron/exon borders. This however may not identify certain types of variants (e.g., deep intronic variants). Updated data regarding the different variants, including their interpretation as pathogenic or not, can be found in the website of the International Agency for Research on Cancer (IARC) TP53 mutation database (listed below). It compiles TP53 mutation data that have been reported in the published literature since 1989, and is updated annually. In addition to somatic TP53 mutations, it includes the largest datasets on individuals carrying a sequenced TP53 germline mutation or affected by cancer and belong to an “LFS family.”

Management Surveillance Implementation of a surveillance protocol for germline TP53 mutation carriers enables the presymptomatic detection of malignancies. Such early tumor detection can potentially allow definitive localized treatment approaches that may obviate or minimize exposure to systemic or radiation treatment. In 2011, and later in 2016, Villani et al. (2016) first reported feasibility and potential survival benefit of a comprehensive surveillance strategy in patients with LFS. This surveillance protocol, termed the “Toronto Protocol,” consisted of regular physical examinations together with scheduled imaging and biochemical studies, and was later proven to be beneficial in terms of survival. Among the initial cohort of 59 patients who chose to undertake this surveillance protocol, 40 neoplasms were detected pre-symptomatically in 19 patients, over a median follow-up period of 32 months. Two additional cancers were diagnosed between surveillance assessments in one patient. In contrast, 61 neoplasms presented clinically in 43 of 49 patients who did not undergo surveillance. Furthermore, 25 of 40 neoplasms on the surveillance “arm” were low grade or premalignant at the time of detection, suggesting that early detection through surveillance may identify lesions before malignant transformation. Five years OS was 88.8% in the surveillance arm, versus 59.6% in individuals not undergoing surveillance. Several other studies have since suggested improved clinical outcomes for TP53 mutation carriers with intensive screening, specifically with the utility of WBMRI.

Li–Fraumeni Syndrome Table 2

365

Recommended LFS screening protocol

Adults General assessment • Complete physical examination every 6 months • Prompt assessment with primary care physician for any medical concerns Breast cancer • Breast awareness (age 18 years onward) • Clinical breast examination twice a year (age 20 years onward) • Annual breast MRI screeninga (ages 20–75) • Consider risk-reducing bilateral mastectomy Brain tumor (age 18 years onward) • Annual brain MRI (first MRI with contrast; thereafter without contrast if previous MRI normal) Soft tissue and bone sarcoma (age 18 years onward) • Annual WBMRIa • US of abdomen and pelvis every 12 months Gastrointestinal cancer (age 25 years onward) • Upper endoscopy and colonoscopy every 2–5 years Melanoma (age 18 years onward) • Annual dermatologic examination Children (birth to age 18 years) General assessment • Complete physical examination every 3–4 months, including blood pressure, growth curve, Cushingoid appearance, signs of virilization, and full neurologic assessment • Prompt assessment with primary care physician for any medical concerns ACC • US of abdomen and pelvis every 3–4 months • In case of unsatisfactory US, blood testsb,c, may be performed every 3–4 months: total testosterone, dehydroepiandrosterone sulfate, and androstenedione Brain tumor • Annual brain MRI (first MRI with contrast; thereafter without contrast if previous MRI normal and no new abnormality) Soft tissue and bone sarcoma • Annual WBMRI WBMRI, whole-body MRI; US, ultrasound; ACC, adrenocortical carcinoma. a Breast MRI/US of abdomen and pelvis to alternate with annual WBMRI (at least one scan every 6 months). b The efficacy of biochemical surveillance for detection of adrenocortical carcinoma has not been shown. c Serial specimens obtained at the same time of day and processed in the same laboratory. Kratz, C.P., Achatz, M.I. and Brugières, L. (2017). Cancer screening recommendations for individuals with Li–Fraumeni syndrome. Clinical Cancer Research 23(11), e38–e45. http:// clincancerres.aacrjournals.org/content/23/11/e38.

In 2016, international consensus surveillance recommendations were generated for individuals carrying a pathogenic TP53 variant, and those fitting the classic clinical definition of LFS. These were based on the “Toronto protocol” with slight modifications and is depicted in Table 2. Based on prediction of timing of cancer initiation for the various tumor types, the surveillance approach is proposed according to age groups. Also considered are adverse effects of “over” surveillance including use of contrast, risks from invasive medical procedures or anesthesia, and false positive or false negative results. The importance of regular thorough physical examination and targeted history is emphasized, as well as the prompt assessment of any medical concern. For early identification of ACC, ultrasound is recommended until 18 years of age, with or without ACC-specific blood tests every 3–4 months. Given its low incidence in adults, there is no need for specific ACC surveillance after childhood. The lifelong brain tumor risk justifies annual brain MRI in all age groups. In order to minimize the potential risk for gadolinium accumulation in individuals undergoing multiple enhanced MRIs, if the initial MRI performed with a gadolinium-based contrast agent (GBCA) shows normal results, the following MRIs may be conducted without GBCA unless an abnormality is seen. For solid tumor surveillance, specifically the lifelong risk of sarcoma, an annual rapid whole body MRI (WBMRI) including limbs, is recommended to all individuals. In addition, abdominal and pelvic ultrasound is recommended annually in adults (while children will have such imaging every 3–4 months as above). The adult abdominal imaging should be every 6 monthsdalternating between WBMRI and ultrasound. The annual WBMRI cannot replace dedicated brain MRI, as was demonstrated by Ballinger et al. (2017) in a meta-analysis, and it is generally suggested that the two examinations should alternate. In adults, as the risk for gastrointestinal cancers increases, a dedicated periodic exam (colonoscopy/upper endoscopy) is indicated starting at age 25. Annual dermatology exams should start no later than at 18 years of age, as are breast cancer prevention strategies. While this protocol is currently widely accepted, changes will evolve with more universal uptake and as more experience is reported. Specifically, genotype: phenotype correlations will be clarified, permitting differential recommendations by specific organ risk and disease penetrance.

366

Li–Fraumeni Syndrome

Treatment of Malignancies in the Context of LFS In multiple tumor types, somatic TP53 mutation was proven to confer poorer prognosis in comparison to wild type. Likewise, the prognosis of some specific tumors was reported to be worse among TP53 germline carrier in comparison to the same tumors when occurring sporadically. Since the key role of p53 is in response to DNA damage, one could make an argument of the need to minimize exposure of these patients to radiotherapy and some chemotherapy. Indeed, the deleterious effect of radiation in animal models of LFS is well established. Data among LFS patients is not vast, yet is growing. There are several case reports and small case series that integrate a growing volume of evidence suggesting an increased risk of second malignancies in the radiation field of patients with LFS. For example, Bougeard et al. (2015) reported development of tumors within the radiotherapy fields in 30% of mutation carriers. Hisada et al. (1998) studied 200 LFS patients of whom 30 had multiple primary cancers. Nine of those 30 had radiation in the past, in six of which the secondary tumor was within the radiation field. Hence, trying to avoid radiation therapy when feasible, especially in high dose and large volumes, is recommended. However, it should be noted that in the context of treatment discussions for large or high-grade malignancies, omitting radiation may not be a viable therapeutic option. More knowledge of the effects of radiation as well of chemotherapy and other agents in this setting is needed. An exciting new direction in treatment options for LFS patients is coming from studies of specific treatments aimed at restoring the function of p53. PRIMA-1 (p53 reactivation and induction of massive apoptosis) and its derivate, APR-246 (PRIMA-1MET), were studied for their ability to restore some of the functions of mutant p53. In various tumor cell lines they were shown to convert p53 to its active conformation, inhibit proliferation and promote apoptosis. Antitumor activity has been demonstrated in several xenograft models and the first human studies showed a good safety profile. Other promising agents such as pyrazoles (e.g., PK7088) bind to specific p53 mutated proteins, and restore their conformation and activity to some extent. Zinc-metallochaperones can transport zinc to specific zinc-deficient p53 mutant proteins, so that the protein’s folding and function are restored. Metformin, a commonly used antidiabetic drug, was shown to impact several pathways in cancer, one of which is p53. Metformin is known to inhibit mitochondrial respiration, which is essential to cancer progression. Wang et al. (2017) took mice with a known Trp53 germline mutation and generated an additional mutation in their mitochondria, causing a decreased oxygen consumption rate. These mice had increased cancer free survival as compared to the other Trp53 mutated mice. Metformin was given to the mutant mice leading to the same effect on mitochondria as the double mutant ones, of reduced oxygen consumption and induction of antiproliferation signaling cascade. Metformin given to LFS (human) patients, resulted in decreased oxygen consumption rate in mononuclear cells, with decreased mitochondrial activity and similar anticell proliferation signaling. These are all examples of the evolving approach to make p53 a targetable genetic lesion, that will hopefully change the current treatment of LFS specifically, and cancer in general.

Ethical and Psychological Considerations The potential ethical and psychosocial impact of being able to predictively screen unaffected family members or children for disease predisposition raises a number of complex issues. While this article is not meant to discuss these in detail, one would be remiss to ignore their significance. Predictive testing for germline TP53 mutations (or any other disease-predisposing gene) must be based on theoretical principles of respect for the autonomy of the patient and freedom from coercion, benefit to the patient, universal accessibility and freedom from discrimination, and maintenance of confidentiality of results. In the evolving flood of disease and cancer-predisposing genes that are being made available for mutation screening, guidelines addressing the following issues must now be established using TP53 testing as a model: (1) who is to have access to the information, (2) what safeguards are to be established to protect employees and others from discrimination, (3) at what stage in the development of these genetic tests should mass screening be advocated, and (4) when might mandatory genetic screening be acceptable practice? For several reasons, testing cannot at present be offered to the general population. Even within the general cancer population the carrier rate is demonstrably low. However, mass screening can be considered in populations that carry a high mutation rate. Such an exceptional example is the southern Brazilian population. Indeed, in a Brazilian study, neonatal screening for the p.R337H mutation followed by surveillance in mutation carriers led to the detection of lower stages of ACCs. The sensitivity and specificity of screening methods to identify mutations are yet not well known and both false-positive and false-negative results have been encountered. The sequence-confirmed presence of a base pair alteration itself must be interpreted carefully to ensure that its protein product has significance in that in some way it inactivates normal p53 function. Furthermore, effective ways must be developed to provide informed consent to those being screened for a TP53 germline mutation, about potential beneficial and harmful psychologic and social sequelae. Multidisciplinary services should be incorporated, including oncology, psychiatry, psychology, genetic counseling, medical ethics, and molecular and medical genetics. Such services require skills in communicating risk information and how to motivate adherence to cancer prevention and early detection. Yet ethical considerations should be taken into account: Is there an obligation to inform family members of individuals who are identified as having such a genetic susceptibility? How aggressively should people at genetic risk for cancer be recruited into surveillance protocols, or clinical trials? The issue of prenatal testing is also challenging. Risks for future children in the family should be addressed, and discussion with families should include options for preimplantation genetic diagnosis, prenatal diagnosis or testing of infants after birth. The

Li–Fraumeni Syndrome

367

potential to reduce risk of losing a child or young adult makes prenatal testing compelling to some families. For other parents, prenatal diagnosis will not change the fate of the pregnancy but can be beneficial in term of very early initiation of surveillance protocol. Furthermore, the option of preimplantation genetic diagnosis (PGD) is also a possible with the use of in vitro fertilization (IVF) techniques, and should be discussed when feasible. The psychological burden of the identification of a cancer predisposition syndrome has to be taken into consideration. While some individuals who are de novo carriers do not have personal of familial experience of cancer, others are very familiar with malignancies and their outcomes. The perceptions are hence different. While some individuals will prefer to disregard such diagnosis, other advocate for a more active route. Although surveillance confers physical, psychosocial, and financial challenges, and sometimes increased anxiety, many individuals believe in the value of this approach. In fact, some of the patients going through surveillance report enhanced sense of control, security, and empowerment facing this very poor prognostic syndrome.

Concluding Remarks Decades after its first description, and years after the identification of its cause, our knowledge of Li–Fraumeni syndrome has expanded dramatically. However, the prognosis for individuals with LFS is still grim. Comprehensive data of the molecular causes, clinical manifestations and better understanding of genotype: phenotype interactions will allow for better prediction of disease. Improving surveillance, which in turn will allow for earlier detection of tumor will greatly influence the longevity and quality of life of the patient. Most importantly, discovering ways to make p53 a drugable target will make the most substantial contribution. Not only will we be able to treat premalignant lesions in the setting of germline mutations, the impact on all cancers, sporadic and predisposed, will be enormous.

See also: Cell Responses to DNA Damage. TP53.

References Ballinger, M.L., et al., 2017. Baseline surveillance in Li-Fraumeni syndrome using whole-body magnetic resonance imaging: A meta-analysis. JAMA Oncology 3 (12), 1634–1639. https://doi.org/10.1001/jamaoncol.2017.1968. Bond, G.L., et al., 2004. A single nucleotide polymorphism in the MDM2 promoter attenuates the p53 tumor suppressor pathway and accelerates tumor formation in humans. Cell 119 (5), 591–602. Bougeard, G., et al., 2015. Revisiting Li–Fraumeni syndrome from TP53 mutation carriers. Journal of Clinical Oncology 33 (21), 2345–2352. https://doi.org/10.1200/ JCO.2014.59.5728. Chompret, A., et al., 2001. Sensitivity and predictive value of criteria for p53 germline mutation screening. Journal of Medical Genetics 38 (1), 43–47. Chène, P., 1998. In vitro analysis of the dominant negative effect of p53 mutants. Journal of Molecular Biology 281 (2), 205–209. Donehower, L.A., et al., 1992. Mice deficient for p53 are developmentally normal but susceptible to spontaneous tumors. Nature 356, 215. Eeles, R.A., 1995. Germline mutations in the TP53 gene. Cancer Surverys 25, 101–124. Hisada, M., et al., 1998. Multiple primary cancers in families with Li-Fraumeni syndrome. Journal of the National Cancer Institute 90 (8), 606–611. Hwang, S.J., et al., 2003. Germline p53 mutations in a cohort with childhood sarcoma: Sex differences in cancer risk. American Journal of Human Genetics 72, 975–983, 975. Kuperwasser, C., et al., 2000. Development of spontaneous mammary tumors in BALB/c p53 heterozygous mice. A model for Li-Fraumeni syndrome. American Journal of Pathology 157 (6), 2151–2159. Lang, G.A., et al., 2004. Gain of function of a p53 hot spot mutation in a mouse model of Li-Fraumeni syndrome. Cell 119 (6), 861–872. Li, F.P., Fraumeni Jr., J.F., 1969. Rhabdomyosarcoma in children epidemiologic study and identification of a familial cancer syndrome. Journal of the National Cancer Institute 43, 1365–1373. Liu, G., et al., 2000. High metastatic potential in mice inheriting a targeted p53 missense mutation. Proceedings of the National Academy of Sciences 97 (8), 4174–4179. Mai, P.L., et al., 2016. Risks of first and subsequent cancers among TP53 mutation carriers in the National Cancer Institute Li-Fraumeni syndrome cohort. Cancer 122 (23), 3673–3681. Malkin, D., et al., 1990. Germ line p53 mutations in a familial syndrome of breast cancer, sarcomas, and other neoplasms. Science 250 (4985), 1233–1239. Olive, K.P., et al., 2004. Mutant p53 gain of function in two mouse models of Li-Fraumeni syndrome. Cell 119 (6), 847–860. Olivier, M., Goldgar, D.E., Sodha, N., 2003. Li–Fraumeni and related syndromes: Correlation between tumor type, family structure, and TP53 genotype. Cancer Research 63, 6643–6650. Rausch, T., et al., 2012. Genome sequencing of pediatric medulloblastoma links catastrophic DNA rearrangements with TP53 mutations. Cell 148 (1–2), 59–71. https://doi.org/ 10.1016/j.cell.2011.12.013. Shlien, A., et al., 2008. Excessive genomic DNA copy number variation in the Li–Fraumeni cancer predisposition syndrome. Proceedings of the National Academy of Sciences of the United States of America 105 (32), 11264–11269. https://doi.org/10.1073/pnas.0802970105. Tabori, U., et al., 2007. Younger age of cancer initiation is associated with shorter telomere length in Li-Fraumeni syndrome. Cancer Research 67 (4), 1415–1418. Tinat, J., et al., 2009. 2009 version of the Chompret criteria for Li–Fraumeni syndrome. Journal of Clinical Oncology 27, 108–109. Villani, A., et al., 2016. Biochemical and imaging surveillance in germline TP53 mutation carriers with Li–Fraumeni syndrome: 11 year follow-up of a prospective observational study. The Lancet Oncology 17 (9), 1295–1305. https://doi.org/10.1016/S1470-2045(16)30249-2. Wang, P.Y., et al., 2017. Inhibiting mitochondrial respiration prevents cancer in a mouse model of Li–Fraumeni syndrome. The Journal of Clinical Investigation 127 (1), 132–136. https://doi.org/10.1172/JCI88668.

Further Reading Ariffin, H., et al., 2014. Whole-genome sequencing analysis of phenotypic heterogeneity and anticipation in Li–Fraumeni cancer predisposition syndrome. Proceedings of the National Academy of Sciences of the United States of America 111 (43), 15497–15501. https://doi.org/10.1073/pnas.1417322111.

368

Li–Fraumeni Syndrome

Deb, S., Deb, S.P., 2014. Mutant p53 and MDM2 in cancer. Springer, Berlin. Duffy, M.J., Synnott, N.C., Crown, J., 2017. Mutant p53 as a target for cancer treatment. European Journal of Cancer 83, 258–265. https://doi.org/10.1016/j.ejca.2017.06.023. Hainaut, P., Olivier, M., Wiman, K.G., 2013. p53 in the clinics. Springer, Berlin. Kool, M., et al., 2014. Genome sequencing of SHH medulloblastoma predicts genotype-related response to smoothened inhibition. Cancer Cell 25 (3), 393–405. https://doi.org/ 10.1016/j.ccr.2014.02.004. Kratz, C.P., et al., 2017. Cancer screening recommendations for individuals with Li–Fraumeni syndrome. Clinical Cancer Research 23 (11), e38–e45. https://doi.org/10.1158/ 1078-0432.CCR-17-0408. Zhang, J., et al., 2015. Germline mutations in predisposition genes in pediatric Cancer. New England Journal of Medicine 373 (24), 2336–2346. https://doi.org/10.1056/ NEJMoa1508054. Zhukova, N., et al., 2013. Subgroup-specific prognostic implications of TP53 mutation in medulloblastoma. Journal of Clinical Oncology 31 (23), 2927–2935. https://doi.org/ 10.1200/JCO.2012.48.5052.

Relevant Websites http://p53.iarc.fr/TP53GeneVariations.aspxdIARC p53 updated database. https://www.lfsassociation.orgdLi–Fraumeni association. http://p53.fr/dThe TP53 website. http://exac.broadinstitute.org/gene/ENSG00000141510dThe Exac browser.

Lipid Metabolism Fatima Ameer, Rimsha Munir, and Nousheen Zaidi, University of the Punjab, Lahore, Pakistan © 2019 Elsevier Inc. All rights reserved.

Introduction This is widely recognized that tumors display an increased ability to synthesize lipids, and that this lipogenesis is tightly coupled to glucose metabolism. The endogenously synthesized lipids fuel membrane biogenesis in rapidly proliferating cancer cells. Lipids may also play substantial role in cell signaling, operating as second messengers and hormones. Changes in lipid metabolism may also affect cell growth, proliferation, differentiation, and motility. Major lipid synthesis pathways include de novo fatty acid synthesis pathway and mevalonate pathway; that leads to the synthesis of cholesterol and isoprenoids. The significance of these pathways in tumorigenesis is widely reported.

De Novo Fatty Acid Synthesis Pathway De novo fatty acid synthesis or de novo lipogenesis is a coordinated series of enzymatic reactions that moderates the flow of carbons from different sources to fatty acids (Fig. 1). The first step of de novo lipogenesis is the conversion of cytoplasmic citrate to acetyl-CoA by ATP-citrate lyase (ACLY). The resulting acetyl-CoA is carboxylated to malonyl-CoA by acetyl-CoA carboxylase (ACACA). AcetylCoA and malonyl-CoA are then coupled to the acyl-carrier protein domain of the rate-limiting enzyme fatty acid synthase (FASN).

G3P Pyruvate

PDK

Glutamate IDH2

Isocitrate

Extracellular Fatty Acid Synthesis Pathway

G6P

Glutamine

Acetate Metabolism

Lactate

SLC1A5 SLC1A5

Glutamine Metabolism

GLUT-1

Glycolysis

Glucose

HIF

Acetate

Glutamine

MCT

Glucose

Complex Fatty acids

Acetate

Citrate

Palmitate C75 & derivatives

FASN

Malonyl-CoA SoraphenA

ACACA

Acetylation

Acetyl-CoA

ACLY

PDH HMG-CoA SB-204990

Acetyl-CoA

Statins

HMGCR

Mevalonate

Citrate

Oxaloacetate

GPP Bisphosphonates

Malate

TCA Cycle

FDPS

Isocitrate

Fumarate

Mevalonate Pathway

FPP IDH1 α-KG

Zaragozic acid

Succinate

SQS Squalene

Succinyl-CoA Cholesterol

Mitochondrion Steroids

Cytoplasm

Sterol Branch

• • •

Ubiquinone Dolichol Haeme A

•

Farnesylated proteins (Ras) • Geranylated Proteins (Rho, Rac) Non-Sterol Isoprenoid Branch

Fig. 1 Overview of cellular lipid metabolism. De novo lipogenesis and mevalonate pathway are fueled by acetyl-CoA derived from glucose, glutamine and/or acetate. Intermediates of these pathways yield diverse biosynthetic precursors for macromolecule production necessary for cell proliferation and survival. Key metabolic enzymes known to be upregulated during cancer are represented in blue. Some of the common pharmacological inhibitors of these pathways are indicated in red. Many reactions and their ability to be reversed are omitted for simplicity. Abbreviations used: a-KG, a-ketoglutarate; G3P, glyceraldehyde-3-phosphate; G6P, glucose-6-phosphate; HIF, hypoxia-inducible factor; MCT, monocarboxylate transporter; IDH, isocitrate dehydrogenase; ACACA, acetyl-CoA carboxylase; ACLY, ATP-citrate lyase; FASN, fatty acid synthase; HMGCR, 3-hydroxy-3-methylglutaryl (HMG)-coenzyme A (CoA) reductase; FDPS, farnesyl disphosphate synthase; FPP, farnesyl pyrophosphate; GLUT, glucose transporter; PDH, pyruvate dehydrogenase; PDK, pyruvate dehydrogenase kinase; SLC1A5, solute carrier family 1 member 5; TCA, tricarboxylic acid.

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.64998-X

369

370

Lipid Metabolism

Repeated condensations of acetyl groups generate palmitic acid (16:0)da basic 16’carbon saturated fatty acid. Palmitic acid is further elongated and desaturated to generate complex fatty acids (e.g., stearate). Cancer cells are shown to display up-regulated de novo lipogenesis, whereas most nonmalignant counterparts preferentially acquire fatty acids from exogenous sources. Certain types of nonmalignant cells/tissuesdincluding adipocytes, hepatocytes, hormone-sensitive cells and fetal lung tissuedalso display active de novo lipogenesis. However, in general de novo lipogenesis is suppressed in most nonmalignant cells. Several lines of evidence suggest that activation of the de novo fatty acid synthesis pathway is required for carcinogenesis. Chemical or RNA interference-mediated inhibition of the key enzymes involved in fatty acid synthesis, including FASN, ACACA and ACLY, has been shown to attenuate cancer cell growth and to induce cancer cell death. Moreover, enhanced expression of FASN is correlated with poor prognosis in cancer patients. The mammalian cells have a limited ability to synthesize polyunsaturated fatty acids de novo, because they lack the D12 desaturasedan enzyme which introduces a second cis double bond into monounsaturated fatty acids. Therefore, enhanced de novo lipogenesis enriches the cancer cell membranes with saturated and/or mono-unsaturated fatty acids. As these fatty acids are less prone to lipid peroxidation than polyunsaturated acyl chains, de novo fatty acid synthesis is suggested to make cancer cells more resistant to oxidative stress-induced cell death. Moreover, as saturated lipids pack more densely, the increased lipogenesis also alters lateral and transverse membrane dynamics that may limit the uptake of drugs, rendering the cancer cells more resistant to therapy. The altered saturation index of cancer cell membrane also affects fundamental cellular processes including signal transduction, gene expression and ciliogenesis. Furthermore, de novo fatty acid synthesis may also regulate the formation and composition of membrane structures that orchestrate signal transduction and motility (e.g., lipid rafts, invadopodia, and blebs). Lipogenesis can also control biosynthesis of lipid signaling effectors implicated in cancer genesis and progression, including phosphatidylinositols, lysophosphatidic acid, and prostanoids. In nonmalignant cells fatty acids are also used to supply energy via a highly regulated process-lipolysis. Three lipasesdadipose triglyceride lipase (ATGL), hormone sensitive lipase (HSL) and monoacylglycerol lipase (MAGL)dact in a stepwise series of reactions in order to generate diacyl glycerol (DAG), monoacylglycerol (MAG), and glycerol respectively, with subsequent release of three fatty acids. Free cytoplasmic fatty acids are then activated, by coupling to coenzyme A followed by exchange of acyl chain to carnitine via carnitine acyltransferase, to be transported into the mitochondrial matrix. Repeated rounds of this process (known as b-oxidation) guarantees the balance between lipogenesis and lipid digestion in nonmalignant cells. Certain types of tumors, including prostate tumors, display increased dependence on b-oxidation of fatty acids as their main source of energy. Human leukemia cells are also shown to require b-oxidation for their proliferation and survival.

Mevalonate Synthesis Pathway The mevalonate pathway leads to the synthesis of sterols and isoprenoids that are shown to be crucial for tumor-growth. Multiple enzymes of this pathway are recognized to be essential for proliferation and survival of various types of cancer cells. The first committed step of the mevalonate pathway is conversion of 3-hydroxy-3-methylglutaryl-CoA (HMG-CoA) to mevalonic acid (mevalonate) by HMG-CoA reductase (HMGCR) (Fig. 1). In the next step of the pathway mevalonate is metabolized to isopentenyl pyrophosphate (IPP) and dimethylallyl pyrophosphate (DMAPP). Farnesyl pyrophosphate synthase catalyzes sequential condensation reactions of DMAPP with two units of IPP to form farnesyl pyrophosphate (FPP) and geranylgeranyl pyrophosphate synthase catalyzes yet another condensation reaction to form geranylgeranyl pyrophosphate (GGPP). FPP is the branch point for several pathways leading to various end-products including cholesterol, steroid and dolichols. Additionally, FPP can also be converted into geranylgeranyl pyrophosphate (GGPP) by GGPP synthase. Squalene synthase catalyzes the first reaction of the pathway committed exclusively to cholesterol biosynthesis and plays a crucial role in directing intermediates to either sterol or non-sterol branches of this metabolic pathway. The oncogenic potential of the mevalonate pathway has also been investigated. Mevalonate pathway is shown to induce proliferation in primary leukemia cells. HMGCR is also suggested to be a candidate metabolic oncogene. It was shown that ectopic expression of HMGCR accentuates growth of transformed and non-transformed cells and cooperates with RAS to drive the transformation of primary mouse embryonic fibroblasts cells. Moreover, direct administration of mevalonate via miniosmotic pumps into mice harboring breast cancer cell line xenografts resulted in increased tumor-growth. Higher expression of different genes of the mevalonate pathway has also been correlated with poor prognosis in breast cancer patients. As mentioned above the mevalonate pathway generates various metabolites that are implicated in carcinogenesis. For instance, cholesterol is shown to be involved in cancer cell proliferation and protection of cancer cells against immune surveillance as well as various therapeutic agents. In addition to that cholesterol serves as the precursor for synthesis of steroid hormones and oxysterols that are suggested to play various roles in cancer progression. Farnesyl-diphosphate and geranylgeranyl-diphosphate are respectively involved in farnesylation and geranylgeranylation of a variety of proteins. Farnesylation and geranylgeranylation are required for the ability of Ras and Rho proteins to induce malignant transformation, invasion, and metastasis. Dolichol is an essential component of the N-glycosylation of nascent polypeptide. Protein N-glycosylation is often altered in cancer and may contribute in tumor formation progression. On the other hand, complex branching of N-glycans also leads to tumor-suppressive properties in some cancers. Coenzyme Q is crucial for ATP production in cancer cells that rely on oxidative phosphorylation to produce energy.

Lipid Metabolism

371

Therapeutic Targeting Studies on significance of de novo lipid synthesis pathways in cancer progression suggest that the enzymes involved in these pathways would be rational therapeutic targets for cancer treatment. Multiple studies on therapeutic targeting of de novo fatty acid synthesis pathways mainly focused on pharmacological inhibition of FASN (Fig. 1). The effects of FASN-inhibitors have been examined in diverse preclinical tumor models. These inhibitors are effective in chemoprevention of breast cancer in HER2/neu transgenic mice. They also reduce incidence of chemically induced lung tumorigenesis. Several studies have shown that inhibition of other key enzymes involved in fatty acid synthesis pathwaydincluding ACACA and SCDdlimit the growth and proliferation of cancer cells. Silencing of ACLY has also been reported to prevent cancer cell growth both in vivo and in vitro. The significance of inhibitors of the mevalonate pathway as antitumor agents has also been investigated. Statinsdpotent competitive inhibitors of HMGCRdhave been shown to induce growth arrest and apoptosis in cancer cells both in vitro and in vivo. Moreover, bisphosphonatesdinhibitors of FPP synthasedare also approved for the treatment of multiple myeloma breast and prostate cancer. Zaragozic acid Adinhibitor of squalene synthasedattenuates proliferation and induces death of prostate cancer cells. Table 1 provides a list of pharmacological inhibitors of de novo lipid synthesis in clinical or preclinical trials.

Table 1

List of pharmacological inhibitors of lipid metabolism pathways Drug development stage

Cancer type

Selected references

Fatty acid synthesis pathway SB-204990 ACLY

Preclinical

Solid and non solid tumors

TOFA

ACACA

Preclinical

Lung, colon, and breast cancer

Soraphen A

ACACA

Preclinical

Prostate cancer

Metformin GSK2194069 Triclosan

ACACA FASN FASN

FDA approved Preclinical Preclinical

Solid tumors Prostate cancer Prostate cancer, breast cancer

Fasnall FAS31 C247

FASN FASN FASN

Preclinical Preclinical Preclinical

BZ36 A939572

SCD SCD

Preclinical Preclinical

NA Ovarian cancer Non-small-cell lung cancer and breast cancer Prostate cancer Solid tumors

Aisenberg (1961), Alli et al. (2005), Baenke et al. (2013), and Baron et al. (2004) Aisenberg (1961), Alli et al. (2005), Bauer et al. (2005), Beckers et al. (2007), and Beloribi-Djefaflia et al. (2016) Aisenberg (1961) and Boroughs and DeBerardinis (2015) Chajes et al. (2006) Aisenberg (1961), and Das and Hoefler (2013) Aisenberg (1961), Alli et al. (2005), DeBerardinis et al. (2008), Effert et al. (1996), and Fackler and Grosse (2008) Fritz et al. (2010) Aisenberg (1961) and Gao and Zhang (2008) Aisenberg (1961), Hanai et al. (2011), and Hatzivassiliou et al. (2005) Aisenberg (1961) and Kridel et al. (2004) Aisenberg (1961), Alli et al. (2005), Kuhajda (2006), Kuhajda et al. (2000), and Kusakabe et al. (2000)

Preclinical

Prostate cancer

Compound/inhibitor

Target

Transcriptional regulators Fatostatin SREBPs Betulin Mevalonate pathway Statins

SREBPs

Preclinical

Multiple cancers

Aisenberg (1961), Liu et al. (2010), and Mashima et al. (2009) Aisenberg (1961) and Mason et al. (2012)

HMGCR

Solid tumors

Alli et al. (2005) and Menendez and Lupu (2007)

Simvastatin FAs oxidation Etomoxir

HMGCR

Clinical and Preclinical Preclinical

Prostate cancer

Alli et al. (2005) and Mullen et al. (2016)

CPT1

Preclinical

Prostate cancer, Leukemia and myeloma

CPT1 ACS

Preclinical Preclinical

Leukemia Solid tumors

Aisenberg (1961), Alli et al. (2005), Orita et al. (2008), Pavlova and Thompson (2016), and Price et al. (2002) Aisenberg (1961) and Rohrig and Schulze (2016) Aisenberg (1961) and Rysman et al. (2010)

MAGL AGPAT

Preclinical Preclinical

Solid tumors Solid tumors

Fritz et al. (2010) and Samudio et al. (2010) Fritz et al. (2010) and Stylli et al. (2008)

Perhexiline Triacscin C Lipolysis JZL184 CT-32501

Key: ACACA, (acetyl-CoA carboxylase A); ACLY (ATP-citrate lyase); ACP, (acyl carrier protein); ACS, (acyl-CoA synthetases); AGPAT, (acylglycerolphosphate acyltransferase); AMPK, (AMP-activated protein kinase); CPT1 (carnitine palmitoyltransferase); FASN, (fatty acid synthase); HMGCR, (3-hydroxy-3-methylglutaryl (HMG)-coenzyme A (CoA) reductase); FDPS, (farnesyl disphosphate synthase); MAGL, (monoacylglycerol lipase); SCD, (stearoyl-CoA desaturase); SCAP, (SREBP cleavage-activating protein); SREBPs, (sterol regulatory element binding protein); TOFA, (5-tetradecyl-oxy-2-furoic acid).

372

Lipid Metabolism

See also: Dietary Factors and Cancer.

References Aisenberg, A.C., 1961. The glycolysis and respiration of tumors. Academic Press, New York, pp. pp. 12–13. Alli, P.M., Pinn, M.L., Jaffee, E.M., McFadden, J.M., Kuhajda, F.P., 2005. Fatty acid synthase inhibitors are chemopreventive for mammary cancer in neu-N transgenic mice. Oncogene 24, 39–46. Baenke, F., Peck, B., Miess, H., Schulze, A., 2013. Hooked on fat: The role of lipid synthesis in cancer metabolism and tumor development. Disease Models and Mechanisms 6 (6), 1353–1363. Baron, A., Migita, T., Tang, D., Loda, M., 2004. Fatty acid synthase: A metabolic oncogene in prostate cancer? Journal of Cellular Biochemistry 91, 47–53. Bauer, D.E., Hatzivassiliou, G., Zhao, F., Andreadis, C., Thompson, C.B., 2005. ATP citrate lyase is an important component of cell growth and transformation. Oncogene 24, 6314–6322. Beckers, A., Organe, S., Timmermans, L., Scheys, K., Peeters, A., et al., 2007. Chemical inhibition of acetyl-CoA carboxylase induces growth arrest and cytotoxicity selectively in cancer cells. Cancer Research 67, 8180–8187. Beloribi-Djefaflia, S., Vasseur, S., Guillaumond, F., 2016. Lipid metabolic reprogramming in cancer cells. Oncogenesis 5, e189. Boroughs, L.K., DeBerardinis, R.J., 2015. Metabolic pathways promoting cancer cell survival and growth. Nature Cell Biology 17 (4), 351–359. Chajes, V., Cambot, M., Moreau, K., Lenoir, G.M., Joulin, V., 2006. Acetyl-CoA carboxylase alpha is essential to breast cancer cell survival. Cancer Research 66, 5287–5294. Das, S.K., Hoefler, G., 2013. The role of triglyceride lipases in cancer associated cachexia. Trends in Molecular Medicine 19, 292–301. DeBerardinis, R.J., Lum, J.J., Hatzivassiliou, G., Thompson, C.B., 2008. The biology of cancer: Metabolic reprogramming fuels cell growth and proliferation. Cell Metabolism 7 (1), 11–20. Effert, P.J., Bares, R., Handt, S., Wolff, J.M., Bull, U., et al., 1996. Metabolic imaging of untreated prostate cancer by positron emission tomography with 18fluorine-labeled deoxyglucose. The Journal of Urology 155, 994–998. Fackler, O.T., Grosse, R., 2008. Cell motility through plasma membrane blebbing. The Journal of Cell Biology 181, 879–884. Fritz, V., Benfodda, Z., Rodier, G., Henriquet, C., Iborra, F., et al., 2010. Abrogation of de novo lipogenesis by stearoyl-CoA desaturase 1 inhibition interferes with oncogenic signaling and blocks prostate cancer progression in mice. Molecular Cancer Therapeutics 9, 1740–1754. Gao, X., Zhang, J., 2008. Spatiotemporal analysis of differential Akt regulation in plasma membrane microdomains. Molecular Biology of the Cell 19, 4366–4373. Hanai, J., Doro, N., Sasaki, A.T., Kobayashi, S., Cantley, L.C., et al., 2011. Inhibition of lung cancer growth: ATP citrate lyase knockdown and statin treatment leads to dual blockade of mitogen-activated protein Kinase (MAPK) and Phosphatidylinositol-3-kinase (PI3K)/AKT pathways. Journal of Cellular Physiology 227, 1709–1720. Hatzivassiliou, G., Zhao, F., Bauer, D.E., Andreadis, C., Shaw, A.N., et al., 2005. ATP citrate lyase inhibition can suppress tumor cell growth. Cancer Cell 8, 311–321. Kridel, S.J., Axelrod, F., Rozenkrantz, N., Smith, J.W., 2004. Orlistat is a novel inhibitor of fatty acid synthase with antitumor activity. Cancer Research 64, 2070–2075. Kuhajda, F.P., 2006. Fatty acid synthase and cancer: New application of an old pathway. Cancer Research 66, 5977–5980. Kuhajda, F.P., Pizer, E.S., Li, J.N., Mani, N.S., Frehywot, G.L., et al., 2000. Synthesis and antitumor activity of an inhibitor of fatty acid synthase. Proceedings of the National Academy of Sciences of the United States of America 97, 3450–3454. Kusakabe, T., Maeda, M., Hoshi, N., Sugino, T., Watanabe, K., et al., 2000. Fatty acid synthase is expressed mainly in adult hormone-sensitive cells or cells with high lipid metabolism and in proliferating fetal cells. The Journal of Histochemistry and Cytochemistry 48, 613–622. Liu, Y., Zuckier, L.S., Ghesani, N.V., 2010. Dominant uptake of fatty acid over glucose by prostate cells: A potential new diagnostic and therapeutic approach. Anticancer Research 30, 369–374. Mashima, T., Seimiya, H., Tsuruo, T., 2009. De novo fatty-acid synthesis and related pathways as molecular targets for cancer therapy. British Journal of Cancer 100, 1369–1372. Mason, P., Liang, B., Li, L., Fremgen, T., Murphy, E., et al., 2012. SCD1 inhibition causes cancer cell death by depleting mono-unsaturated fatty acids. PLoS One 7, e33823. Menendez, J.A., Lupu, R., 2007. Fatty acid synthase and the lipogenic phenotype in cancer pathogenesis. Nature Reviews Cancer 7, 763–777. Mullen, P.J., Yu, R., Longo, J., Archer, M.C., Penn, L.Z., 2016. The interplay between cell signaling and the mevalonate pathway in cancer. Nature Reviews Cancer 16 (11), 718–731. Orita, H., Coulter, J., Tully, E., Kuhajda, F.P., Gabrielson, E., 2008. Inhibiting fatty acid synthase for chemoprevention of chemically induced lung tumors. Clinical Cancer Research 14, 2458–2464. Pavlova, N.N., Thompson, C.B., 2016. The emerging hallmarks of cancer metabolism. Cell Metabolism 23 (1), 27–47. Price, D.T., Coleman, R.E., Liao, R.P., Robertson, C.N., Polascik, T.J., et al., 2002. Comparison of [18F] fluorocholine and [18F]fluorodeoxyglucose for positron emission tomography of androgen dependent and androgen independent prostate cancer. The Journal of Urology 168, 273–280. Rohrig, F., Schulze, A., 2016. The multifaceted roles of fatty acid synthesis in cancer. Nature Reviews Cancer 16 (11), 732–749. Rysman, E., Brusselmans, K., Scheys, K., Timmermans, L., Derua, R., et al., 2010. De novo lipogenesis protects cancer cells from free radicals and chemotherapeutics by promoting membrane lipid saturation. Cancer Research 70, 8117–8126. Samudio, I., Harmancey, R., Fiegl, M., Kantarjian, H., Konopleva, M., et al., 2010. Pharmacologic inhibition of fatty acid oxidation sensitizes human leukemia cells to apoptosis induction. The Journal of Clinical Investigation 120, 142–156. Stylli, S.S., Kaye, A.H., Lock, P., 2008. Invadopodia: At the cutting edge of tumor invasion. Journal of Clinical Neurosciences 15, 725–737.

Further Reading Swinnen, J.V., Brusselmans, K., Verhoeven, G., 2006. Increased lipogenesis in cancer cells: New players, novel targets. Current Opinion in Clinical Nutrition & Metabolic Care 9 (4), 358–365. Wagle, S., Bui, A., Ballard, P.L., Shuman, H., Gonzales, J., et al., 1999. Hormonal regulation and cellular localization of fatty acid synthase in human fetal lung. The American Journal of Physiology 277, L381–390. Willemarck, N., Rysman, E., Brusselmans, K., Van Imschoot, G., Vanderhoydonc, F., et al., 2010. Aberrant activation of fatty acid synthesis suppresses primary cilium formation and distorts tissue development. Cancer Research 70, 9453–9462. Zaidi, N., Royaux, I., Swinnen, J.V., Smans, K., 2012a. ATP citrate lyase knockdown induces growth arrest and apoptosis through different cell- and environment-dependent mechanisms. Molecular Cancer Therapeutics 11, 1925–1935. Zaidi, N., Swinnen, J.V., Smans, K., 2012b. ATP-citrate lyase: A key player in cancer metabolism. Cancer Research 72, 3709–3714. Zaidi, N., Lupien, L., Kuemmerle, N.B., Kinlaw, W.B., Swinnen, J.V., Smans, K., 2013. Lipogenesis and lipolysis: The pathways exploited by the cancer cells to acquire fatty acids. Progress in lipid research 52 (4), 585–589. Zha, S., Ferdinandusse, S., Hicks, J.L., Denis, S., Dunn, T.A., et al., 2005. Peroxisomal branched chain fatty acid beta-oxidation pathway is upregulated in prostate cancer. Prostate 63, 316–323.

Lipid Metabolism

373

Zhan, Y., Ginanni, N., Tota, M.R., Wu, M., Bays, N.W., et al., 2008. Control of cell growth and survival by enzymes of the fatty acid synthesis pathway in HCT-116 colon cancer cells. Clinical Cancer Research 14, 5735–5742. Zhang, F., Du, G., 2012. Dysregulated lipid metabolism in cancer. World Journal of Biological Chemistry 3 (8), 167–174. Zhou, W., Han, W.F., Landree, L.E., Thupari, J.N., Pinn, M.L., et al., 2007. Fatty acid synthase inhibition activates AMP-activated protein kinase in SKOV3 human ovarian cancer cells. Cancer Research 67, 2964–2971.

Lynch Syndrome Pa¨ivi Peltoma¨ki, University of Helsinki, Helsinki, Finland © 2019 Elsevier Inc. All rights reserved.

Glossary Epigenetic change Heritable alteration in the function (expression) of a gene without a change in DNA sequence. Epigenetic mechanisms include DNA methylation, histone modification, and noncoding RNA (e.g., micro-RNA)-based gene regulation. Gene penetrance The proportion of individuals with a particular variant (genotype) that express an associated trait (phenotype). Haploinsufficiency Half of the normal level of the protein product of a gene (as a result of a heterozygous mutation, for example) is insufficient for normal function. Loss of heterozygosity (LOH) The absence of one of the two parental alleles in tumor tissue from an individual who is constitutionally heterozygous for the marker used to measure LOH. LOH can expose a mutant tumor suppressor gene in the remaining allele. Microsatellite instability (MSI) The number of repeat units of a microsatellite repeat is different in tumor DNA compared to normal DNA from the same individual, resulting in a difference in the microsatellite length between normal and tumor tissue. During replication, DNA polymerase may slip from the DNA strand at the repeat region, causing deletion or addition of repeat units, and the errors remain if postreplicative mismatch repair is defective.

Abbreviations CIMP CpG island methylator phenotype CMMRD Constitutional mismatch repair deficiency EMAST Elevated microsatellite alterations at selected tetranucleotide repeats FCCTX Familial colorectal cancer, type X LLS Lynch-like syndrome LS Lynch syndrome MMR DNA mismatch repair MSI Microsatellite instability MSI-H High-degree MSI MTS Muir–Torre syndrome TCGA The Cancer Genome Atlas TS Turcot syndrome

Lynch Syndrome: Definitions and Phenotypes Lynch syndrome (LS) is an autosomal dominant disorder caused by a defect in one of the DNA mismatch repair (MMR) genes. The syndrome is characterized by the development of colorectal carcinoma, endometrial carcinoma, and other cancers. Three main sets of clinical criteria have been formulated to select families suspected of having LS for molecular and other studies (Table 1). These include the stringent Amsterdam criteria I and II and the less stringent Bethesda criteria. Absent MMR protein(s) and DNA microsatellite instability (MSI) serve as tumor markers for LS. These are not, however, specific for LS since similar abnormalities occur in 15% of sporadic cancers as well. A panel of five microsatellite markers, the so-called Bethesda panel (BAT25, BAT26, D2S123, D5S346, and D17S250) has been recommended for screening purposes. Instability at two or more microsatellite loci indicates high-degree MSI (MSI-H) (see section “MMR Gene Dysfunction and MSI”). Table 2 summarizes the clinical and genetic features of LS and its variants. The identification of a pathogenic germline mutation in one of the DNA MMR genes MLH1 (MutL Homolog 1), MSH2 (MutS Homolog 2), MSH6 (MutS Homolog 6), or PMS2 (postmeiotic segregation increased 2 (S. cerevisiae)) is required for LS diagnosis. The genetic definition of LS covers the Muir–Torre syndrome, where sebaceous gland tumors co-occur with LS-type internal malignancy. Rare individuals who inherit defective alleles from both parents in a recessive manner resulting in homozygosity or compound heterozygosity for predisposing mutation have constitutional mismatch repair deficiency (CMMRD). Clinical presentation is severe and includes childhood cancers such as hematological malignancies and brain tumors, and early-onset colorectal cancer. Germline mutations in MMR genes may also be responsible for Turcot syndrome (TS), in which primary brain tumors, usually glioblastomas, are accompanied by multiple colorectal

374

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65048-1

Lynch Syndrome Table 1

375

The Amsterdam I and II and revised Bethesda criteria for the clinical diagnosis of LS

Amsterdam criteria I There should be at least three relatives with colorectal cancer (CRC); all the following criteria should be present: (1) One should be a first-degree relative of the other two (2) At least two successive generations should be affected (3) At least one CRC should be diagnosed before age 50 (4) Familial adenomatous polyposis should be excluded (5) Tumors should be verified by pathological examination Amsterdam criteria II There should be at least three relatives with a Lynch syndrome-associated cancer (colorectal cancer (CRC), cancer of the endometrium, small bowel, ureter, or renal pelvis): all of the following criteria should be present: (1) One should be a first-degree relative of the other two (2) At least two successive generations should be affected (3) At least one should be diagnosed before age 50 (4) Familial adenomatous polyposis should be excluded in the CRC case (s) if any (5) Tumors should be verified by pathological examination Revised Bethesda criteria The criteria can be used to select individuals for MSI analysis. Individuals that meet one of these criteria are suspected of having Lynch syndrome. (1) Colorectal cancer diagnosed in a patient MSH6

AD

Constitutional mismatch repair deficiency (CMMRD)

276300

PMS2 > MSH6 > MSH2, MLH1

AR

Turcot syndrome (TS)

276300

See CMMRD

AR (AD)c

Multiple sebaceous gland adenomas co-occurring with visceral malignancies, such as colorectal carcinoma Childhood cancers, mainly hematological malignancies and/or brain tumors, combined with early-onset colorectal cancers Primary brain tumors co-occurring with multiple colorectal adenomas

Syndrome

MIM no. a

Lynch syndrome (LS)

120436 120435

Muir–Torre syndrome (MTS)

Clinical features

AD, autosomal dominant; AR, autosomal recessive. a Mendelian Inheritance in Man (www.omim.org/). b The designation A > B indicates that mutations in gene A are more frequent than in gene B. c TS arising as a variant of CMMRD shows recessive transmission. TS may also be associated with germline mutation of the Adenomatous Polyposis Coli (APC) gene (MIM no. 175100), in which case it is dominantly inherited.

adenomas. TS is a variant of LS if associated with a heterozygous MMR gene mutation, CMMRD if associated with a homozygous or compound heterozygous MMR gene mutation, and familial adenomatous polyposis if associated with a heterozygous mutation of the gene for adenomatous polyposis coli (APC). LS is among the most prevalent hereditary cancer syndromes in man. It accounts for 1%–3% of unselected colorectal or endometrial carcinomas and some 15% of those with MSI or absent MMR protein. The population incidence of LS is around 1:400.

LS Genes and Their Functions Postreplicative DNA Mismatch Repair LS genes are conserved in evolution (Table 3). The postreplicative MMR system in E. coli includes three main proteins: MutS (recognizes the mismatch), MutH (marks the nascent strand by making a nick), and MutL (couples mismatch recognition to the

376

Lynch Syndrome Table 3

Lynch syndrome genes in evolution and phenotypes resulting from germline mutations

E. coli

S. cerevisiae

H. sapiens

Phenotype resulting from germline mutation

MutS

Msh2 Msh6 Msh3 Msh1 Msh4 Msh5 Mlh1 Pms1 Mlh2 Mlh3 Not identified

MSH2 MSH6 MSH3 Not identified MSH4 MSH5 MLH1 PMS2 PMS1 MLH3 Not identified

LS, MTS, CMMRDa LS, CMMRDa Attenuated adenomatous polyposis/CMMRDa N/A No Lynch syndrome-associated mutations known No Lynch syndrome-associated mutations known LS, MTS, CMMRDa LS with reduced penetrance, TS/CMMRDa No Lynch syndrome-associated mutations known LS? N/A

MutL

MutH

LS, Lynch syndrome; MTS, Muir–Torre syndrome; CMMRD, constitutional mismatch repair deficiency; TS, Turcot syndrome; N/A, not applicable. a Biallelic germline mutations.

downstream events). MMR acts on the newly synthesized strand; therefore, a mechanism of strand discrimination is necessary. In E. coli, a methyl group on the template strand directs MutH to make a nick on the opposite strand. Thus, hemimethylated DNA is corrected on the unmethylated strand and the methylated strand serves as a template. Multiple homologues of MutS and MutL exist in eukaryotes. In humans, five MutS homologues (MSH2, MSH6, MSH3, MSH4, and MSH5) and four MutL homologues (MLH1, PMS2, PMS1, and MLH3) have been identified and at least six of them play a direct role in MMR (Table 3, Fig. 1). MSH4 and MSH5 are necessary for meiotic (and possibly mitotic) recombination but do not participate in MMR. The role of human PMS1 in MMR is unclear (see later). The MMR proteins form heterodimers in different combinations and have different substrate specificities (Table 4). The main mismatch-binding factor is hMutSa, consisting of MSH2 and MSH6, which recognizes single-base mispairs and insertion–deletion loops (IDLs). Another mismatch-binding heterodimer is hMutSb, formed by MSH2 and MSH3, acting mainly on IDLs. According to

T

Insertion/deletion loop TG TG TG GTG TGT

G

CACAC ACACA

Single-base mismatch

hMSH2 hMSH6 or hMSH3

hMutS Mismatch recognition -> Formation of a sliding clamp

Key:

ATP

hMLH1 hPMS2 or hMLH3

ADP hMutL Assembly of larger complex of MMR proteins

Strand discrimination and removal of mismatch

PCNA RFC EXO1 Helicase(s) PCNA RPA DNA polymerase δ/ε DNA ligase C

hMutS

GTGTGTGTGT

Resynthesis and ligation G

CACACACACA

Fig. 1 DNA mismatch repair. The main biochemical steps necessary for the correction of replication errors, single-base mismatches, and insertion/ deletion loops in the newly synthesized strand (red), are shown. PCNA, proliferating cell nuclear antigen; RFC, replication factor C; EXO1, exonuclease 1; RPA, replication protein A.

Lynch Syndrome Table 4

377

Complexes of human MMR proteins and their substrate specificities

Heterodimer

Components

Type of mismatches repaired

Predominant type of MSI (defective gene)

hMutSa hMutSb hMutLa hMutLb hMutLg

MSH2 þ MSH6 MSH2 þ MSH3 MLH1 þ PMS2 MLH1 þ PMS1 MLH1 þ MLH3

Single-base mismatches, ins/del loops Ins/del loops Single-base mismatches, ins/del loops ? Single-base mismatches, small loops

MSI-H (MSH2 or MSH6) MSI-L/EMAST (MSH3) MSI-H (MLH1 or PMS2) ? MSI-L/EMAST or MSI-H (MLH3)

the “molecular switch” or “sliding clamp” model, the hMutS heterodimer binds to mismatched DNA in an ADP-bound form. Mismatch binding induces an ADP to ATP exchange that allows the hMutS complex to convert to a sliding clamp. The purpose of the sliding clamp that diffuses away from the mismatch is thought to be to empty the mismatch site for the loading of additional hMutS clamps and to recruit the hMutL heterodimer and the necessary downstream factors to the site of repair. The hMutL heterodimer coordinates the interplay between the mismatch recognition complex and other proteins necessary for MMR. The main hMutL complex is hMutLa, consisting of MLH1 and PMS2 and responsible for single-base mismatch and IDL repair. Alternative hMutL heterodimers are hMutLg (MLH1 and MLH3), which may predominantly contribute to IDL repair, and hMutLb (MLH1 and PMS1), which does not appear to participate in MMR. When one of the component proteins (e.g., MSH2) is absent, its pair in the heterodimer (MSH6 in hMutSa) becomes unstable as well, whereas the expression of components belonging to other heterodimers (MLH1 and PMS2) is unaffected. Defined patterns of absent MMR proteins in immunohistochemical evaluations of tumor tissues can help prioritize individual MMR gene(s) for germline mutation screens in suspected LS. No MutH homolog has been identified in yeast or mammalian cells, and two main alternative mechanisms for strand discrimination have been proposed. First, nicks that separate Okazaki fragments on the lagging strand may mark the nascent strand during replication. Second, MutLa has a proliferating cell nuclear antigen-dependent endonuclease activity that allows strand-specific cleavage. The predominant role of MLH1 and MSH2 as LS predisposition genes (Table 2) is easy to understand given that their protein products are obligatory components in all types of heterodimers (Table 4), followed by MSH6 and PMS2. MLH3 mutations are rare (the protein is functionally redundant with PMS2). No LS-predisposing heterozygous germline mutations are known for MSH3 (functionally redundant with MSH6); however, biallelic (homozygous) mutations may cause susceptibility to attenuated adenomatous polyposis. No convincing LS-associated germline mutations are known for PMS1, MSH4, or MSH5.

MMR Gene Dysfunction and MSI Substrate specificities of the individual MMR proteins are reflected in different MSI phenotypes in tumors (Table 4). MSI-H is detected with the Bethesda panel of two mononucleotide repeats (BAT25 and BAT26) and three dinucleotide repeats (D2S123, D5S346, and D17S250); at least two unstable markers indicate MSI-H. (If only one marker or none is unstable, a tumor is said to have low-degree MSI (MSI-L) or be microsatellite-stable (MSS), respectively.) It has been suggested that the mononucleotide repeat markers BAT25 or BAT26 alone are sensitive and specific indicators of MSI-H. MSH2 and MLH1 mutations are associated with MSI-H involving mononucleotide and dinucleotide (and other short tandem) repeats. The same is true for PMS2 mutations. In the case of MSH6 mutations mononucleotide repeats are predominantly affected. In tumors from MLH3 mutation carriers, mononucleotide repeats may be less informative than dinucleotide and tetranucleotide repeats and MSI phenotypes may range from MSI-high to the absence of MSI. Loss of MSH3 function may underlie a distinct form of MSI called EMAST (elevated microsatellite alterations at selected tetranucleotide repeats). EMAST and MSI-L are likely to be the same. EMAST can be frequent as an acquired defect in colorectal cancers or result from rare biallelic germline mutations in MSH3 associated with polyposis predisposition. While no consensus definition for EMAST exists to date, a commonly used definition is 1–2 tetranucleotide repeat markers mutated out of 5 or more markers examined.

Other Functions of MMR Genes Genomic instability is an “enabling characteristic” that, by generating genetic diversity, expedites the acquisition of the fundamental hallmarks of cancer that are prerequisites for tumor growth and metastasis. Apart from the correction of replication errors, MMR proteins have replication-independent (so-called noncanonical) functions whose failure may play important roles in conferring selective advantage on subclones of mutant cells. MMR proteins recognize diverse types of endogenous and exogenous damage, such as heterocyclic amine (e.g., PhIP) DNA adducts and damage induced by oxidation or alkylation. If possible, the lesions are corrected; otherwise, MMR proteins signal DNA damage to cell cycle arrest or apoptosis. In these processes, MMR proteins interact with other players of the DNA damage response pathway, including ATM and ATR, protein kinases involved in apoptosis and cell cycle regulation.

378

Lynch Syndrome

Recombination between DNA sequences that are related but nonidentical (homeologous) generates mismatches that act as substrates for the MMR pathway. MMR proteins correct mismatches occurring in recombination and suppress recombination between homeologous sequences. If recombination-related mismatches are not corrected, gene conversion is reduced in mitotic cells and “postmeiotic segregation” (PMS) is increased in meiotic cells (S. cerevisiae). The MMR system can also promote mutations when appropriate. For example, MSH4 and MSH5 facilitate meiotic crossover between homologous chromosomes. MMR proteins also promote somatic hypermutation and class switch of antibody genes.

LS Genes and Knudson’s Two-Hit Hypothesis Generally, both gene copies of a given MMR gene need to be inactivated for a phenotype in agreement with Knudson’s two-hit hypothesis for tumor suppressor genes. As a consequence, protein expression of the affected MMR gene is absent in tumor tissue and microsatellite repeats become unstable (Fig. 2, upper panel). The frequencies of immunohistochemical aberrations and especially MSI display some organ-specific variation (Fig. 2, lower panel), which is discussed in greater detail in the section “LS Tumor Spectrum and Its Molecular Basis.” In LS, the first hit inactivating one allele of the responsible MMR gene is inherited, whereas the second hit inactivating the remaining wild-type allele occurs in, and is restricted to, the somatic cancer progenitor cell in a target tissue. Either one of the two hits, or both of them, may be genetic or epigenetic. Both hits are usually genetic in LS-associated cancer and consist of a point mutation or a large rearrangement for the first hit and loss of the wild-type allele (loss of heterozygosity, LOH), gene conversion, or point mutation for the second hit (Fig. 3). Occasionally, promoter methylation of a MMR gene (MLH1 or MSH2) provides the first hit (see section “Constitutional Epimutations Predisposing to LS”). Promoter methylation can also provide the second hit, although this is relatively rare. In sporadic tumors with MSI, two inactivating hits, one in each allele of a MMR gene, occur somatically prior to tumor initiation. In the most common situation, the two hits are epigenetic and consist of biallelic methylation of MLH1. In approximately half of MMR-deficient tumors without germline mutations or promoter methylation, two somatic mutations of a MMR gene, or a somatic mutation and LOH, explain the MMR deficiency (Lynch-like syndrome (LLS), see section “Differential Diagnosis of LS”). Although MMR genes are typically recessive on cellular level, a single allele is sometimes sufficient to initiate tumorigenesis (haploinsufficiency). MMR gene dosage makes a clear difference regarding phenotypic consequences. Heterozygosity versus homozygosity for the predisposing mutation causes drastically different constitutional phenotypes (LS vs. CMMRD, Table 2). In somatic target tissues, the amount of available MMR protein may regulate tumorigenesis in a function-specific manner. For example, it has been shown that DNA damage signaling requires higher dosage of the MLH1 protein than DNA MMR. Furthermore, MLH1 haploinsufficiency in sporadic cancers and cell lines (such as pancreatic and renal cell cancers) prone to MLH1-LOH has been reported to cause a distinctive form of genomic instability involving increased rates of insertion/deletion mutations without

MSI analysis (BAT26)

Immunohistochemical analysis (MLH1)

Normal Cancer

Normal

A26 Cancer

A14

A26

r do de m r et riu m Ki dn ey Br ea st Br ai n

te

ad

En

U re

Bl

ry

on ol

C

va O

St

om ac

h

100 % 80 % 60 % 40 % 20 % 0%

MSI-H

MMR protein inactivation

Fig. 2 Microsatellite instability and absent MMR protein as hallmarks of LS tumors. Top left: BAT26 includes a stretch of 26 consecutive adenine nucleotides (A26); shortened alleles (A14 in this example) indicate MSI. Top right: Staining with an MLH1 antibody reveals absent MLH1 expression (blue color) in colorectal carcinoma cells from an MLH1 mutation carrier, whereas MLH1 is normally expressed (brown color) in the patient’s nonneoplastic mucosa. Bottom: The frequencies of LS tumors with MSI and MMR protein inactivation show organ-specific variation.

Lynch Syndrome

M

P

M

P

379

- Point mutation - Large rearrangement - Epimutation

First hit (germline)

M

P

M

P

M

P

M

P

Second hit (somatic)

Deletion of wild-type allele (LOH)

Gene conversion

Point mutation

Promoter methylation

None or unknown (haploinsufficiency)

Fig. 3 Mechanisms of two-hit inactivation of MMR genes in LS (see text). M refers to the maternal and P to the paternal allele (the maternal allele was randomly chosen as the target of the first hit).

conventional MSI. The truncating insertion/deletion mutations can inactivate tumor suppressor genes and thereby drive sporadic tumor development.

Germline Mutations Predisposing to LS The International Society for Gastrointestinal Hereditary Tumors (InSiGHT) has collected information of LS-associated MMR gene mutations in a public database established in 1994. MLH1, MSH2, MSH6, and PMS2 account for 40%, 34%, 18%, and 8%, respectively, of the over 3000 unique germline sequence variants of MMR genes deposited to the database to date (Fig. 4, left, and www. insight-group.org). Mutations are scattered throughout the genes. Most MLH1 and MSH2 mutations and an important proportion of MSH6 mutations are truncating nonsense or frameshift mutations. Missense changes, which lead to single amino acid substitutions without altering the reading frame, also account for a significant share (30%–60%) of the known mutations in all four genes. It was precisely the abundance of missense-type changes that prompted a recent large-scale effort by InSiGHT to classify each MMR gene variant according to pathogenicity (Fig. 4, right). The classification is based on variant, family, and tumor characteristics

Classification for pathogenicity

No. of mutations

Predictive testing

Lynch syndrome mutations

Surveillance of variant carriers

5 (Pathogenic)

Nonsense, frameshift, large rearrangements, missense with proven pathogenicity

Yes

Full high-risk guidelines

4 (Likely pathogenic)

Splicing aberrations (untested)

Yes

Full high-risk guidelines

3 (Uncertain)

Missense, small inframe changes

No

Tailored for each family

2 (Likely non pathogenic)

Intronic and synonymous variants

No

Treat as no mutation detected

No

Treat as no mutation detected

InSiGHT class

1200 1000 800 600 400 200 0 MLH1

MSH2

Nonsense or frameshift Splicing change Missense

MSH6

PMS2

Large rearrangement In-frame change

Examples of variants

1 Polymorphisms (Non pathogenic)

Fig. 4 Mutations predisposing to LS and translation of mutation information to clinical recommendations. Left: Based on information deposited in the InSiGHT database, the number of unique mutations discovered for each main LS susceptibility gene is shown, along with a breakdown into five mutation types within each gene. Right: An individual variant can be assigned a pathogenicity class from 1 (non-pathogenic) to 5 (pathogenic) on the basis of evaluation criteria formulated by the InSiGHT. Typical examples of changes representing each category are given. Possibilities for clinical use in predictive testing of at-risk relatives or regular surveillance of variant carriers vary according to the class of pathogenicity.

380

Lynch Syndrome

on the one hand and a variety of functional assays on the other hand. It is linked to clinical recommendations as follows. Classes 5 and 4 indicate a “pathogenic” and “likely pathogenic” variant, respectively, implying that a causative mutation was detected. Surveillance according to full high-risk guidelines is warranted and predictive testing of at-risk relatives should be made available. Nonsense and frameshift mutations constitute a majority of class 5 and 4 variants. Class 3 comprises variants of unknown significance (VUS), predominantly missense changes. Clinical management has to be tailored case by case. Classes 2 and 1 indicate a “likely nonpathogenic” and “nonpathogenic” variant, respectively, suggesting that no causative mutation was found. Carriers of class 2 and 1 variants should be clinically treated as “no mutation detected.” Intronic variants and synonymous (silent) variants with no associated RNA aberration are typical representatives of category 2 and variants with frequency  1% in the general population (polymorphisms) of category 1. Most MMR gene mutations are inherited from either parent and de novo mutations are rare. A majority of MMR gene mutations are unique (specific to a single family). Nevertheless, some prevalent recurrent mutations are known and may arise de novo or represent founder mutations. Founder mutations originate from a single ancestor, revealed in present-day carriers by shared haplotypes of various sizes depending on the age of the mutation. Founder mutations are enriched in isolated populations, such as the Finns, Newfoundland population, or Ashkenazi Jews. A genetic point mutation is the main type of germline mutation in MMR genes. Analysis of LS cohorts from different geographic locations indicates a frequency of 15% for large genomic rearrangements. Mutations typically abrogate one or several capabilities necessary for a functional MMR system, such as MMR protein expression, transport to the nucleus, heterodimer formation, DNA mismatch binding, and ATP–ADP cycling (Fig. 1). In a small but significant percentage of LS-suspected families with MMRdeficient tumors and negative for point mutations and large rearrangements, the predisposing defect is regulatory and consists of a constitutional epimutation in an MMR gene. The latter mechanism is discussed in more detail in the next section.

Constitutional Epimutations Predisposing to LS Constitutional epimutation can be defined as constitutional hypermethylation at the promoter of one allele of a given (nonimprinted) gene leading to silencing of expression from that allele in all main somatic tissues. If induced by a genetic alteration (usually located in cis), epimutation is secondary and otherwise primary. Primary and secondary epimutations of MLH1 have been described (Fig. 5A and B). The only known type of constitutional epimutation of MSH2 is a secondary epimutation caused by deletions of the 30 end of the EPCAM gene (Fig. 5C). After removal of stop codon, transcription of EPCAM reads into the (A)

(B)

Primary epimutation of MLH1

Secondary epimutation of MLH1 TF

Normal MLH1 allele

Exon 1

Exon 2

Epimutationcontaining MLH1 allele

Exon 1

Exon 2

(C)

C

Exon 1

Exon 2

Exon 1

Exon 2

5′UTR

Exon 1

Exon 2

5′UTR

Exon 1

Exon 2

A

Secondary epimutation of MSH2 Stop

5′UTR Exon 1

Exon 8

5′UTR Exon 1

Exon 8 EPCAM

Exon 9

3′UTR

MSH2

Fig. 5 Mechanisms of constitutional epimutations in MMR genes. In A to C, the upper allele is normal (open lollipops denote unmethylated CpG sites and an arrow depicts active transcription) whereas the lower allele contains an epimutation (methylated CpG sites and transcriptionally inactive). (A) CpG sites at the promoter of an epimutation-containing MLH1 allele are methylated (black lollipops) blocking transcription from that allele (crossed arrow). No apparent reason (e.g., a genetic change) for methylation can be identified, which is why the epimutation is designated primary. (B) A single nucleotide polymorphism (C to A) abolishes transcription factor (TF) binding and leads to secondary methylation at the MLH1 promoter. (C) Owing to a genomic deletion, the stop codon of the EPCAM gene is lost and transcription continues into the adjacent MSH2 gene, causing secondary methylation of the structurally normal MSH2 promoter.

Lynch Syndrome

381

adjacent, structurally normal MSH2 gene inducing promoter methylation of MSH2. In tumor tissues of constitutional epimutation carriers, the remaining wild-type allele is inactivated by a genetic or epigenetic mechanism according to Knudson’s two-hit hypothesis (Fig. 3). Carriers of constitutional epimutations of MLH1 or MSH2 are clinically indistinguishable from carriers of conventional (structural) MMR gene mutations (Table 5). However, family features and segregation patterns make a distinction. Epigenetic changes are erased on passage through the germline, preventing regular transmission from parent to child. Therefore, primary constitutional epimutations segregate in a non-Mendelian manner and are seldom associated with any remarkable family history of cancer (the revised Bethesda criteria may be fulfilled, Table 1). Examination of the families of primary epimutation carriers has revealed variable patterns of transmission, including apparent heritability, mosaic epigenetic inheritance, and reversion of the methylated allele to normal active state. Importantly, it has been shown that asymptomatic low-level somatic mosaicism for a MLH1 epimutation in a parent can produce a nonmosaic constitutional epimutation in the offspring, resulting in early-onset colorectal cancer. Such observations make genetic counseling of MLH1 epimutation carriers and their families quite challenging. In contrast, secondary epimutations of MLH1 or MSH2 give rise to classical LS families and regularly cosegregate with their cis’acting genetic changes in a dominant Mendelian fashion. Nevertheless, the mechanism of transmission may display distinct features compared to that of a genetic mutation. Secondary epimutation of MLH1 has been shown to undergo a complete but transient reversion in the germline, being erased in spermatozoa but reinstated in somatic cells of the next generation. To date, more than 50 index cases with primary or secondary constitutional epimutation of MLH1 are known. Constitutional epimutations may explain 1%–10% of mutation-negative Lynch-suspected families with silenced MLH1 expression in tumors (Table 5). A comparable number of index cases with EPCAM deletion-induced MSH2 epimutation have been described. Their share of Lynch suspected families without conventional germline mutations and with absent MSH2 protein in tumor tissues ranges between 0% and 40% depending on possible founder effects.

Tumorigenesis in LS Precursor Lesions Colorectal tumorigenesis in LS is traditionally considered to follow the adenoma–carcinoma sequence, which forms the rationale of using colonoscopy screening and removal of adenomas as an important means of cancer prevention in MMR gene mutation carriers. In sporadic colorectal tumorigenesis, the complete adenoma–carcinoma sequence can take 15 years or more. The presence of one defective copy of the predisposing MMR gene in every cell of LS individuals may shorten the process to only a few years or less. MMR genes are “caretaker” genes which when mutated, may accelerate tumor progression in particular, while the incidence of polyps is not markedly different from the average population. This is in contrast to “gatekeeper” genes such as APC that accelerate tumor initiation and underlie polyposis syndromes. Compatible with the underlying MMR defect, adenomas in LS often show aggressive features such as villosity and poor differentiation (high grade). Additionally, tumor-infiltrating lymphocytes (TILs) are common because of neoantigens resulting from mutant proteins. Recent evidence suggests that adenomas in LS may not always be easily detectable or may not exist prior to colorectal carcinoma development. For example, a report from a prospective LS database based on 10 centers from Europe and Australia

Table 5

Constitutional epimutations of MLH1 and MSH2 in LS predisposition

Type of epimutation Gene defects associated with secondary epimutation reported in the literature Tumor phenotype Clinical phenotype

Pattern of transmission Share of LSa a

MLH1

MSH2

Primary or secondary • Genomic deletion of MLH1 exons 1–2 • Genomic deletion of MLH1 exon 1 • Duplication of MLH1 and flanking regions • c.-27C > A in the MLH1 promoter MSI Lack of MLH1 (and PMS2) protein May present as a sporadic case (primary epimutation) or as a classical Lynch syndrome family (secondary epimutation). Age at onset and tumor spectrum are similar to Lynch syndrome. Atypical phenotypes (e.g., increased number of polyps) are possible Variable, from apparent lack of heritability (primary epimutation) to autosomal dominant (secondary epimutation) 1%–10%

Secondary • Deletions of EPCAM removing stop codon

MSI Lack of MSH2 (and MSH6) protein Presents as a classical Lynch syndrome family. Colorectal cancer risk is high. Endometrial cancer risk is increased with deletions extending close to the MSH2 promoter Autosomal dominant 10%–40%

Estimated from Lynch-suspected cases with MMR-deficient tumors and negative screens for conventional germline mutations in MMR genes.

382

Lynch Syndrome

(http://LScarisk.org) shows that colorectal cancer is surprisingly frequent despite colonoscopic surveillance with 1–3-year intervals. Furthermore, a long-term (mean 56 months) follow-up analysis of close to 900 MMR gene mutation carriers enrolled in the Colorectal Adenoma/Carcinoma Prevention Program 2 (CAPP2) surprisingly reveals that aspirin exhibits delayed protection against colorectal cancer without reducing adenoma incidence. These observations imply that colorectal carcinomas in LS may arise from lesions other than adenomas. Such possible alternative lesions remain to be identified and defined histologically and molecularly. Endometrial cancer, the most common extracolonic cancer in LS, is thought to develop via endometrial hyperplasias as precursor lesions. Accordingly, endometrial cancer prevention in LS is based on screening for endometrial hyperplasias by regular endometrial biopsies. Epidemiological studies suggest that atypical and complex hyperplasias are associated with increased risks of malignant transformation. According to the current knowledge, ovarian carcinomas of endometrioid and clear cell types, the predominant histological types of ovarian carcinoma in LS, also arise from endometrial epithelium, with atypical endometriosis as a recognized precursor. The fact that the abovementioned precursor lesions of endometrial and ovarian carcinoma share many genetic and epigenetic characteristics of the corresponding malignancies, such as MMR defects, tumor suppressor promoter methylation, and mutations in ARID1A, PIK3CA, or PTEN strongly supports their malignant potential.

Somatic Mutations in LS Tumors Owing to defective MMR, LS tumors belong to the “hypermutated” category defined by The Cancer Genome Atlas (TCGA) for unselected colorectal, endometrial, and other cancers. Whole-exome sequencing of MSI colorectal tumors by the TCGA network has revealed somatic nonsynonymous mutation rates above 12 per 106 bases, with a median number of
700 mutations per tumor. This is 10–100-fold compared to MMR-proficient tumors (median 58 mutations per tumor). The profiles of mutant genes differ in hypermutated and nonhypermutated tumors. For example, frameshift mutations in coding repeats within ACVR2A, TGFBR2, and other genes are characteristics of MSI tumors. The karyotypes are usually diploid and copy number changes are relatively rare. Molecular features typical of MSI tumors broadly apply to LS tumors as well. Additionally, certain universal signatures of tumor types, such as mutations in the adenoma–carcinoma progression sequence genes APC, KRAS, and TP53 in colorectal tumors and ARID1A, PIK3CA, and PTEN genes in endometrial and nonserous ovarian cancers, are largely preserved in LS tumors, although mutation frequencies of individual genes may vary. The BRAF oncogene is an important exception to the generally similar patterns of mutant genes between sporadic and LS tumors with MSI. Analysis of large cohorts of colorectal cancers indicates that BRAF V600E mutation occurs with a frequency of 1.4% in tumors from MMR gene mutation carriers (LS), compared with 5% in MSS cases and 63% in sporadic MSI colorectal cancers. BRAF V600E mutation coincides with MLH1 promoter “C region” methylation and strongly argues against LS. BRAF V600E mutation can therefore be utilized as a biomarker for predicting MMR-negative mutation status in MSI colorectal cancers.

Epigenetic Alterations in LS Tumors The MMR system can influence the epigenetic machinery and vice versa. For example, deficient MMR can induce mutations in genes with key roles in epigenetic regulation (such as ARID1A and other chromatin regulator genes), therefore having the potential to alter epigenetic modifications. Coding repeats, common targets of frameshift mutations in MMR-deficient tumors, are part of several epigenetic regulator genes and can be one reason why epigenetic regulation is commonly affected in LS and sporadic cancers with MSI. Conversely, MMR genes can be epigenetically regulated by promoter methylation (MLH1) and micro-RNAs (MSH2, MSH6, and MLH1). Moreover, the histone mark H3K36me3 is required to recruit hMutSa to chromatin in the process of MMR. In light of this interplay, it is not unexpected that apart from somatic mutations, coordinated methylation of CpG islands that are normally unmethylated (CpG island methylator phenotype, CIMP) and global hypomethylation (with the long interspersed element 1, LINE-1, as a surrogate marker) are integral parts of LS tumorigenesis. CIMP has the potential to silence hundreds of tumor suppressor genes per cancer cell. DNA hypomethylation in turn can activate oncogenes and increase chromosomal instability via LINE-1 retrotransposition. Sporadic MSI colorectal cancers with MLH1 promoter methylation typically display high CIMP phenotypes. More or less the same loci are involved in LS-associated colorectal cancers, but the frequencies of tumors with methylation are lower. Promoter methylation appears early and increases along with tumor progression. Promoter methylation of SFRP1 and SLC5A8, established tumor suppressor genes of colorectal carcinogenesis, has been shown to be significantly elevated in normal mucosa from MMR gene mutation carriers previously diagnosed with colorectal carcinoma, as compared to mutation carriers of a comparable age with no prior colorectal carcinoma diagnosis, suggesting that methylation changes may form cancer-prone “fields” in histologically normal mucosa. Moreover, methylation increases with histological dysplasia, with 0% of normal mucosae, 15% of adenomas with lowdegree dysplasia, 23% of adenomas with high-degree dysplasia, and 50% of carcinomas from LS individuals reported to display CIMP-high with established marker genes. Analogous observations are available from endometrial tumorigenesis, except that different tumor suppressor genes (e.g., RASSF1 and CHD13) are preferentially affected given the tissue-specific nature of epigenetic regulation. Retrospective examination of serial endometrial biopsy samples from a long-term follow-up of LS mutation carriers has revealed MMR defects and elevated promoter methylation at tumor suppressor gene loci up to 12 years before endometrial carcinoma development. Furthermore, promoter methylation of CpG islands associated with tumor suppressor genes and microRNA genes stratifies endometrial lesions into a low-methylator group (normal endometrium and simple hyperplasia) and a high-methylator group (complex hyperplasia

Lynch Syndrome

383

with and without atypia and endometrial carcinoma) suggesting that increased methylation accompanies endometrial tumorigenesis. The findings apply to both LS-associated and sporadic endometrial lesions emphasizing their universal nature. The highest levels of global hypomethylation occur in sporadic MSS colorectal cancers. Extensive hypomethylation is not a feature of LS-associated colon cancers; however, when present, LINE-1 hypomethylation has been reported to be associated with a worse prognosis compared with LS patients without significant hypomethylation. In line with observations from LS colorectal cancers, LS-ovarian carcinoma shows a significantly higher average number of methylated tumor suppressor genes and lower degree of LINE-1 hypomethylation compared with the corresponding sporadic disease, which may contribute to the favorable prognosis of LS-ovarian cancer (see section “Molecular Profiles in Tumors Versus Disease Outcome”).

Temporal Relationship Between MMR Deficiency and Other Tumorigenic Events MMR defects are among the earliest detectable alterations in cancer-prone target tissues of LS individuals. Intestinal resections for small or large bowel cancer from LS mutation carriers display lesions termed MMR-deficient crypt foci with a frequency of one focus per 1 cm2 of nontumorous mucosa, as compared to none in non-LS control patients. MMR protein is absent and MSI is present in such lesions, indicating biallelic MMR gene inactivation. The abundance of MMR-deficient crypt foci contrasts with the low number of adenomas or carcinomas observed in LS and suggests that most lesions do not progress to malignancy. In colorectal polyps from MMR gene mutation carriers, the prevalence of MMR deficiency increases with the size and dysplasia of adenomatous polyps, from 67% in adenomas with low-degree dysplasia to 100% in adenomas with high-degree dysplasia; the latter frequency is similar to colorectal carcinomas. Hyperplastic polyps from LS individuals rarely (< 5%) display MMR defects and their malignant potential is considered low. In analogy to colorectal tumorigenesis, the prevalence of MMR defects increases with endometrial tumor progression in LS. Decreased MMR protein expression has been reported to occur in 7% in normal endometrium, 40% in simple hyperplasia, and
100% in complex hyperplasia with or without atypia and likewise in endometrial carcinoma; the frequencies of MSI are somewhat lower. The observations of abundant MMR-deficient crypt foci without progression to visible tumors on the one hand and adenoma development without biallelic MMR gene inactivation on the other hand imply the need of other oncogenic events besides MMR deficiency. As mentioned earlier, DNA methylation changes can be early events in tumorigenesis. Promoter methylation of SFRP1 and SLC5A8 may form field defects in histologically normal colonic mucosa from LS individuals. Studies on sporadic cases has revealed frequent hypermethylation of SFRP1 and SLC5A8 promoters in aberrant crypt foci, the earliest detectable morphological lesions of colorectal tumorigenesis that usually lack APC mutations, and methylation is accompanied by decreased expression of the corresponding proteins. The repair gene MGMT encoding O6-methylguanine DNA methyltransferase is a further gene whose loss of expression, usually by promoter methylation, can form field defects in normal mucosa from LS and sporadic cases. The resulting failure to process mutagenic methyl adducts may trigger cellular transformation by inducing mutations in cancer-related genes such as KRAS or by inactivating the MMR genes by mutations or promoter methylation. It has been postulated that methylation tolerance due to MGMT field defects may initiate sporadic or LS-associated MSI-colorectal cancer prior to MMR deficiency. As, however, no such studies are available that would have examined the earliest lesions (aberrant crypts) for both MMR and methylation abnormalities, it remains unsettled which aberrations come first. Mutations of APC, an important gatekeeper of colon tumorigenesis, are known to occur early in colon tumorigenesis; therefore, this gene has been used in studies addressing the temporal relationship between MMR defects and other molecular events. Investigations on mice heterozygous for the Apc Min mutation and knockouts for selected MMR genes (Min/þ, Msh2 / and Min/þ, Mlh1 / mice) demonstrate that MMR deficiency changes the spectrum of somatic Apc alterations from LOH, the usual second hit, to point mutations, suggesting that deficient MMR exerts its effect before Apc. On the other hand, comparison of the APC mutation spectra in sporadic MSI and MSS colon cancers has failed to identify in the former tumors any clear excess of changes characteristic of MMR defects, such as preferential involvement of repeat sequences, suggesting that MMR deficiency occurs after APC mutations. Taken together, studies conducted to date have remained conflicting or inconclusive in an attempt to determine the chronological order of MMR defects relative to other molecular events in multistep tumorigenesis.

Molecular Profiles in Tumors Versus Disease Outcome Evidence from a prospective LS database (http://LScarisk.org) on MMR gene mutation carriers without previous cancer and under regular surveillance shows that the 10-year survival after the first cancer is excellent: 91% for colorectal cancer, 98% for endometrial cancer, 89% for ovarian cancer, and 87% for any cancer. Although early detection probably plays a role, LS cancers are additionally likely to have intrinsic factors that contribute to the favorable outcome compared to cancers arising in the general population. Early studies already found that despite aggressive histological features (poor differentiation) of LS-associated colorectal carcinoma, LS patients experience a definite survival advantage over average colorectal carcinoma patients. Since the same applies to unselected colorectal carcinoma with MSI, it is logical to reason that MMR deficiency offers a plausible explanation. Increased mutational load triggering apoptosis of tumor cells can be one possible mechanism. The favorable effects of deficient MMR may also be related to the generation of neoantigens that can elicit strong antitumor immune responses. Multiple genes encompass coding microsatellite repeats prone to insertion/deletion mutations, resulting in neopeptides with altered carboxy-terminal sequences that are recognized as foreign by the immune system. Some peptides derived from mutant genes (TGFBR2, AIM2, HT001, and

384

Lynch Syndrome

TAF1B in one study and ASTE1, HNF1A, and TCF7L2 in another study) have been identified as especially immunogenic and may help develop vaccination-based immunotherapy in LS. Pronounced immune responses provoked by various neopeptides may eliminate MMR-deficient crypt foci present in nontumorous mucosa of LS individuals. Progression and metastasis of tumors that have already developed may be prevented by similar mechanisms. Furthermore, the active immune microenvironment is counterbalanced by immune inhibitory signals, resulting in overexpression of several immune checkpoint molecules, such as PD-1 and PD-L1. This may make metastatic MMR-deficient cancers particularly susceptible to pembrolizumab and other immune checkpoint inhibitors. Ovarian carcinomas arising in LS individuals are mainly of nonserous histology (endometrioid and clear cell type) and well or moderately differentiated early-stage tumors at diagnosis, as opposed to sporadic ovarian carcinomas, a majority of which are serous and advanced at presentation. Although the abovementioned clinicopathological differences are likely to contribute to differential disease outcomes, the 5-year survival of LS patients with even stage III or IV ovarian cancer is 59% compared to 28% in sporadic cases. A novel molecular profile for LS-associated ovarian carcinoma has been described, consisting of normal p53 expression, lack of KRAS mutation, MMR deficiency, significantly higher rate of methylated tumor suppressor genes and lower degree of LINE-1 hypomethylation compared to sporadic endometrioid and clear cell ovarian cancers. This markedly different overall molecular profile compared to sporadic ovarian cancers may in part explain the surprisingly high survival rates of LS-associated ovarian carcinoma.

Genotype-Phenotype Correlations in LS Cancer Risks Associated With Germline Mutations in MMR Genes The life-time risks of cancer vary according to the mutant gene, which may reflect the degree of functional redundancy of the corresponding proteins in MMR (Table 4). MLH1 and MSH2 mutations have high penetrance, with 60%–80% of mutation carriers developing some form of cancer during their lifetime. The mean age of onset is
45 years for colorectal cancer and
45–55 years for other cancers. MSH6 and especially PMS2 mutations have lower penetrance, 25%–54% for MSH6 and
20%–30% for PMS2. The mean age at onset is 55–60 years for colorectal cancer and 54 years for endometrial cancer in MSH6 mutation carriers. Heterozygous carriers of PMS2 mutation have a mean age at onset of around 50 years for colorectal and other cancers. The type (truncating vs. missense) or location (relative to different functional domains) of MMR gene mutations is poorly correlated with clinical phenotype.

LS Tumor Spectrum and Its Molecular Basis The Amsterdam II criteria (Table 1) acknowledge cancers of the colon and rectum, endometrium, small bowel, ureter, and renal pelvis as LS-associated cancers, because these are significantly more frequent in LS compared to the average population. Since the Amsterdam II criteria were formulated, significantly increased standardized incidence ratios have repeatedly been reported for cancers of the stomach, ovaries, and pancreas as well as for certain other cancers (footnote in Table 1), which combined with molecular profiles characteristic of LS, such as MMR protein loss and MSI, justifies their inclusion in the LS spectrum. Recently, breast cancer and prostate cancer have also been suggested to belong to the LS tumor spectrum, based on MMR deficiency in tumors from MMR gene mutation carriers, in addition to some published evidence of increased risk in LS compared to the average population. While confirmation from additional study series is still warranted for reliable conclusions, a recent investigation compared the rate of MMR deficiency in breast carcinomas from proven mutation carriers versus noncarriers from the same families, resulting in frequencies of 65% versus 0% (P < 0.001). Moreover, the age at onset of breast carcinoma in mutation carriers was earlier if the tumor was MMR-deficient, suggesting that MMR deficiency does play a role in breast cancer development in LS. Why MMR gene mutation carriers are predisposed to a narrow spectrum of cancers, considering that they carry one defective copy of a given MMR gene in all their cells, is not well understood. Some possible mechanisms of organ selectivity are depicted in Fig. 6. First, the individual MMR genes may be associated with different tumor spectra. For example, the risk of extracolonic cancers has been suggested to be higher in MSH2 than MLH1 mutation carriers. As another example, female carriers of MSH6 mutations are at a higher risk of endometrial than colorectal cancer. Second, the dosage of the MMR gene or protein is important for phenotype. As discussed earlier (Table 2), homozygosity or compound heterozygosity for germline mutation results in a separate syndrome, CMMRD. At present,
150 cases with CMMRD are known. Childhood cancers of the hematological system and brain, signs of neurofibromatosis type 1 (café-au-lait spots), and coexistence of colorectal and brain tumors (TS) are common manifestations of CMMRD. The distinct tumor spectrum may reflect the sensitivity of neural and hematological progenitor cells to MMR deficiency via specific somatic target genes, such as NF1. In contrast to LS with heterozygous MMR gene mutations (Fig. 2, upper right), CMMRD patients lack expression of the MMR protein(s) in question not only in cancer tissue but in normal tissue as well. Despite biallelic MMR gene inactivation, MSI is not usually detectable by standard methods in peripheral blood lymphocytes because of clonal heterogeneity; however, immortalized lymphoblastoid cells, which are oligoclonal, can reveal MSI. Another line of evidence to support the significance of MMR gene or protein dosage for phenotypic consequences comes from observations of haploinsufficiency in somatic cells (see section “LS Genes and Knudson’s Two-Hit Hypothesis”). Different organs may have different requirements for MMR gene dosage or

Lynch Syndrome

385

Inactivation by • genetic mutation • epimutation

Normal functions of MMR proteins • Correct replication errors • Act on endogenous and exogenous damage • Suppress mitotic recombination between homeologous sequences • Facilitate meiotic crossover between homologous chromosomes (MSH4, MSH5) • Promote somatic hypermutation and class switch of antibody genes

Genomic instability (MSI) + Selective mechanism • Predisposing MMR gene • Dosage of MMR gene or protein • Tissue-specific patterns of MMR deficiency combined with tissuespecific target genes • Exposure to endogenous or exogenous damage • Rate of cellular proliferation • Immune defense

Lynch syndrome phenotype • Colorectal, endometrial, and other cancers • Age at onset ~45 years (varies) • Lifetime cancer risk up to 80% (varies)

Fig. 6 Possible mechanisms that may lead to organ selectivity in LS. MMR gene inactivation by genetic or epigenetic mechanisms (see Fig. 3) abrogates normal functions of MMR proteins and results in genomic instability (see Fig. 2), which together with various selective mechanisms can target specific organs for tumor development and contribute to the overall LS phenotype (see text).

protein, which may affect their susceptibility to tumor development. For example, analysis of second hits in MLH1 mutation carriers shows that deletion of the wild-type MLH1 allele is more common in colorectal than endometrial cancer. Third, tissue specificity for MMR deficiency and genes targeted by failing MMR may be a factor behind organ selectivity. In LS, immunohistochemical analysis of cancers regularly demonstrates absent MMR protein corresponding to the gene mutant in the germline, but the frequency of MSI-high varies from 80% to 100% (stomach, ovary, colon, and ureter) to
50% (bladder, endometrium, and kidney) and even less (35% for breast and 0% for brain tumors) (Fig. 2, lower panel). Clonal heterogeneity is a feature of LS and sporadic MMR-deficient tumors and can partly explain the different frequencies of MSI between tumor types. Observations of BAT markers showing shorter allelic shifts in endometrial than colorectal cancers from LS patients offer another indication of tissue-specific patterns of clonal growth. Genes with coding repeats are preferential targets in MMR-deficient cancers (see earlier) and different genes confer selective advantage for different cancers, such as the TGFb superfamily for gastrointestinal cancers and PTEN for endometrial cancers. Organ-specific differences in the manifestation of MMR deficiency can therefore contribute to LS tumor spectrum via tissue-specific target genes with mutation-prone sequences. Fourth, exposure to endogenous and exogenous damages can have serious consequences when the ability of the MMR system to repair such damage is compromised due to inherited mutations. Such damage can be exogenous (many heterocyclic amines are of dietary origin) or endogenous (such as inflammation-induced oxidation). The accumulation of damage could facilitate cancer development in exposed organs, such as the gastrointestinal tract and endometrial epithelium. Western-style diet has been shown to increase colorectal adenomas in LS mutation carriers and colon adenomas and carcinomas in Mlh1 þ/ mice. Likewise, tobacco smoking appears to increase colorectal adenoma risk in LS mutation carriers. Such observations imply a reduced capacity of MMR gene mutation carriers to repair dietary and tobacco-associated damage. Fifth, tumors initiated by unrepaired mutations or carcinogen-derived adducts may show differential progression rates in different tissues depending on cellular proliferation. Colon and many other epithelial cells have fast turnovers, and short cell cycles may allow less time to repair errors. Stem cell divisions in the epithelium continue throughout life. Hematopoietic tissue, which is likewise highly proliferative, may be less cancer-prone because of fewer stem cell divisions during lifetime. Proliferation rates may explain why a majority of LS-associated tumors are epithelial and why MMR deficiency can occasionally (in CMMRD) cause predisposition to hematological malignancies. Finally, immunological processes may play a role in organ-specific tumor susceptibility. MMR deficiency induces frameshift mutations that can have differential effects on the immune system. Cell surface proteins responsible for antigen processing and presentation may lose expression as a consequence of frameshift mutations of genes encoding them and this may make it possible for tumor cells to escape from immune surveillance. For example, mutations in the B2M gene coding for beta 2 microglobulin can abrogate HLA class I antigen expression on the surface of tumor cells, allowing evasion from the attacks by cytotoxic T cells. Mutations in other types of genes such as ACVR2 or TGFBR2 may result in the formation of neoantigens and induce a strong immune response against cancer cells. High levels of TILs in colon, gynecological, and other tumors from LS patients are indicators of active immune surveillance. Differential manifestation of MMR defects (see earlier) combined with possible variations in the inherent efficacy of immune surveillance in different organs may contribute to organ selection in LS.

Differential Diagnosis of LS A suspicion of LS typically arises because of a family history of LS-spectrum cancers or because of MSI and/or immunohistochemical loss of MMR protein in tumor tissue. Starting from family or tumor studies, a number of alternative diagnoses are possible and need to be considered in differential diagnosis (Fig. 7, Table 6).

386

Lynch Syndrome

Clinical Phenotype Hereditary non-polyposis colorectal cancer (HNPCC)

Monoallelic LS associated with LS constitutional epimutation

Sporadic MSI cancer

Germline mutation in a MMR gene? No Yes Biallelic CMMRD

Yes No MLH1 promoter hypermethylation?

FCCTX

Two somatic mutations in a MMR gene? Yes LLS

MMR-deficient MMR-proficient Tumor phenotype

Sporadic MSS cancer

Fig. 7 Differential diagnosis of LS. Clinical and tumor phenotypes are common starting points for molecular evaluation to differentiate LS from overlapping conditions (see Table 6). When two options exist, a thick arrow denotes the most likely alternative.

Table 6

Syndromes that overlap with or mimic LS

Syndrome

Share among unselected CRCs (%)

Responsible gene

Family features

Tumor characteristics No MMR defects in a family that fulfills the Amsterdam criteria MSI-H or absent MLH1 protein due to MLH1 methylation MSI-H or absent MMR protein not explained by methylation of MLH1 or germline mutation of MMR genes. Attributable to two inactivating somatic events of MMR genes Rate of nonsynonymous mutations comparable to that caused by MMR defects, but MSI-H is absent. Somatic mutations result in “ultramutated” tumors, whereas germline mutations predispose to polymerase proofreading-associated polyposis (PPAP)

Familial colorectal cancer type X (FCCTX) Sporadic MSI


2

Mostly unknown

Hereditary/familial


15

MLH1

Sporadic

Lynch-like syndrome (LLS)

2.5

MLH1, MSH2, MSH6, or PMS2

Sporadic (some familial clustering reported)

POLE/POLD1-associated hypermutability

1–2

POLE or POLD1

Sporadic or hereditary

CRC, colorectal cancer; MSI-H, high-degree of microsatellite instability.

Familial Colorectal Cancer Type X Before the discovery of the LS genes in the 1990s, the Amsterdam Criteria (Table 1) were used to identify families with a presumably inherited form of colorectal carcinoma. Families that meet these clinical criteria were called hereditary non-polyposis colorectal cancer (HNPCC). Molecular analyses have revealed that in about half of such families, MSI or absent MMR protein in tumors or a germline MMR gene defect cannot be identified. Such families are referred to as familial colorectal cancer “type X” (FCCTX), because the genetic basis of this disease is largely unknown. FCCTX forms the MMR-proficient subset and LS with pathogenic germline mutations in MMR genes, the MMR-deficient subset of HNPCC. FCCTX families show site-specific (mainly distal) colorectal cancer, as opposed to colonic (mainly proximal) and extracolonic cancers seen in LS; furthermore, the mean age at cancer onset is 10 years higher (55 years in FCCTX vs. 45 years in LS). A number of novel candidates for FCCTX susceptibility genes have recently been identified, including GALNT12, BMPR1A, RPS20, SEMA4, and FAN1. The genes function in different biological pathways and each seems to be restricted to a small number of families. Colorectal carcinomas from FCCTX families display a remarkably low LINE-1 methylation (i.e., strong LINE-1 hypomethylation) as a common denominator. As low LINE-1 methylation is also a feature of normal mucosae from FCCTX patients, it may represent a field defect or a constitutional alteration. The mechanistic basis of FCCTX-associated hypomethylation remains to be uncovered by future studies.

Lynch Syndrome

387

Sporadic MSI-H Colorectal Cancer Associated With MLH1 Promoter Methylation If a colorectal cancer is MMR-deficient, the main subgroup (
15% of all colorectal cancers) to be considered is sporadic colorectal cancer with acquired biallelic methylation of the MLH1 promoter. Like LS-associated colorectal cancer, sporadic colorectal cancer with MLH1 methylation preferentially affects the proximal colon; however, the age at onset is higher (
70 years) compared to LS and, for reasons that remain to be clarified, there is a clear female predominance. The presence of BRAF V600E mutation combined with MLH1 promoter methylation identifies the sporadic MSI-H subgroup of colorectal cancers. A comparable (or even higher, up to 30%) share of unselected endometrial carcinomas may be MMR-deficient because of biallelic MLH1 methylation. While BRAF V600E mutation is absent in LS-associated endometrial carcinoma in analogy to LS-colorectal cancer, BRAF mutations are so infrequent in endometrial carcinomas in general that BRAF V600E testing has no role in differentiating sporadic MMR-deficient endometrial carcinoma from LS-associated disease.

Lynch-Like Syndrome Approximately 2.5% of all colorectal cancers are MMR-deficient in the absence of promoter methylation or causal germline mutations, which is referred to as LLS. In half of the cases, two somatic events that inactivate MLH1, MSH2, MSH6, or PMS2 are detectable. These events can consist of two point mutations or a point mutation and LOH. The mean age at onset of colorectal cancer and predominant location in the proximal colon resemble LS. Unlike LS, LLS is an acquired disease, although some family clustering by unknown mechanisms has been reported. The profiles of molecular alterations in MMR-deficient tumors may depend on the underlying defect. While hypermutability with repeat sequences as preferential targets is a feature common to all tumors with MMR deficiency, the CIMP phenotype combined with BRAF mutation is a particular characteristic of sporadic MSI colorectal cancers with MLH1 methylation, as discussed earlier. Preliminary evidence suggests that BRAF mutations are absent LLS colorectal carcinomas, thus resembling LS tumors. More frequent mutations in PIK3CA may distinguish LLS colorectal and endometrial tumors from LS tumors.

POLE or POLD1-Associated Hypermutability Hypermutability without MSI-H in tumors can result from proofreading mutations of the replicative DNA polymerases Pol ε and Pol d (the polymerases are depicted in MMR in Fig. 1). Somatic POLE proofreading domain mutations occur in 1%–2% of colorectal cancers and 7%–12% of endometrial cancers (somatic POLD1 mutations are rare). Germline mutations affecting the POLE and POLD1 proofreading domains predispose to polyposis that is often relatively mild (polymerase proofreading-associated polyposis). Interestingly, faulty proofreading owing to somatic or germline mutations in POLE or POLD1 can induce mutations in MMR genes and thereby explain some LLS cases with MSI-H.

Prospective Vision The fact that LS cannot be diagnosed on the basis of clinical or tumor phenotype alone emphasizes two important aspects, first, that the syndrome serves as a general model for common cancers and second, that LS and related phenotypes are breaking down to separate subsets by modern molecular methods. At present, germline mutations in MMR genes are detectable in up to 88% of LS families fulfilling the Amsterdam criteria and showing MSI in tumors. Smaller or atypical families and families not prescreened by MSI or immunohistochemical analyses of tumor tissues show mutation frequencies of 10%–40% depending on the method of ascertainment. Deep sequencing by targeted gene panels and whole exomes or genomes, which will be replacing current single gene-based approaches, is likely to increase the fraction of LS with detectable predisposing mutations and to unravel the genetic basis of colon cancer families unrelated to MMR defects. The complex phenotype of LS is associated with multiple puzzles that remain to be solved by future investigations. Why does the penetrance of MMR gene mutations vary so much even between individuals with identical predisposing mutations? Related to the former question: what is the relative contribution of genetic versus nongenetic factors to LS phenotype? What is the clinical and functional significance of MMR gene variants of the VUS type? Which tumors are unequivocal components of the LS tumor spectrum and by which mechanisms? With a better understanding of the pathogenesis of LS, will cancer eventually become preventable in a large fraction of mutation carriers? Apart from defining the spectrum of genes and mutations predisposing to LS, genome-wide screens (as opposed to previous targeted approaches) can identify low-penetrance alterations that may modify the LS phenotype. Comprehensive genetic, epigenetic, and expressional profiling of tumor tissues may define tumorigenic mechanisms in LS and identify clinically actionable somatic alterations in analogy to efforts on sporadic cancers (TCGA). One in ten carriers of MMR gene mutations may never develop cancer during their lifetime and identification of possible protective factors, inherited or acquired, would be of particular interest. Clinical and molecular studies need to be combined with epidemiological investigations to explore the impact of environmental and lifestyle factors on LS phenotypedan area that is a largely uncharted in LS compared to sporadic cancer. By integrating clinical, molecular, and epidemiological information, a complete picture of LS can be built, which is likely to advance cancer research and facilitate the management of LS individuals.

388

Lynch Syndrome

Acknowledgments This work was supported by Jane and Aatos Erkko Foundation, the Academy of Finland (grant no. 294643), the Finnish Cancer Organizations, and the Sigrid Juselius Foundation.

See also: Colorectal Cancer: Pathology and Genetics. DNA Mismatch Repair: Mechanisms and Cancer Genetics. Genetic Instability.

Further Reading Bak, S.T., Sakellariou, D., Pena-Diaz, J., 2014. The dual nature of mismatch repair as antimutator and mutator: for better or for worse. Frontiers in Genetics 5, 287. Burn, J., Mathers, J.C., Bishop, D.T., 2013. Chemoprevention in Lynch syndrome. Familial Cancer 12, 707–718. Carethers, J.M., 2017. Microsatellite instability pathway and EMAST in colorectal cancer. Current Colorectal Cancer Reports 13, 73–80. Hampel, H., de la Chapelle, A., 2013. How do we approach the goal of identifying everybody with Lynch syndrome? Familial Cancer 12, 313–317. Kloor, M., Huth, C., Voigt, A.Y., Brenner, A., Schirmacher, P., von Knebel Doeberitz, M., Bläker, H., 2012. Prevalence of mismatch repair-deficient crypt foci in Lynch syndrome: a pathological study. Lancet Oncology 13, 598–606. Lynch, H.T., Snyder, C.L., Shaw, T.G., Heinen, C.D., Hitchins, M.P., 2015. Milestones of Lynch syndrome: 1895–2015. Nature Reviews Cancer 15, 181–194. Peltomäki, P., 2014. Epigenetic mechanisms in the pathogenesis of Lynch syndrome. Clinical Genetics 85, 403–412. Sijmons, R.H., Hofstra, R.M., 2016. Review: Clinical aspects of hereditary DNA mismatch repair gene mutations. DNA Repair (Amst) 38, 155–162. Sloane, M.A., Nunez, A.C., Packham, D., Kwok, C.T., Suthers, G., Hesson, L.B., Ward, R.L., 2015. Mosaic epigenetic inheritance as a cause of early-onset colorectal cancer. JAMA Oncology 1, 953–957. Thompson, B.A., Spurdle, A.B., Plazzer, J.-P., Greenblatt, M.S., Akagi, K., Al-Mulla, F., Bapat, B., Bernstein, I., Capellá, G., Den Dunnen, J., du Sart, D., Fabre, A., Farrell, M.P., Farrington, S.M., Frayling, I.M., Frebourg, T., Goldgar, D.E., Heinen, C.D., Holinski-Feder, E., Kohonen-Corish, M., Lagerstedt-Robinson, K., Leung, S.Y., Martins, A., Moller, P., Morak, M., Nystrom, M., Peltomaki, P., Pineda, M., Qi, M., Ramesar, R., Rasmussen, L.J., Royer-Pokora, B., Scott, R.J., Sijmons, R., Tavtigian, S.V., Tops, C.M., Weber, T., Wijnen, J., Woods, M.O., on behalf of InSiGHT, Macrae, F., Genuardi, M., 2014. Application of a five-tiered scheme for standardized classification of 2,360 unique mismatch repair gene variants lodged on the InSiGHT locus-specific database. Nature Genetics 46, 107–115. van Duijnhoven, F.J., Botma, A., Winkels, R., Nagengast, F.M., Kampman, E., 2013. Do lifestyle factors influence colorectal cancer risk in Lynch syndrome? Familial Cancer 12, 285–293. Vasen, H.F., Tomlinson, I., Castells, A., 2015. Clinical management of hereditary colorectal cancer syndromes. Nature Reviews Gastroenterology and Hepatology 12, 88–97.

Relevant Websites www.insight-group.orgdInternational Society for Gastrointestinal Hereditary Tumours. www.LScarisk.orgdProspective Lynch syndrome database. www.omim.org/dOnline Mendelian Inheritance in Man.

Malignant Skin Adnexal Tumors: Pathology and Genetics Katharina Wiedemeyer, University of Heidelberg, Heidelberg, Germany Thomas Brenn, University of Calgary, Calgary, AB, Canada; and Alberta Health Services, Calgary, AB, Canada © 2019 Elsevier Inc. All rights reserved.

Abbreviations EMPD Extramammary Paget’s disease EMPSGC Endocrine mucin-producing sweat gland carcinoma MAC Microcystic adnexal carcinoma MTS Muir-Torre-Syndrome NOS Not otherwise specified PCACC Primary cutaneous adenoid cystic carcinoma PPT Proliferating pilar tumor

Malignant Sweat Gland Tumors Malignant sweat gland tumors comprise the largest group of skin adnexal carcinomas. They are classically subdividing according to their behavior into low- and high-grade carcinomas. Low-grade tumors may show locally destructive growth with risk for local recurrences and rare metastasis to loco-regional lymph nodes while disseminated metastasis and disease-related mortality is a complication of high-grade neoplasms. Morphologically, sweat gland carcinomas frequently mimic adenocarcinomas of visceral primary sites, especially of the breast. They should not be mistaken for cutaneous metastasis but reliable separation is often difficult requiring careful clinical correlation and work-up.

Low-Grade Sweat Gland Carcinomas Primary Cutaneous Cribriform Carcinoma Primary cutaneous cribriform carcinoma is a rare tumor of indolent behavior. No local recurrences or metastases have been described to date and it is unclear whether this tumor should be regarded as a true carcinoma. It presents as slowly growing nodules of few centimeters with a preference for the lower extremities of middle-aged adults. Females are twice as commonly affected as males. Complete removal is curative. Histological features: The tumors are well circumscribed and situated in the dermis without connection with the overlying epidermis (Fig. 1A). They are composed of interconnecting strands of medium-sized cuboidal tumor cells with florid duct

Fig. 1 Primary cutaneous cribriform carcinoma. This well circumscribed but unencapsulated nodular tumor is located within dermis and composed of interconnecting epithelial nests and strands separated by a loose fibrous stroma (A). Extensive ductal differentiation leads to its cribriform appearance. The tumor cells are cuboidal without significant cytological atypia (B).

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65096-1

389

390

Malignant Skin Adnexal Tumors: Pathology and Genetics

formation, resulting in a cribriform architecture (Fig. 1B). The mitotic activity is low and there is no nuclear pleomorphism or tumor necrosis. Focal decapitation secretion may be found. Immunohistochemistry: Expression of cytokeratins AE1/3, MNF116, Cam5.2, and CK7 is present. EMA and CEA staining illustrates luminal differentiation. No myoepithelial cell layer is present. Differential diagnosis: Primary cutaneous cribriform carcinoma may resemble tubular adenoma. These benign tumors show a tubular growth pattern with intervening stroma. They lack the cribriform architecture and show a preserved myoepithelial layer. Separation from the more aggressive adenoid cystic carcinoma is of greater importance. Cutaneous adenoid cystic carcinoma is characterized by a diffusely infiltrative growth, and the cribriform architecture is due to mucinous pseudocysts rather than true duct formation.

Endocrine Mucin-Producing Sweat Gland Carcinoma Endocrine mucin-producing sweat gland carcinoma (EMPSGC) is a rare disease, closely related to solid papillary carcinoma of the breast or endocrine ductal carcinoma in situ. The tumors may arise in association with and may represent a precursor of mucinous carcinoma. EMPSGC presents as a solitary, slowly growing lesion in the periorbital region or cheek, with a strong predilection for the eyelids. The patients are elderly adults (median: 70 years) with a female predilection (F:M ¼ 2:1). Local recurrence following complete excision is a rare event and no metastases or disease related mortality have been described. Histological features: EMPSGC is a well-circumscribed and lobulated dermal based tumor with solid, cystic and papillary differentiation in varying proportions (Fig. 2A). The solid tumor lobules are composed of monomorphous, medium-sized round and oval cells with a pale eosinophilic cytoplasm that may contain mucin (Fig. 2B). The nuclei are placed centrally and show stippled chromatin. Cytological atypia is limited and nuclear pleomorphism is rare. Mitotic activity is noted. The presence of extracellular mucin with the formation of mucinous pseudocysts is a characteristic finding (Fig. 2B). True cystic structures are present in varying amounts and may show decapitation secretion. A subset of tumors shows transition to invasive mucinous carcinoma with small tumor islands suspended in large mucin lakes. Tumor necrosis is usually not a feature. Immunohistochemistry: The tumor cells express neuroendocrine markers (synaptophysin, chromogranin, neuron-specific enolase, CD57) and low molecular cytokeratins (Cam5.2) and CK7. Epithelial membrane antigen (EMA) highlights luminal differentiation. EMPSGC also expresses estrogen and progesterone receptors. The tumor cells are negative for CK20 and S100. No myoepithelial differentiation is found. Differential diagnosis: Mucinous carcinoma is less circumscribed with invasion of deeper tissues. Mucin filled lakes containing scattered epithelial islands dominate the histological pattern. Cutaneous metastases from a visceral primary, particularly of breast origin, are an important differential diagnosis, which requires clinico-pathological correlation. The adenoid-cystic variant of basal cell carcinoma contains more pronounced cytological atypia, shows peripheral palisading and lacks duct differentiation. Merkel cell carcinoma is characterized by a sheet-like or trabecular architecture and clearly infiltrative growth. It lacks mucin and is cytokeratin 20 positive.

Primary Cutaneous Mucinous Carcinoma Mucinous carcinoma arising primarily in the skin is a rare sweat gland tumor that affects the scalp and face, especially the eyelids of elderly adults (median: 76 years) with a predilection for females. The nodular tumors grow slowly and often show a violaceous discoloration. They are of low-grade malignant potential with local recurrence rates of 20%–26% but low risk for metastases, mainly

Fig. 2 Endocrine mucin-producing sweat gland carcinoma. This multilobular dermal based tumor shows a basophilic appearance (A). It is composed of medium sized round to polygonal cells containing eosinophilic cytoplasm and centrally located nuclei with indistinct nucleoli. Intracytoplasmic mucin droplets and extracellular mucin pools are also present (B).

Malignant Skin Adnexal Tumors: Pathology and Genetics

391

Fig. 3 Primary cutaneous mucinous carcinoma. The tumor is composed of nests and strands in an abundant mucinous matrix separated by thin fibrous septa (A). The tumor cells are polygonal and contain eosinophilic cytoplasm and irregular vesicular to hyperchromatic nuclei (B).

to loco-regional lymph nodes. Disseminated spread and disease related mortality is exceptional. Importantly, the diagnosis requires an investigation for a possible underlying mucinous adenocarcinoma of visceral sites, especially the breast, ovary, gastrointestinal and genitourinary tract. Histological features: The tumors show an invasive growth in dermis and subcutaneous fat. They are lobulated showing fine, delicate fibrous septa and are composed of irregularly shaped epithelial islands and strands embedded in pools of mucin (Fig. 3A). The tumor cells are cuboidal containing eosinophilic cytoplasm occasionally with mucin droplets (Fig. 3B). Nuclear pleomorphism is variable. Areas of cribriform as well as solid growth are present in some tumors. Glandular structures with papillary projections and decapitation secretion may also be seen. Immunohistochemistry: The tumor cells are positive for cytokeratins AE1/AE3, Cam 5.2., CK7, and estrogen (ER) and progesterone receptors (PR). Duct differentiation is highlighted by EMA and CEA staining. They are negative for CK20. Differential diagnosis: Primary cutaneous mucinous carcinoma needs to be distinguished from cutaneous metastases from visceral primaries. Tumors of colonic origin can be distinguished by the presence of so-called “dirty” necrosis and expression of CK20. Visceral primaries of other sites, especially breast, require careful clinic-pathological correlation and work-up.

Microcystic Adnexal Carcinoma Microcystic adnexal carcinoma (MAC, syringomatous carcinoma, malignant syringoma, sclerosing sweat duct carcinoma) is a rare carcinoma that usually affects adults in their 5th and 6th decade without gender bias. It typically involves the face, especially the nasolabial and periorbital regions. Other sites are rarely affected but it may be related to infiltrating syringomatous adenoma of the nipple. The tumors present as slowly growing, poorly circumscribed plaques measuring several centimeters. They are locally destructive and show high local recurrence rates of around 30%–40%. Metastasis to regional lymph nodes and systemic disease are exceptionally rare. Wide local excision or Mohs surgery are the treatment of choice to ensure complete removal and prevent against local recurrence. Histological features: The tumor is located in the dermis and is characterized by a diffusely infiltrative architecture with invasion of subcutaneous fat, skeletal muscle, fascia and rarely bone (Fig. 4A). Morphologically it shows both follicular and sweat duct differentiation. The tumor consists of solid cords and strands of small to medium sized basaloid cells in a dense fibrotic stroma (Fig. 4B). Cytological atypia is minimal and mitoses are rare. Superficially, keratocysts and dystrophic calcifications are present. Ductal differentiation is most evident in the deeper parts of the tumor (Fig. 4C). Perineural infiltration is commonly present. Immunohistochemistry: The tumor cells express pan-cytokeratins. EMA and CEA highlight duct differentiation. Differential diagnosis: MAC may resemble syringoma, desmoplastic trichoepithelioma, infiltrative basal cell carcinoma and desmoplastic squamous cell carcinoma. Separation from syringoma and desmoplastic trichoepithelioma is important as these are benign skin adnexal tumors. Both tumors are confined to the superficial and mid dermis and lack the diffusely infiltrative and deep growth of MAC. Reliable separation is often challenging and may be impossible on superficial biopsies. In these cases, a deeper repeat biopsy or complete excision is recommended for a definitive diagnosis. Sclerosing basal cell carcinoma and squamous cell carcinoma lack ductal differentiation and show more pronounced cytological atypia.

Primary Cutaneous Adenoid Cystic Carcinoma Primary cutaneous adenoid cystic carcinoma (PCACC) is rare. It affects middle-aged to elderly adults (median: 62 years) without sex predilection. It presents as slowly growing nodules and plaques, often measuring multiple centimeters. There is a predilection for the head and neck area, especially the scalp, but the anatomic distribution is wide, also including the genital area, in particular the

392

Malignant Skin Adnexal Tumors: Pathology and Genetics

Fig. 4 Microcystic adnexal carcinoma. The tumor is characterized by a diffusely infiltrative growth within dermis (A). It consists of cords and strands showing hair follicular differentiation in a sclerotic stroma. Small keratocysts are also present (B). In the deeper aspects, ductal differentiation is evident. The tumor invades skeletal muscle (C).

Fig. 5 Primary cutaneous adenoid cystic carcinoma. This basophilic tumor is poorly circumscribed and shows an infiltrative growth within dermis with invasion of subcutaneous adipose tissue (A). The tumor cells are medium sized without marked nuclear pleomorphism. Mucin filled pseudocyst give rise to the typical cribriform architecture (B). Ductal differentiation is a focal finding (C).

vulva. In contrast to its visceral counterparts the behavior of primary cutaneous tumors is less aggressive, characterized by locally destructive growth and risk for local recurrence. The risk for metastatic spread and mortality is however low. Histological features: The tumor is poorly demarcated with an infiltrative growth in dermis and subcutaneous fat (Fig. 5A). It is composed of variably sized and shaped nests, tumor lobules and cords and strands of basaloid epithelioid cells with little cytoplasm and hyperchromatic nuclei. Cytological atypia is mild to moderate but mitotic activity is readily identified. The individual tumor islands are surrounded by stromal mucin leading to the formation of mucinous pseudocysts and the characteristic cribriform appearance (Fig. 5B). Ductal differentiation is focally present and perineural infiltration is usually seen (Fig. 5C). Immunohistochemistry and genetics: The tumor expresses cytokeratins AE1/3 and Cam5.2. Myoepithelial differentiation is present and highlighted by SMA, SOX10 and S100 staining. Most tumors show positivity for CD117. Ductal structures stain with CEA and EMA. The t(6;9)(q22-23;p23-24) translocation leading to the MYB-NFIB fusion gene and resulting in MYB overexpression characteristic of visceral adenoid cystic carcinoma has also been found in primary cutaneous tumors. Differential diagnosis: Primary cutaneous tumors need to be distinguished from cutaneous extension or metastasis from visceral adenoid cystic carcinoma. This requires careful clinico-pathological correlation and work-up as the histological, immunohistochemical and genetic features are identical. Adenoid cystic basal cell carcinoma can be differentiated from PCACC by the presence of palisading of tumor cells, the stromal cleft artifact and lack of duct differentiation.

Squamoid Eccrine Ductal Carcinoma Squamoid eccrine ductal carcinoma is a rare and likely under-recognized entity, closely related if not identical to adenosquamous carcinoma of the skin. The tumor mainly affects the head and neck area of elderly males. It presents as solitary and often ulcerated nodules and plaques. The tumors are characterized by local recurrence rates of 25% and occasional metastasis to regional lymph nodes. Systemic disease and mortality are rarely reported events.

Malignant Skin Adnexal Tumors: Pathology and Genetics

393

Fig. 6 Squamoid eccrine ductal carcinoma. The tumor is situated in the dermis, is poorly circumscribed and shows a diffusely infiltrative growth invading subcutaneous fat (A). In the superficial aspects of the tumor squamous cell differentiation is typical and may be morphologically identical to squamous cell carcinoma. A connection with the overlying epidermis is often present and there may be surface ulceration (B). Eccrine ductal differentiation is found in the deeper reaches of the tumor consisting of irregular cords and strands of pleomorphic cuboidal cells in a desmoplastic stroma. Mitotic activity is high (C).

Histological features: The tumors are situated in the dermis with a diffusely infiltrative growth, frequently invading subcutaneous fat, skeletal muscle and fascia (Fig. 6A). They are characterized by a dual differentiation towards squamous cell carcinoma in the superficial aspects and eccrine ductal carcinoma in the deeper reaches. A connection with the overlying epidermis is often present and there may be surface ulceration. The tumors are indistinguishable from squamous cell carcinoma superficially (Fig. 6B). In the deeper areas, they are composed of irregular cords and strands of pleomorphic cuboidal cells in a desmoplastic stroma (Fig. 6C). Mitotic activity is brisk and ductal differentiation is present. Tumor necrosis, perineural infiltration and lymphovascular invasion may be present. Immunohistochemistry: Duct differentiation is highlighted by immunohistochemistry for CEA and EMA. Differential diagnosis: Squamoid eccrine ductal carcinoma may be mistaken for squamous cell carcinoma on superficial biopsies. Eccrine porocarcinoma may show focal squamous differentiation but lacks the zonation. Microcystic adnexal carcinoma also shows duct and squamoid differentiation but lacks the pronounced cytological atypia.

High-Grade Sweat Gland Carcinomas Porocarcinoma Porocarcinoma (malignant eccrine poroma) affects elderly people in their 7th and 8th decade without gender bias. It is typically found on the lower limbs but can also be seen on the trunk, head and neck area and upper extremities. Risk factors include trauma, burning, radiotherapy and potentially also immunosuppression. Porocarcinoma presents as long-standing solitary verrucous tumors, polypoid nodules or plaques measuring several centimeters. Ulceration is common. The behavior is similar to squamous cell carcinoma if adjusted for tumor thickness. The risk for local recurrences is 17% with potential for lymph nodes metastasis. Although rare, the presence of distant metastatic spread is an adverse prognostic factor associated with high mortality rates. Histopathological features: Porocarcinoma is a poorly delineated dermal based tumor showing multifocal epidermal connection and epidermal ulceration (Fig. 7A). Its borders may be pushing or diffusely infiltrative and invasion of subcutis and deeper structures may be present. A preexisting benign poroma is noted in 11% of cases. The tumors are composed of irregularly shaped and interconnecting strands and islands of polygonal epithelioid cells with varying degrees of cytological atypia and nuclear pleomorphism (Fig. 7B). The identification of duct differentiation is a prerequisite for the diagnosis (Fig. 7C). Tumor necrosis, perineural infiltration and lymphovascular invasion may also be present. Other rare features include squamous or clear cell change and sarcomatoid differentiation. The tumors can be pigmented by melanocytic colonization. Immunohistochemistry: The tumor cells express pancytokeratins. P16 is overexpressed in most porocarcinomas as is p53. Duct differentiation is highlighted by EMA and CEA staining. Differential diagnoses: Squamous cell carcinoma can be separated by the absence of duct differentiation. Squamoid eccrine ductal carcinoma shows a zonation with areas indistinguishable from squamous carcinoma and eccrine ductal carcinoma. When adjusted for tumor thickness the behavior of porocarcinoma, squamous cell carcinoma and squamoid eccrine ductal carcinoma is similar.

394

Malignant Skin Adnexal Tumors: Pathology and Genetics

Fig. 7 Porocarcinoma. This large dermal based tumor shows ulceration and connection with the overlying epidermis (A). It consists of solid nests and islands of poroid polygonal cells with varying degrees of cytologic atypia (B). Ductal differentiation is invariably present and its identification is necessary for the diagnosis (C).

Hidradenocarcinoma Hidradenocarcinoma (clear cell hidradenocarcinoma, malignant hidradenoma) is a rare tumor with a wide anatomic distribution and a predilection for adults. The face and the extremities appear to be most frequently affected. The tumors have potential for aggressive behavior with high local recurrence rates (50%–75%) and a high potential to metastasize to lymph nodes, lung and bone. Wide local excision is the treatment of choice. No clear guidelines exist regarding sentinel lymph node biopsy. The possibility should be considered and discussed on an individual basis. Histopathological features: Hidradenocarcinoma is located in the dermis as a multinodular tumor with an infiltrative growth and invasion of deeper structures (Fig. 8A). The tumor lacks an epidermal connection and is composed of polygonal epithelioid cells with eosinophilic cytoplasm and variable nuclear pleomorphism (Fig. 8B). Clear cell change and squamoid features may also be present. Duct differentiation or cystic elements are invariably seen and their recognition is necessary for the diagnosis (Fig. 8C). In addition, mitotic activity, tumor necrosis, perineural infiltration and lymphovascular invasion may be seen. A preexisting benign hidradenoma is observed in a small subset of tumors. Immunohistochemistry and genetics: The tumor cells express cytokeratin and frequently Her-2/neu. Ductal differentiation is demonstrated by EMA and CEA stains. Immunohistochemical overexpression of P53 is seen in the majority of tumors despite low frequency of mutations in the TP53 gene. The translocation t (11;19) can be detected in some hidradenocarcinomas analogous to hidradenoma. Differential diagnoses: Morphologically low-grade tumors are a particular diagnostic challenge and need to be separated from hidradenoma. Tumors with infiltrative margins and cytological atypia need to be regarded with care and require complete excision. In this context, it is also important to be aware that morphologically and biologically benign hidradenomas may rarely show lymphovascular invasion and even lymph node deposits. Morphologically high-grade tumors require separation from cutaneous metastases from visceral primaries by careful clinical correlation, work-up and imaging techniques.

Apocrine Carcinoma Apocrine carcinoma is a rare adenocarcinoma mainly affecting the axilla of adults with a mean age of 60 years and no significant gender predilection. It presents as solitary, slowly growing erythematous or violaceous nodules. The disease course can be aggressive

Fig. 8 Hidradenocarcinoma. The tumor shows a solid growth and arises from a preexisting hidradenoma seen on the right-hand side (A). Marked cytological atypia, nuclear pleomorphism and mitotic figures are present (B). In areas duct differentiation is appreciated (C).

Malignant Skin Adnexal Tumors: Pathology and Genetics

395

Fig. 9 Apocrine carcinoma. This dermal-based tumor shows a mixture of cystic and papillary elements (A). It is composed of large epithelioid cells with abundant eosinophilic cytoplasm containing vesicular nuclei and prominent eosinophilic nucleoli. Mitotic activity is identified (B).

with metastatic disease to lymph nodes and distant organs (lungs, liver, bone) and associated mortality. Histological grading appears to be a good predictor of outcome. The treatment consists of wide local excision. Histopathological features: Apocrine carcinoma is lobulated and located in the dermis, frequently also invading subcutaneous fat (Fig. 9A). It is poorly circumscribed with infiltrative borders. The tumor cells are polygonal containing abundant eosinophilic cytoplasm and vesicular nuclei with prominent nucleoli. Cytological atypia and mitotic activity are variable and tumor necrosis may be seen (Fig. 9B). The growth pattern may show solid, tubular and cystic structures in a dense fibrotic stroma. Papillary structures are also frequently seen and decapitation secretion is consistently present. Adjacent apocrine glands are usually found close to the tumor. The tumors can be graded according to the modified Bloom Richardson grading system, which correlates with outcome. Immunohistochemistry and genetics: The tumor cells express CAM5.2, AE1/AE3, EMA, CEA, GCDFP-15, mammaglobin, ER and PR. They are negative for HER2/neu and adipophilin. Differential diagnosis: Apocrine carcinoma needs to be differentiated from metastatic adenocarcinoma, especially of breast origin. This differential diagnosis is challenging due to the morphologic overlap and presentation in the axilla. It requires careful clinical correlation. In addition, an immunohistochemical panel of ER and PR, HER2/neu, adipophilin, CK5/6, mammaglobin may be helpful: cutaneous apocrine carcinoma is likely to be adipophilin and Her2/neu negative, ER and PR positive and also shows a strong positive staining with CK5/6 and mammaglobin.

Digital Papillary Adenocarcinoma Digital papillary adenocarcinoma is a rare but distinctive sweat gland carcinoma with a narrow anatomic distribution and a marked male predominance. The tumor shows a strong predilection for the distal extremities, particularly the digits. It presents as small erythematous or brown nodules, ranging from few millimeters to multiple centimeters. A wide age range is affected, including children and adolescents. The tumors have potential for aggressive behavior independent of the histological degree of differentiation. The local recurrence rates are high (20%–40%) and metastatic disease is seen in approximately 15%, mainly to lymph nodes and lung. Wide local excision or amputation is the recommended treatment. It may positively influence the disease course with reduced local recurrence and distant metastatic rates. Histological features: Digital papillary adenocarcinoma shows a wide histological spectrum ranging from bland and innocuous appearing tumors to poorly differentiated examples with high-grade morphology. The majority of tumors are well-circumscribed and located in the deep dermis and superficial subcutaneous adipose tissue (Fig. 10A). They are solid and cystic with additional papillary, micropapillary and tubular growth patterns (Fig. 10B and C). The tumor cells are cuboidal basaloid cells with varying degrees of cellular atypia and nuclear pleomorphism, ranging from mild to severe. Mitotic figures are usually present. A second myoepithelial cell layer is present. Tumor necrosis and a more infiltrative growth are rare additional findings (Fig. 10C). Immunohistochemistry and genetics: Tumor cells express cytokeratins. The luminal differentiation can be highlighted by EMA and CEA staining. S100, p63, podoplanin, SMA and calponin staining demonstrates a second myoepithelial cell layer. Differential diagnosis: In view of the bland histological features and the preserved myoepithelial cell layer digital papillary adenocarcinoma may easily be mistaken for benign skin adnexal tumors, especially apocrine cystadenoma, hidradenoma and tubular adenoma. Apocrine cystadenoma is exceptionally rare on acral sites and this diagnosis should be made only after careful consideration. In reality, the vast majority of cystic and papillary tumors thought to be apocrine cystadenomas are in fact digital papillary adenocarcinomas. Hidradenoma shares many of the histological features with digital papillary adenocarcinoma. Papillary differentiation is usually not a prominent finding and the presence of any cytological atypia should be a warning sign. Tubular adenoma is

396

Malignant Skin Adnexal Tumors: Pathology and Genetics

Fig. 10 Digital papillary adenocarcinoma. The tumor is well-circumscribed and situated in the dermis with a solid and cystic architecture (A). In areas duct formation and a tubular growth are noted (B). The cystic structures show micropapillae and a second myoepithelial cell layer can be seen (C).

characterized by a striking tubular growth with intervening stroma. A solid growth component is not observed. Metastases from papillary adenocarcinomas of visceral origin can be excluded by the demonstration of the myoepithelial differentiation.

Malignant Neoplasms Arising from Preexisting Spiradenoma, Cylindroma or Spiradenocylindroma These rare tumors are defined by the presence of a malignant component arising in the background of a preexisting spiradenoma (spiradenocarcinoma), cylindroma (cylindrocarcinoma) of a hybrid spiradenoma-cylindroma. They present as solitary nodules with a preference for the head and neck and the extremities of elderly adults without gender bias. A long-standing history with recent enlargement may be given. The tumors may develop in association with the autosomal dominant Brooke Spiegler syndrome. The syndrome is characterized by the development of multiple adnexal neoplasms, especially cylindroma, spiradenoma, spiradenocylindroma and trichoepithelioma. The Brooke Spiegler syndrome is linked to mutations resulting in the inactivation of the tumor suppressor gene CYLD. Morphologically the tumors arising from preexisting spiradenoma and cylindroma are divided into low- and high-grade, which correlates with outcome. Low-grade carcinomas have a 20% risk of local recurrence but distant metastasis and disease-related mortality is seen almost exclusively in high-grade carcinomas. Histological features: These multinodular tumors are located in the deep dermis and subcutis and often show pushing rather than diffusely infiltrative borders. Recognition of a preexisting spiradenoma, cylindroma or hybrid tumor is essential for the diagnosis. Morphologically low-grade tumors show a similar architecture to spiradenoma on low power examination (Fig. 11A). They are composed of a monotonous population of uniform basaloid cells with mild to moderate cytological atypia and increased mitotic activity (Fig. 11B). A hallmark feature is the lack of the dual cell population, typical of spiradenoma and cylindroma. Additional findings include clear cell change and the formation of squamous eddies. Tumor necrosis and ulceration may also be observed. Morphologically high-grade tumors show a wide range of histological features. They are characterized by high-grade cytological atypia and nuclear pleomorphism and atypical mitoses, resembling poorly differentiated carcinoma or adenocarcinoma, NOS (Fig. 11C). Necrosis is a frequent finding.

Fig. 11 Spiradenocarcinoma. The malignant nodular tumor (on the left-hand side) arises in association with a preexisting benign spiradenoma (on the right-hand side) (A). Morphologically low-grade tumors show a monotonous proliferation of uniform basaloid cells with mild to moderate degrees of cytological atypia and increased mitotic activity. The dual cell population characteristic of benign spiradenoma is lost (B). Morphologically highgrade spiradenocarcinoma is characterized by marked cytological atypia and resembles poorly differentiated adenocarcinoma (C).

Malignant Skin Adnexal Tumors: Pathology and Genetics

397

Immunohistochemistry and genetics: The tumor cells express cytokeratins, duct differentiation is highlighted by EMA or CEA staining. Immunohistochemical overexpression of MYB is used as surrogate marker for the t(6;9) translocation seen in benign spiradenomas and cylindromas. MYB expression is lost, at least in the low-grade malignant tumors and serves as a helpful diagnostic marker to distinguish these tumors from their benign counterparts. Adnexal neoplasms arising in patients with Brooke Spiegler syndrome are likely to show mutations in the CYLD gene mapped on chromosome locus 16q12–13. Differential diagnosis: The separation of morphologically low-grade spiradenocarcinoma from benign spiradenoma is particularly challenging. It requires careful examination with recognition of loss of the dual cell population, the presence of cytological atypia and mitotic activity. Immunohistochemical loss of MYB expression is helpful in this setting. High-grade tumors resemble metastases from visceral primaries. The recognition of a benign precursor is essential for the correct diagnosis.

Extramammary Paget’s Disease Extramammary Paget’s disease (EMPD) is a rare adenocarcinoma. The majority of tumors develop primary to the skin which is subject of this article. Less commonly, EMPD presents secondary to the skin as cutaneous spread of an underlying visceral carcinoma, especially of the rectum, the urogenital tract, endocervix and stomach. Primary cutaneous EMPD develops in areas rich in apocrine sweat glands, especially in the vulva and perianally. Other sites include the penis, scrotum, axillae, umbilicus, and very rarely the face, scalp, trunk and extremities. It shows a predilection for females in their sixth to eighth decade of life. EMPD presents as erythematous, pigmented or erosive plaques accompanied by pruritus or pain. Early lesions can resemble eczema. In late stages nodules and ulceration are noted. Primary cutaneous EMPD is usually diagnosed in the intraepidermal state (carcinoma in-situ). It shows a high risk for local recurrences but the overall prognosis is excellent. In contrast, invasion is seen in approximately 10% of EMPD and is associated with potential for aggressive behavior with regional and distant metastasis resulting in disease related mortality. Treatment for localized disease includes radical excision and radiotherapy. In-situ disease may also respond to photodynamic therapy and local immunomodulators (imiquimod). Histological features: Primary cutaneous EMPD presents as an intraepidermal carcinoma in most cases. It is composed of mediumsized to large round to epithelioid cells arranged singly and in clusters (Fig. 12A). The tumor cells are scattered throughout the epidermis and may also involve hair follicular epithelium and sweat ducts. They contain abundant cytoplasm and vesicular nuclei showing variable pleomorphism. Intracytoplasmic mucin may be seen. Signet ring cell and glandular differentiation have been described. The involved epidermis may be hyperplastic and ulceration may be a feature. An invasive component is present in approximately 10% of cases. The invasive component shows a diffusely infiltrative growth involving dermis and deeper structures. It may be solid or nested and can show single cell infiltration. Signet ring or acinar differentiation may also be evident (Fig. 12B). Immunohistochemistry: The tumor cells are PAS-positive and express low molecular weight cytokeratins (CAM5.2) and cytokeratin 7. Tumor cells are also positive for EMA, CEA, GCDFP-15 and mucicarmine 1. Androgen receptors and Her-2/neu are variably positive. Differential Diagnoses: Most importantly, an epidermotropic metastasis of an internal adenocarcinoma needs to be excluded. Metastasis from colonic carcinoma is usually positive for both CK20 and CDX-2 in contrast to EMPD. Melanoma in situ and superficial spreading melanoma may show morphological similarities but can be distinguished by staining for melanocytic markers (S100, MelanA, HMB45). Clear cell squamous cell carcinoma is negative for GCDFP-15 and CEA but positive for p63.

Fig. 12 Extramammary Paget’s disease. The epidermis is diffusely infiltrated by medium sized tumor cells with abundant eosinophilic to clear cytoplasm and vesicular nuclei containing prominent eosinophilic nucleoli. The tumor cells are arranged singly and in small clusters, there is surrounding epidermal acanthosis (A). The invasive component consists of irregular nests of tumor cells. The invasive tumor cells are morphologically identical to the intraepidermal cells. Mitoses are readily spotted (B).

398

Malignant Skin Adnexal Tumors: Pathology and Genetics

Eccrine Ductal Carcinoma NOS Eccrine ductal carcinoma (eccrine adenocarcinoma) NOS is a diagnosis of exclusion. These rare tumors do not fit with any of the entities described above, and there are no systematic studies in the literature. Histologically they may be identical to metastases from visceral primaries, particularly of breast origin. Their diagnosis is only possible after a visceral primary has been excluded by clinical work-up and imaging studies. The tumors present as firm erythematous nodules involving a wide range of anatomical locations, especially of middle aged and elderly adults. The disease course is aggressive with high recurrence, metastasis and mortality rates. Histological features: Eccrine ductal carcinoma is located in the deep dermis and may invade subcutaneous fat and deeper structures (Fig. 13A). The tumors are poorly circumscribed and consist of nests and strands of pleomorphic cuboidal cells with prominent ductal differentiation in a sclerotic stroma (Fig. 13B). Mitotic activity is brisk and tumor necrosis may be present. Perineural infiltration and lymphovascular invasion are common features. Immunohistochemistry: The tumor cells express cytokeratins and CEA. Hormone receptors, GCDFP-15 and S100 are variably positive. Differential diagnosis: Eccrine ductal carcinoma is morphologically identical to cutaneous metastasis from adenocarcinomas of visceral sites. The diagnosis of eccrine ductal carcinoma is one of exclusion requiring a careful clinical work-up and imaging of the patient.

Malignant Tumors with Follicular Differentiation Pilomatrix Carcinoma Pilomatrix carcinoma (malignant pilomatricoma, matrical carcinoma) is a rare neoplasm of low-grade malignancy and a predilection for male adults. The tumor presents as nodules of a few to multiple centimeters with a wide anatomical distribution. The head and neck area and the back are most commonly involved. Local recurrences are common but loco-regional and distant metastases are rare. Histological features: Pilomatrix carcinoma is a multinodular dermal based tumor and may show invasion of subcutis, skeletal muscle and fascial tissues (Fig. 14A). The irregularly shaped and sized tumor lobules are composed of basophilic tumor cells with varying degrees of cytological atypia and a brisk and atypical mitotic activity. In addition, the tumor shows the distinctive hair matrix keratinization with ghost or shadow cell differentiation (Fig. 14B). Confluent necrosis, perineural infiltration and lymphovascular invasion may be present. Occasionally, colonization by melanocytes is seen. Immunohistochemistry and genetics: Pilomatrix carcinoma is associated with mutations in beta-catenin that have also been identified in pilomatricoma, its benign counterpart. By immunohistochemistry, the tumor cells show nuclear expression of beta-catenin and cyclin D. Differential diagnosis: Differentiation from the benign pilomatricoma is challenging. An infiltrative growth pattern, the presence of cytological atypia and tumor necrosis are the most reliable differentiating features. Basal cell carcinoma with matrical differentiation shows a peripheral palisade and a stromal cleft artifact.

Fig. 13 Eccrine ductal carcinoma, NOS. The dermal based tumor is composed of irregularly shaped basaloid nests and lobules (A). There is marked nuclear pleomorphism and prominent duct differentiation. A desmoplastic stroma surrounds the tumor nests (B).

Malignant Skin Adnexal Tumors: Pathology and Genetics

399

Fig. 14 Pilomatrix carcinoma. This large tumor is composed of irregular lobules containing sheets of basophilic cells and keratin (A). The tumor cells show a monotonous growth pattern and keratinization with ghost- or shadow-cell differentiation (B).

Trichilemmal Carcinoma Trichilemmal carcinoma is a rare carcinoma of low-grade malignancy. It affects sun-exposed skin, particularly of the face, scalp, neck and hands of the elderly with a mean age of 71 years. The tumors are solitary erythematous nodules and plaques that may ulcerate. After complete excision trichilemmal carcinoma recurs very rarely. Regional lymph node metastases and systemic disease are rare events, but have been described, especially in immunocompromised patients. Histopathological features: Trichilemmal carcinoma is a dermal based tumor with pushing borders composed of irregular tumor lobules, nests and strands (Fig. 15A). A diffusely infiltrative architecture is rarely observed. The tumor may extend into deep dermis and invades subcutaneous fat. A multifocal connection with the overlying epidermis or with hair follicles is typically present and there may be epidermal ulceration. The neoplastic cells show trichilemmal characteristics with clear to palely eosinophilic cytoplasm. Trichilemmal keratinization is invariably present either in the form of intracytoplasmic hyaline droplets, in form of small horn cysts or as large confluent areas (Fig. 15B). Cytological atypia and nuclear pleomorphism are variable and range from mild to severe. Nuclear palisading may be seen in the periphery of tumor lobules with surrounding hyaline basement membrane material. Mitoses are often numerous including atypical forms. Tumor necrosis may be an additional finding. Immunohistochemistry: The tumor cells express cytokeratins and the cytoplasm is PAS-diastase positive. With rare exception, there is no expression of EMA and CEA. Differential diagnoses: Differentiation from clear cell squamous cell carcinoma is difficult. The trichilemmal keratinization and hyaline basement membrane material as well as EMA negativity are diagnostic clues. Basal cell carcinoma with outer root sheath differentiation shows palisading of peripheral cells and retraction artifacts between tumor lobules and tumor stroma. Porocarcinoma and hidradenocarcinoma are separated by the presence of ductal differentiation.

Fig. 15 Trichilemmal carcinoma. The tumor is connected with the surface epithelium and shows a multinodular growth pattern with pushing borders (A). It consists of large epithelial cells with abundant palely eosinophilic cytoplasm and trichilemmal keratinization. Nuclear pleomorphism is marked (B).

400

Malignant Skin Adnexal Tumors: Pathology and Genetics

Proliferating Pilar Tumor Proliferating pilar tumor (PPT) is a rare neoplasm that shows a predilection for elderly women. The tumors are usually located on the scalp. Other locations include eyelids, face, neck, vulva, trunk and extremities. PTT presents as slowly growing subcutaneous nodules measuring several centimeters. The majority of proliferating pilar tumors is benign. Local destructive growth and metastases to regional lymph nodes and distant organs is rare and seen in tumors with adverse histological features as listed below. Histopathological features: The tumor has a multinodular growth pattern in deep dermis and may extend into the subcutaneous fat, skeletal muscle and fascia (Fig. 16A). It shows solid and cystic differentiation composed of large epithelioid cells with palely eosinophilic cytoplasm, vesicular nuclei and prominent trichilemmal keratinization (Fig. 16B). The periphery of tumor lobules is composed of smaller, more basophilic cells that may show palisading and an outer hyaline basement membrane. The presence of marked cytological atypia, atypical mitotic activity, an infiltrative growth pattern, tumor necrosis, perineural infiltration and lymphovascular invasion confers risk for more aggressive behavior. Immunohistochemistry: Immunohistochemistry plays no role in the diagnosis of these tumors. Differential diagnosis: Squamous cell carcinoma and trichilemmal carcinoma are characterized by an epidermal connection. They both lack the classical solid and cystic multilobular growth pattern characteristic of proliferating pilar tumor.

Sebaceous Carcinoma Sebaceous carcinoma is a rare neoplasm that is classically divided into periocular (sebaceous carcinoma of the eyelids) and extraocular sebaceous carcinoma. Periocular sebaceous carcinoma affects the elderly and appears to be more common in Asia. It has potential for aggressive behavior with local recurrence rates of 30%–40% and metastases to regional lymph nodes and distant organs in up to 25%. Metastatic disease is associated with high mortality and poor 5-year survival. Extraocular sebaceous carcinoma is rare. It is mainly found in the head and neck area of elderly adults and presents with erythematous and often ulcerated nodules and plaques. The presentation outside the head and neck area, young age at presentation and multiple sebaceous neoplasms raise suspicion for the Muir-Torre-Syndrome (MTS), a variant of hereditary nonpolyposis colorectal cancer syndrome (Lynch syndrome). The behavior of extraocular sebaceous carcinoma is thought to be similar to that of periocular tumors. Treatment of choice is complete excision and long-term follow-up. Histological features: Sebaceous carcinomas are circumscribed and multinodular or diffusely infiltrative dermal based tumors, often invading underlying subcutaneous tissues (Fig. 17A). They may show multifocal epidermal connection and ulceration (Fig. 17B). The tumor cells are epithelioid with basophilic cytoplasm and vesicular nuclei. Admixed in varying proportions are pale staining tumor cells with a vacuolated cytoplasm and indented nuclei (Fig. 17C). Poorly differentiated tumors show a predominance of the basaloid cells and sebaceous differentiation is difficult to appreciate on histology (Fig. 17D). There may be marked cytological atypia and nuclear pleomorphism. Mitoses are numerous and tumor necrosis is frequently seen. Lymphovascular or perineural invasion may be present. An intraepidermal growth with florid pagetoid spread is a feature seen particularly in the periocular tumors. Immunohistochemistry: Tumor cells are positive for cytokeratin 7, p63 and p40. Sebaceous differentiation is highlighted by staining for EMA and adipophilin. The immunohistochemical demonstration of sebaceous differentiation may be difficult in poorly differentiated tumors. Differential diagnosis: Sebaceous carcinoma needs to be distinguished from basal cell carcinoma, basaloid squamous cell carcinoma and Merkel cell carcinoma. The demonstration of sebaceous differentiation on morphology and immunohistochemistry is the clue to the correct diagnosis. In contrast to basal cell carcinoma sebaceous carcinoma does not show palisading of peripheral

Fig. 16 Proliferating pilar tumor. The lobulated tumor is composed of solid epithelial elements and abundant trichilemmal keratinization (A). There is cytological atypia and numerous mitotic figures are present (B).

Malignant Skin Adnexal Tumors: Pathology and Genetics

401

Fig. 17 Sebaceous carcinoma. This basaloid tumor is characterized by a multinodular growth within dermis (A). Multifocal epidermal connection and ulceration are present (B). The tumor is composed of basaloid cells with cellular atypia and admixed pale staining cells with a vacuolated cytoplasm (C). Sebaceous differentiation is difficult to appreciate in poorly differentiated examples (D).

cells, no retraction artifacts and is usually BerEp4 negative. Merkel cell carcinoma expresses CK20 which is consistently negative in sebaceous carcinoma.

Prospective Vision Insight into the clinical behavior and underlying molecular events of skin adnexal carcinomas is still limited. Larger and more comprehensive studies with genetic analysis are necessary to better define these tumors and develop novel targeted therapeutic options, especially for disseminated disease.

See also: Squamous Cell and Basal Cell Carcinoma of the Skin: Diagnosis and Treatment.

Further Reading Brenn, T., 2015. Malignant sweat gland tumors: An update. Advances in Anatomic Pathology 22, 242–253. Calonje, E., Brenn, T., Lazar, A., McKee, P.H., 2012. McKee’s pathology of the skin, fourth ed. Vol. 2. Elsevier, Edinburgh. Flux, K., 2017. Sebaceous neoplasms. Surgical Pathology Clinics 10, 367–382. Flux, K., Brenn, T., 2017. Cutaneous sweat gland carcinomas with basaloid differentiation: An update with emphasis on differential diagnoses. Clinics in Laboratory Medicine 37, 587–601. Kazakov, D., Michal, M., Kacerovska, D., McKee, P.H., 2012. Cutaneous adnexal tumors. Wolters Kluwer Health, Philadelphia. Patterson, J., 2015. Weedon’s skin pathology, 4th ed. Elsevier, Edinburgh. Van der Horst, M.P.J., Brenn, T., 2017. Update on malignant sweat gland tumors. Surgical Pathology Clinics 10, 383–397.

Malignant Tumors of the Eye, Conjunctiva, and Orbit: Diagnosis and Therapy Laurence Desjardins, Nathalie Cassoux, and Alexandre Matet, Curie Institute, Paris, France © 2019 Elsevier Inc. All rights reserved.

Primary Intraocular Malignant Tumors Uveal Melanoma Diagnosis Uveal melanoma is the most frequent primary intraocular tumors of adults. The age at diagnosis is around 60 years. For most patients, conservative management by proton beam therapy, plaque brachytherapy or transcleral resection is possible with equal survival than enucleation. Metastasis occurs in 50% of the patients mostly in the liver and survival with liver metastasis is short. For many years, there has been little progress in the management of metastatic disease but there is now increasing research and interest in many centers. Blue and green eyes are more frequently involved and uveal melanoma is very rare in melanodermic patients. Usual visual symptoms include scotomas, phosphenes, loss of vision if the tumor is involving the macula or if the central retina is detached. Large peripheral tumors can cause visual field defects. Diagnosis relies mainly on fundus examination. Location of the tumor is very variable from ciliary body to posterior pole. Small ciliary body masses are often asymptomatic at first. They can invade the iris angle but are only visible with dilated pupil. Most tumor appears as a choroidal or ciliochoroidal mass with variable amount of pigmentation (sometimes very dark, sometimes amelanotic). The typical shape is the mushroom shape due to rupture of Bruch’s membrane. Orange pigment can be present on the surface representing lipofuscin deposits. Serous retinal detachment is often associated sometimes visible only on OCT, sometimes clinically visible and associated to inferior exudative retinal detachment. Useful ancillary test are ultrasonography of the eye and fluorescein angiography. MRI can be useful when there is invasion of optic disk or suspected extra scleral extension. Ultrasonography usually shows typical attenuation of ultrasounds due to the very dense tumor tissue. The tumors appear hypoechogenic at the base with frequent choroidal excavation. On fluorescein angiography, it is possible to see a typical double circulation in achromic tumors in the early phase. Later tumors all show hyperfluorescence with surface pin points. In the absence of appropriate treatment, uveal melanoma can invade the sclera and grow in the orbit. Extrascleral extension can be detected by ultrasonography or MRI. MRI is also helpful to verify the absence of optic nerve invasion for tumors invading the optic disk. If extrascleral extension is detected it can be treated by brachytherapy or proton beam but large extrascleral extension requires enucleation followed by external beam radiation. Differential diagnosis includes choroidal naevi and other benign pigmented lesions such as melanocytomas and congenital hypertrophy of the retinal pigment epithelium, choroidal metastasis, choroidal hematomas, choroidal hemangiomas and choroidal osteomas. The most difficult and frequent situation is to differentiate between suspicious choroidal naevi and melanoma. The risk factors for a nevus to grow into a melanoma have been extensively described by JA Shields and CL Shields. They are diameter of > 7 mm, thickness of > 2 mm, orange pigment, subretinal fluid, visual symptoms, absence of drusen and location close to the optic disk. In studies performed with fluorescein angiography it has been shown that the presence of pin points is a significant risk factor. Risk for growth increases when the number of risk factors increases. The decision to treat depends on the number of risk factors, the location of the tumor and age of the patients. Decision should always be made with informed consent of the patient. Unique metastasis can sometimes be difficult to differentiate from achromic melanoma. Choroidal metastasis can be seen in all cancer but more frequently in breast cancer and lung cancer. In lung cancer, metastasis is the first symptom of the disease in many cases. It is recommended to perform thoracoabdominal CT in all patients with achromic fundus mass.

Prognosis Clinical risk factors for metastatic disease are well known: largest tumor diameter and anterior tumor location are important. Histological risk factors are the presence of epithelioid cells, mitotic count, vascular loops and presence of intravascular tumor cells. Largest tumor diameter should be correlated with histological and cytogenetic studies. There has been a lot of progress in the understanding of genetics of uveal melanoma during the past 10 years. Since the first papers on genetic abnormalities based on karyotype, chromosome comparative genomic hybridization (CGH) and fluorescent in situ hybridization (FISH) analyses monosomy 3 has been recognized as bad prognostic factor. Other chromosomal imbalances, especially gain of 8q, have been reported as important events for metastasis. Molecular techniques such as CGH- and single nucleotide polymorphism (SNP)-arrays, and multiplex-ligation-dependent probe arrays (MLPA), has made it possible to reliably assess the status of genomic markers of prognostic value. These studies have highlighted genetic heterogeneity with isodisomy on chromosome 3. For others gene expression profile is the most precise way to predict the prognosis of uveal melanoma. All these advances were made possible thanks to the collaboration of ophthalmologist oncologists and geneticists in multidisciplinary teams.

Local treatment The treatment of uveal melanoma by radiotherapy was popularized by Stallard who initiated the use of cobalt plaque brachytherapy in the 1950s with encouraging results. In the 1970s Zimmerman and colleagues showed that enucleation of the eye does not prevent metastatic dissemination compared to conservative management of uveal melanoma. Since then treatment by brachytherapy or

402

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65323-0

Malignant Tumors of the Eye, Conjunctiva, and Orbit: Diagnosis and Therapy

Fig. 1

403

Uveal melanoma (fundus).

accelerated heavy particles has been used extensively. Compared to brachytherapy and to fractionated stereotactic radiotherapy, proton beam therapy has the advantage of delivering a homogenous dose of radiation to the entire tumor. This is due to the physical properties of accelerated proton beam which deliver the dose precisely with sharp decrease after the delivery point called the Bragg peak (Fig. 1). The first step to treat uveal melanoma with proton beam is radiological spotting of the tumor by clip positioning. During a surgical procedure, tantalium rings of 2.5 mm in diameter with two holes in the middle are sutured to the sclera around the tumor base that has been visualized by transillumination of the globe. Proton beam therapy is planned and performed 2 weeks after surgery. Head immobilization during radiotherapy is ensured by a custom-made thermoplastic mask associated with a bite block. All the data collected during surgery (axial length, clip measurements, tumor diameter, tumor thickness) are sent to the physicist in charge of the planning of therapy. It is mandatory to have good pictures of the fundus, ultrasonography of the eye and biometry. EYEPLAN software is used for dosimetry based on three-dimensional (3D) reconstruction of the eye, including the tumor. The irradiated volume is the tumor volume plus a safety margin of 2.5 mm around the tumor. The standard dose delivered is 60 Grays in four fractions. During treatment, patients are seated on a robot chair facing the beam and are asked to aim at small target light placed at a known angle adopted during the treatment planning process. Local control is excellent. The local recurrence rate at 10 years is usually around 5%. Secondary enucleation is performed in 10% to 15% of patients either for complications or for local recurrence. Complications include retinal detachment, maculopathy, papillopathy, cataract, glaucoma, vitreous hemorrhage, and dryness can occur. The most severe complication that usually leads to secondary enucleation is neovascular glaucoma and it is encountered after irradiation of large to extra-large tumors. The toxic tumor syndrome was recently described. It is supposed to result from the residual tumor scar producing proinflammatory cytokines and VEGF, leading to intraocular inflammation and neovascular glaucoma. Additional treatments after proton beam such as TTT (trans pupillary thermotherapy), endoresection of the tumor scar or intravitreal injections of anti-VEGF antibody may reduce the rate of these complications. Plaque radiotherapy is the best option when proton beam therapy is not available, when the patient’s general health precludes proton beam, for small anterior tumors or for tumors situated anteriorly in the superior and external quadrant (where the dose at the entrance of proton beam can cause lacrimal gland atrophy). Iodine 125 seeds can be used. This is a low gamma energy radiation emitting isotope and this radiation can be stopped by heavy metals such as gold or lead. Gold or lead plaques (12–20 mm diameter) are prepared with iodine 125 seeds and sutured on the sclera at the location of the tumor. The usual calculated dose is 90 Grays at the apex of the tumor. The time duration is calculated by the physicist and the plaque is then removed surgically. Ruthenium is a Beta emitting isotope with less penetration and less diffusion. Plaques are already prepared and can be used for multiples applications. It is recommended to avoid the use of ruthenium for tumor with a thickness of > 6 mm.

Role of biopsy at the time of local treatment Ocular oncologists have recently changed their approach to uveal melanoma in most centers. During many years, it has been considered that biopsy was not necessary in the management of uveal melanoma. Diagnostic techniques were reliable enough to allow performing conservative treatment by radiotherapy without histopathological confirmation of diagnosis. More recently it has become evident intraocular biopsies at the time of local treatment are needed to better characterize the disease at cell and molecular level. The techniques of transcleral and transvitreal fine needle aspiration biopsies were published on large series of patients, with very few side effects. Today most centers are able to offer fine needle aspiration biopsies for cytogenetic studies despite in principle risk of tumor dissemination, vitreous hemorrhage and retinal detachment; On the other hand, the benefits for the patient are still

404

Malignant Tumors of the Eye, Conjunctiva, and Orbit: Diagnosis and Therapy

limited as there is still a lack of efficient chemotherapy to prevent metastasis in high-risk patients. Others benefits of genetic studies are the possibility of entering a closer follow-up protocol to detect early liver metastasis and to participate to adjuvant therapy protocols.

Management and follow-up Besides genetic studies of the primary tumor, the presence of circulating tumor cells could be an indicator for metastatic risk and is currently under evaluation. A few centers have started to use adjuvant therapy protocols. In the past, there has been three published studies on adjuvant therapies in uveal melanoma, one randomized with DTIC, one nonrandomized with interferon and one with intraarterial fotemustine. None of these trials gave statistically significant results. There have been a lot of controversies on how to follow patients that have been treated for uveal melanoma. Early detection of metastasis by repeated liver ultrasonography can allow the surgical resection of localized metastasis and therefore increase survival. Some of the new targeted therapies are active in 10% of metastatic patients. Ultrasonography of the liver every 6 months is usually recommended. Liver MRI is useful, in particular to confirm liver metastasis but TEP scan are sometimes not detecting metastasis from uveal melanoma though they are very sensitive in the case of cutaneous melanoma. Oncologists have become recently more interested in this disease and many new drugs or protocols are already under trial for metastatic disease. Chemoembolization is employed for the liver but is not useful when the patient has other localizations of the metastatic disease. Vaccine protocols are available in HLA2 patients, with limited success but there is still ongoing work. Current research is investigating the prognosis and predictive significance of transcriptomic analysis. Proteins involved in the migration of tumor cells from the eye to the liver and adhesion and proliferation of the metastasis are being studied. Other primary intraocular malignancies in adults are very rare. Some rare cases of adenocarcinoma of the pigmented epithelium have been published. Oculocerebral lymphoma is a rare disease due to large cells lymphomas. The vitreoretinal lymphoma is associated generally with central nervous system lymphoma. A typical retinal infiltrate is associated with the presence of abnormal cells in the vitreous. Diagnosis can usually be made by vitreous cytology or more rarely by biopsy of the cerebral lesions. Treatment must be performed in reference centers.

Retinoblastoma Diagnosis Retinoblastoma is the most common malignant intraocular tumor in children with an estimated incidence of 1/20,000 live births. Retinoblastoma is the first tumor for which a genetic origin was demonstrated. According to the Knudson hypothesis, retinoblastoma carcinogenesis is dependent upon inactivating mutation on the two alleles of the RB1 tumor suppressor gene located on chromosome 13q14.2.2. The tumor can arise in a child carrying a germline mutation on one of the two alleles. A second hit mutation of the other allele in a precursor cell is responsible for the appearance of a retinal tumor. This predisposition can be transmitted by a parent carrying the mutation (familial form) or can occur as a de novo event. In this setting, retinoblastoma is usually bilateral or unilateral multifocal and the child is at risk of developing other extra ocular tumors (soft tissue sarcoma, osteosarcoma, carcinoma, high-grade glioma, malignant melanoma). Treatment must take into account the risk of developing second tumors, which can be increased by treatment with mutagenic agents (chemotherapy), X-rays, and radiotherapy. Retinoblastoma occurs almost exclusively in children under the age of 15 years and the median age at diagnosis is lower for the bilateral form (< 12 months) than for the unilateral unifocal form (24 months). Early diagnosis is a priority, as it allows the child to be cured with preservation of almost normal vision in at least one of the two eyes. However, despite well-known clinical signs, retinoblastoma is still often diagnosed at an advanced stage, requiring aggressive therapy with a high-risk of enucleation of the affected eye in unilateral forms and of the two eyes in bilateral forms. The clinical presentation of retinoblastoma is well known. Leukocoria is the presenting sign in 60% of cases, but is often a late sign. At the beginning, leukocoria can be intermittent, only visible in specific direction of gaze or with specific light. If no diagnosis is made, leukocoria become more and more visible and permanent. It can easily be detected on photos with flash especially if the antired eye system is disconnected. The second most common early sign of retinoblastoma is strabismus, generally related to a macular tumor. Provided that strabismus is not confused with accommodative strabismus, treatment can be curative and allow preservation of the eye. Less common clinical presentations may also be encountered, generally indicating advanced forms (buphthalmia, neovascular glaucoma, orbital inflammation, phthisis bulbi, etc.), which generally do not allow globe-conserving treatment. The clinical presentation varies according to the stage of the disease and the presence of one or several tumors in either or both eyes, with symmetrical or asymmetrical lesions. The tumor is visible as a white retinal mass with a variable blood supply depending on the size of the tumor and may present calcifications. The tumor (or tumors) may present as an endophytic form, in which the tumor extends into the vitreous, an exophytic form, associated with retinal detachment and subretinal infiltration; or a mixed form, comprising vitreous involvement and retinal detachment. Diffuse infiltrating retinoblastoma is a rare form of retinoblastoma with infiltration of the retina without tumor mass. Examination shows vitreous infiltration and sometimes hypopyon (Fig. 2). Retinoma (or retinocytoma) is a spontaneously arrested or spontaneously regressive form that resembles treated retinoblastoma with a grayer color, intense calcification, and a zone of atrophy around the lesion. These lesions can be detected on ocular fundus examination of a parent of a child with retinoblastoma or in a child consulting for leukocoria or strabismus. These lesions must be carefully monitored in young children, as they can progress to active retinoblastoma (Fig. 3).

Malignant Tumors of the Eye, Conjunctiva, and Orbit: Diagnosis and Therapy

Fig. 2

Leukocoria.

Fig. 3

Retinoblastoma (fundus).

405

Prognosis There are also several well-known differential diagnoses (Coats disease, persistent hyperplastic primary vitreous, medulloepithelioma, uveal melanoma in children, retinal detachment due to other causes, Toxocara canis infection, etc.). Examination in a reference center and modern imaging comprising duplex Doppler ocular ultrasound and magnetic resonance imaging (MRI) allow confirmation of the diagnosis in the vast majority of cases. Several classifications have been proposed to address various issues, but none of them are fully satisfactory. The first classification proposed by Reese-Ellsworth was developed in the age of external beam radiotherapy to evaluate the chances of globe-conserving radiotherapy. The second classification, the International Intraocular Retinoblastoma Classification (ABC classification), was developed by Murphree in 2005 in the age of systemic chemotherapy. This classification, ranging from A to E, can be used to estimate the possibility of preserving the eye by systemic chemotherapy in combination with local therapy. This classification remains widely used. The new TNM classification (eighth American Joint Committee on Cancer TNM classification) estimates survival and the chances of preserving the eye by taking into account clinical risk factors (cT), histological risk factors (pT), the presence or absence of a germline mutation (H), and of course the presence of lymph node (N) or distant metastasis (M). Modern MRI imaging is also used for intraorbital and central nervous system staging. The treatment of retinoblastoma was very much improved during the past 30 years. The cure rate in developed countries is closed to 100%. Radiotherapy has been the treatment of choice until early 1990s but has now been phased out because of the risk of second cancer in the field or radiation that can be as much as 30% in patients carrying RB1 mutation. Chemotherapy is now the treatment of choice. The combination of intravenous carboplatin, vincristine and etoposide found to be active on metastasis is also very efficient on intraocular retinoblastoma. Depending on the severity, patients with bilateral retinoblastoma can be treated with a combination of two to three drugs. More recently the use of intraarterial chemotherapy with melphalan, topotecan, and carboplatin has allowed very impressive regression of the tumors. Chemotherapy can also be injected directly in the vitreous cavity in selected cases. Disseminated metastatic disease or cerebral involvement (pinealoma or metastases via the optic tract) requires intensive chemotherapy with autologous bone marrow transplantation. Cerebral involvement is associated with a poorer prognosis.

406

Malignant Tumors of the Eye, Conjunctiva, and Orbit: Diagnosis and Therapy

The second factor that has improved survival and eye preservation rates is the use of local therapy. Most teams specialized in the treatment of retinoblastoma agree that chemotherapy alone cannot ensure perfect control of intraocular disease but must be combined with local therapy (diode laser transpupillary thermotherapy, laser, cryoapplication, brachytherapy). Early diagnosis, as a result of information campaigns directed to the general population, healthcare professionals, and ophthalmologists, allows ablation of small tumors exclusively by local therapy or a combination of carboplatin and transpupillary thermotherapy. Local therapies can also be also used as second-line treatment for local relapse. Another important factor for improved patient survival is the establishment of specialized centers comprising multidisciplinary teams composed of ophthalmologists, oncologists, pediatric oncologists, radiation physicians, radiologists, geneticists, anesthetists, and so on. This high level of multidisciplinary competence allows tailored treatments based on the findings of the initial ophthalmological examination, which must be performed under general anesthesia. In centers treating retinoblastomas, genetic information concerning the parents of an affected child and patients treated for retinoblastoma during childhood is essential and germline mutation screening must be systematically proposed, with due consideration to mosaicism.

Chemotherapy Various drugs are used for intravenous, intraarterial, subconjunctival, or intravitreal chemotherapy depending on the team (most commonly carboplatin, vincristine, etoposide, topotecan, melphalan). Intravenous chemotherapy alone cannot cure retinoblastoma but chemoreduction with intravenous chemotherapy followed or combined with local therapy is still widely used by many teams with recognized efficacy and acceptable systemic toxicity. The intraarterial route of administration developed by Kaneko, using melphalan or topotecan or carboplatin alone or in combination, allows drug administration directly into the ophthalmic artery, thereby limiting systemic distribution of chemotherapy and its adverse effects. Consequently, local administration cannot be used to treat micrometastases in advanced forms of the disease. This route of administration is not devoid of local toxicity on the choroidal and retinal blood supply and must be performed by an experienced team of neuroradiologists. In a large meta-analysis based on 208 publications on the subject, only 28 articles with sufficient follow-up could be analyzed. Most of these studies were retrospective and noncomparative. Enucleation was avoided in 66% of cases and 20 cases of metastases were reported. These cases of metastases probably occurred in the context of treatment of advanced retinoblastoma. Nevertheless, this treatment is very effective in unilateral forms with visual potential (cT2 or group C or D). Some authors also use intraarterial chemotherapy to treat bilateral forms. The place of this treatment remains controversial in advanced unilateral forms (cT3 or V) regardless of the subgroup in the absence of visual potential. No clear consensus has been reached, as the indications for intraarterial chemotherapy vary from one team to another. No multicenter prospective studies comparing intraarterial chemotherapy and systemic chemotherapy are available and the published studies are all too often retrospective and present excessive bias to allow reliable scientific results. There is also insufficient follow-up concerning the sensitivity of bone and soft tissues to repeated exposure to low doses of X-rays necessary for catheter positioning by radiologists: these doses are not negligible in the context of constitutional alteration of the RB1 gene, associated with an increased risk of second tumor (essentially sarcoma) in the irradiated field. Intravitreal administration of melphalan or topotecan is particularly effective treatment for vitreous seeding. A rigorous technique is essential due to the risk of tumor cell dissemination along the needle track. Some cases of extraocular tumor spread have been described. This phenomenon is probably rare, but several cases that have been reported to the author and have never been published. Compliance with the most commonly accepted doses (20–30 mg of melphalan) is important to avoid major complications, including phthisis bulbi for doses higher than 50 mg.

Radiotherapy and other local therapies External beam radiotherapy (EBRT) is now rarely used due to the risk of local complications and the major risk of sarcoma in the irradiated field. The incidence of second cancer in the irradiated field, but also outside of the irradiated field, was estimated to be 51% at 50 years by a team in New York on the basis of a very large series. Brachytherapy is generally used for peripheral tumors, usually as second-line treatment. Orbital brachytherapy using rows of iodine 125 plus a plaque can be used instead of EBRT after enucleation in the presence of extraocular tumor spread or positive optic nerve resection margins (pT4), in combination with systemic chemotherapy. Other focal Treatments include triple freeze-thaw cryotherapy with freezing as far as the tip of the tumor for peripheral tumors inaccessible to laser ablation not exceeding a thickness of several millimeters. Laser thermotherapy is used as an adjuvant to chemotherapy or alone for small tumors or localized relapses. We use a diode laser (810 nm) with continuous wavelength delivered via an operating microscope or via the indirect ophtalmoscope. The laser is absorbed at the level of the pigment epithelium and produces energy and hyperthermia of which is synergistic with carboplatin (chemothermotherapy). Duration and intensity is adapted to the size of the tumor and degree of pigmentation (5–20 min, 400–800 mW). A complete ophthalmological assessment, including bilateral dilated ocular fundus examination under general anesthesia, MRI, and a pediatric assessment, is essential before any treatment decisions and should be done in emergency. Genetic information is essential in familial forms. Prenatal diagnosis is possible if the mutation is identified. Embryo selection and in vitro fertilization of embryos not carrying a constitutional alteration of the RB1 gene is possible in some countries. Parents must be informed about this possibility. A child born to a parent with a history of retinoblastoma must be examined during the first week of life and, in the absence of visible tumor, must be examined at least monthly under general anesthesia. Small cT1a tumors detected on clinical examination or optical coherence tomography can be treated by transpupillary thermotherapy alone.

Malignant Tumors of the Eye, Conjunctiva, and Orbit: Diagnosis and Therapy

407

Retinoblastoma survival is around 95% in high-resource countries but is much less in low-resource countries where diagnosis is often made at late stage and access to care constrained. Programs to improve retinoblastoma care are ongoing in many countries. In low-resource countries the rate of ocular conservation is improving but careful evaluation is needed for treatment, balance of benefits and risk, side effects and long term results and metastatic rate.

Medulloepithelioma Medulloepitheliomas are very rare, almost always unilateral congenital tumors arising from the ciliary epithelium usually becoming symptomatic between 2 and 4 years of age. They are most often localized in the ciliary body. They often are slow growing but some of them can be very aggressive with possible orbital recurrence or metastasis.

Malignant Conjunctival Tumors The conjunctiva is the outermost layer of the ocular surface, above the sclera, and also forms a continuous membrane with the innermost layer of the eyelids behind the tarsal plane. Anatomically these regions are described as bulbar and tarsal conjunctiva, and they merge in the conjunctival fornices. Histologically, the conjunctiva is composed of an epithelium, overlying a basal membrane that separates it from the conjunctival stroma, that is densely vascularized, innervated, and contains immune elements. This structure has clinical consequences since most primary conjunctival neoplasia develop initially within the epithelium, and may become invasive when they progress into the stroma, where there is a risk of local spread and distant metastases.

Conjunctival invasive squamous cell carcinoma Invasive squamous cell carcinoma is a continuation of the spectrum of conjunctival intraepithelial neoplasia (CIN), a minimally aggressive lesion localized to the conjunctival epithelium. Severe CIN, also termed in situ squamous cell carcinoma, can progress through the basement membrane, invade adjacent tissues and become malignant. Invasive squamous cell carcinoma can also metastasize (in < 2% of cases), usually to the regional and cervical lymph nodes. Therefore, CIN is a major risk factors for invasive squamous cell carcinoma. Other identified factors include UV-light exposure, human papillomavirus infection (especially HPV-16, that may be associated to a better prognosis), immunosuppression, especially by human immunodeficiency virus (HIV), associated to more aggressive tumors, and conditions that predispose to epithelial malignancies, such as atopic eczema and xeroderma pigmentosum, a rare genetic skin disorder characterized by an extreme sensitivity to UV light. The reported incidence of invasive squamous cell carcinoma ranges from 0.02 to 3.5 per 100,000, and the disease seems more frequent in men (estimated male–female ratio, 3:1) and in elderly subjects (Fig. 4). The main diagnostic challenge is the similar appearance of CIN and invasive squamous cell carcinoma, which are sometimes difficult to differentiate clinically. Both lesions develop in the interpalpebral region of the conjunctiva, often adjacent to the limbus, with part of the lesion overlapping the cornea, and can more rarely extend to the conjunctival fornices. Although most frequently affecting elderly Caucasian males, they can be diagnosed at all ages, in both genders, and in patients of all ethnicities. While CIN

Fig. 4

Conjunctival invasive carcinome.

408

Malignant Tumors of the Eye, Conjunctiva, and Orbit: Diagnosis and Therapy

presents as a fleshy, with moderate elevation and a sessile base, invasive squamous cell carcinoma can have multiple presentations, from a localized gelatinous mass to a large elevated, papillomatous lesion. It is often associated with dilated conjunctival feeder vessels. A frequent feature is the multilobular aspect of the lesion. Variants of squamous cell carcinoma need to be recognized. The diffuse form, flat and poorly demarcated, may be underdiagnosed. Two aggressive forms, the mucoepidermoid and spindle cell variants (< 5% of conjunctival squamous cell carcinoma) carry a more severe prognosis. The mucoepidermoid form may progress intraocularly with the development of large supra-uveal mucinous cysts. Spindle cell carcinoma is more aggressive locally with a higher risk of distant organ metastases. Diagnosis should rely on pathological examination after primary complete surgical excision. The characteristic histological features include abnormal keratin-producing epithelial cells with mitotic activity. A small proportion of cases are poorly differentiated with atypical pleomorphic giant cells and dyskeratosis. The main characteristic is that the tumor extends beyond the basal conjunctival membrane into the subepithelial connective tissue. The mucoepidermoid form harbors an epidermoid component with vacuolated cytoplasms and variable mucin accumulation. The spindle cell variant presents with pleomorphic spindle cells of epithelial origin. Differential diagnoses, besides CIN, include all nonpigmented conjunctival malignancies may resemble invasive conjunctival squamous cell carcinoma. Notable entities include amelanotic conjunctival melanoma; mucoepidermoid and spindle cell carcinoma, detailed above; conjunctival basocellular carcinoma, a very rare entity; and diffuse conjunctival extension of eyelid sebaceous carcinoma, that carries a higher metastasis and mortality rates than invasive squamous cell carcinoma and should be recognized. The local prognosis is related to the risk of local recurrence, which is globally quite high (40%), in the absence of neoadjuvant treatments, such as irradiation, as detailed below. The systemic prognosis is generally favorable and related to the risk of developing metastases. Although rare in invasive squamous cell carcinoma, metastatic disease is conditioned by tumor size, the depth of local invasion, and presence of immunosuppression (especially HIV infection). According to these factors, locoregional extension should be explored at regular intervals by ultrasound exploration of head and neck lymph nodes (for lower risk cases) or Positron Emission Tomography (for higher risk cases). Regarding treatment, a schematic drawing or preferably a color photograph of the lesion at diagnosis, before any therapeutic intervention is mandatory. Initial management consists in a surgically complete excision, that should be performed with a “no touch” approach similar to pigmented lesions (avoiding direct contact of instruments with the lesion, and replacing instruments for conjunctival suture/reconstruction), given the possibility of presumed epithelial malignancy to be eventually identified as amelanotic melanoma upon pathological analysis. For the same reason, surgery is preferably performed under general anesthesia to avoid orbital seeding. Cryotherapy of tumor margins combined to excision contributes to reduce the risk of local recurrence. If pathological examination confirms the diagnosis of squamous cell carcinoma (or other malignant epithelial tumor), neoadjuvant radiotherapy of the tumoral site by either conventional external radiotherapy, plaque brachytherapy or proton beam irradiation should be performed, according to the initial tumor topography. These treatments allow to reduce dramatically the recurrence rate from 40% to 2%, and their major adverse effect is the acceleration of cataract formation. Topical chemotherapy by mitomycin C eyedrops is indicated as neoadjuvant therapy for CIN. The initial recommended dose is 0.02%, four times daily, two 2-week cycles. In case of relapse, the same treatment at the higher dose of 0.04% is indicated. Interferon eyedrops, administered four times daily for 6 months, are an alternative to mitomycin C, in patients with severe corneal alterations or in those with limited disease. With adjuvant, topical chemotherapy having widely improved local control and recurrence rates, future developments rely in the higher integration of treatment approaches. In a near future, advances in diagnosis techniques, for instance by impression cytology, and by tumoral genome detection, may improve management of conjunctival epithelial malignancies by allowing their detection and referral to tertiary centers at earlier stages.

Conjunctival melanoma Conjunctival melanoma is a rare, malignant invasive melanocytic tumor, affecting the bulbar conjunctiva, and more rarely the conjunctival fornices or the palpebral conjunctiva. Its incidence increased with age, and is higher in individuals in their 50–60s or older. It only rarely occurs in subjects younger than 30 years. There seem to be a recent increase in the incidence of conjunctival melanoma, according to data from the United States and Northern Europe. Conjunctival melanoma has a variable, but usually aggressive course, characterized by a tendency to recur often after simple surgical excision. It therefore requires adjuvant treatments to optimize local control. Moreover, conjunctival melanomas frequently metastasize. This tumor is composed by variably pigmented malignant melanocytes (Fig. 5). Commonly recognized risk factors for conjunctival melanoma include lighter skin (although it has been scarcely studied in black-skinned subjects), exposure to UV light (this explains the preferential localization to exposed parts of the conjunctiva, nasal and temporal to the limbus, but does not account for the possibility of melanoma localized in the conjunctival fornices), primary acquired melanocytosis (75% of conjunctival melanoma arise from primary acquired melanocytosis) and conjunctival nevi (25% of conjunctival melanoma arise from conjunctival nevi). At clinical presentation, conjunctival melanoma usually consists in a nodular pigmented elevation, most often located close or adjacent to the limbus, with possible extension over the corneal epithelium, and usually surrounded by a fine, planar, granular pigmentation of the conjunctiva. Usually sharply demarcated, the borders of the pigmented tumor may be ill-defined, taking the appearance of diffuse of multiple nodules, which often occurs in cases arising over areas of primary acquired melanocytosis. It can occasionally develop in the conjunctival fornices or palpebral conjunctiva, in which case it may remain undetected until the diagnosis is made when the tumor is advanced. Conjunctival melanoma can also be nonpigmented (amelanotic), making it

Malignant Tumors of the Eye, Conjunctiva, and Orbit: Diagnosis and Therapy

Fig. 5

409

Conjunctival melanoma.

difficult to distinguish clinically from malignancies of the conjunctival epithelium. In addition, seeding of the palpebral conjunctiva by continuous contact with a bulbar lesion may occur. Diagnosis relies on histopathological evaluation after surgical excision, identifying the morphology of malignant melanocytes, that may vary from low-grade spindle cells to epithelioid cells. It affects the basal conjunctival epithelium at early stages, and secondarily invades the stroma. Morphological evaluation may be challenging, for instance in purely fusiform or amelanotic presentations. Immunostaining can be used to identify S-100 protein (sensitive but not specific), HMB-45 or melan-A (more specific), SOX10 (very sensitive and specific), and other proteins expressed by melanoma cells. Ki-67 labeling evaluates the depth of the proliferative activity and help differentiate malignant from naevic pigmented lesions. The pathological evaluation also recognizes preexisting naevi or primary acquired melanocytosis. Differential diagnoses include conjunctival nevus, primary acquired melanocytosis (with are also risk factors and may coexist with melanoma), ethnic conjunctival melanocytosis in dark-skin subjects, squamous cell carcinoma (difficult to differentiate from completely amelanotic melanomas), ciliary or very peripheral choroidal melanoma with extrascleral extension, conjunctival foreign body (especially long-standing metallic elements progressively surrounded by a brown halo due to perilesional oxidation), and most other benign or malign conjunctival tumors that may harbor variable pigmentation. Prognosis is hampered by the difficulty to locally control the tumor due to frequent recurrences and the tendency to spread by simple contact, and by the high rate of metastasis (up to 40% 10 years after diagnosis). Preferred sites of metastasis are regional lymph nodes, and follow-up by serial ultrasonography of cervical lymph nodes is mandatory. The particularity of conjunctival melanoma is the serious risk of aggravating the prognosis in cases of mismanagement, especially by improper steps during surgery. Contact of surgical instruments with the tumor itself must be avoided, and new forceps and scissors must be used after excision for repair of the conjunctival defect by either direct suture, or amniotic membrane graft for wider defects. To date, intraorbital disease is treated by orbital exenteration, illustrating the catastrophic course that may result from improper surgical management. Although difficult to estimate, the overall mortality related to conjunctival melanoma is close to 25% at 10 years. Regarding treatment, as for epithelial malignancies, a color photograph of the lesion at diagnosis, before any therapeutic intervention, is mandatory. The treatment of conjunctival melanoma consists in removing surgically the malignant tissue, and is associated to additional multidisciplinary treatment modalities to limit the risk of recurrence. Surgical excision of the whole tumor with safety margins is the first step. Due to the tendency of melanoma cells to disseminate and develop in seeded tissues, the excision of any pigmented or nonpigmented conjunctival tumors (due to the possibility of amelanotic melanoma resembling spindle cell

410

Malignant Tumors of the Eye, Conjunctiva, and Orbit: Diagnosis and Therapy

carcinoma) should be performed using “no touch” technique under general anesthesia to avoid orbital seeding by local anesthesia infiltration. The tumor should be totally removed, with 2-mm security margins macroscopically devoid of any tumor. In order to limit tumor spread, biopsies must also be avoided, except for diagnostic confirmation of advanced cases requiring orbital exenteration. Additional cryotherapy can be performed on tumor margins to destroy microscopic lesions or invisible primary acquired melanosis. A double freeze-thaw process is recommended, but overtreatment should be avoided due to the destructive effect on underlying ocular structures, and the inflammatory conjunctival reaction. Adjuvant irradiation by plaque brachytherapy or proton beam therapy is performed for invasive forms of conjunctival melanoma. Adjuvant topical chemotherapy by 0.04% mitomycin C (four times daily, two 2-week cycles) aims at treating primary acquired melanosis lesions. The recent advent of neoadjuvant therapy after surgical excision seems to dramatically reduce recurrences of conjunctival melanoma. Radiotherapy by either proton beam irradiation or plaque brachytherapy has provided ocular oncologists with an unprecedented tool to treat conjunctival melanoma and reduce the rate of recurrences. With now a few years’ follow-up, topical adjuvant chemotherapy by mitomycin seems efficient in controlling primary acquired melanocytosis, a major predisposing risk factor that also favors recurrences. Finally, progress remains to be made in the search for efficient treatment strategies for metastatic disease, a frequent course of this aggressive conjunctival neoplasia.

Caruncular malignant tumors Due to the particular histological origin of the caruncle, an evolutionary remnant of the third eyelid still present in certain species of birds or reptiles, it harbors cutaneous elements such as hair follicle and sebaceous glands, and can be affected by tumors types usually found on eyelids or conjunctiva. Sebaceous gland carcinoma is a rare malignant tumor arising from sebaceous glands of the caruncle, similarly to eyelid sebaceous gland carcinoma that arises from sebaceous glands of the eyelids. A purely conjunctival variant also exists that develops insidiously as a flat diffuse “pagetoid” lesion at the surface of the conjunctiva. Both forms may pose diagnostic dilemmas. The caroncular/palpebral form may be initially mistaken and managed as a chalazion (a benign meibomian gland cyst), and the diffuse conjunctival form as an ocular surface inflammation (such as chronic conjunctivitis or episcleritis). Histologically, the lesions are composed of pleomorphic cells with prominent nucleoli. A characteristic but not systematical aspect consists in disseminated malignant cells within the conjunctival epithelium, termed “pagetoid infiltrate.” If advanced, this infiltrate can replace completely the normal epithelium (in situ carcinoma). Sebaceous gland carcinoma is aggressive and can metastasize, with an estimated mortality of 25% at 5 years. Local treatment relies on surgical excision, that must be complete, and additional cryotherapy or mitomycin C topical chemotherapy. Caruncular melanoma presents as variably pigmented, noncystic solid lesions. They are diagnosed and managed similarly to conjunctival melanoma. Due to the high risk of seeding into the lacrimal drainage system, because of the proximity of the caruncle to the lacrimal ducts, lacrimal plugs may be employed.

Malignant Orbital Tumors The orbit is a circumscribed anatomical region where skeletal, muscular, vascular, and neural elements coexist. A wide array of lesions can produce orbital masses, including inflammatory syndromes, thyroid-related extraocular muscle dilation, and benign or malignant tumors of muscular, skeletal, vascular or neural origin. Although rare, malignant orbital tumors should be recognized and appropriately managed. Three of the most important malignancies localizing to the orbit will be detailed in this section. Rhabdomyosarcoma represents the most frequent orbital malignancy in childhood and teenage. Median age at diagnosis is 8 years but the tumor can develop during the first two decades of life. This tumor derives from skeletal muscle. It is favored by local irradiation, which is of particular importance for the follow-up of children previously irradiated for retinoblastoma, a common treatment modality until the early 1990s. These patients are at higher risk of developing secondary tumors, including rhabdomyosarcoma, with devastating consequences. Clinical manifestations are nonspecific and result from the orbital volume occupation by the developing tumor. They include proptosis, lateral or vertical globe displacement in primary gaze position, variable eye movement limitations, ptosis of the superior eyelid, eyelid or conjunctival swelling, pain and detection of a palpable protruding mass in most advanced cases. Most often, the mass is localized in the superonasal part of the orbit and the globe is displaced inferiorly and temporally. Diagnosis relies on radiological suspicion and pathological confirmation. Imaging by computed tomography (CT) scan shows an irregular, moderately demarcated mass generally not involving the extraocular musculature. Magnetic resonance (MR) imaging reveals on T1 an hypointense lesion, as compared to extraocular muscles, and a hyperintense T2 images, enhanced after gadolinium injection. If accessible to biopsy, rhabdomyosarcoma may harbor different cell morphologies, the most frequent being the embryonal type, and the more aggressive the alveolar type. Embryonal type displays skeletal muscle cells at different stages of embryogenesis. The alveolar type displays separated cells with septae resembling pulmonary alveolar structures. Differential diagnoses include anterior or posterior orbital cellulitis, nonspecific orbital inflammation, capillary hemangioma, lymphangioma, rupture dermoid cyst, and any orbital may present as an orbital space-occupying process and should be differentiated from rhabdomyosarcoma. Prognosis depends on the presence of metastases in distant organs, such as lung or lymph nodes. The local long-term prognosis may also be influenced by the treatment strategy, with patients receiving radiotherapy to the facial and orbital region in their childhood or teenage years at higher risk of developing secondary neoplasias of skin, bone, or soft tissue.

Malignant Tumors of the Eye, Conjunctiva, and Orbit: Diagnosis and Therapy

411

Regarding treatment, initial management is surgical and consists in a biopsy of the orbital mass for diagnosis conformation. When possible, the entire tumor should be removed. Once the diagnostic is ascertained by histopathological evaluation, the treatment relies on radiotherapy and chemotherapy, alone or combined, according to guidelines. The optimization of irradiation treatment strategies (total dose, dose fractioning, irradiation modality) will hopefully reduce the burden of secondary tumors arising years or even decades after successful management. Progresses in chemotherapeutic approaches may contribute to reduce partially or totally the amount of radiation administered.

Malignant tumors of the lacrimal gland Approximately 10% of orbital lesions involve the lacrimal gland. Malignant lesions of several histological type can affect the lacrimal gland, associated with different prognosis. Primary epithelial tumors arise from epithelial structures of the lacrimal gland. Nonepithelial neoplasia include inflammatory, infiltrative, lymphoid lesions. Adenoid cystic carcinoma, pleomorphic adenocarcinoma and primary ductal carcinoma are the most frequent epithelial malignant tumors affecting the lacrimal gland, among several histological diagnoses, and will be detailed in this section. Lacrimal gland malignancies do not have any gender or ethnic predilection. They occur more frequently in middle-aged individuals, with the exception of adenoid cystic carcinoma that showing incidence peak in childhood. The clinical presentation of all tumor types is very similar, reflecting the space-occupying orbital process located superiorly and temporally. The main sign is a progressive, unilateral proptosis, with medial and inferior displacement of the eyeball, possibly associated with pain in the more aggressive types due to local sensitive nerve invasion. Frequently, patients experience progressive diplopia. Posterior neural invasion induces cutaneous hypoesthesia in the territory of the 5th cranial nerve (upper cheek, periocular area). Diagnosis relies on imaging by CT scan and MRI to visualize a round or oval-shaped mass, located in the superior temporal orbit, possibly displaying irregular contours. Bone erosion is visible on CT scan in case of large, aggressive tumors. These imaging modalities provide certain features suggestive of malignancy, like intralesional calcifications, but cannot distinguish between pathological types. MRI show lesions with low T1 signal, high T2 signal, and moderate enhancement by contrast agents. Diagnosis also requires pathological examination after ultrasound-guided or surgical biopsy, or preferably whole lesion removal. Each histological type harbors distinctive features. Adenoid cystic carcinoma can be cribriform (a pattern described as “Swiss cheese”), sclerosing, basaloid (associated to a worse prognosis), tubular or comedocarcinoma. Pleomorphic adenocarcinoma harbors a proportion of malignant epithelial cells within a benign pleomorphic adenoma. Ductal carcinoma presents proliferative epithelial cells within lacrimal ducts, resulting in cystic dilation. Differential diagnoses include symptoms and signs related to malignant tumors tend to progress more rapidly than in benign lesions. Pain is also more frequently associated to malignant than benign lesions, due to neural invasion. Benign lacrimal gland lesions include pleomorphic adenoma, also termed “benign mixed tumor,” and ductal epithelial cysts, also termed “dacryops.” Epithelial malignant neoplasia of the lacrimal gland are aggressive tumors. Prognosis depends mostly on histological subtype, the main factor influencing local invasiveness and the tendency for distant metastases. Treatment includes Irradiation, chemotherapy, and in most advanced cases exenteration. Improvements in the care of rhabdomyosarcoma patients rely on two major axes. First, better screening of at risk individuals, namely those previously treated for another malignancy in childhood, is needed. Prolonged and sustained follow-up in tertiary centers with standardized imaging protocols will improve the detection of new tumors, and the understanding of risk factors and progression patterns of these secondary tumors. Second, current improvements in therapeutic modalities by mini-invasive or endoscopic orbitofacial surgery, brachytherapy, optimized external irradiation, and pharmaceutical approaches may hopefully improve the local and systemic prognosis of these aggressive tumors.

Orbital metastases The tumors that have the higher propensity for seeding metastases to the orbit are carcinomas of the breast, lung, kidney, prostate, and gastrointestinal system. Recognizing orbital metastases is of importance since in 20% of cases patients have no history of cancer and orbital metastasis is the revealing manifestation of the primary neoplasia. Orbital metastases follow the age distribution of solid organ tumors and are more common in elderly subjects. They are very rare in children. They manifest with the classical signs of orbital space-occupying lesions, with a very rapid onset and progression, and frequent pain. Paradoxically, certain tumors induce endophthalmos rather than proptosis, like breast or gastric neoplasia, that tend to induce tissue contraction. Regarding diagnosis, some primary cancer types have preferential localizations within the orbit, visible on imaging. For instance, breast cancer localizes to soft orbital tissue and prostate cancer tends to localize to bone. Several cancer types generate demarcated, oval-shaped orbital metastases that resemble benign lesions. A biopsy is mandatory to ascertain the nature of an orbital mass and its benign or malign characteristics. The final diagnosis is made by the pathologist (or cytologist, when only needle aspiration is feasible instead of surgical biopsy), who identifies the histological nature of the primary neoplasia. Differential diagnoses include all intraorbital lesions and should be explored by cytological, or preferably histological analysis. Ultrasound-guided biopsy for cytological analysis is a diagnostic option for lesions not accessible surgically, or for patients unfit for invasive orbital surgery. Certain metastases from epithelial tumors in distant organs pose a diagnostic challenge because they may lack specific histological features to distinguish them from orbital, conjunctival or eyelid epithelial neoplasia. Orbital metastases indicate advanced metastatic disease and are classically associated to a poor prognosis. They are frequently associated to brain metastases. Nonetheless, advances in chemotherapy, especially drugs modulating the immune system, seems to

412

Malignant Tumors of the Eye, Conjunctiva, and Orbit: Diagnosis and Therapy

improve the survival of metastatic patients. Treatment involves irradiation, chemotherapy, or hormone treatment are possible management strategies for orbital metastases, depending on the primary tumor type, stage and general status of the patient. Progress in fine-needle, ultrasound-guided biopsy techniques should improve diagnostic procedures. Novel chemotherapeutic agents may bring new possible treatment lines for advanced metastatic disease in many different primary neoplasia, variably effective on metastatic disease. However, despite of the rich vascularization of the orbit supplying the eye and extraocular musculature, intraorbital spaces are moderately accessible to oral or intravenous drug administration. Progress in drug delivery to the orbit will improve the management and prognosis of orbital metastases.

See also: Eye and Orbit Cancer: Pathology and Genetics.

Further Readings Abramson, D.H., Shields, C.L., Munier, F.L., Chantada, G.L., 2015. Treatment of retinoblastoma in 2015: Agreement and disagreement. JAMA Ophthalmology 133 (11), 1341–1347. Aerts, I., Sastre-Garau, X., Savignoni, A., Lumbroso-Le Rouic, L., Thebaud-Leculee, E., Frappaz, D., et al., 2013. Results of a multicenter prospective study on the postoperative treatment of unilateral retinoblastoma after primary enucleation. Journal of Clinical Oncology 31 (11), 1458–1463. Ahmed, S., Goel, S., Khandwala, M., Agrawal, A., Chang, B., Simmons, I.G., 2006. Neuroblastoma with orbital metastasis: Ophthalmic presentation and role of ophthalmologists. Eye (London, England) 20 (4), 466–470. Al-Jamal, R.T., Cassoux, N., Desjardins, L., Damato, B., Konstantinidis, L., Coupland, S.E., et al., 2015. The pediatric choroidal and ciliary body melanoma study: A survey by the European ophthalmic oncology group. Ophthalmology 123 (4), 898–907. American Brachytherapy SocietydOphthalmic Oncology Task Force, 2014AmericanBrachytherapySocietydOphthalmicOncologyTaskFor. American Brachytherapy Society consensus guidelines for plaque brachytherapy of uveal melanoma and retinoblastoma. Brachytherapy 13 (1), 1–14. Bakhshi, S., Singh, P., Chawla, N., 2008. Malignant childhood proptosis: Study of 104 cases. Journal of Pediatric Hematology/Oncology 30 (1), 73–76. https://doi.org/10.1097/ MPH.0b013e31815a0599. Bellaton, E., Bertozzi, A.I., Behar, C., Chastagner, P., Brisse, H., Sainte-Rose, C., et al., 2003. Neoadjuvant chemotherapy for extensive unilateral retinoblastoma. The British Journal of Ophthalmology 87 (3), 327–329. Boutroux, H., Levy, C., Mosseri, V., Desjardins, L., Plancher, C., Helfre, S., et al., 2015. Long-term evaluation of orbital rhabdomyosarcoma in children. Clinical & Experimental Ophthalmology 43 (1), 12–19. Brisse, H.J., de Graaf, P., Galluzzi, P., Cosker, K., Maeder, P., Goricke, S., et al., 2015. Assessment of early-stage optic nerve invasion in retinoblastoma using high-resolution 1.5 Tesla MRI with surface coils: A multicentre, prospective accuracy study with histopathological correlation. European Radiology 25 (5), 1443–1452. Brouwer, N.J., Marinkovic, M., van Duinen, S.G., Bleeker, J.C., Jager, M.J., Luyten, G.P.M., 2017. Treatment of conjunctival melanoma in a Dutch Referral Centre. The British Journal of Ophthalmology. https://doi.org/10.1136/bjophthalmol-2017-311082. Cassoux, N., Charlotte, F., Sastre, X., Orbach, D., Lehoang, P., Desjardins, L., 2009. Conservative surgical treatment of medulloepithelioma of the ciliary body. Archives of Ophthalmology 128 (3), 380–381. Cassoux, N., Cayette, S., Plancher, C., Lumbroso-Le Rouic, L., Levy-Gabriel, C., Asselain, B., et al., 2013. Choroidal melanoma: Does endoresection prevent neovascular glaucoma in patient treated with proton beam irradiation? Retina 33 (7), 1441–1447. Cassoux, N., Rodrigues, M.J., Plancher, C., Asselain, B., Levy-Gabriel, C., Lumbroso-Le Rouic, L., et al., 2014. Genome-wide profiling is a clinically relevant and affordable prognostic test in posterior uveal melanoma. The British Journal of Ophthalmology 98 (6), 769–774. Desjardins, L., Lumbroso-Le Rouic, L., Levy-Gabriel, C., Cassoux, N., Dendale, R., Mazal, A., et al., 2012. Treatment of uveal melanoma by accelerated proton beam. Developments in Ophthalmology 49, 41–57. Finger, P.T., 2009. The 7th edition AJCC staging system for eye cancer: An international language for ophthalmic oncology. Archives of Pathology & Laboratory Medicine 133 (8), 1197–1198. Gombos, D.S., Kelly, A., Coen, P.G., Kingston, J.E., Hungerford, J.L., 2002. Retinoblastoma treated with primary chemotherapy alone: The significance of tumour size, location, and age. The British Journal of Ophthalmology 86 (1), 80–83. Harbour, J.W., 2009. Molecular prognostic testing and individualized patient care in uveal melanoma. American Journal of Ophthalmology 148 (6), 823–829.e1. Hirst, L.W., 2007. Randomized controlled trial of topical mitomycin C for ocular surface squamous neoplasia: Early resolution. Ophthalmology 114 (5), 976–982. Kheir, W.J., Tetzlaff, M.T., Pfeiffer, M.L., Mulay, K., Ozgur, O., Morrell, G., et al., 2016. Epithelial, non-melanocytic and melanocytic proliferations of the ocular surface. Seminars in Diagnostic Pathology 33 (3), 122–132. Kim, I.K., Lane, A.M., Jain, P., Awh, C., Gragoudas, E.S., 2016. Ranibizumab for the prevention of radiation complications in patients treated with proton beam irradiation for choroidal melanoma. Transactions of the American Ophthalmological Society 114, T2. Kivela, T., Kujala, E., 2013. Prognostication in eye cancer: The latest tumor, node, metastasis classification and beyond. Eye (London, England) 27 (2), 243–252. Kivela, T.T., Piperno-Neumann, S., Desjardins, L., Schmittel, A., Bechrakis, N., Midena, E., et al., 2016. Validation of a prognostic staging for metastatic uveal melanoma: A collaborative study of the European ophthalmic oncology group. American Journal of Ophthalmology 168, 217–226. Lane, A.M., Egan, K.M., Harmon, D., Holbrook, A., Munzenrider, J.E., Gragoudas, E.S., 2009. Adjuvant interferon therapy for patients with uveal melanoma at high risk of metastasis. Ophthalmology 116 (11), 2206–2212. Lumbroso-Le Rouic, L., Savignoni, A., Levy-Gabriel, C., Aerts, I., Cassoux, N., Salviat, F., et al., 2015. Treatment of retinoblastoma: The Institut Curie experience on a series of 730 patients (1995 to 2009). Journal Français d’Ophtalmologie 38 (6), 535–541. Munier, F.L., Gaillard, M.C., Balmer, A., Soliman, S., Podilsky, G., Moulin, A.P., et al., 2012. Intravitreal chemotherapy for vitreous disease in retinoblastoma revisited: From prohibition to conditional indications. The British Journal of Ophthalmology 96 (8), 1078–1083. https://doi.org/10.1136/bjophthalmol-2011-301450. Nguyen, D.T., Houillier, C., Choquet, S., Cassoux, N., Soussain, C., Le Cossec, C., et al., 2016. Primary Oculocerebral lymphoma: MTX Polychemotherapy alone on intraocular disease control. Ophthalmology 123 (9), 2047–2050. Pandey, K., Singh, P., Singh, A., Pandey, H., 2011. Primary sebaceous gland carcinoma of the bulbar conjunctiva without involvement of eyelid: A clinical dilemma. Oman Journal of Ophthalmology 4 (2), 97–99. https://doi.org/10.4103/0974-620X.83665. Papakostas, T.D., Lane, A.M., Morrison, M., Gragoudas, E.S., Kim, I.K., 2017. Long-term outcomes after proton beam irradiation in patients with large choroidal melanomas. JAMA Ophthalmology 135 (11), 1191–1196. Sant, D.W., Tao, W., Field, M.G., Pelaez, D., Jin, K., Capobianco, A., et al., 2017. Whole exome sequencing of lacrimal gland adenoid cystic carcinoma. Investigative Ophthalmology & Visual Science 58 (6), BIO240–BIO246.

Malignant Tumors of the Eye, Conjunctiva, and Orbit: Diagnosis and Therapy

413

Sellam, A., Desjardins, L., Barnhill, R., Plancher, C., Asselain, B., Savignoni, A., et al., 2014. Fine needle aspiration biopsy in uveal melanoma: Technique, complications, and outcomes. American Journal of Ophthalmology 162, 28–34.e1. Shields, C.L., Shields, J.A., Kiratli, H., De Potter, P., Cater, J.R., 1995. Risk factors for growth and metastasis of small choroidal melanocytic lesions. Ophthalmology 102 (9), 1351–1361. Shields, C.L., Shields, J.A., Cater, J., Gunduz, K., Miyamoto, C., Micaily, B., et al., 2000. Plaque radiotherapy for uveal melanoma: Long-term visual outcome in 1106 consecutive patients. Archives of Ophthalmology 118 (9), 1219–1228. Shields, C.L., Jorge, R., Say, E.A., Magrath, G., Alset, A., Caywood, E., et al., 2016. Unilateral retinoblastoma managed with intravenous chemotherapy versus intra-arterial chemotherapy. Outcomes based on the International Classification of Retinoblastoma. Asia-Pacific Journal of Ophthalmology 5 (2), 97–103. Shields, C.L., Chien, J.L., Surakiatchanukul, T., Sioufi, K., Lally, S.E., Shields, J.A., 2017. Conjunctival tumors: Review of clinical features, risks, biomarkers, and outcomesdThe 2017 J. Donald M. Gass Lecture. Asia-Pacific Journal of Ophthalmology 6 (2), 109–120. Yousef, Y.A., Soliman, S.E., Astudillo, P.P., Durairaj, P., Dimaras, H., Chan, H.S., et al., 2016. Intra-arterial chemotherapy for retinoblastoma: A Systematic Review. JAMA Ophthalmology. https://doi.org/10.1001/jamaophthalmol.2016.0244.

Melanoma: Pathology and Genetics Lorenzo Salvati, University of Florence, Florence, Italy Giuseppe Palmieri, National Research Council (CNR), Sassari, Italy Vincenzo De Giorgi, University of Florence, Florence, Italy Antonio Cossu, University Hospital Health Unit/Azienda Ospedaliero-Universitaria (AOU), Sassari, Italy Mario Mandala`, Papa Giovanni XXIII Hospital, Bergamo, Italy Daniela Massi, University of Florence, Florence, Italy © 2019 Elsevier Inc. All rights reserved.

Glossary Radial growth phase (RGP) An atypical melanocytic proliferation that enlarges along the radii of an imperfect circle and is limited to epidermis and superficial dermis. Clinically, the lesion tends to be relatively flat with oval or circular borders. Clark used the term RGP to describe the clinical peripheral enlargement of melanoma, but not the growth direction of melanocytic cells, as they may grow upward in the epidermis, downward into the papillary dermis, or peripherally, determining enlargement of the margins. Melanoma in situ and invasive melanoma that has not yet metastasized, are in the radial growth phase. Vertical growth phase (VGP) An atypical melanocytic proliferation that expands as a spheroidal nodule in the dermis and is associated with the risk of subsequent metastatic disease. Clark used the term VGP to describe the focal appearance within the radial growth phase of a new clone of melanocytic cells, which is characterized by a distinct expansive growth preference, similar to that of a metastasis.

Nomenclature AKT AKT serine/threonine kinase ALK ALK receptor tyrosine kinase ARAF A-Raf proto-oncogene, serine/threonine kinase ARID2 AT-rich interaction domain 2 ATM ATM serine/threonine kinase BAP1 BRCA1 associated protein 1 BRAF B-Raf proto-oncogene, serine/threonine kinase BRCA1 BRCA1, DNA repair associated BRCA2 BRCA2, DNA repair associated CCND1 Cyclin D1 CDH1 Cadherin 1 CDH11 Cadherin 11 CDK4 Cyclin-dependent kinase 4 CDKN2A Cyclin-dependent kinase inhibitor 2A CHEK2 Checkpoint kinase 2 CIMP CpG island methylator phenotype CM Cutaneous melanoma CN Congenital nevus CRAF Raf-1 proto-oncogene, serine/threonine kinase CSD Cumulative solar damage CTNNB1 Catenin beta 1 CYSLTR2 Cysteinyl leukotriene receptor 2 DDX3X DEAD-box helicase 3, X-linked EGFR Epidermal growth factor receptor EIF1AX Eukaryotic translation initiation factor 1A, X-linked ERBB1 Erb-B2 receptor tyrosine kinase 1 ERBB2 Erb-B2 receptor tyrosine kinase 2 ERK Extracellular signal-regulated kinase EZH2 Enhancer of zeste 2 polycomb repressive complex 2 subunit FBXW7 F-box and WD repeat domain containing 7

414

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65095-X

Melanoma: Pathology and Genetics

FISH Fluorescent in situ hybridization FoxP3 Forkhead box P3 GAB2 GRB2 associated binding protein 2 GDP Guanosine diphosphate GNA11 G protein subunit alpha 11 GNAQ G protein subunit alpha q GTP Guanosine triphosphate HER2 Human epidermal growth factor receptor 2 HRAS HRas proto-oncogene, GTPase IDH1 Isocitrate dehydrogenase (NADP þ) 1, cytosolic IDO Indoleamine-2,3-dioxygenase IkB Inhibitor of kappa B KDR Kinase insert domain receptor KIT KIT proto-oncogene receptor tyrosine kinase KRAS KRAS proto-oncogene, GTPase LM Lentigo maligna LMM Lentigo maligna melanoma MAP2K1 Mitogen-activated protein kinase kinase 1 MAPK Mitogen-activated protein kinase MBN Melanoma arising in blue nevus MC1R Melanocortin-1 receptor MCN Melanoma arising in congenital nevus MEK MAPK/ERK kinase MET MET proto-oncogene, receptor tyrosine kinase MITF Microphthalmia-associated transcription factor mTOR Mechanistic target of rapamycin kinase NF1 Neurofibromin 1 NFKBIE NFKB inhibitor epsilon NF-kB Nuclear factor kappa-light-chain-enhancer of activated B cells NGS Next-generation sequencing NM Nodular melanoma NMSC Nonmelanoma skin cancer NRAS NRAS proto-oncogene, GTPase NTRK1 Neurotrophic receptor tyrosine kinase 1 NTRK3 Neurotrophic receptor tyrosine kinase 3 PALB2 Partner and localizer of BRCA2 PDGFRA Platelet derived growth factor receptor alpha PD-L1 Programmed death ligand 1 PI3K Phosphatidylinositol 3-kinase PIK3CA Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit alpha PIP2 Phosphatidylinositol 4,5-bisphosphate PIP3 Phosphatidylinositol 3,4,5-trisphosphate PLCB4 Phospholipase C beta 4 POT1 Protection of telomeres 1 PPP6C Protein phosphatase 6 catalytic subunit PREX2 Phosphatidylinositol-3,4,5-trisphosphate dependent Rac exchange factor 2 PTEN Phosphatase and tensin homolog PTPN11 Protein tyrosine phosphatase, nonreceptor type 11 RAC1 Rac family small GTPase 1 RAD50 RAD50 double strand break repair protein RAF1 Raf-1 proto-oncogene, serine/threonine kinase RB1 RB transcriptional corepressor 1 RET Ret proto-oncogene RGP Radial growth phase

415

416

Melanoma: Pathology and Genetics

RICTOR RPTOR independent companion of MTOR complex 2 ROS Reactive oxygen species ROS1 ROS proto-oncogene 1, receptor tyrosine kinase RTK Receptor tyrosine kinase SF3B1 Splicing factor 3b subunit 1 SNX31 Sorting nexin 31 SOS1 SOS Ras/Rac guanine nucleotide exchange factor 1 SPRED1 Sprouty related EVH1 domain containing 1 SSM Superficial spreading melanoma STK19 Serine/threonine kinase 19 TACC1 Transforming acidic coiled-coil containing protein 1 TCGA The Cancer Genome Atlas TERT Telomerase reverse transcriptase TET Tet methylcytosine dioxygenase TME Tumor microenvironment TP53 Tumor protein p53 UM Uveal melanoma UV Ultraviolet VEGFR2 Vascular endothelial growth factor receptor 2 VGP Vertical growth phase WT Wild-type WT1 Wilms tumor 1

Introduction The approach to the classification of melanoma has evolved from the traditional clinicopathologic to a novel molecular classification. Based on recent knowledge, it is generally agreed that melanoma definition cannot be exclusively confined to the cytoarchitectural morphology observed under light microscope, since melanoma molecular features are much more intricate, and genetics has a remarkable clinical impact for effective systemic therapy. Cutaneous melanoma is the malignant tumor with the highest mutation load and a prevailing cytosine to thymine nucleotide transition signature, which is attributable to UV radiation. While such a signature has been found to be a characteristic of melanoma arising on sun-exposed skin, acral and mucosal melanomas have been shown to have structural changes and mutational signatures not related to UV radiation damage. The most significantly mutated genes in cutaneous melanoma are BRAF, NRAS, CDKN2A and TP53, in acral melanoma NRAS and NF1, in mucosal melanoma SF3B1, and in 80%–90% of uveal melanomas GNAQ or GNA11, whose mutations are mutually exclusive and happen early in the pathogenesis. Overall, the most frequent mutations are those in noncoding DNA sequences and involve the TERT promoter. The TERT gene encodes the catalytic subunit of telomerase, and its transcriptional regulation is usually the limiting step in telomerase activity. Telomerase activity is usually silenced in normal tissues, causing telomeres to shorten with each successive round of cell division. The expression of telomerase is considered a hallmark of tumorigenesis, as over 90% of human cancers express the enzyme, including melanoma. TERT mutation results in cell immortalization by extension of the replicative lifespan of cells, allowing them to avoid replicative senescence in response to critical telomere shortening. In addition, whole-genome sequencing of melanoma has unveiled how different mutational pathways are implicated in melanoma pathogenesis contributing to the development of targeted therapies. The molecular perspective has developed as a reinterpretation of the clinicopathological classification on the basis of genotype– phenotype correlation. As proposed by Bastian, the taxonomy of melanocytic neoplasia is structured according to a dichotomous approach that differentiates lesions that originate from intraepithelial melanocytes, and are variably associated with UV radiation damage, from those lesions arising from nonintraepithelial melanocytes that can be uveal or nonuveal (brain, visceral organs, and proliferation in the dermis). While the first type of lesion is correlated with patient’s age, the second type does not show strict correlation with age distribution, with the exception of melanoma on congenital nevus, which occurs primarily in children. Melanomas can be considered as a group of biologically distinct categories showing differences in age and ethnic distribution, clinical and histological features, pattern of metastasization, etiopathogenetic role of UV radiation, primary oncogenic mutations, somatic alterations, and pathways of mutational mechanisms. Furthermore, analysis of The Cancer Genome Atlas (TCGA), mainly in metastatic melanoma dataset, has indicated that, independently of the oncogenic mutations (BRAF, NRAS, NF1), melanomas with immune signature have a better survival than those without immune signature. According to presence or not of an immune

Melanoma: Pathology and Genetics

417

signature, two major subsets of melanomas characterized by the presence or absence of a gene expression profile indicative of a preexisting T cell-inflamed tumor microenvironment (TME) have been described. The T cell-inflamed subset of tumors is characterized by a type I Interferon transcriptional profile, the presence of T cell transcripts, T cell recruitment through chemokines and macrophage activation markers. Accordingly, immunohistochemistry confirms the positivity for CD8 þ T cells, macrophages, and some B cells and plasma cells. While a subset of T cells specific for tumor antigens is present in tumor microenvironment, functional analysis indicates various degrees of dysfunction of these T cells due to immune suppressive mechanisms acting at the level of the TME. Transcripts encoding IDO, PD-L1, and FoxP3 and positivity for PD-L1 and IDO protein expression, and also nuclear FoxP3 þ CD4þ cells by immunohistochemistry, are present within T cell-inflamed melanomas in the same region as CD8þ T cells. This immune adaptive resistance feature represented a clinical challenge recently addressed by the adoption of immune checkpoint inhibitors, which unleash the immune system against melanoma.

Genetic Features Cutaneous melanoma originates in melanocytes, whose biology and role in melanin production have been well studied and characterized. Melanin plays a role in the dispersion of UV radiation and in the absorption of reactive oxygen species (ROS); physiologically, the protein product of the TP53 gene plays a crucial role in melanocytes for the control of cell response to DNA damage related to increased UV exposure and ROS levels. In particular, the damage to DNA induced by UV rays determines specific mutational signatures, many of which are driven by TP53, which therefore define the frequent events of melanomagenesis in the skin. Considering data from TCGA and the results of the main initial studies with screening approaches based on mutational analysis of the entire exome/genome by next-generation sequencing approaches, cutaneous melanoma has been shown to possess a very high prevalence of somatic mutations, both in primary lesions anddat an even higher leveldin metastatic lesions (Fig. 1). The mean mutation rate of cutaneous melanoma, calculated by TCGA from whole-exome sequencing of 318 primary and metastatic melanomas originating from nonglabrous (hair-bearing) skin, has been estimated ranging from 18 to 21 mutations per megabase, which is among the highest rates reported for any type of cancer, with the exception of colorectal cancer carrying deficient mismatch repair (MMR) mechanisms. Approximately three-fourths (range, 69%–82%) of these genomic variations are represented by C > T substitutions, which is associated with another small fraction (about 4%–5%) of CC > TT transitions; all these variants were strictly dependent on a mutagenic effect of UV rays and typically cause the so-called UV signature. Overall, the mutation rate in melanomas

M U TAT I O N S

11%

19%

89%

81%

TOTAL CASES

PRIMARY MELANOMAS

9%

WILD-TYPE MUTATED

91%

METASTATIC MELANOMAS

C O P Y N U M B E R VA R I AT I O N S

55%

62%

54%

45%

38%

46%

TOTAL CASES

PRIMARY MELANOMAS

METASTATIC MELANOMAS

NORMAL DELETED/AMPLIFIED

Fig. 1 Mutations and copy number variations in primary and metastatic melanomas. Nine out of ten melanomas are genetically mutated. Copy number alterations are present in 45% of cases. The frequency of mutations and copy number variations increases significantly from primary to metastatic lesions; indeed, metastatic melanomas have a higher mutation burden.

418

Melanoma: Pathology and Genetics

arising on chronically sun-exposed skin was approximately five times higher compared to those on skin not subject to chronic damage by the sun. Finally, UV signature appears to be slightly more prevalent in metastatic lesions (about 85% of the total amount of detected mutations) than primary tumors (about 75%). About 10% of melanomas occur in families with increased recurrence of the disease, sometimes associated with the presence of other neoplastic diseases (such as pancreatic carcinoma or brain tumors). Approximately, 40% of these familial melanomas are due to hereditary transmission of mutations in high penetrance susceptibility genes, including cyclin-dependent kinase inhibitor 2A (CDKN2A; the most frequently mutated gene for predisposition to melanoma, located on chromosome 9p21), cyclin-dependent kinase 4 (CDK4; 12q14), breast cancer gene 1 (BRCA1)-associated protein 1 (BAP1; 3p21), telomerase reverse transcriptase (TERT; 5p15), protection of telomeres 1 (POT1; 7q31), and microphthalmia-associated transcription factor (MITF; 3p13dsusceptibility is quite exclusively associated with the intermediate risk variant pE318K). Carriers of BAP1 mutations show an intermediate risk for cutaneous melanoma (CM), which becomes much higher in families with coexistence of CM cases and uveal melanoma (UM): 3% in UM cases (population based), 15% in CM families, and 28% in CM–UM families. Overall, carriers of germline mutations in CDKN2A/CDK4 and BAP1 genes may present an increased predisposition to either cutaneous melanoma (for BAP1, also to other melanoma subtypes such as UM; for CDKN2A/CDK4, also to higher prevalence of atypical melanocytic nevi), either to an increased incidencedthough with a penetrance markedly lower than melanomadof other types of cancer (particularly, pancreatic and renal carcinoma, brain tumor and mesothelioma). In about 60% of families with melanoma recurrence, the predisposition to the disease is probably due to the combination of variantsdboth mutations and functional polymorphisms (the latter often associated with specific environmental exposure conditions)din susceptibility genes with a lower penetrance. In other words, the increased predisposition to melanoma would seem to be linked to the inheritance of specific patterns of genetic alterations rather than mutations in single, stronger susceptibility genes (hence, the difficulty encountered so far in identifying additional melanoma-predisposing genes). In addition to the model of dominant inheritance described above, melanoma may represent a subordinate malignancy in the context of other tumor syndromes. Members of families with rare hereditary syndromesdsuch as Cowden syndrome (due to mutations in PTEN), Li Fraumeni syndrome (TP53 mutations), Lynch syndrome (inactivation of genes controlling DNA mismatch repair mechanisms) and xeroderma pigmentosum (mutations in multiple predisposing genes)dor relatively more common genetic conditions, such as hereditary breast and ovary cancers (mainly, mutations in BRCA1–2, but also mutations in ATM, CDH1, CHEK2, PALB2, PTEN, RAD50, TP53), present a poorly defined but higher risk of melanoma. All these observations strongly support the hypothesis that the pathogenesis of melanoma is complex and involves both genetic and nonhereditary or extrinsic factors.

Molecular Pathways Summarizing the data generated by the main NGS analyses on the whole-exome/genome, the molecular pathways involved in the melanoma pathogenesis depend on the following genes: BRAF (able to activate the pathway downstream, which includes two effector proteins MEK and ERK), NRAS (whose activation is able to promote the activity of downstream pathways: BRAF-MAPK and PI3K-AKT), NF1 (able to induce the activation of NRAS), CDKN2A (able to activate the pathways dependent on the two proteins encoded by this tumor suppressor gene: p16CDKN2A-CDK4-RB and p14CDKN2A-MDM2-p53 pathways), and PTEN (activating the PI3K-AKT-mTOR pathway). From a therapeutic point of view, besides BRAFV600, there are no alterations currently considered relevant in clinical decision making. Testing for NRAS or NF1 mutations is largely a matter of identifying patients who are potential candidates for clinical trials. Preliminary evidence supports the importance of CDKN2A and/or PTEN inactivation as mediators of acquired resistance to BRAF inhibitor-based therapy, but these alterations cannot be currently considered routinely in clinical decision making.

BRAF The RAF kinase family consists of three proteins (ARAF, BRAF, and CRAF), all of which are part of the signal transduction cascade named mitogen-activated protein kinase (MAPK) pathway. In melanoma, the BRAF gene is mutated in 45%–50% of cases; the most prevalent mutation (about 90% of cases) is represented by a substitution of a valine with glutamic acid at codon 600 (BRAFV600E). The remaining BRAF mutations often occur at the same codon (V600-K/D/R); therefore, mutations in codons other than V600 are not very common (among them, K601 is the most prevalent). The constitutive oncogenic activation of BRAFdthrough a process of phosphorylation of MEK and ERK, the effectors immediately downstream of the MAPK pathwaydpromotes a continuous, uncontrolled stimulation of cell proliferation. There is an inverse relationship between BRAF mutation prevalence and age-decade. The majority of patients < 30 years and 25%  70 years harbor a BRAF-mutant melanoma. Among BRAF-mutant melanomas, the frequency of non-V600E genotypes (including V600K) raises with increasing age. Non-V600E genotypes have been found in < 20% of patients < 50 years and > 40% in those  70 years. A higher degree of cumulative sun-induced damage correlates with V600K but not V600E melanoma. Melanomas that carry BRAF mutation are characterized by an enrichment of alterations in PTEN, corroborating the frequent coexistence of BRAF and PTEN mutations in melanoma. At the same time, a large fraction of melanomas with mutated BRAF exhibits a reduced or absent expression of p16CDKN2A or p53 proteins, due to the coexistence of mutations inactivating the corresponding genes (CDKN2A and TP53, respectively). The demonstration that BRAF is mutated

Melanoma: Pathology and Genetics

419

in a large fraction (65%–80%) of common nevi suggests that its oncogenic activation is a necessary but not sufficient condition for the development of melanoma (in fact, it is considered an initiation event in melanocyte transformation). Therefore, the onset of alterations in other genes able to cooperate with BRAF seems to act as the main mechanism underlying the transformation and neoplastic progression in the melanocytic lineage. It has finally been demonstrated that some genetic variants of the melanocortin-1 receptor membrane receptor (MC1R) are not able to adequately stimulate the production of melanin protecting the skin from the damage of UV rays. The combination of the presence of such variants of MC1R and exposure to UV rays (especially, when intermittent) was also considered responsible for the oncogenic activation of BRAF in this specific subgroup of melanomas.

RAS The products of the RAS family are composed of three tissue-specific isoforms: HRAS, KRAS and NRAS. NRAS mutations are found in 20%–25% of melanoma cases; again, they occur almost exclusively in a single gene codon (Q61, about 90% of cases); in the remaining 10% of cases, the mutated codon is G12 or G13. The oncogenic stimulation of RAS is able to activate specific cytoplasmic downstream proteins with kinase function: RAF and phosphatidylinositol 3-kinase (PI3K). RAS mutations have been demonstrated to be mutually exclusive with BRAF mutations in nearly all cases (coexistence of the two genes mutated in a constitutive manner are reported in a fraction of cutaneous melanomas, < 2%–3% of cases). It has been observed that, in the presence of NRAS activation (both for the acquisition of mutations or functional oncogenic induction), the translation of the mitogenic signal in the MAPK pathway can be switched to CRAF, which therefore acquires a key role in maintaining cell proliferation stimulation in this subset of melanomas. Interestingly, an increased activation of CRAF, as well as the presence of NRAS mutation, have been described as responsible for the acquired resistance to BRAF inhibitors. In addition to BRAF, genetic alterations (mutations and gene copy number changes) in other important components of pathways downstream of RAS that control cell proliferation and survivaldincluding PIK3CA (altered in 5% of cases) and AKT3 (8%)d occur at lower frequency. In fact, a vast majority of the PI3K-PTEN/AKT genes downstream of RAS are functionally silenced in melanoma. In this pathway, an exception is represented by a markedly higher frequency of mutations (about one fifth of melanomas in NGS studies) in the PREX2 gene regulating PTEN. A limited fraction of melanomas that are not mutated in BRAF and RAS may carry activating mutations in KIT, a tyrosine kinase receptor of the cell membrane, resulting in a continuous induction of cell proliferation, through functional stimulation, mainly of the MAPK pathway. In particular, KIT mutations have been described in acral melanomas (about 10% of cases) and in those from chronically sun-exposed skin areas (5% of cases) among cutaneous melanomas, showing, however, the highest prevalence in mucosal melanomas, particularly in anal melanomas (15%–20% of cases).

NF1 Germline mutations in NF1 cause an inherited multisystem genetic disorder, neurofibromatosis type 1, which is characterized by changes in skin coloring pigmentation, as well as the growth of both benign and malignant tumors. In preclinical models, NF1 mutation suppresses BRAF-induced senescence in melanocytes, promoting melanocyte proliferation and enhancing melanoma development. The Cancer Genome Atlas has also identified a subset of melanomas (approximately 15% of cases) in which activation of the MAPK pathway is due to inactivating mutations of the NF1 gene. Physiologically, NF1 encodes for neurofibromin, a RASGTPase-activating protein, which negatively regulates RAS signaling by facilitating hydrolysis of RAS-GTP to the RAS-GDP-inactive form; therefore, mutations functionally silencing NF1 result in RAS activation. In this perspective, the activation of several RASopathy genesdsuch as SOS1, PTPN11, RAF1, and SPRED1dappears to occur more frequently in subtypes of melanomas with NF1 mutations. At the same time, these data are a clear demonstration that NF1 cooperates with other RASopathy genes in melanomagenesis. Although the mutations of NF1 are the most prevalent alterations in the group of BRAF and RAS wild-type melanomas (about two-fifths of these cases, which, however, represent 4%–5% of the total cutaneous melanomas), they are present at almost the same frequency in BRAF- and RAS-mutated melanomas (3%–4% of cases). NF1 is a key driver of melanoma, and this is suggested by three biological features: (i) a high frequency of nonsilent exonic mutations, (ii) a low frequency of synonymous or intronic mutations, (iii) it occurs in a considerable portion of studied melanoma specimens. Melanomas with NF1 mutations generally occur on chronically sun-exposed skin or in elder individuals and show a high mutation burden. Additionally, NF1 mutations characterize certain clinicopathologic melanoma subtypes, specifically desmoplastic melanoma. Unlike BRAF-mutated melanomas, those with the NF1 subtype have a stronger correlation with the UV-induced mutagenesis. Overall, occurrence of mutations in NF1, alone or associated with BRAF or RAS mutations, induces a higher mutational load and, consequently, a greater probability of generating neoantigens. Due to this increased antigen capacity, tumors with mutations in NF1 and/or constitutive activation of RASopathy genes are thought to be addressed to immunotherapeutic treatment with immune checkpoint inhibitors. In several melanoma cell lines, there is evidence that NF1 ablation decreases the sensitivity of melanoma to BRAF inhibitors. NF1 mutations are present in BRAF-mutant tumors intrinsically resistant to BRAF inhibitors, as well as in melanomas of patients showing resistance to BRAF inhibitors. Correspondingly, NF1/BRAF-mutant murine tumors are resistant to BRAF inhibitors but sensitive to combined inhibition of MAPK/ERK and mTOR pathways.

420

Melanoma: Pathology and Genetics

Triple Wild-Type Melanomas According to TCGA classification, cutaneous melanoma can be categorized into four genetic subgroups on the basis of MAPK driver mutations: BRAF, RAS (N-H-K), NF1 (lacking a BRAF p.V600 or RAS p.G12, G13 and Q61 hotspot mutation) and Triple wild-type (WT) melanomas. Triple wild-type melanomas represent a genetic distinct entity for several reasons: (i) only  30% of Triple WT melanomas harbor a UV signature compared with over 90% considering the other three molecular subtypes; (ii) somatic copy number analysis shows that Triple WT melanomas comprise a significant number of cases with copy number amplifications; amplicons include the 4q12 minimal common region containing KIT, PDGFRA and KDR (also known as VEGFR2), as well as amplifications in loci encompassing TERT, CDK4 and CCND1; (iii) Triple WT melanomas have more complex structural rearrangements and candidate fusion drivers, although few recurrent fusions have been described so far.

CDKN2A The CDKN2A gene encodes two proteins: p16CDKN2A and p14CDKN2A. It acts as tumor suppressor gene in a recessive manner; inactivation of both alleles is necessary for the development of melanoma. In patients with melanoma familiarity, 20%–40% of probands may carry germline mutations in CDKN2A. At the somatic level, about two-thirds of melanoma patients instead present a CDKN2A gene inactivation; these melanomas seem to be resistant to BRAF inhibitors, or to BRAF and MEK inhibitors. In particular, activation of the downstream CDK4-RB effectors through inactivation of p16CDKN2A seems to be correlated with disease progression; indeed, its prevalence significantly increases during transition from primary tumors to metastatic melanomas and melanoma cell lines. Similarly, the inactivation of p14CDKN2A causes the reduction of the p53 protein levels, with consequent impairment of the cell-cycle progression control and inhibition of the apoptosis, contributing to increase the aggressiveness of tumor cells and their refractoriness to therapy. Activation of the CDKN2A-dependent pathway may also be associated with the amplification of the CyclinD1 (CCND1) gene, which is generally found in melanomas negative for BRAF and NRAS mutations. However, in a fraction (about 15%) of mutated BRAF melanomas, the coexistence of CCND1 amplification seems to confer resistance to BRAF inhibitors.

PI3K-PTEN In addition to MAPK, the second cell regulating pathwaydmainly dependent on RAS activationdconsists of the PI3K protein pathway which is characterized by signal transduction through the PTEN, PI3K, and AKT (in particular, the AKT3 isoform in melanoma) protein cascade. Oncogenic activation of the PI3K pathway can occur through several mechanisms, including: mutation and/ or amplification of genes encoding RTKs (e.g., EGFR (ERBB1) and HER2 (ERBB2)), subunits of PI3K (e.g., p110a, p110b, p85a, and p85b), AKT (AKT1), or activating isoforms of RAS. Within the PI3K/AKT/mTOR pathway, AKT plays a major role, being at the crossroad and implicated in the oncogenic mutation, the immunotolerance, and the developing of de novo and acquired treatment resistance in various tumor types treated with targeted therapies. Furthermore PTEN (encoding phosphatase and tensin homolog, the major PIP3 phosphatase) is frequently mutated. In physiological conditions, the phosphatase activity of the PTEN protein reduces the intracellular level of the PIP2 and PIP3 phosphoinositoles, which are produced by the activation of PI3K, in order to regulate the activation level of downstream AKT and its mTOR substrate, thus modulating the synthesis of proteins involved in apoptosis and cell survival. In melanoma, the PTEN gene is deleted in about a third of cases, with complete loss of expression of the corresponding protein in 10%–20% of primary melanomas; the level of this loss increases during neoplastic progression, up to 40%–50% in melanoma cell lines. The combined effect of the PTEN loss and the PI3K-AKT pathway activation results in aberrant growth of melanoma cells and increased survival capacity with the acquisition of apoptosis resistance. All of these findings clearly indicate that there are distinct molecular subtypes of melanoma; the well-characterized ones are summarized in Table 1. From a practical point of view, the characterization of these subtypes becomes extremely important for an increasingly correct therapeutic approach and an appropriate definition of clinical–biological behavior in patients with melanoma.

Epigenetic Events Alongside the genetic alterations, data are emerging that increasingly support a role of epigenetic events in the development and progression of melanoma. The latter are defined as inheritable alterations in gene expression without a change in the DNA sequence. Main epigenetic modifications include: - posttranslational modifications of histones and remodeling of nucleosomal complexes; - methylation of cytosine–guanine (CpG) dinucleotides from accessible DNA; - gene silencing by noncoding RNAs, either short or long, whose expression can in turn be directly regulated through epigenetic modifications of the corresponding genes.

Melanoma: Pathology and Genetics

421

Main molecular subtypes of melanoma: frequent mutations and rearrangements

Table 1 Subtype

Most frequently mutated genes ( 10%)

BRAF mutant

TP53, CDKN2A, PTEN, ARID2

RAS mutant

CDKN2A, TP53, ARID2, NF1, PPP6C

BRAFneg/RASneg mutant

NF1, TP53, ARID2, RAC1

Less frequently mutated genes ( 40 cm). The size corresponds to the largest diameter that can be expected in adulthood (predicted adult size or PAS) because congenital nevi extend in proportion to the child’s growth. Risk factor for the development of melanoma is the lesion size: in giant congenital nevus, the risk increases up to 10%–15%. In pre-pubertal patients with large congenital nevus, the risk for the development of melanoma is 1%–2%, which is ten-thousand times more frequent than same-aged general population. Large congenital nevi have a 2%–42% risk of malignant transformation and a 6%–14% lifetime risk of developing melanoma. MCN presents as a nodular growth in the context of congenital nevus, though rapidly growing nodules are more often mimickers of melanoma, including atypical

424

Melanoma: Pathology and Genetics

proliferative nodules, as they have a benign behavior. MCN arises within the cutaneous lesion intradermally or subcutaneously before puberty, at the dermo-epidermal junction in adulthood. Gain-of-function somatic mutation in NRAS at codon 61 with the substitution of glutamine by lysine (Q61K) or by arginine (Q61R) have been found in > 90% of large and giant CN, and 70% of small and medium CN. In contrast to large-giant CN, small-medium ones harbor in 30% of cases BRAFV600E mutation.

Melanoma Arising in Blue Nevus Blue nevus is a frequent dermal melanocytosis. The name derives from the color blue that results from the Tyndall effect. The two most important variants are common blue nevus and cellular blue nevus. The common blue nevus consists of an exclusively dermal proliferation and it most frequently arises on the dorsal surface of the hands, albeit ubiquitous. The cellular blue nevus is a proliferation of spindled melanocytes arranged in bundles involving reticular dermis and hypodermis. MBN usually arises on the cellular blue nevus variant and appears as a dermal proliferation of large atypical spindled and/or epithelioid melanocytes with moderate to severe cytologic atypia and necrosis. MBN has conceptual affinity with uveal melanoma, as both originate from melanocytes that are not intraepithelial, and it is not a coincidence that their genetic mutations extensively overlap. Indeed, melanomas arising on blue nevus show GNAQ and GNA11 gain-of-function mutations (Q209 in Exon 5). Additional mutations have been described in SF3B1 and BAP1. Chromosomal aberrations, such as gains in chromosome 6p and/or 8q, monosomy of 3, and loss in 1p, are similar to those described in uveal melanoma mutational pathway.

Uveal Melanoma Uveal melanoma, the most common intraocular tumor in adults, arise from melanocytes in the uvea, the middle vascular layer of the eye consisting of three parts: iris, ciliary body, and choroid. Uveal melanomas are not related to sun exposure, as demonstrated by the absence of UV signature. The mutation load is lower than CSD cutaneous melanoma. In 80%–90% of uveal melanomas, there are GNAQ or GNA11 mutations which are mutually exclusive and happen early in the pathogenesis. Other mutations involving the Gaq pathway include PLCB4 and CYSLTR2. BAP1 mutation, which affects up to 50% of cases, SF3B1 (18%), and EIF1AX occur while the tumor progresses, and correlate with different risks of metastasis. Chromosome 8 and 3 can be studied via FISH for additional information on prognostication, since monosomy of chromosome 3 and isochromosome 8q correlate with poor prognosis with higher risk of liver metastasis.

Prospective Vision Melanoma research is now moving toward a holistic vision, defining melanoma with specific histopathologic and genetic features in a unique and complex individual, aimed at achieving the most effective treatment and the best outcome. Genotype–phenotype correlation is a key lever in that sense. The efforts in cataloging the genomic architecture of different subtypes of melanoma have led to the study of a distinct set of genetic alterations, even in noncoding regions, and such developments open new frontiers in our understanding of melanoma pathogenesis. High-throughput technologies have revolutionized melanoma research, and will continue to do so, especially in defining the genetic progression of melanoma subtypes from their precursors, as well as characterizing metastatic melanoma polyclonality.

Acknowledgments This work was supported by Italian Melanoma Intergroup (IMI), the Italian network for melanoma treatment and research (www.melanomaimi.it), MIUR COFIN (2015 2015HAJH8E), and Fondazione Ente Cassa di Risparmio di Firenze.

Further Reading Bastian, B.C., 2014journal. The molecular pathology of melanoma: An integrated taxonomy of melanocytic neoplasia. Annual Review of Pathology: Mechanisms of Disease 9, 239–271. Clark Jr., W.H., Ainsworth, A.M., Bernardino, E.A., et al., 1975journal. The developmental biology of primary human malignant melanomas. Seminars in Oncology 2 (2), 83–103. Clark Jr., W.H., Elder, D.E., Van Horn, M., 1986journal. The biologic forms of malignant melanoma. Human Pathology 17 (5), 443–450. Curtin, J.A., Fridlyand, J., Kageshita, T., et al., 2005journal. Distinct sets of genetic alterations in melanoma. New England Journal of Medicine 353, 2135–2147. Girotti, M.R., Gremel, G., Lee, R., et al., 2015journal. Application of sequencing, liquid biopsies, and patient-derived xenografts for personalized medicine in melanoma. Cancer Discovery 6 (3), 286–299. Hayward, N.K., Wilmott, J.S., Waddell, N., et al., 2017journal. Whole-genome landscapes of major melanoma subtypes. Nature 545 (7653), 175–180. Hodis, E., Watson, I.R., Kryukov, G.V., et al., 2012journal. A landscape of driver mutations in melanoma. Cell 150 (2), 251–263. Leachman, S.A., Lucero, O.M., Sampson, J.E., et al., 2017journal. Identification, genetic testing, and management of hereditary melanoma. Cancer Metastasis Reviews 36 (1), 77–90. Liu, F., Wang, L., Perna, F., Nimer, S.D., 2016journal. Beyond transcription factors: How oncogenic signaling reshapes the epigenetic landscape. Nature Reviews. Cancer 16 (6), 359–372.

Melanoma: Pathology and Genetics

425

Palmieri, G., Ombra, M.N., Colombino, M., et al., 2015journal. Multiple molecular pathways in melanomagenesis: Characterization of therapeutic targets. Frontiers in Oncology 5, 183. Shain, A.H., Yeh, I., Kovalyshyn, I., et al., 2015journal. The genetic evolution of melanoma from precursor lesions. New England Journal of Medicine 373, 1926–1936. The Cancer Genome Atlas Network, 2015journal. Genomic classification of cutaneous melanoma. Cell 161 (7), 1681–1696. Whiteman, D.C., Pavan, W.J., Bastian, B.C., 2011journal. The melanomas: A synthesis of epidemiological, clinical, histopathological, genetic, and biological aspects, supporting distinct subtypes, causal pathways, and cells of origin. Pigment Cell & Melanoma Research 24 (5), 879–897. Woodman, S.E., Lazar, A.J., Aldape, K.D., Davies, M.A., 2012journal. New strategies in melanoma: Molecular testing in advanced disease. Clinical Cancer Research 18 (5), 1195–1200. Zhang, T., Dutton-Regester, K., Brown, K.M., Hayward, N.K., 2016journal. The genomic landscape of cutaneous melanoma. Pigment Cell & Melanoma Research 29 (3), 266–283.

Metastatic SignaturesdThe Tell-Tale Signs of Metastasis Christina Ross and Kent Hunter, National Institutes of Health, Bethesda, MD, United States © 2019 Elsevier Inc. All rights reserved.

Introduction Metastasis is defined as the spread of a disease from one organ, or “primary site”, to another non-connected “secondary site”. In the context of cancer, metastasis refers to the process of a single cell leaving the original “primary” tumor, traveling through the circulatory or lymphatic system, and invading into a distant organ to create a second tumor or metastasis. For a cancer cell, this is an incredibly challenging multi-step feat that most cannot achieve. Indeed, while millions of cells are shed from a tumor every day, very few clinically detectable metastatic colonies are usually formed. Briefly, the steps of metastasis are as follows: 1. 2. 3. 4.

Cells separate from the primary tumor Tumor cells invade through the basal membrane into a blood or lymphatic vessel. Tumor cells intravasate into the circulatory or lymphatic system Tumor cells travel throughout the body as circulating tumor cells (CTCs) until halting at a distant organ, at which point they are referred to as disseminated tumor cells (DTCs) 5. Tumor cells extravasate from the vessels into the secondary organ and re-initiate proliferation

Early Work Defining the Metastatic Process Metastases have been observed in mummies estimated to be around 2400 years old, and have been described in manuscripts as far back as 200 CE. However, the term metastasis was not coined until 1829 by French surgeon Joseph Récamier. It was taken from the Greek meaning removal, migration, change or revolution. At the time, very little was understood about cancer, but Récamier was considered an expert on breast malignancy. In his book “Sur le traitment de cancer,” one of the earliest medical texts on oncology, he describes the suffering of cancer patients of the day and in particular, several breast cancer patients with metastasis to the brain. In his text, Récamier was the 1st to propose that metastasis was due to the translocation of cells by local infiltration at the primary site, invasion of the veins, and distant growth in the brain. The theory of metastasis continued to take shape through the 1800s with the work of Rudolf Virchow, who in 1858 suggested that metastatic tumor dissemination to secondary sites was determined largely by mechanical factors. More specifically Virchow hypothesized that circulating tumor cells somewhat randomly became arrested at a distant organ because they were stuck in the vasculature. However, in 1889 Stephen Paget proposed a slightly different theory in his Lancet paper “Distribution of secondary growths in cancer of the breast”. Paget, who is now referred to as the father of metastatic theory, analyzed 735 fatal cases of breast cancer and argued the distribution of metastases could not be due to chance alone. The patterns he observed did not simply follow blood flow distribution as suggested by Virchow. Instead Paget’s theory referred to tumor cells as “seeds”, and stated that their distribution throughout the body must be understood by studying the properties of the “soil” of distant organs. Paget drew parallels with botany, stating that seeds can only live and grow if they fall on congenial soil. While this theory is currently widely accepted, it was challenged as recently as 1928 by James Ewing who theorized that location of metastases was primarily determined by the anatomy of the vascular and lymphatic channels that drain the primary tumor. These theories were ultimately merged and supplemented by several metastasis studies published by Isaiah Fidler. His groundbreaking studies demonstrated that metastasis occurred in a series of sequential steps that involve cancer cells with different metastatic capabilities. Dr. Fidler suggested that different “seeds” interact with their microenvironment, leave the primary tumor, and ultimately either land in a “congenial soil” or perish. Dr. Fidler’s work also showed that tumor cells could leave the primary tumor by entering the bloodstream or the lymphatics. This theory of the metastatic process took into account Virchow, Paget and Ewing’s work and suggested that successful cancer clones metastasized in a stochastic manner. It also hinted at the notion that while many cancer cells leave the primary site, very few are successful at creating a metastasis, as cells with differing characteristics are selected for or lost along the way. Today’s current understanding of the metastatic process is built upon the observations and studies of clinicians and researchers going back centuries, and while considerable strides have been made in the last few decades, metastatic disease remains an incredible challenge to treat.

Current Clinical Observations and Challenges in Treating Metastasis Metastasis is the primary clinical challenge in the treatment of cancer as it is unpredictable in onset and it exponentially increases the clinical impact to the patient. It is a key element in cancer staging systems such as the tumor/node/metastasis (TNM) staging system, where metastasis places a cancer in Stage IV. The possibilities of curative treatment are greatly reduced, or often entirely removed, when a cancer has metastasized. For example, in breast cancer approximately 20–50% of patients diagnosed at an early stage are expected to develop metastatic disease, and because the treatment of primary tumors is quite good, around 90% of breast cancer deaths in the U.S. are instead due to untreatable metastases. It remains a major challenge to eliminate, identify, or remove metastases, and there are several factors that have led to this disparity in curative treatments between localized and metastatic tumors.

426

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65027-4

Metastatic SignaturesdThe Tell-Tale Signs of Metastasis

427

Metastatic cancer is multifaceted and therefore complex to study and treat. Firstly, not all cancers metastasize, but those that most commonly do - lung, breast, skin (melanoma), colon, kidney, prostate, pancreas, and liver – don’t all metastasize to the same distant organs. For example, breast cancer usually spreads to the brain, bone, or lungs, yet colorectal cancer will primarily spread to the liver. Furthermore, the sensitivity of clinical imaging techniques cannot always expose small metastatic lesions. Another key characteristic that confounds treatment is the drug resistance often observed in metastatic lesions. The current treatments available target the biology of the primary tumor, for which ample patient tissue is available for study. Metastatic tissue is often not resected from patients and therefore not widely available to study; because of this the current therapies are often unable to eliminate the metastatic cells which can be significantly altered during the metastatic process. Finally, another major challenge in the treatment of metastatic cancer is cancer cell dormancy. Once shed from the primary tumor, dormant cancer cells cease dividing but survive in a quiescent state, often in the bone or other organs, waiting for appropriate environmental conditions to trigger proliferation again. This period of quiescence can last from months to many years. Dormancy creates a large hurdle for treatment as individual dormant cells cannot be detected in the clinic and most therapies target proliferating cells. Furthermore, the creation of drugs to target cells in this state is incredibly difficult as the outcome of effectively targeting dormant cells may not be measurable for upwards of 10 years and may only be recognized as the absence of further metastasis. Ultimately, in order to improve patient outcomes, clinicians must be able to identify the spread of cancer or a patient’s potential for metastasis early in their disease. The identification and use of such “metastatic signatures” will be described in this chapter.

Tumor-Intrinsic Signatures of Metastasis Gene Expression Profiling of the Primary Tumor Molecular profiling of tumors following patient surgery has enabled oncologists to categorize tumor types beyond the organ of origin to several subtypes, ultimately guiding physicians’ prognoses and treatments. A major factor in this subcategorization has been the assessment of somatic gene mutations, amplifications, and tumor-specific gene expression patterns. Interestingly, epidemiological studies combining the genetic analysis of patient tumors with patient outcomes has also revealed which subtypes, and which genetic alterations in the primary tumor, are likely to result in metastatic disease. The categorization of breast cancers into subtypes based on gene expression is a prime example of using primary tumor characteristics to predict patient outcomes. Breast cancer subtypes have been extensively categorized by the expression levels of the estrogen and progesterone hormone receptors, and cell surface marker HER2. The 4 major subtypes of breast cancer are Luminal A which expresses both hormone receptors but is negative for HER2, Luminal B which expresses the hormone receptors and either HER2 or Ki-67, Triple negative/Basal-like which is hormone-receptor negative and HER2 negative, and finally HER2-enriched tumors which are hormone-receptor negative but HER2 positive. Analysis published by the American Society of Clinical Oncology has revealed that the Luminal A type has the lowest rate of metastasis, while Luminal B HER2 positive has the highest. When metastasis does occur, subtypes can also be used to predict patterns of metastatic spread. For example, high rates of brain metastases are seen among HER2-enriched, basal-like, and Triple negative types, whereas brain metastases are less frequently seen in the luminal/ HER2 groups. Interestingly, bone is the predominant site of metastasis for the luminal/HER2 groups but the least common site for the basal subtype. In addition to using the expression of specific markers, primary tumor metastatic propensity can be evaluated by different gene expression programs. With recent advances in technology, it has become possible to compare gene expression patterns between primary tumors and metastases. Over the last decade microarray and RNA sequencing studies of patient tissue, mouse metastasis models, and cell lines have informed numerous biomarker panels that can be used to predict metastasis and patient outcome. Several cancer gene signature tests are used clinically to guide personalized treatment decisions for breast, colon, or prostate cancer. Such panels can typically assess expression of 10–70 genes that span cellular pathways critical for cell adhesion, migration, cell-cell signaling, differentiation, and metabolism. As well as creating robust biomarker panels that can direct treatment for individual cancer types, there is also an ongoing effort to create multi-cancer gene expression biomarker panels for cancer metastasis. Many studies have demonstrated that microRNAs, and the altered expression of specific microRNAs, may play essential roles as either promoters or suppressors of metastatic progression. By altering the expression of networks of genes, microRNAs can be powerful determinants of pathway activation and cellular phenotype. For example, by down regulating a specific network of target transcripts, miR-155 promotes the dissolution of tight junctions via TGFb-induced cell plasticity and RhoA suppression, ultimately increasing cell migration and metastasis. In addition, sustained expression of miR-155 through positive feed-back of TGF-b signaling has been associated with dysregulation of cell polarity in solid tumors. Conversely, miR-29b has previously been reported as a metastasis suppressor. miR-29b targets transcripts encoding prometastatic effector molecules involved not only in angiogenesis and collagen remodeling such as VEGFA, ANGPTL4, PDGF, and MMP9, but also in differentiation and plasticity such as ITGA6, ITGB1, and TGF-b. miR-29b is enriched in luminal breast cancers and loss of miR-29b increases metastasis and promotes a mesenchymal phenotype. As well as contributing to each tumor cell’s metastatic capacity, microRNAs can also function to induce tumor cell non-autonomous alterations. In section “Systemic Signatures of Metastasis” of this chapter we will discuss, microRNA involvement in the preparation of the distant pre-metastatic niche.

428

Metastatic SignaturesdThe Tell-Tale Signs of Metastasis

Somatic Mutations of the Primary Tumor Several high-quality genomic studies have explored the genetic relationship between primary and metastatic lesions in order to understand metastatic evolution. It is now well accepted that metastases arise from a clone of unique cells within the primary tumor. DNA copy number analysis of metastatic prostate cancers has indicated a common clonal origin in most cases, although sub-clonal alterations have also been observed in metastases. Whole-genome sequences of a basal-like primary breast tumor and corresponding brain metastases have consistently shown that while copy number alterations and overall mutational spectra within the genome are not significantly different, specific mutations revealed a subset of cancer cells from the primary tumor were preferentially enriched in metastatic lesions. Genomic sequencing analyses of pancreatic cancer lesions also suggested that metastatic lesions are clonal in nature, but also likely require additional driver mutations that are not found in the primary tumor. Phylogenetic analysis of metastatic progression in breast cancer using somatic mutations and copy number aberrations (CNAs) has shed yet further light on the timing of the evolution of metastases. In agreement with earlier studies, descent of metastases from a common origin was observed in patients diagnosed with early stage breast cancer, however multiple seeding events from the primary tumor were observed in patients diagnosed with advanced stage disease. Interestingly, the number of late single nucleotide variants and CNAs increased as distant metastases evolved and can therefore give an indication of the time elapsed since they diverged from their common ancestor. Not surprisingly, there is a correlation between overall patient survival and the phylogenetic branch lengths of the metastases. This phylogenetic analysis ultimately revealed that metastases from patients with longer cancer histories are genetically more distant from their primary tumor tissue of origin than those of patients with a shorter cancer history. Despite a large number of studies into the evolution of metastatic tumor cells, no specific driving mutations of metastasis have been identified. However, there are several tumor progression-promoting genetic alterations that often result in more aggressive and metastatic disease. In the case of metastatic colorectal cancer, associations have been made between mutational activation of the KRAS, BRAF, PIK3CA, and NRAS oncogenes and metastatic patterns in patients. The presence of a KRAS mutation is associated with decreased liver metastases and increased lung, brain, and bone metastases. BRAF mutation, a poor prognostic factor, is associated with decreased liver-limited metastasis and increased peritoneal and distant lymph node metastases. Conversely, PIK3CA and NRAS mutations do not clearly affect outcomes in the metastatic setting, although PIK3CA is associated with concurrent KRAS mutations. Our expanding understanding of how somatic mutations correlate with metastatic patterns contributes to the clinical care of patients by focusing treatment and guiding surveillance strategies for metastasis.

Inherited Predisposition to Metastasis The inherited basis for many cancers has been thoroughly studied. For example, inherited predisposition to breast and ovarian cancer with a BRCA1 mutation is commonly assessed in the clinic. However, the inherent diversity in the host genetic background has also been shown to influence metastatic risk. In 1998, a pioneering study showed that an identical oncogenic event in the mammary tissue of mice with different genetic backgrounds led to the development of mammary tumors with similar primary tumor properties but with distinct propensity for metastasis. Linkage analysis has further mapped several genomic loci that modify metastatic potential in mice. With recent advances in global transcript analysis, several studies have now demonstrated the influence of germline polymorphisms and gene networks on the susceptibility to metastatic disease. Inherited predisposition to metastasis often impacts metastasis in a manner specific to cancer type and subtype. However, recent work suggests that inherited variants can function to modify metastasis by tumor cell-autonomous and non-autonomous mechanisms. Consider that while a patient may have inherent specific expression of genes that benefit tumor cell proliferation or survival, they may also harbor variants in immune surveillance or metabolism pathways that alter the tumor cell’s road to the secondary site. In a series of genetics studies in mice, several inherited tumor-intrinsic factors have been identified as metastasis modifiers. One such factor is human circadian rhythm gene Arntl2, for which polymorphisms associated with its expression correlate with metastatic prognosis of ER- breast cancer patients by a tumor intrinsic mechanism. Ultimately, future and ongoing studies will extend this area of research to human patients with the application of genome-wide association study (GWAS) methods. The goal of such studies is to enable early screening of patients for inherited metastatic modifying genes, and develop targeted therapies for specific groups.

Epigenetic Signatures of Metastasis The study of genetic lesions alone has been unable to explain the complexity of the aberrations that arise in a cancer cell. It is now recognized that cancer is both a genetic and an epigenetic disease. Epigenetics is defined as the inheritance of changes in gene activity that are independent of DNA sequence. The two main epigenetic events involved in gene regulation, development, and carcinogenesis are histone modifications and DNA methylation. In cancer cells, the fine control of epigenetic mechanisms is lost and the disruption of epigenetic patterns promotes the expression of the tumoral phenotype. The alteration of epigenetic marks on the genome is a key tool in successful metastasis, as these marks are reversible and allow the cell to adapt as it travels to the metastatic site. A noteworthy example of such a phenomenon is the E-Cadherin gene, which is silenced in some human cancers due to DNA hypermethylation. It has been shown that in some primary tumors displaying E-cadherin hypermethylation, the corresponding metastases are unmethylated at the E-cadherin gene and regain gene expression at this locus. Demethylation of E-cadherin and re-expression are thought to be indispensable for the incorporation of metastatic cells into their new cellular environment.

Metastatic SignaturesdThe Tell-Tale Signs of Metastasis

429

Recent work in metastatic pancreatic cancer has provided tremendous insight into the epigenetic changes that arise in metastases that seed different sites. Using matched primary and metastatic samples, researchers revealed differences in epigenetic reprogramming within subclones of the primary tumor that seeded local (peritoneal) and distant (liver and lung) metastases. Differences in the epigenetic reprogramming between local and distant metastatic lesions were also observed. Interestingly, no significant changes were seen in the tumors of patients whose cancer spread only locally. However, tumors from patients with distant metastases showed substantial epigenetic changes both in the distant metastases and in the subclone of the primary tumors from which they arose. The genes impacted by these epigenetic alterations in lung metastases were associated with oxidoreductase activity as well as DNA damage, rendering corresponding cell cultures resistant to oxidative stress and chemotherapy. Also, genes encoding epithelial and mesenchymal identity were reciprocally expressed between the local and distant metastases, such that peritoneal metastases were epithelial and lung metastases were poorly differentiated. This work suggests that epigenetic regulation by metastatic cells is paramount to reaching and growing secondary tumors in specific organs. It also highlights the complexity of metastatic cancers and the multifaceted treatment strategy that must be implemented to eradicate it.

Metabolic Changes in Metastatic Cells It has long been appreciated that metabolism is altered in cancer cells. In 1955, Otto Warburg presented his work showing the nuanced manner in which cancer cells utilize glucose and oxygen. Since this discovery, metabolism in cancer has been thoroughly studied and the conditions to which tumor cells can adapt and thrive by metabolic modifications is wide and varied. In recent years, further specialization of metabolic changes has been observed during tumor cell dissemination into the circulation and establishment of metastatic lesions. The dissemination of malignant cells is an integral characteristic of advanced stages in transformation. While contributions of peroxide signaling to metastasis may be context-dependent, elevated levels have been found to mediate metabolic changes that facilitate anchorage-independent survival and consequent metastatic spread. The mitochondria generate reactive oxygen species (ROS), which are required for KRAS-induced anchorage-independent growth through regulation of the ERK/MAPK signaling pathway, resulting in the transduction of a survival signal. The peroxide signal underlying anchorage-independent expansion has now been tied to ATP generation, which is increased to satisfy energy requirements for cells that lose contact with the substratum. The elevation of peroxide levels must be tightly regulated however, as intracellular ROS can be limiting to cell survival. This is just one example of the specific metabolic changes wielded by tumor cells to aid in the process of dissemination from the primary tumor. Interestingly it has also been revealed that the tropism of metastatic cells may in part be dependent on their ability to adapt to the metabolic environment of the secondary organ. For example, the liver is largely hypoxic with uneven glucose levels due to competing hepatocytes. Both glucose and oxygen are key substrates in both glycolytic and oxidative metabolism and therefore metastatic cells in the liver are forced to adapt. It has been observed in patients with colon cancer that metastatic cells are able to escape this potential metabolic stress by capitalizing on the higher levels of creatine in the liver, which is synthesized and secreted by hepatocytes. Through the modulation of miRNAs these cells are able to upregulate and secrete Creatine Kinase B (CKB) into the extracellular space. Once secreted, CKB catalyzes phosphocreatine from ATP and creatine. Phosphocreatine can then be imported back into the tumor cells through specific upregulated transporters, and used to regenerate ATP. Through this mechanism the cells bypass glycolytic and oxidative metabolism pathways and essentially scavenge energy from the secondary organ’s extracellular environment. Pharmacological inhibition of CKB has been shown to substantially decrease the level of liver metastasis in mouse models and may be a useful therapeutic strategy in the clinic.

Signatures of Metastasis in the Tumor Microenvironment Tumor Vasculature Not only is tumor vasculature important for tumor growth, it is also key in tumor cell dissemination in the initial steps of metastasis. It is well established that tumor vasculature is distinct to that of healthy tissue, and these distinctive qualities can influence the spread of tumor cells into the circulation. Normal vasculature is maintained through a careful balance of pro- and antiangiogenic factors and is composed of a hierarchical distribution of arterioles, capillaries, and venules. In contrast, tumor vasculature lacks this hierarchical structure and is mainly composed of immature differentiated and undifferentiated vessels with high permeability. Undifferentiated tumor vessels frequently present with either collapsed or an absent lumen, and consequently tumor vasculature is inefficient in carrying blood flow. Due to this phenomenon, and despite increased intra-tumoral microvessel density, the tumor microenvironment is often hypoxic when compared to healthy tissue. This hypoxia can promote the generation of new vessels and also induce metastasis. Tumor blood vessels can form by the common processes of neovascularization, but often also through alternative mechanisms. “Vascular mimicry” or VM is one such alternate process and involves the creation of tumor celllined channels for fluid transport independent of typical endothelial cell-driven angiogenesis. It is this process of VM that has most recently become the focus of attention as a signature of metastatic cancer. The initial morphological, clinical, and molecular characterization of VM was performed using human melanoma as a model. These tumor cells were shown to express endothelial, embryonic/stem cell, and tumor markers; they were also shown to form channels, networks, and tubular structures rich in laminin, collagens, and heparin sulfate proteoglycans and contain plasma as well as

430

Metastatic SignaturesdThe Tell-Tale Signs of Metastasis

red blood cells. Since these initial characterizations in the late 1990s VMs have been hypothesized to function as “escape routes” for metastatic cells, and indeed several studies have demonstrated that VM is implicated in poor patient clinical prognosis. In a 2015 study, Wagenblast and colleagues showed VM can indeed pave a way for breast cancer metastasis. Using a polyclonal mouse model of breast tumor heterogeneity and a systems genetics approach, they identified discrete clones within the primary tumor that displayed phenotypic specializations and their corresponding gene expression profiles. Clones that efficiently entered the vasculature expressed two secreted proteins, Serpine2 and Slpi, which are necessary and sufficient in programing tumor cells for vascular mimicry. This study suggests that vascular mimicry drives the ability of breast tumor cells to contribute to distant metastases. Furthermore, this study also revealed that in the clinic SERPINE2 and SLPI are overexpressed preferentially in patients that have lung-metastatic relapse. Thus, these two secreted proteins, and the VM phenotype they promote, may be a relevant signature of metastatic progression in human cancer.

Tumor-Associated Immune Cells A substantial amount of clinical data has indicated that tumor infiltration of certain immune cell types correlates with poor prognosis of cancer patients. Inflammation in the tumor microenvironment can contribute to the initiation, promotion, and progression of malignant disease by fostering a nurturing environment for tumor progression and metastasis. Specifically, chronic inflammation modifies the transformed tissue by providing an abundant source of growth factors, cytokines, chemokines, prostaglandins, and reactive oxygen species, which promote angiogenesis and metastasis. As a result, the tumor microenvironment is frequently rich with immune cells including innate immune cells, such as mast cells, tumor-associated macrophages (TAMs), and myeloidderived suppressor cells (MDSCs), as well as adaptive immune cells including T and B lymphocytes and dendritic cells that can have tumor and metastasis-promoting properties. Accumulating evidence has indicated that myeloid cells, such as TAMs and MDSCs, contribute to early steps of metastasis by promoting tumor cell migration and invasion into blood vessels. Macrophages are considered to be highly plastic cells and have distinct functions in response to environmental signals. Accumulating data suggest that the tumor microenvironment polarizes recruited macrophages from a potentially tumorreactive state to a tumor-promoting state. These ‘tumor-educated’ macrophages influence the early steps of the metastatic cascade by promoting tumor cell invasion of the surrounding tissue and intravasation into the circulation. Research has shown that in breast cancer mouse models the ablation of macrophages inhibits tumor angiogenesis, suggesting that TAMs promote metastasis by increasing the density of leaky blood vessels that provide pro-tumorigenic and pro-invasive factors, as well as a route of escape for tumor cells. TAMs have also been found to secrete factors such as epidermal growth factor (EGF), which activate tumor cell receptors and enhance invasion and motility by increasing invadopodium formation and matrix degradation. Along with this, TAMs also secrete many cytokines that recruit tumor cells to the blood vessels and remodel the extracellular matrix. Interestingly, intravital imaging has also revealed that perivascular TAMs aid in tumor cell invasion of surrounding tissues by directly associating with tumor cells as they intravasate. In the clinic, breast cancer patients are given a ‘tumor microenvironment for metastasis’ (TMEM) score, which is an indication of the number of TAM, endothelial cell, and tumor cell interactions observed within a tumor. A high TMEM score, and thus high number of interactions, is a signature associated with an increased risk of metastasis.

Tissue Rigidity and Metastasis Mechanical properties of the extracellular matrix (ECM) have become increasingly recognized as functionally important during tumor progression. In particular, tissue rigidity has emerged as an important factor during metastasis. Briefly, recent work shows that mesenchymal-like tumor cells enhance ECM remodeling through deposition of additional ECM molecules, which alters the local ECM, increases the pool of available growth factors, and potentiates the mesenchymal phenotype. Tumor cells have also been shown to hijack the normal functions of stromal cells to facilitate ECM remodeling. Take for example cancer-associated fibroblasts, which can drive matrix stiffening through collagen crosslinking. Force mapping of human and mouse breast tumors revealed invasive tumors have heterogeneous mechanical properties. More specifically, when compared to the tumor core, the periphery is relatively stiffer which is consistent with the observation that cells from the invasive front of tumors are prone to be more mesenchymal before dissemination. While research continues into this phenomenon, such a finding does suggest that local mechanical properties of a niche may control the movement of tumor cells. Indeed, studies in 3D culture systems and mouse tumor models have shown that increased tissue rigidity has a functional role in driving tumor invasion and malignancy. More notably, in breast cancer patients, increased tissue rigidity due to dense clusters of collagen fibers and fibroblast at the tumor site correlates with metastatic recurrence and poor patient survival. In 2011, Conklin et al. performed a thorough study of the relationship between the long-term survival rate of human patients and the tumor-associated collagen signature 3 (TACS-3). TACS-3 is defined as bundles of straightened and aligned collagen fibers that are oriented perpendicular to the tumor boundary which specifically increase matrix stiffness through increased collagen deposition and matrix alignment. In this study, the presence of TACS-3 was evaluated in tissue biopsies from 196 breast cancer patients. Remarkably, the presence and extent of TACS-3 observed in primary tumor samples was confirmed as a prognostic indicator regardless of tumor grade or subtype. In this study, even the presence of small TACS-3 sites, and therefore only niche regions of stiffness, carried an increased risk of long-term relapse. This study produced strong statistical evidence for poor survival in patients with TACS3, and showed that routine TACS assessment can be performed with clinical histopathological samples. Ultimately, while much is

Metastatic SignaturesdThe Tell-Tale Signs of Metastasis

431

left to ascertain about the role of ECM and tissue stiffness in metastasis, it is clear that TACS-3 is a strong biomarker for the prediction for breast cancer survival.

Systemic Signatures of Metastasis Tumor-Secreted Factors Cancer is able to grow and spread by implementing unique survival and proliferative pathways. This subsequently results in the production and release of specific metabolic, hormonal, and exosomal signatures. With recent advances in blood and liquid biopsies, many different tumor-secreted factors have been identified within the peripheral blood of cancer patients. By using liquid biopsy, many of the unique signatures produced by tumors can now be observed serially – a current limitation – and without invasive tissue biopsies. While tumors produce specific factors distinct from other organs, they also have signatures unique to their type and even molecular subtype. Not surprisingly, tumors undergoing metastatic spread have also been shown to produce measurable factors indicative of unique metastatic characteristics such as preferred metastatic site or “organotropism”, and therapeutic sensitivity or response. The specific ways in which tumor-secreted factors are being used as signatures of metastatic spread are discussed in the following sections.

Hormones and metabolites For some cancer types, the use of circulating metabolites and hormones as biomarkers for diagnosis has become common in the clinic. One example of a cancer prognosis marker is Prostate-specific antigen (PSA), for which serum levels are used to screen for prostate cancer in older male patients. Another serum test in early clinical use is a three protein signature indicative of glioblastoma for which the levels of CRP, LYAM1, and BHE40 are assessed. Recent work has also revealed that some metastatic tumors can be identified by changes in circulating metabolites and hormones. Currently this is a young field, however with the increasing use of liquid biopsy in the clinic, the number of metabolite and hormone-based serum screens for metastasis will likely continue to grow. In several cancer types, including breast, lung, and prostate cancers, Parathyroid hormone-related protein (PTHrP) has been identified as a bone metastasis-specific serum biomarker. In normal physiology PTHrP is widely expressed in embryos and is critical to organ development. PTHrP is also critical during lactation, where it is secreted from the mammary tissue into the circulation to mobilize skeletal calcium for milk production. PTHrP also contributes to the pathophysiology of several cancers, and its production by disseminated tumor cells in the bone microenvironment has been shown to promote osteoclastic activity and contribute to osteolytic bone metastases. PTHrP appears to also play a central role in cancer-associated hypercalcemia. Thus the serum levels of PTHrP and calcium are together very important biomarkers for bone metastasis of several cancers. As well as a biomarker, PTHrP inhibitors have attractive prospects for clinical treatment. Two small nucleotide analogs have been reported to inhibit production of PTHrP by tumor cells and reduce bone lesions with higher survival rates in animal models, but are not yet in use for human patients.

Cell-free nucleic acids Due to advances in nucleotide sequencing, along with a better understanding of tumor and metastasis mutational signatures, the assessment of circulating DNA and RNA has recently become an exciting new tool for the diagnosis of cancer and metastasis. Circulating, cell-free DNA (cfDNA) are fragments of DNA found in the blood. cfDNA was first described in 1948, but cfDNA fragments originating from tumor cells (ctDNA) were not characterized until the 1980s. The origin of ctDNA has not been well defined yet, but is thought to result from cell death. Nucleosomes play essential roles in the fragmentation of DNA during programmed cell death and genome-wide nucleosome mapping has shown that ctDNA fragments can be distinguished from non-tumor derived cfDNA by distinct patterns of nucleosome spacing. ctDNA as a prognostic biomarker is currently used for several different types of cancers, including melanoma, colorectal cancer, cervical cancer, and pancreatic cancer. Through the assessment of ctDNA, somatic oncogenic Ras, p53 and other metastasis-related gene mutations, plus epigenetic signatures can be detected. As well as providing insight into metastasis promoting genetic changes occurring in tumors, the relative abundance of ctDNA can also be used as a prognostic marker. It was recently shown that ctDNA is found at a relatively high concentration in the peripheral circulation in patients with metastatic cancer compared with localized disease: ctDNA detection in patients with no radiographic evidence of metastasis varies between 49 and 78%, compared with 86–100% in metastatic disease. Circulating RNAs (cf-RNA) in human cancer patients were first described in the 1990s in patients with different types of cancers. As mRNA is critical in intracellular protein translation, extracellular mRNAs provide a rare glimpse into the status of the intracellular processes. Non-coding RNAs which largely function to regulate gene expression also make up a large percentage of cf-RNAs. While the majority of research to date has focused on circulating microRNAs and long non-coding RNAs, the number of studies of other RNA classes has recently grown. Interestingly, despite the unstable nature of RNA and the presence of RNA degrading factors in the blood, cf-RNA is stable enough for use as a biomarker. This is likely because tumor derived cf-RNA is often released in vesicles from tumor and apoptotic cells. cf-RNA is used as a biomarker in several ways. In serum obtained pre-surgically from patients with early stage colorectal cancers, a panel of 6 circulating miRNAs can predict cancer recurrence. Similarly, changes in cf-miRNA patterns within the same patients can be monitored over time to assess response to therapy. In addition, several studies have indicated the usefulness of cf-RNA as a marker of host response to treatment. In terms of metastatic cancer, cf-RNA is currently evaluated to reflect the gene expression within the

432

Metastatic SignaturesdThe Tell-Tale Signs of Metastasis

primary tumor, and as such its use in the prognosis of metastatic disease is limited by the current understanding of cancer spread. Essentially, cf-RNA is screened for distinct primary tumor gene expression signatures indicative of metastatic spread. As the use and study of cf-RNA increases, it is likely that the nuances of metastatic disease will be further revealed in circulating nucleotides, increasing their use in the clinic for metastatic prognosis.

Tumor-secreted exosomes Exosomes are spherical or cup-shaped nanovesicles 40–100 nm in diameter that are secreted by many cell types and can be found in most body fluids such as urine and blood. They are end-products of the recycling endosomal pathway which involves endocytosis, formation of early and late endosomes, multivesicular bodies, and secretion with the help of specific cellular pathways. Exosomes are distinct from other secreted lipid entities such as microvesicles and ectosomes, which are microvesicles derived from neutrophils or monocytes and apoptotic bodies. The basic makeup of exosomes consists of a lipid bilayer containing transmembrane and nonmembrane bound proteins and nucleic acids. Each of these components give insight into the state of the cell of origin, and, not surprisingly, exosomes released from tumor cells have been found to contain proteins involved in cancer pathogenesis such as mutant KRAS. Interestingly, exosomes can transfer their constituents and cargo to neighboring or distant cells, and it is in this way that tumor exosomes are thought to aid in tumor cell invasion of local stroma, immune cell suppression, and preparation of the metastatic site for colonization. As a signature of metastasis, exosomes can be evaluated to determine organotropism of tumor cells. In a seminal study published in 2015, David Lyden’s group showed that exosomes and their contents play a large role in determining the secondary site of metastasis. In their study, exosomes from lung-, liver- and brain-tropic tumor cells were shown to fuse preferentially with resident fibroblasts and epithelial cells at their predicted destinations. In addition, by taking up tumor-derived exosomes, the organ-specific cells were altered to create a pre-metastatic niche (section “The Premetastatic Niche”). This was revealed when treatment of mice with exosomes from lung-tropic models redirected the metastasis of bone-tropic tumor cells to the lung. Assessment of the protein signatures within exosomes from different organotropic tumors revealed distinct integrin expression patterns. Specifically, exosomal integrins a6b4 and a6b1 were associated with lung metastasis, while exosomal integrin avb5 was linked to liver metastasis. Ultimately this study reveals a new tool for predicting and monitoring the site of cancer recurrence, as exosomal integrin signatures can be used to predict organ-specific metastasis.

Circulating Tumor Cells The hypothesis that circulating tumor cells (CTCs) are an essential prerequisite to metastasis was first proposed by the Australian pathologist Thomas Ashworth in the mid-19th century. Today, the identification and molecular characterization of CTCs in cancer patients remains key to further understanding the metastatic process and developing new therapeutic targets for treatment. Fortunately, recent advances in technology have provided a means to reliably identify and isolate tumor cells in the peripheral blood stream of cancer patients, opening up new avenues of research into their biological significance. Release of CTCs into the circulation is frequently termed shedding. A study using a single rat model reported that CTCs are shed from solid tumors at a daily rate of 3.2 to 4.1  106 per gram of tissue. However, this cell loss via blood comprised about 10% of the tumor weight and resulted in a CTC count of approximately 20,000 CTCs/mL blood, a rate much higher than that observed in human cancer patients. Clinical data published in 2016 suggests a much lower rate of tumor cell shedding in humans at around 3–5 CTCs/ 7.5 mL blood for breast, prostate cancer, and colon cancer patients. To date, the majority of clinical studies have focused on CTC enumeration in guiding prognosis in metastatic cancer patients. This work aims to use CTC number as a signature to discriminate which patients are more likely to benefit from adjuvant treatment, as well as a method to monitor the efficacy of therapy. CTC number surveillance, as a marker of metastatic potential, is also predicted to detect relapse at a much earlier stage than current clinical or radiological tests. Pioneering studies in breast, prostate, and colorectal cancer have shown that higher CTC number predicts for worse prognosis, and a change in number following initiation of therapy can be predictive of survival outcome. Essentially, the characterization of CTCs from a ‘simple’ blood test could serve as a non-invasive ‘real-time tumor biopsy’ permitting an up-to-date snapshot of a patient’s tumor biology. In this case, molecular characterization of CTCs will also be essential for understanding the status of treatment. The characterization of CTCs on a molecular level is also useful is determining the course of treatment for recurrent and metastatic disease. For example, HER2 is a targetable factor in breast cancer and the expression of HER2 has been detected in CTCs of metastatic breast cancer patients in cases where primary tumors were negative at the original diagnosis. Subsequent clinical studies have shown that despite a HER2-negative status of the original tumor biopsy, patients with HER2-positive CTCs do benefit if treated with HER2-specific therapy to target CTCs and consequent metastatic lesions. Similarly, acquired genetic mutations can be detected in CTCs, and as next generation sequencing and whole cell proteomics become more accessible, the signatures associated with metastasis will likely become more obvious. The recent advances in isolating and evaluating CTCs are exciting and have greatly accelerated progress in understanding, treating, and monitoring metastasis.

Disseminated Tumor Cells As described in the previous section, tumors are able to shed cells into the circulation in large numbers. However, simply entering the blood stream alone is not enough to initiate metastasis. A fraction of CTCs are also capable of entering distant sites and

Metastatic SignaturesdThe Tell-Tale Signs of Metastasis

433

persisting as DTCs. Interestingly, research has revealed that the presence of disseminated tumor cells can be observed before the detection of a primary tumor, suggesting that shedding and colonization of distant tissues can occur at very early stages of the disease. DTCs are currently of great interest to the metastasis research community, as they can remain dormant for many years before initiating growth of a metastasis. However, the exact “trigger” that pushes DTCs from dormancy to metastatic outgrowth is still unclear. The use of DTC characteristics as a prognostic indicator, or even a sign of likely metastasis, is also in its infancy for several reasons. Firstly, the study of DTCs is difficult because their isolation in the clinic invariably requires invasive procedures. Current strategies are limited to isolating DTCs from tissue that is convenient to collect and lacks common epithelial tumor markers, such as lymph nodes or bone marrow. Markers such as cytokeratins and proliferation-associated Ki67 are then used to detect epithelial cells foreign to that tissue. As new markers are identified and verified, they are increasingly being applied in a multi-marker fashion. Detection of DTCs in lymph nodes and bone marrow is primarily used before and after therapy, however their use as an indicator for relapse remains weak. Despite DTCs predicting adverse outcomes, the majority of DTC-positive patients do not develop metastases. This suggests that the ability to take up residence in bone marrow or lymph node is not sufficient to initiate the growth of a metastatic lesion. The presence of DTCs in these tissues also does not indicate which other organs may also be occupied. Rather, it is thought that a change in the microenvironment of these cells must also occur to initiate the switch from dormant to proliferative state. Furthermore, as well as the technical caveats in analyzing DTCs, the timing of dissemination remains a confounding factor in their use as biomarkers for relapse or sensitivity to therapy. Regardless, the study of such cells with more refined detection, isolation, and characterization technologies has the potential to make DTCs clinically useful in metastatic disease management and also as therapeutic targets.

The Premetastatic Niche As described in section “Current Clinical Observations and Challenges in Treating Metastasis” of this chapter, the “seed and soil” hypothesis of metastasis was proposed by Paget in the late 1800s. Modern research continues to expand on this concept, with a rapidly growing emphasis on the “soil”. Today, the “soil” of metastasis describes not only the distant organ but also a suitable microenvironment for metastatic growth, often referred to as the premetastatic niche. The microenvironment within distant organs is critical in the initiation of metastatic growth, as the dissemination of tumor cells alone is not sufficient for metastasis (section “Disseminated Tumor Cells”). First described by Kaplan and colleagues in 2005, the premetastatic niche is conceived to form before tumor cells arrive in the tissue. It is populated with bone marrow-derived cells (BMDCs) that promote the chemoattraction and attachment of disseminated cancer cells, as well as numerous other cell types that together impact cancer cell extravasation, survival, colonization, and aggressive growth. Revolutionary work by the Lyden group has revealed that tumor-derived exosomes, which are described in section “Tumor-secreted exosomes”, initiate the formation of the premetastatic niche by reprograming stromal cells and determining organ specificity. Further work has found that a premetastatic niche can also be further enhanced in a tumor-independent manner by tissue damage or systemic physiological changes such as pregnancy. The unique features of the premetastatic niche can change surrounding tissue remarkably, revealing several signatures of possible metastatic development. There are various key alterations that have been shown to prime tissue for the growth of metastatic lesions. One of the first observed phenomena is the recruitment of BMDCs to the premetastatic niche by a myriad of tumor secreted factors that can both mobilize the BMDCs from the bone marrow and attract them to specific sites. One example is primary tumor STAT3 signaling which induces secretion of factors such as IL6 and IL10 from tumor cells. This secretion results in widespread STAT3 activation in distant organs. In premetastatic lungs, STAT3 activation stimulates fibroblasts to produce fibronectin (FN) and induces the clustering of a specific BMDC type, the myeloid-derived suppressor cells (MDSCs), which together create a favorable site for metastatic growth. Interestingly, MDSCs are not the only immune cell found in the premetastatic niche, however there is widespread heterogeneity among research model systems. Neutrophils, macrophages, T-Cells, and MDSCs are all reported to infiltrate premetastatic tissues, and contribute to the favorable conditions for growth. As well as the recruitment of several foreign cell types, the premetastatic niche is also prepared by the reprogramming of resident stromal cells. In bone tissue, senescence is induced in osteoblasts which results in their elevated secretion of IL6 and other factors that promote NFATc1-driven osteoclastogenesis. Osteoclasts function to remodel bone, a process that creates osteolytic lesions and provides a suitable environment for metastatic growth. Yet another example can be found in the premetastatic lung. Normal lung fibroblasts express the miR-30 family of microRNAs to limit MMPs and stabilize vasculature. Cancer cells, however, secrete exosomes whose contents can reprogram miR-30 family expression in lung fibroblasts, resulting in elevated MMPs and ultimately increased vascular permeability and metastasis. Vascular permeability is key to the establishment of metastasis (section “Tumor Vasculature”) and is targeted by exosomes for tumor-derived adaptation in preparation of the premetastatic niche. Specifically, tumor-secreted exosomes can promote dissemination of CTCs by transporting regulators of the vascular endothelial barriers to distant tissues. This modulation of the blood vessels in the premetastatic niche is key, as they directly control the arrest and extravasation of CTCs into the tissue. For example, brain metastasis by breast cancer cells is partially facilitated through the secretion of exosomes carrying miR-181c, which modulates actin to promote systemic vascular leakiness and the destruction of the blood-brain barrier. Metabolic manipulation within target tissues of metastasis is another specific form of tumor-induced stromal cell reprogramming. Upon dissemination into a distant site, cancer cells must compete with the surrounding tissue for nutrients to establish a metastatic colony. Breast cancer cells have been shown to secrete exosomes containing miR-122, which reprograms lung

434

Metastatic SignaturesdThe Tell-Tale Signs of Metastasis

fibroblasts to decrease their glucose consumption by suppressing pyruvate kinase. By decreasing glucose consumption of resident cells, the cancer cells have increased glucose availability, which facilitates their proliferation and survival. Finally, alterations in ECM are observed early in the formation of the premetastatic niche. One such change is the accumulation of FN, which can be induced by primary tumor-secreted factors. Hypoxia in the primary tumor has been shown to induce the secretion of FN and lysyl oxidase (LOX), which colocalize in the lung premetastatic niche to recruit myeloid cells in the lung. LOX can also mediate collagen cross-linking, which in turn increases MMP2 activity to also recruit myeloid cells. This phenomenon functions as a positive feedback loop because myeloid cells break down collagen IV, releasing it into the blood to function as a chemoattractant to draw yet more myeloid cells to the niche. Activated fibroblasts are another cell within the premetastatic niche that also excretes LOX, which results in collagen deposition and ECM stiffening to promote metastatic cell survival and engraftment. The majority of premetastatic niche characteristics have been observed in the laboratory and are not routinely assessed clinically as they would require potentially unnecessary invasive procedures. However, several biomarkers of premetastatic niche formation can be identified by non-invasive blood biopsy. Current focus is being placed on circulating exosome contents, as tumor- secreted exosomes play a large role in adapting the distant stromal cells at the site of metastasis. By analyzing tumor-secreted factors clinicians may be able to determine the likelihood of metastatic spread, the organ of metastasis, or target mediators of premetastatic niche development.

Prospective Vision Early detection and the interruption of metastasis pathways holds promise for the treatment of cancer patients with, or at risk of, metastatic disease. Target discovery efforts are currently focused on blocking metastasis at several steps, as the degree of metastatic progression in patients often cannot be determined. The obstacles that must be overcome for a tumor cell to achieve metastasis provide several vulnerabilities to exploit. The ability to enter the circulatory system, evade immune surveillance, extravasate into distant tissues, switch from dormancy to growth phase once disseminated, and establish a premetastatic niche are all essential steps of interest to target therapeutically. Detection and screening for metastatic pre-disposition are also important aspects of the current efforts in treating patients. The use of liquid biopsies continues to improve physicians’ abilities to predict secondary site location and measure treatment response. In addition, studies of inherited predisposition have potential to begin early screening of cancer patients to determine their likelihood of developing secondary tumors. Finally, as new drugs are tested pre-clinically and in patients, researchers have begun to consider a specific shift in the end points of clinical trials for metastasis therapeutics. Preclinical metastasis therapies often provide a significant delay in the development of metastases rather than reducing the size of established lesions. It is thought that “time to new lesion” end points will be more beneficial in the prevention of an initial metastasis, or additional metastases. In summary, with sights focused squarely on improving detection and treatment, creative minds and the advancement of technology continues to accelerate our grasp of the nuanced biology of metastasis.

See also: Cell Adhesion During Tumorigenesis and Metastasis.

Further Reading Hendrix, M.J.C., Seftor, E.A., Hess, A.R., Seftor, R.E.B., 2003. Angiogenesis: vasculogenic mimicry and tumour-cell plasticity: lessons from melanoma. Nature Reviews. Cancer 3, 411–421. Hunter, K., 2006. Host genetics influence tumour metastasis. Nature Reviews. Cancer 6, 141–146. Karagiannis, G.S., Goswami, S., Jones, J.G., Oktay, M.H., Condeelis, J.S., 2016. Signatures of breast cancer metastasis at a glance. Journal of Cell Science 129, 1751–1758. Liu, Y., Cao, X., 2016. Characteristics and significance of the pre-metastatic niche. Cancer Cell 30 (5), 668–681. Paget, S., 1889. The distribution of secondary growths in cancer of the breast. Lancet 1, 99–101. Siravegna, G., Marsoni, S., Siena, S., Bardelli, A., 2017. Integrating liquid biopsies into the management of cancer. Nature Reviews. Clinical Oncology. Published online 02 march 2017. Steeg, P.S., 2016. Targeting metastasis. Nature Reviews. Cancer 16, 201–218. Smith, H.A., Kang, Y., 2013. The metastasis-promoting roles of tumor-associated immune cells. Journal of Molecular Medicine (Berlin, Germany) 91 (4), 411–429. Valastyan, S., Weinberg, R.A., 2011. Tumor metastasis: molecular insights and evolving paradigms. Cell 147 (2), 275–292.

Metformin Matteo Lazzeroni and Sara Gandini, European Institute of Oncology, Milan, Italy © 2019 Elsevier Inc. All rights reserved.

Glossary

AMPK 50 AMP-activated protein kinase or AMPK is an enzyme that plays a role in cellular energy homeostasis. It is expressed in a number of tissues, including the liver, brain, and skeletal muscle. The net effect of AMPK activation is stimulation of hepatic fatty acid oxidation, ketogenesis, stimulation of skeletal muscle fatty acid oxidation and glucose uptake, inhibition of cholesterol synthesis, lipogenesis, and triglyceride synthesis, inhibition of adipocyte lipolysis and lipogenesis, and modulation of insulin secretion by pancreatic beta-cells. Recently AMPK is emerging as a possible metabolic tumor suppressor and target for cancer prevention and treatment. Activation of AMPK has been found to oppose tumor progression in several cancer types and offers a promising cancer therapy. AMPK activity opposes tumor development and progression in part by regulating inflammation and metabolism. IGFs Insulin-like growth factors I (IGFs) are important mediators of growth, development, and survival. They are synthesized by almost any tissue in the body, and their action is modulated by a complex network of molecules, including binding proteins, proteases, and receptors, which all comprise the IGF system. IGFs may promote cell cycle progression and inhibition of apoptosis either by directly associating with other growth factors or indirectly by interacting with other molecular systems which have an established role in carcinogenesis and cancer promotion. In addition, increased serum levels of IGFs and/or altered levels of their binding proteins are associated with increased risk for developing several malignancies. MAPK A mitogen-activated protein kinase (MAPK or MAP kinase) is a type of protein kinase that is specific to the amino acids serine, threonine, and tyrosine. MAPKs are involved in directing cellular responses to a diverse array of stimuli, such as mitogens, osmotic stress, heat shock, and proinflammatory cytokines. They regulate cell functions including proliferation, gene expression, differentiation, mitosis, cell survival, and apoptosis. mTOR The mammalian target of rapamycin is a kinase that in humans is encoded by the MTOR gene. mTOR is a member of the phosphatidylinositol 3-kinase-related kinase family of protein kinases. mTOR functions as a serine/threonine protein kinase that regulates cell growth, cell proliferation, cell motility, cell survival, protein synthesis, autophagy, and transcription. Furthermore, mTOR also promotes the activation of insulin receptors and insulin-like growth factor 1 receptors. REDD1 The protein regulated in development and DNA damage response 1 (REDD1) is a protein that in humans is encoded by the DDIT4 gene. REDD1 acts as a negative regulator of mTOR, and its clinical interest is based primarily on its effect on mTOR, which has been associated with aging and linked with diseases such as diabetes and cancer.

Introduction Metformin is an orally administrated drug commonly used to lower blood glucose concentrations in patients with type II diabetes mellitus (T2DM). There is intense interest in the cancer prevention research community regarding the potential use of metformin (1,1-dimethylbiguanide) to decrease cancer incidence or cancer-related mortality. A series of recent metaanalyses confirmed the association of diabetes or prediabetes (impaired fasting glucose and/or impaired glucose tolerance) with an increased risk of cancer. Such metaanalyses have estimated a relative risks (RR) of 1.1–2.5 for cancer risk at various organ sites in patients with T2DM, as shown in Table 1. Insulin-resistance and hyperinsulinemia are the principal conditions associated with T2DM. Various mechanisms have been reported to explain the relationship of T2DM with cancer incidence. In particular, insulin can exert its oncogenic effect through the stimulation of growth factors, such as insulin-like growth factor-I (IGF-I), and the concomitant reduction of the binding proteins (BPs) for these growth factors. Metformin reduces gluconeogenesis and hepatic glucose production and improves insulin sensitivity by increasing peripheral glucose uptake. Thus, it does not increase the circulating insulin, and it is a relative safe drug with a good pharmacokinetic profile and mild toxicity that is usually well tolerated. The most common side effects are nausea and diarrhea, usually self-limited and transient. The only potential major adverse event during therapy is lactic acidosis, but a direct association between this condition and metformin use has not yet been clearly demonstrated. Moreover, this disease is rare and primarily confined to patients with concomitant renal and hepatic disorders. Despite many epidemiological studies and metaanalyses showing an association between metformin use and reduced cancer incidence and mortality (fully discussed later), only a minority of these associations have a robust supporting evidence without suggestion of bias. In particular, time-related biases have been recently recognized as a major potential source of error in some epidemiologic studies, imposing a reevaluation of the literature linking metformin use to reduced cancer incidence, and indicating the need to conduct properly designed randomized clinical trial. The chapter summarizes the scientific evidence of metformin use to

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65133-4

435

436

Metformin

Table 1

Metaanalyses on cancer incidence in patients with diabetes

Cancer

First author (PY)

No. of studies

SRR

95% CI

Any cancer Breast

Noto et al. (2011) Boyle et al. (2012) Liao et al. (2011) Larsson et al. (2007) Hardefeldt et al. (2012) Larsson et al. (2006) Zhu et al. (2013) Jing et al. (2012) Deng et al. (2012) Jiang et al. (2011) Friberg et al. (2007) Huang et al. (2012) Ren et al. (2011) Tian et al. (2012) Ge et al. (2011) Larsson and Wolk (2011) Bao et al. (2013) Gao et al. (2010) Wang et al. (2012) Castillo et al. (2012) Lee et al. (2013) Castillo et al. (2012) Mitri et al. (2008) Castillo et al. (2012) Lee JY Wang et al. (2017) Ben et al. (2011) Kasper and Giovannucci (2006) Bansal et al. (2013) Schmid et al. (2013)

12 39 10 20 43 16 36 5 24 41 16 17 21 25 21 9 24 14 18 26 34 34 15 26 19 13 35 19 45 6

1.10 1.27 1.23 1.20 1.20 1.24 1.35 1.60 1.26 1.27 2.10 1.30 1.43 1.11 1.09 1.42 1.40 3.33 2.01 1.22 1.11 1.22 1.19 1.22 1.17 1.19 1.94 0.84 0.86 1.17

1.04–1.17 1.16–1.39 1.18–1.27 1.12–1.28 1.13–1.29 1.08–1.42 1.17–1.56 1.38–1.87 1.20–1.31 1.21–1.34 1.75–2.53 1.12–1.50 1.18–1.72 1.00–1.24 0.98–1.22 1.06–1.91 1.16–1.69 1.82–6.10 1.61–2.51 1.03–1.44 1.02–1.20 0.98–1.53 1.04–1.35 1.07–1.39 1.02–1.33 1.06–1.34 1.66–2.27 0.76–0.93 0.80–0.92 0.99–1.39

Bladder Biliary duct Colorectum Endometrium Esophagus Gallbladder Gastric Kidney Liver Liver (HCC) Leukemia Lung Multiple Mieloma Non-Hodgkin lymphoma Ovary Pancreas Prostate Thyroid

PY, Publication year; SRR, Summary relative risk; 95% CI, 95% confidence interval.

reduce cancer incidence and mortality, from putative mechanisms of action through critical evaluation of observational studies and the limited data on clinical trials.

Molecular Mechanisms of Metformin in Cancer and Diabetes Despite being introduced clinically in the 1950s, the exact mechanism of action of metformin has not been fully elucidated. This applies not only to its antihyperglycemic and antihyperinsulinemic effects but also to its anticancer properties. Two major mechanistic categories are relevant to the anticancer action of metformin: an indirect and a direct mechanism, not mutually exclusive and converging on the same molecular pathways.

Indirect Mechanisms This category involves “endocrine-type effects” related to its insulin-lowering activity. Metformin acts on the liver to inhibit hepatic production of glucose, antagonizing glucagon-mediated effects that lead to hyperglycemia in diabetics. To a lesser extent metformin increases insulin-mediated uptake of glucose in skeletal muscle, though not by increasing insulin levels, as the drug actually decreases the hyperinsulinemia that is seen in diabetes. Since high insulin levels are known to stimulate proliferation of some common cancers, this decrease in insulin levels may slow tumor proliferation in hyperinsulinemic patients.

Direct Mechanisms In this category metformin exerts its effects directly on the target cells by an insulin-independent mechanism that leads to suppression of ATP production via metformin’s inhibition of mitochondrial complex I. NADH-ubiquinone oxidoreductase (Complex I) plays an essential role in biosynthesis and redox control during proliferation, resistance to cell death, and metastasis of cancer cells. Both routes of metformin activity converge on the mTOR (mammalian target of rapamycin) pathway resulting in inhibition of its proproliferative activity.

Metformin Table 2

Possible pathways involved in the anti-cancer effects of metformin

Mechanism

Function

AMPK-dependent

Inhibition of cell mitosis and proliferation Upregulation of the p53 axis DNA synthesis Growth inhibition Cell apoptosis REDD1 mediates cell cycle arrest Induced apoptosis Inhibition of cell growth Induction of apoptosis Induction of autophagy Inhibition of tumor promotion Suppression of HER2 overexpression Prevention of IGF-1R upregulation Reduction of cell proliferation and invasion Reduction of circulating insulin and IGF-1 Arrest of cell growth and proliferation Arrest of tumor cells migration and invasion Inhibition of cell growth, colony formation Induction of cell cycle arrest Survival signals blockade

AMPK-independent Suppression of mTOR

Suppression of IGF signaling

The MAPK signaling pathway

Fig. 1

437

Main signaling pathways involved in the anti-cancer effects of metformin.

Table 2 and Fig. 1 show some possible signaling pathways involved in the anticancer effects of metformin. Some effects are mediated by the activation of adenosine monophosphate protein kinase (AMPK), an intracellular energy sensor whose activation inhibits cell mitosis and proliferation by directly influencing the dynamics of cell division. Metformin stimulates AMPK through upregulation of the p53 axis. However, certain antitumor effects of metformin are independent of the AMPK signaling pathway. Metformin directly influences mTOR in a p53-dependent manner through an AMPK-independent mechanism increasing the level of REDD1, a negative regulator of mTOR, thus contributing to the cell circle arrest. mTOR regulates cellular energy homeostasis by modulating protein synthesis and autophagy, and exerting significant positive effect on cell proliferation and tumorigenesis. An imbalanced activation of mTOR is associated with malignant tumor progression, drug resistance, and a worse prognosis. mTOR metformin-induced inhibition of the mTOR pathway has been demonstrated in different types of cancer and through AMPKdependent and -independent pathways.

438

Metformin

The suppression of IGF signaling is another possible mechanism of action: insulin and IGFs are key regulators of metabolism and cancer progression by activating signaling pathways associated with cell growth and proliferation. There are two subtypes of IGF, IGF-1 and IGF-2, which are both mitogenic and antiapoptotic. IGF-1 receptor (IGF-1R) binds to the ligand IGF-1, IGF-2, or insulin to promote autophosphorylation of tyrosine at its kinase domain and with a consequent activation of signaling through the phosphatidylinositol-3-kinase PI3K/Akt/mTOR and RAS/RAF/mitogen-activated protein kinase (MAPK) pathways. Emerging evidence suggests that metformin decreases IGF-1 by indirectly downregulating insulin and insulin-binding proteins to reverse hyperinsulinemia.

Metaanalyses of Observational Studies and Randomized Controlled Trials Examining the Effect of Metformin on Cancer Incidence and Mortality Multiple metaanalyses of observational studies have reported the use of metformin is associated with a decrease in cancer incidence ranging from 14% to 40%, along with a decrease in mortality in the same range (Table 3). Organ sites include the breast, colon, liver, pancreas, prostate, endometrium, and lung, among others. However, these studies have a multitude of study designs, from cohort and case-control studies through randomized controlled trials. As the level of scientific evidence increases with the study design, the effect of metformin on cancer incidence appears to decrease. Thakkar et al. showed that while cancer incidence was decreased by 30% in cohort studies and 10% in case-control studies, there was no clear effect in randomized controlled trials. Gandini et al. performed a metaanalysis with particular attention to biases and confounders and also found that the risk reduction for cancer incidence decreased from 31% for all studies to 29% for prospective studies to 5% for randomized clinical trials. Several authors suggested that the existing metformin literature might be subject to important sources of time-related bias, which potentially overestimate cancer-reducing effects. Time-related bias is a form of differential misclassification bias that can be avoided by appropriate accounting of follow-up time and exposure status in the design and analysis of studies. There are different types of time-related biases: (i) immortal time bias, when unexposed time is misclassified as drug-exposed time; (ii) time-window bias, when there are differential exposure opportunity time windows between exposed and unexposed subjects; (iii) time-lag bias, when there’s a comparison of treatment given during different stages of the disease. Immortal time bias refers to a period of cohort follow-up time during which a cancer event (that determines end of follow-up) cannot occur. Immortal time bias can arise when the period between cohort entry and date of first exposure to metformin, during which cancer has not occurred, is either misclassified or excluded and not accounted for in the analysis. In particular this is frequently found in studies that compare “ever users” against “nonusers” and do not take into account, in the analysis and in the study design, of time of use (when patients’ start and stop metformin). Caution should be used also when analyzing cohort studies where a first-line therapy with metformin is compared with second- or third-line therapies. Patients are unlikely to be at the same stage of diabetes, which can induce confounding of the association with an outcome (e.g., cancer incidence) by disease duration. An outcome related to the first-line therapy may also be attributed to the second-line therapy if it occurs after a long period of exposure. Such a situation requires matching on disease duration and consideration of latency time windows in the analysis.

Table 3

Metaanalyses of observational studies and randomized controlled trials examining the effect of metformin on cancer incidence and mortality

Author (PY)

No. of studies included

Cancer incidence SRR (95%CI)

Cancer mortality SRR (95% CI)

DeCensi et al. (2010) Noto et al. (2012) Soranna et al. (2012) Thakkar et al. (2013)

11 6 17 24

0.70 (95% CI, 0.51–0.96) 0.66 (95% CI, 0.49–0.88) NR NR

Franciosi et al. (2013)

53

Gandini et al. (2014)

47

Wu et al. (2015)

265

Ma et al. (2017)

23

0.68 (0.52–0.88) 0.67 (0.53–0.85) 0.61 (0.54–0.70) 0.70 (0.67–0.73) Co 0.90 (0.84–0.98) CC 1.01 (0.81–1.26) RCT 0.73 (0.61–0.88) 0.98 (0.81–1.19) RCT 0.69 (0.52–0.90) 0.95 (0.69–1.30) RCT 0.86 (0.83–0.90) 0.88 (0.83–0.92) Co 0.71 (0.63–0.80) CC 1.05 (0.94–1.18) RCT 0.52 (0.40–0.68) 0.64 (0.48–0.86) Co 0.50 (0.36–0.70) CC 0.84 (0.10–6.83) RCT

0.65 (95% CI, 0.53–0.80) 0.66 (95% CI, 0.54–0.81) 0.70 (95% CI, 0.53–0.94) 0.66 (95% CI, 0.49–0.89) Co 0.91 (95% CI, 0.37–2.23) RCT NR

PY, Publication year; SRR, Summary relative risk; 95% CI, 95% confidence interval; NR, Not reported; Co, cohort study; CC, case-control study; RCT, randomized control trial.

Metformin

439

In the metaanalysis performed by Gandini et al., time-related biases were taken into account, resulting in a diminution of the protective effect on cancer incidence. The risk reduction in cancer incidence at all sites decreased from 30% to 10%, albeit still statistically significant. When focusing on specific organ sites, metformin use was significantly associated with decreased breast cancer incidence, but the magnitude of the effect was not clinically relevant. For colon cancer, 8% decrease in incidence was calculated. For prostate and pancreatic cancers, metformin use was not associated with a statistically significant protective effect. For lung cancer, examination of all studies showed a 18% decrease in cancer incidence. Adjustment for time-related biases decreased this protective effect to a statistically significant 12%, but adjustment for smoking, which is by far the most relevant lung cancer risk factor, resulted in no association between metformin use and cancer incidence. Similarly, the strong protective effect of metformin use against liver cancer became nonsignificant when only unbiased studies were examined. It should be emphasized, however, that the total number of studies and patients was rather small for the liver cancer associations, and thus these results should be interpreted with caution. Notably, a very recent metaanalysis of studies using metformin as a reducer for liver cancer risk in diabetic patients reported that metformin use reduced the ratio of liver cancer by 48% compared with nonusers. The protective effect was validated in all the exploratory subgroup analyses, except that pooled result of post hoc analyses of two randomized controlled trials found no significant difference between subjects with metformin and those without. After adjusting for hepatitis B/C virus infection, cirrhosis, obesity, behavioral factors, and time-related bias, the association was stable, pooled OR ranged from 0.42 to 0.75. Another potential source of confounding that is often not adequately explored in the existing literature is the effect of obesity and its surrogate, body mass index (BMI). Obesity is intimately linked to increased risk of multiple cancer types. Potential mechanisms include both direct and indirect effects of obesity on insulin, IGF-1, sex hormones, adipokines, and inflammation, many of which are directly impacted by metformin. Metformin, unlike several other antidiabetic agents, is associated with weight loss. Furthermore, a recent clinical trial showed that metformin affected breast cancer proliferation differentially according to insulin resistance status and BMI, with a trend toward inhibiting proliferation only in women with insulin resistance or high BMI. Gandini et al. examined BMI as a potential confounder and found that adjustment for BMI decreased the protective effect of metformin for all-cancer incidence from 31% to 18%, which was still statistically significant. For breast cancer, adjustment for BMI revealed a borderline association of metformin use with cancer incidence. For colon and prostate cancer, there was no statistically significant association between metformin use and cancer incidence. Other organ sites could not be assessed due to a small number of published studies with adjusted estimates. It should be noted, however, that BMI is dynamic and therefore adjusting for a single BMI assessment might be inadequate to account for confounding by BMI dynamics over time. Furthermore, Gandini et al. were unable to assess the effect of BMI and time-related biases simultaneously due to inadequate numbers of studies that controlled for both factors. The effect of metformin on cancer mortality is influenced by a limited literature and thus Gandini et al. were unable to determine the effect by cancer sites. However they found that metformin use was associated with a 34% decrease mortality that remained significant when the analysis was limited to prospective studies. Adjustment for BMI maintained the magnitude of the effect although adjustment for time-related biases, limited to only three studies, resulted in loss of statistical significance. Different mechanisms may be responsible for effect of metformin on cancer mortality compared with cancer incidence. Several retrospective analyses have suggested that diabetics treated with metformin during chemotherapy have longer survival than individuals treated with other antidiabetic agents. A previous mouse xenograft study showed that metformin targets breast cancer stem cells and synergizes with doxorubicin to prevent relapse. Increasing the effectiveness of chemotherapy could result in improved survival. Taken together, these results point out the limitations of the current literature regarding the association between metformin use and cancer incidence and mortality. While all the studies, including metaanalyses, suggest that metformin use is associated with a reduced risk of cancer and death, the effect may be far smaller than previously believed.

Published Clinical Trials of Metformin in Cancer Table 4 summarizes the main characteristics of the completed clinical trials (mainly phase II) in cancer patients with metformin. Two studies examined various doses of metformin given for 3–6 months to women after therapy for breast cancer. Goodwin et al. phase II study (2008) examined women with breast cancer who completed adjuvant therapy and whose plasma levels of insulin were at least 45 pmol/L. Metformin decreased circulating insulin levels by 22.4% significantly. Campagnoli et al. study (2012) enrolled women with breast cancer, who had completed adjuvant therapy but had elevated testosterone, and compared two different doses of metformin. The higher dose (1500 mg/day) significantly reduced serum testosterone levels and free androgen index compared to the lower dose. In 2015, Goodwin and colleagues analyzed 492 patients enrolled in the NCIC Clinical Trials Group MA.32 phase III trial (metformin vs placebo effect on invasive disease-free survival and other outcomes in early breast cancer) to explore whether metformin may improve metabolic factors (insulin, glucose, leptin, highly sensitive C-reactive protein [hs-CRP]). At 6 months, decreases in weight and blood variables were statistically significantly greater in the metformin arm (vs placebo) in univariate analyses: weight – 3.0%, glucose – 3.8%, insulin – 11.1%, homeostasis model assessment – 17.1%, leptin – 20.2%, hs-CRP – 6.7%. There was no statistically significant interaction of change in these variables with baseline BMI or insulin. Five trials examined the short-term effects (1–4 weeks) of various doses of metformin on cell proliferation (Ki-67) in women awaiting surgery for breast cancer (presurgical, window of opportunity trial). Hadad et al. (2011) found a 3.4% reduction in Ki-67 in the metformin arm, but 29% of the patients on metformin withdrew due to gastrointestinal side effects. Bonanni et al.

440

Table 4

Completed phase II–III clinical trials of metformin in cancer patients Phase Organ

Number of patients

Patients Characteristics

Goodwin et al. (2008)

II

Breast cancer

32

Insulin levels >45 pmol/L

Hosono et al. (2010)

II

Colon

23

Nondiabetic

Hadad et al. (2011)

II

Breast cancer

8 þ 47

Nondiabetic

Campagnoli et al. (2012)

II

Breast cancer

125

Bonanni et al. (2012)

II

Breast cancer Presurgical

Cazzaniga et al. (2013) DeCensi et al. (2014) Niraula et al. (2012)

II

Kalinsky et al. (2014)

II

Laskov et al. (2014)

II

Mitsuhashi et al. (2014)

II

Joshua et al. (2014)

II

Goodwin et al. (2015)

Treatment

Endpoint

Results

Insulin level

Significant reduction

200

Nondiabetic

MET versus Placebo

Number of rectal aberrant crypt foci Ki-67; transcriptome analysis; insulin level Insulin; HOMA IR index; testosterone, free-androgen index Ki-67

Significant reduction

Nondiabetic, testosterone levels 0.28 ng/mL

MET versus no treatment MET versus no treatment MET versus no treatment HD MET versus LD MET

Breast cancer Presurgical

39

Nondiabetic

MET versus no treatment

Breast cancer presurgical Endometrial cancer Presurgical Endometrial cancer Presurgical

35 11

Nondiabetic BMI  25 Nondiabetic

31

Nondiabetic

MET versus no treatment MET versus no treatment MET versus no treatment

24

Nondiabetic

MET (single arm)

III

Prostate cancer Presurgical Breast cancer

492

Nondiabetic

MET versus Placebo

Schuler et al. (2015)

II

Endometrial cancer

20

Chak et al. (2015) Higurashi et al. (2016)

II III

Barrett’s esophagus Colorectal adenoma

74 498

Nondiabetic BMI  30 Nondiabetic Nondiabetic

• TUNEL • Ki-67 • BMI, weight, HOMA • Insulin level, leptin, CRP • Symptoms/quality of life • Ki-67 • BMI, cholesterol, leptin

Insulin level, IGF-1, IGFBP-7, Ki-67, Ps6 • Ki-67, Topoisomerase IIa, Ps6,ERK1/2 • AMPK, p27 • Insulin, glucose, IGF-1, leptin Ki-67

Significant reduction Significant reduction Overall nonsignificant reduction; significant in HER2 þ, top hs-CRP; if HOMA > 2.8

• Significant increase • Significant reduction • Significant reduction • Nonsignificant reduction • Nonsignificant alteration • Nonsignificant reduction • Significant reduction Significant reduction

• Significant reduction • Significant increase • Significant reduction Significant reduction Significant reduction

MET (single arm)

Weight, glucose, insulin level, HOMA, leptin, hs-CRP Ki-67

Significant reduction

MET versus Placebo MET versus Placebo

pS6K1 Adenoma recurrence

No reduction Significant reduction in prevalence

Abbreviations: MET, Metformin; AMPK, Phospho-adenosine monophosphate-activated protein kinase; HD, High dose; LD, Low dose; HOMA-IR, Homeostasis model assessment-insulin resistance; IGF-1, Insulin-like growth factor-1; TUNEL, Terminal eoxynucleotidyl transferase dUTP nick end labeling; HOMA, Homeostasis model assessment; hs-CRP, Highly sensitive C-reactive protein; IGFBP-3, Insulin-like growth factor binding protein 3; BMI, Body mass index; CRP, C-reactive protein; Ps6, Phospho-ribosomal protein S6; pS6K1, Phosphorylated S6 kinase; IGFBP-7, Insulin-like growth factor binding protein 7; ERK1/2, Phospho-extracellular signal-regulated kinase 1/2; ACF, Aberrant crypt foci.

Metformin

First Author (year of publication)

Metformin

441

(2009) performed a larger double-blind, placebo-controlled study, where metformin effect on Ki-67 change relative to placebo was not statistically significant, with a mean proportional increase of 4.0%. Women with a HOMA index (Homeostasis Model Assessment, used to quantify insulin resistance) > 2.8 had a nonsignificant decrease of 10.5% while women with a HOMA index < 2.8 had a nonsignificant increase of 11.1%. The interaction between HOMA index and metformin on Ki-67 was statistically significant; a similar trend was seen with BMI although the interaction was not statistically significant. Niraula et al. (2012) in a single-arm trial reported a 3% decrease in Ki-67 (p ¼ 0.016) after a median of 18 days of treatment, and a recently completed study (Kalinsky et al., 2014) showed no reduction in Ki-67. Two studies have been reported in women with early-stage endometrial cancer. Endometrial cancer prevention is of particular interest because of its relationship with obesity and the use of metformin for treatment of polycystic ovary syndrome, a condition associated with increased endometrial cancer risk. In the Laskov study (2011), 11 newly diagnosed, untreated, nondiabetic patients with endometrial cancer received metformin 500 mg tide from diagnostic biopsy to surgery. Ki-67, pAMPK, and pS6 immunohistochemistry staining were performed on the endometrial cancer before and after metformin treatment, and they were compared with a control group of 10 women with endometrial cancer who did not receive metformin. Mean plasma insulin, IGF-1, and IGFBP-7 were significantly reduced after metformin treatment. A clear reduction in Ki-67 and pS6 expression was observed by both conventional light microscope analysis and digital image analysis with a significant mean reduction in percentage of cells staining for Ki-67 and pS6. In the untreated control group expression of ki-67 was similar between the biopsy and the surgical specimens. Mitsuhashi et al. trial (2014) reported that increasing the dosage from 750 mg QD to 1500 mg or 2250 mg for 4 weeks of metformin resulted in a 44.2% decrease in Ki-67. Three additional studies in other organ systems (Barrett’s esophagus, colon, and prostate) have been published. A study in Barrett’s esophagus (Chack et al., 2015) showed no significant change in phosphor-serine 6 kinase (pS6K), cell proliferation, or apoptosis. Hosono et al. (2010) study examined change in the number of aberrant crypt foci (ACF, a putative precursor of colon cancer) in individuals with preexisting ACF, randomizing participants to a very low dose of metformin (250 mg/day) or no treatment for one month. A significant decrease in ACF from 8.78  6.45 to 5.11  4.99 was observed in the metformin arm but not in the control group. Additionally, proliferation was significantly decreased in the metformin group. Recently, Hirugashi et al. (2016) in a phase III trial reported that the administration of low-dose metformin (250 mg/daily) for 1 year to patients without diabetes after colorectal polipectomy has shown to be safe and to reduce the prevalence and number of metachronous adenomas or polyps. In the trial published by Joshua et al. (2014), men with prostate cancer were treated with metformin for 4–12 weeks prior to surgery. The primary endpoint, change in Ki-67, was statistically significant, with an absolute decrease of 1.44%. In a per patient and per tumor analyses, metformin reduced the Ki67 index by relative amounts of 29.5% and 28.6%, respectively.

Ongoing Phase III Trials With Metformin in Cancer Patients and Prevention Table 5 summarizes the main characteristics of ongoing phase III clinical trials in cancer patients with metformin. NCT01864096 trial aims to see if metformin can delay the time to progression in men with low-risk prostate cancer when compared to a placebo. NCT01101438 is looking at whether metformin can decrease or affect the ability of breast cancer cells to grow and whether metformin will work with other therapies to keep cancer from recurring. NCT02614339 trial wants to identify the effect of adjunctive metformin on recurrence of nondiabetes mellitus stage III colorectal cancer. This trial is open-label randomized controlled study, and the primary endpoint is to compare the 3-year disease-free survival between metformin group and nonmetformin group. The secondary endpoint is to compare the 5-year overall survival and disease-specific survival between two group, to identify the safety of metformin, and to compare the recurrence rate of polyps after polypectomy between two groups. NCT01697566 study analyzes the effects of metformin and/or a program called “lifestyle intervention” on the endometrium in postmenopausal women who are also obese (both risk factors for endometrial cancer). Lifestyle intervention is made up of a series of in-person sessions with a coach to discuss strategies for losing weight and ways to increase physical activity. NCT03184493 is a prospective trial to compare the role of celebrex alone, metformin alone, and celebrex plus metformin in preventing hepatocellular carcinoma recurrence after hepatic resection. NCT02065687 is a randomized phase II/III trial to investigate how well paclitaxel, carboplatin, and metformin hydrochloride works and compares it to paclitaxel, carboplatin, and placebo in treating patients with endometrial cancer that is stage III–IV. The hypothesis is that metformin may help paclitaxel and carboplatin to work better by making cancer cells more sensitive to the drugs. NCT02040376 is a placebo-controlled, double-blind crossover trial of metformin in 30 children treated with radiation for medulloblastoma, the most common malignant brain tumor. Tests of thinking and learning and brain imaging techniques will be used to examine whether metformin can enhance cognition or promote brain repair following radiation-induced brain injury. NCT01905046 tests whether metformin is able to get rid of atypia (early cell changes that are thought to be a marker of breast cancer risk) in women at increased risk for breast cancer. A total of 300 patients with atypia (Masood Score 14–17) on random periareolar fine needle aspiration (RPFNA) at baseline are expected to be randomized between Metformin and placebo control arm. The randomization will be stratified for known BRCA mutation (BRCA1 or BRCA2 mutation vs no mutation), prior excisional biopsy (atypical epithelial hyperplasia, atypical lobular hyperplasia, flat epithelial hyperplasia vs DCIS vs LCIS), and baseline fasting insulin levels. The primary objective is to test for the presence or absence of cytological atypia in RPFNA bilateral aspirates after 12 and 24 months.

442 Metformin

Table 5

ClinicalTrials.gov Search Results 07/14/2017

NCT number

Recruitment

Conditions

Interventions

No. of patients

Study designs

Primary endpoint

NCT01864096 NCT01101438

Prostate cancer on active surveillance Breast cancer

Metformin versus placebo Metformin versus Placebo

408 3649

Randomized; double blinded Randomized; double blinded

Stage III colorectal cancer Nondiabetic Healthy women; BMI  30

Metformin versus control

593

Randomized; open label

Time to progression Invasive disease-free survival Disease-free survival

Metformin versus placebo

100

Randomized; double blinded

Ki-67

NCT03184493

Recruiting Active, not recruiting Active, not recruiting Active, not recruiting Recruiting

200

Nonrandomized

1-year tumor recurrence

NCT02065687

Recruiting

Endometrial cancer

540

Randomized; double blinded

Overall survival

NCT02040376

Recruiting

30

Recruiting

Metformin versus Placebo

400

Randomized; double blind; crossover Randomized; double blinded

Fostering brain repair

NCT01905046

Children with cranial-spinal radiation for medulloblastoma Breast ADH; BRCA1/2 mutation carrier; DCIS; LCIS

Metformin plus celebrex versus Metformin versus cerebrex Metformin plus paclitaxel and carboplatin versus paclitaxel and carboplatin Metformin versus placebo

NCT02614339 NCT01697566

Liver cancer

Abbreviations: NCT, Number of clinical trial; ADH, Atypical ductal hyperplasia; DCIS, Ductal carcinoma in situ; LCIS, Lobular carcinoma in situ.

Presence of atypia in RPFNA

Metformin

443

Conclusions The emerging data from preclinical, epidemiologic, and clinical studies suggest that there is a signal for cancer-preventive potential with metformin use. There is biologic plausibility for a cancer-preventive effect given the multiple ways that metformin can interfere with cancer-promoting signaling pathways. However, both animal and epidemiologic studies have shown somewhat mixed effects. Further, the epidemiologic literature reviewed presented results that regard mainly individuals with diabetes and the effect is of lesser magnitude than previously reported once the appropriate adjustments to avoid bias and confounding are made. It remains to be determined whether a similar protective effect on cancer risk and mortality can be confirmed in nondiabetic individuals by properly designed studies. Multiple ongoing clinical trials are addressing which cancer patient cohorts are most likely to benefit from metformin. Although the long history and clinical experience with metformin make it a very attractive candidate for drug repurposing, general recommendations about its use, particularly in nondiabetic populations, need to await the results of the ongoing studies.

See also: Chemoprevention Trials. Diabetes and Cancer. Obesity and Cancer: Epidemiologic Evidence.

References Bansal, D., Bhansali, A., Kapil, G., Undela, K., Tiwari, P., 2013. Type 2 diabetes and risk of prostate cancer: A meta-analysis of observational studies. Prostate Cancer and Prostatic Diseases 16 (151–8), S1. Bao, C., Yang, X., Xu, W., Luo, H., Xu, Z., Su, C., Qi, X., 2013. Diabetes mellitus and incidence and mortality of kidney cancer: A meta-analysis. The Journal of Diabetic Complications 27, 357–364. Ben, Q., Xu, M., Ning, X., Liu, J., Hong, S., Huang, W., Zhang, H., Li, Z., 2011. Diabetes mellitus and risk of pancreatic cancer: A meta-analysis of cohort studies. European Journal of Cancer 47, 1928–1937. Bonanni, B., Puntoni, M., Cazzaniga, M., Pruneri, G., Serrano, D., Guerrieri-Gonzaga, A., Gennari, A., Stella, T.M., Galimberti, V., Veronesi, P., Johansson, H., Aristarco, V., Bassi, F., Luini, A., Lazzeroni, M., Varricchio, C., Viale, G., Bruzzi, P., Decensi, A., 2012. Dual Effect of Metformin on Breast Cancer Proliferation in a Randomized Presurgical Trial. Journal of Clinical Oncology 30, 2593–2600. Boyle, P., Boniol, M., Koechlin, A., Robertson, C., Valentini, F., Coppens, K., Fairley, L.-L., Boniol, M., Zheng, T., Zhang, Y., Pasterk, M., Smans, M., Curado, M.P., Mullie, P., Gandini, S., Bota, M., Bolli, G., Rosenstock, J., Autier, P., 20-9-2012–. Diabetes and breast cancer risk: A meta-analysis. British Journal of Cancer 107, 1608–1617. Campagnoli, C., Pasanisi, P., Abba, C., Ambroggio, S., Biglia, N., Brucato, T., Colombero, R., Danese, S., Donadio, M., Venturelli, E., Zito, G., Berrino, F., 2012. Effect of different doses of metformin on serum testosterone and insulin in non-diabetic women with breast cancer: A randomized study. Clinical Breast Cancer 12, 175–182. Castillo, J.J., Mull, N., Reagan, J.L., Nemr, S., Mitri, J., 2012. Increased incidence of non-Hodgkin lymphoma, leukemia, and myeloma in patients with diabetes mellitus type 2: A meta-analysis of observational studies. Blood 119, 4845–4850. Cazzaniga, M., Decensi, A., Pruneri, G., Puntoni, M., Bottiglieri, L., Varricchio, C., Guerrieri-Gonzaga, A., Gentilini, O.D., Pagani, G., Dell’Orto, P., Lazzeroni, M., Serrano, D., Viale, G., Bonanni, B., 2013. The effect of metformin on apoptosis in a breast cancer presurgical trial. British Journal of Cancer 109, 2792–2797. Chak, A., Buttar, N.S., Foster, N.R., Seisler, D.K., Marcon, N.E., Schoen, R., Cruz-Correa, M.R., Falk, G.W., Sharma, P., Hur, C., Katzka, D.A., Rodriguez, L.M., Richmond, E., Sharma, A.N., Smyrk, T.C., Mandrekar, S.J., Limburg, P.J., 2015. Metformin does not reduce markers of cell proliferation in esophageal tissues of patients with Barrett’s esophagus. Clinical Gastroenterology and Hepatology 13, 665–672. DeCensi, A., Guerrieri-Gonzaga, A., Pruneri, G., Puntoni, M., Cazzaniga, M., Vingiani, A., Serrano, D., Gentilini, O., Johansson, H., Aristarco, V., Lazzeroni, M., Varicchio, C., Petri, M., Viale, G., Bonanni, B, 2014–. Metformin decreases Ki67 in HER2þve ductal carcinoma in situ in a window of opportunity trial. San Antonio Breast Cancer Symposium. Decensi, A., Puntoni, M., Goodwin, P., Cazzaniga, M., Gennari, A., Bonanni, B., Gandini, S., 2010. Metformin and Cancer Risk in Diabetic Patients: A Systematic Review and Metaanalysis. Cancer Prevention Research (Philadelphia, PA) 3, 1451–1461. Deng, L., Gui, Z., Zhao, L., Wang, J., Shen, L., 2012. Diabetes mellitus and the incidence of colorectal cancer: An updated systematic review and meta-analysis. Digestive Diseases and Sciences 57, 1576–1585. Franciosi, M., Lucisano, G., Lapice, E., Strippoli, G.F., Pellegrini, F., Nicolucci, A., 2013. Metformin therapy and risk of cancer in patients with type 2 diabetes: Systematic review. PLoS ONE 8, e71583. Friberg, E., Orsini, N., Mantzoros, C.S., Wolk, A., 2007. Diabetes mellitus and risk of endometrial cancer: A meta-analysis. Diabetologia 50, 1365–1374. Gandini, S., Puntoni, M., Heckman-Stoddard, B.M., Dunn, B.K., Ford, L., Decensi, A., Szabo, E., 2014. Metformin and cancer risk and mortality: A systematic review and metaanalysis taking into account biases and confounders. Cancer Prevention Research (Philadelphia, PA) 7, 867–885. Gao, S., Yang, W.S., Gao, J., Wang, J., Xiang, Y.B., 2010. A meta-analysis of cohort studies on the association between diabetes and the risk of primary liver cancer. Zhonghua Yu Fang Yi. Xue Za Zhi 44, 711–716. Ge, Z., Ben, Q., Qian, J., Wang, Y., Li, Y., 2011. Diabetes mellitus and risk of gastric cancer: A systematic review and meta-analysis of observational studies. European Journal of Gastroenterology & Hepatology 23, 1127–1135. Goodwin, P.J., Parulekar, W.R., Gelmon, K.A., Shepherd, L.E., Ligibel, J.A., Hershman, D.L., Rastogi, P., Mayer, I.A., Hobday, T.J., Lemieux, J., Thompson, A.M., Pritchard, K.I., Whelan, T.J., Mukherjee, S.D., Chalchal, H.I., Oja, C.D., Tonkin, K.S., Bernstein, V., Chen, B.E., Stambolic, V., 2015. Effect of metformin vs placebo on and metabolic factors in NCIC CTG MA.32. Journal of the National Cancer Institute 107. Goodwin, P.J., Pritchard, K.I., Ennis, M., Clemons, M., Graham, M., Fantus, I.G., 2008. Insulin-lowering effects of metformin in women with early breast cancer. Clin Breast Cancer 8, 501–505. Hadad, S., Iwamoto, T., Jordan, L., Purdie, C., Bray, S., Baker, L., Jellema, G., Deharo, S., Hardie, D.G., Pusztai, L., Moulder-Thompson, S., Dewar, J.A., Thompson, A.M., 2011. Evidence for biological effects of metformin in operable breast cancer: A pre-operative, window-of-opportunity, randomized trial. Breast Cancer Research and Treatment 128, 783–794. Hardefeldt, P.J., Edirimanne, S., Eslick, G.D., 2012. Diabetes increases the risk of breast cancer: A meta-analysis. Endocrine-Related Cancer 19, 793–803. Higurashi, T., Hosono, K., Takahashi, H., Komiya, Y., Umezawa, S., Sakai, E., Uchiyama, T., Taniguchi, L., Hata, Y., Uchiyama, S., Hattori, A., Nagase, H., Kessoku, T., Arimoto, J., Matsuhashi, N., Inayama, Y., Yamanaka, S., Taguri, M., Nakajima, A., 2016. Metformin for chemoprevention of metachronous colorectal adenoma or polyps in post-polypectomy patients without diabetes: a multicentre double-blind, placebo-controlled, randomised phase 3 trial. The Lancet Oncology 17, 475–483. Hosono, K., Endo, H., Takahashi, H., Sugiyama, M., Sakai, E., Uchiyama, T., Suzuki, K., Iida, H., Sakamoto, Y., Yoneda, K., Koide, T., Tokoro, C., Abe, Y., Inamori, M., Nakagama, H., Nakajima, A., 2010. Metformin suppresses colorectal aberrant crypt foci in a short-term clinical trial. Cancer Prevention Research (Philadelphia, PA) 3, 1077–1083.

444

Metformin

Huang, W., Ren, H., Ben, Q., Cai, Q., Zhu, W., Li, Z., 2012. Risk of esophageal cancer in diabetes mellitus: A meta-analysis of observational studies. Cancer Causes Control 23, 263–272. Jiang, Y., Ben, Q., Shen, H., Lu, W., Zhang, Y., Zhu, J., 2011. Diabetes mellitus and incidence and mortality of colorectal cancer: A systematic review and meta-analysis of cohort studies. European Journal of Epidemiology 26, 863–876. Jing, W., Jin, G., Zhou, X., Zhou, Y., Zhang, Y., Shao, C., Liu, R., Hu, X., 2012. Diabetes mellitus and increased risk of cholangiocarcinoma: A meta-analysis. European Journal of Cancer Prevention 21, 24–31. Joshua, A.M., Zannella, V.E., Downes, M.R., Bowes, B., Hersey, K., Koritzinsky, M., Schwab, M., Hofmann, U., Evans, A., van der, K.T., Trachtenberg, J., Finelli, A., Fleshner, N., Sweet, J., Pollak, M., 2014. A pilot ‘window of opportunity’ neoadjuvant study of metformin in localised prostate cancer. Prostate Cancer and Prostatic Diseases 17, 252–258. Kalinsky, K., Crew, K.D., Refice, S., Xiao, T., Wang, A., Feldman, S.M., Taback, B., Ahmad, A., Cremers, S., Hibshoosh, H., Maurer, M., Hershman, D.L., 2014. Presurgical trial of metformin in overweight and obese patients with newly diagnosed breast cancer. Cancer Investigation 32, 150–157. Kasper, J.S., Giovannucci, E., 2006. A meta-analysis of diabetes mellitus and the risk of prostate cancer. Cancer Epidemiology, Biomarkers and Prevention 15, 2056–2062. Larsson, S.C., Mantzoros, C.S., Wolk, A., 2007. Diabetes mellitus and risk of breast cancer: A meta-analysis. International Journal of Cancer 121, 856–862. Larsson, S.C., Orsini, N., Brismar, K., Wolk, A., 2006. Diabetes mellitus and risk of bladder cancer: A meta-analysis. Diabetologia 49, 2819–2823. Larsson, S.C., Wolk, A., 2011. Diabetes mellitus and incidence of kidney cancer: A meta-analysis of cohort studies. Diabetologia 54, 1013–1018. Laskov, I., Drudi, L., Beauchamp, M.C., Yasmeen, A., Ferenczy, A., Pollak, M., Gotlieb, W.H., 2014. Anti-diabetic doses of metformin decrease proliferation markers in tumors of patients with endometrial cancer. Gynecologic Oncology 134, 607–614. Lee, J.Y., Jeon, I., Lee, J.M., Yoon, J.M., Park, S.M., 2013. Diabetes mellitus as an independent risk factor for lung cancer: A meta-analysis of observational studies. European Journal of Cancer 49, 2411–2423. Liao, S., Li, J., Wei, W., Wang, L., Zhang, Y., Li, J., Wang, C., Sun, S., 2011. Association between diabetes mellitus and breast cancer risk: A meta-analysis of the literature. Asian Pacific Journal of Cancer Prevention 12, 1061–1065. Ma, S., Zheng, Y., Xiao, Y., Zhou, P., Tan, H., 2017. Meta-analysis of studies using metformin as a reducer for liver cancer risk in diabetic patients. Medicine (Baltimore) 96, e6888. Mitri, J., Castillo, J., Pittas, A.G., 2008. Diabetes and risk of Non-Hodgkin’s lymphoma: A meta-analysis of observational studies. Diabetes Care 31, 2391–2397. Mitsuhashi, A., Kiyokawa, T., Sato, Y., Shozu, M., 2014. Effects of metformin on endometrial cancer cell growth in vivo: A preoperative prospective trial. Cancer 120, 2986–2995. Niraula, S., Dowling, R.J., Ennis, M., Chang, M.C., Done, S.J., Hood, N., Escallon, J., Leong, W.L., McCready, D.R., Reedijk, M., Stambolic, V., Goodwin, P.J., 2012. Metformin in early breast cancer: A prospective window of opportunity neoadjuvant study. Breast Cancer Research and Treatment 135, 821–830. Noto, H., Goto, A., Tsujimoto, T., Noda, M., 2012. Cancer risk in diabetic patients treated with metformin: A systematic review and meta-analysis. PLoS ONE 7, e33411. Noto, H., Tsujimoto, T., Sasazuki, T., Noda, M., 2011. Significantly increased risk of cancer in patients with diabetes mellitus: A systematic review and meta-analysis. Endocrine Practice 17, 616–628. Ren, H.B., Yu, T., Liu, C., Li, Y.Q., 2011. Diabetes mellitus and increased risk of biliary tract cancer: Systematic review and meta-analysis. Cancer Causes Control 22, 837–847. Schmid, D., Behrens, G., Jochem, C., Keimling, M., Leitzmann, M., 2013. Physical activity, diabetes, and risk of thyroid cancer: A systematic review and meta-analysis. European Journal of Epidemiology 28, 945–958. Schuler, K.M., Rambally, B.S., DiFurio, M.J., Sampey, B.P., Gehrig, P.A., Makowski, L., Bae-Jump, V.L., 2015. Antiproliferative and metabolic effects of metformin in a preoperative window clinical trial for endometrial cancer. Cancer Medicine 4, 161–173. Soranna, D., Scotti, L., Zambon, A., Bosetti, C., Grassi, G., Catapano, A., La, V.C., Mancia, G., Corrao, G., 2012. Cancer risk associated with use of metformin and sulfonylurea in type 2 diabetes: A meta-analysis. Oncologist 17, 813–822. Thakkar, B., Aronis, K.N., Vamvini, M.T., Shields, K., Mantzoros, C.S., 2013. Metformin and sulfonylureas in relation to cancer risk in type II diabetes patients: A meta-analysis using primary data of published studies. Metabolism 62, 922–934. Tian, T., Zhang, L.Q., Ma, X.H., Zhou, J.N., Shen, J., 2012. Diabetes mellitus and incidence and mortality of gastric cancer: A meta-analysis. Experimental and Clinical Endocrinology and Diabetes 120, 217–223. Wang, C., Wang, X., Gong, G., Ben, Q., Qiu, W., Chen, Y., Li, G., Wang, L., 2012. Increased risk of hepatocellular carcinoma in patients with diabetes mellitus: A systematic review and meta-analysis of cohort studies. International Journal of Cancer 130, 1639–1648. Wang, L., Wang, L., Zhang, J., Wang, B., Liu, H., 2017. Association between diabetes mellitus and subsequent ovarian cancer in women: A systematic review and meta-analysis of cohort studies. Medicine (Baltimore) 96, e6396. Wu, L., Zhu, J., Prokop, L.J., Murad, M.H., 2015. Pharmacologic Therapy of Diabetes and Overall Cancer Risk and Mortality: A Meta-Analysis of 265 Studies. Science Reporter 5, 10147. Zhu, Z., Wang, X., Shen, Z., Lu, Y., Zhong, S., Xu, C., 2013. Risk of bladder cancer in patients with diabetes mellitus: an updated meta-analysis of 36 observational studies. BMC Cancer 13, 310.

Further Reading Farmer, R.E., Ford, D., Forbes, H.J., et al., 2017. Metformin and cancer in type 2 diabetes: A systematic review and comprehensive bias evaluation. International Journal of Epidemiology 46, 745. Heckman-Stoddard, B.M., Gandini, S., Puntoni, M., Dunn, B.K., et al., 2016. Repurposing old drugs to chemoprevention: The case of metformin. Seminars in Oncology 43, 123–133. Lei, Y., Yi, Y., Liu, Y., et al., 2017. Metformin targets multiple signaling pathways in cancer. Chinese Journal of Cancer 36, 17. Provinciali, N., Lazzeroni, M., Cazzaniga, M., Gorlero, F., Dunn, B.K., Decensi, A., 2015. Metformin: Risk-benefit profile with a focus on cancer. Expert Opinion on Drug Safety 14, 1573–1585.

Mevalonate Pathway Heidrun Karlic, Ludwig Boltzmann Cluster Oncology, Hanusch Hospital, Vienna, Austria Franz Varga, Hanusch Hospital of WGKK and AUVA Trauma Centre Meidling, Vienna, Austria © 2019 Elsevier Inc. All rights reserved.

Glossary Aceto(acetyl)-CoA Acetyl CoA is formed from pyruvate in the final stages of glycolysis and is also a metabolite from fatty acid catabolism (beta oxidation) and a precursor of ketone bodies, which are accumulated in a status of hunger. Addition of an acetyl-residue results in aceto(acetyl)-CoA, a substrate for enzymes of the upper mevalonate pathway. Bisphosphonate Also named diphosphonates, group of chemical compounds which contain two phosphonate groups, this group of drugs is known to inhibit osteoclasts. GTPases A large family of hydrolase enzymes that can bind to and hydrolyze guanosine triphosphate (GTP). The GTP binding and hydrolysis takes place in the highly conserved G domain, which is common to all GTPases. NAD(P)H Nicotinamide adenine dinucleotide phosphate, abbreviated NADPþ, is a cofactor on anabolic reactions, which require NAD(P)H as a reducing agent. NADPþ differs from NADþ in the presence of an additional phosphate group on the 20 position of the ribose ring that carries the adenine moiety. Prenylation The addition of hydrophobic molecules to a protein or to a chemical compound. It is assumed that prenyl groups (3-methyl-but-2-en-1-yl) facilitate attachment to cell membranes. RHOA and RHOB A family of small ( 21 kDa) signaling G proteins. The members of the Rho family have been shown to regulate many aspects of intracellular actin dynamics and metabolic processes. Squalene A 30’carbon unsaturated oily liquid hydrocarbon and an important metabolite in cholesterol biosynthesis. Sterol regulatory element-binding protein (SREBP) SREBPs are transcription factors that bind to the sterol regulatory element DNA sequence TCACNCCAC. Statin A class of drugs that reduce the levels of lipids in the blood by altering the enzyme activity of 3-hydroxy-3methylglutaryl-CoA reductase (HMGCR). Sterol isoprenoids Basically, isoprenoids are a class of organic compounds composed of two or more units of hydrocarbons, with each unit consisting of five carbon atoms arranged in a specific pattern. These compounds include certain sterols, oxysterols, farnesol, and geranylgeraniol, as well as the diphosphate derivatives of isopentenyl, geranyl, farnesyl, geranylgeranyl, and presqualene. They regulate transcriptional and posttranscriptional events that in turn affect lipid synthesis, meiosis, apoptosis, developmental patterning, protein cleavage, and protein degradation.

Introduction The mevalonate pathway (MP) also known as the isoprenoid pathway or 3-hydroxy-3-methylglutaryl-CoA reductase (HMGCR) pathway is an anabolic pathway providing metabolites for multiple cellular processes in eukaryotes, archaea, as well as some bacteria, thus underscoring its importance for nearly all living organisms including humans. The mevalonate which is produced from acetoacetyl-CoA by HMGCR (Fig. 1) is further processed to sterol isoprenoids, such as cholesterol, which is an indispensable precursor of bile acids, lipoproteins, and steroid hormones, and to a number of hydrophobic molecules including nonsterol isoprenoids, such as dolichol, heme-A, isopentenyl tRNA, and ubiquinone. Intermediates of this network play important roles in the posttranslational modification of a multitude of proteins involved in inter- and intracellular signaling. Besides its key role for cholesterol synthesis, the MP has become a challenging topic, when a large number of experimental and clinical studies suggested that inhibition of the MP might have valuable interest in human disease the management of multiple human diseases, besides cardiovascular diseases. Molecules arising from the MP are essential for cell growth and differentiation. They appear to be potential interesting therapeutic targets for many areas of ongoing research: oncology, autoimmune disorders, atherosclerosis, and Alzheimer disease. Also, considerable progress has been made in understanding the pathophysiology of two autoinflammatory disorders resulting from an inherited deficiency of mevalonate kinase (MK), the first committed enzyme of the MP.

Biochemistry of the MP Upper MP The mevalonate–isoprenoid pathway involves first the synthesis of 3-hydroxy-3-methylglutaryl-CoA (HMG)-CoA from acetyl-CoA through acetoacetylCoA. HMGCR, one of the best-regulated enzymes in nature, catalyzes the conversion of HMG-CoA to mevalonic acid. HMGCR is the rate-limiting enzyme of MP.

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65000-6

445

446

Mevalonate Pathway

Fig. 1 Chemical reactions of the mevalonate pathway (MP). MP pathway enzymes condense three acetyl-CoA molecules in a two-step reaction to produce 3-hydroxy-3-methylglutaryl CoA (HMG-CoA). Both reactions are reversible and in equilibria, with the intracellular concentration of acetyl-CoA being the primary driver. HMG-CoA is then reduced by HMG-CoA reductase (HMGCR) to produce mevalonic acid (MA) via an irreversible reaction. MA is then converted into isopentenyl-diphosphate through a series of enzymatic steps, which serves as a monomeric unit for the consequent synthesis of all downstream metabolites. Dashed arrows indicate multiple steps.

In the absence of sterol isoprenoids in the cell, a family of transcription factors, named sterol regulatory element-binding proteins (SREBPs), directly activates HMGCR gene transcription. SREBPs regulate not only HMGCR gene transcription but also every step of cholesterol synthetic pathway by increasing gene expression of all the enzymes acting in the MP. In addition, other regulatory mechanisms can influence the activity of HMGR. The degradation rate of HMGR protein is influenced by cell’s requirements for isoprenoids. Cell’s requirements for isoprenoids will determine the rate of translation of HMGCR mRNA. The investigation of yeast HMGCR accounted for mevalonate production following the reaction: S  HMG  CoA þ 2NADPH þ Hþ /R  mevalonate þ 2NADP þ CoASH The enzyme HMGCR is found in eukaryotes, archaea, and some eubacteria. The conversion of the thioesterified HMGCoA carboxyl to an alcohol represents a two-step reduction, accounting for the stoichiometry of NADPH in the reaction. The reaction thus proceeds through the successive reduction steps to first produce bound mevaldyl-CoA, collapse of the thiohemiacetal to release CoASH and form mevaldehyde; the second reduction step then forms product mevalonate. The eukaryotic proteins (class I HMG-CoA reductases) are associated with the endoplasmic reticulum (ER) and interact through membrane spanning helices in the N-terminal domain. Consequently, the catalytic domain follows this membrane-anchoring sequence. These class I enzymes are potently inhibited by the class of statin drugs that effectively modulate sterol synthesis and, as a result, have been heavily investigated. There is no homologous sequence for membrane association at the N-terminus in the bacterial HMG-CoA reductases (class II), and a few of these (some in recombinant form) have been isolated as soluble proteins. The Pseudomonas mevalonii enzyme has a degradative function, allowing this microbe to grow on mevalonate as a carbon source. In contrast, the Staphylococcus aureus enzyme has a biosynthetic function and is encoded by a gene within a MP gene cluster. Despite the low overall sequence homology (< 20%) and overall protein structure architecture between class I (eukaryotic) and class II (bacterial) HMGCR enzymes, there is considerable similarity between these enzymes in the positioning of active site residues important to catalytic function. Residues from two different subunits contribute to an active site. The histidine proposed to function in protonation of product Coenzyme A is appropriately positioned for this role. The aspartate implicated by mutagenesis is located within the active site and involved in a hydrogen bond network with the lysine and the glutamate that have been identified by functional studies. While both lysine and glutamate are in close proximity to the HMG-CoA thioester carbonyl that is reduced to form

Mevalonate Pathway

447

mevalonate, there are different proposals concerning their precise roles in substrate carbonyl polarization and/or the proton transfers that accompany substrate reduction by NADPH.

Lower MP The lower MP converts mevalonate into the relatively unreactive isopentenyl pyrophosphate (IPP), which further is converted to the more reactive electrophile dimethylallyl pyrophosphate. There exist (at least) three variants of the lower MP: In eukaryotes, mevalonate (MV) is phosphorylated twice in the 5-OH position and then decarboxylated to yield IPP. In some archaea such as Haloferax volcanii, mevalonate is phosphorylated once in the 5-OH position, decarboxylated to yield isopentenyl phosphate (IP), and finally phosphorylated again to yield IPP (Archaeal Mevalonate Pathway I). A third MP variant found in the archaeon Thermoplasma acidophilum phosphorylates mevalonate at the 3-OH position followed by phosphorylation at the 5-OH position. The resulting metabolite, mevalonate-3,5-bisphosphate, is decarboxylated to IP and finally phosphorylated to yield IPP (Archaeal Mevalonate Pathway II). In eukaryotes, the above-mentioned phosphorylation is done by MK. MK also known as MVK or ATP: mevalonate 5phosphotransferase catalyzes the transfer of ATP’s c-phosphoryl to the C5 hydroxyl oxygen of mevalonic acid, resulting in formation of mevalonate 5-phosphate and ADP. The reaction was characterized in yeast; the protein is found in eukaryotes, archaea, and certain eubacteria. The enzyme was highly purified from porcine liver and was demonstrated to catalyze a sequential reaction with mevalonate substrate binding first and MgADP product released. MK is the second essential enzyme of the isoprenoid/cholesterol biosynthesis pathway, after HMGCR, catalyzing the phosphorylation of mevalonic acid into phosphomevalonate. Although MK not has the rate-limiting properties of HMGCR, it is demonstrated that MK activity is regulated via feedback inhibition by intermediates in the isoprenoid/cholesterol pathway geranyl pyrophosphate, farnesyl pyrophosphate (FPP), and geranylgeranyl pyrophosphate (GGPP). High-affinity feedback inhibition has also been observed using a recombinant human protein and contrasted with lowerpotency inhibition (10 5 M) of the Staphylococcus aureus enzyme. Inherited human mevalonate kinase (MVK) mutations are correlated with two diseases, accounting for mevalonic aciduria (MVA) and Hyper-IgD syndrome. Phosphomevalonate kinase (PMK) catalyzes the next step in isoprenoid/sterol biosynthesis, converting mevalonate 5-phosphate and ATP to mevalonate 5-diphosphate and ADP. Activity of this enzyme was demonstrated in pig liver, and the pig liver enzyme has subsequently been isolated and more extensively characterized. PMK is found in eukaryotes and some eubacteria. The amino acid sequences for animal and low-homology invertebrate PMK proteins are not orthologous to those for PMK in plants, fungi, and bacteria. Thus, the proteins that catalyze the enzymatic reaction differ widely, depending on their source. Animal and invertebrate PMK proteins are known for a tertiary and quaternary structure, which is typical of the nucleoside monophosphate kinase family. The other PMK proteins are members of the galactokinase, homoserine kinase, MK, and phosphomevalonate kinase (GHMP) family. The tissue-isolated pig enzyme is reported to catalyze an ordered sequential reaction with mevalonate 5-phosphate assigned as the first substrate bound and ADP as the last product released. A recombinant form of Streptococcus pneumoniae has been characterized and reported to catalyze a random sequential bi-bi reaction. A recombinant form of Enterococcus faecalis PMK has also been isolated and characterized. The sequence of human PMK has been deduced. Functional investigations of the recombinant human enzyme showed that the reaction catalyzed by PMK is a reversible reaction; kinetic constants of human PMK have been determined for both forward (formation of mevalonate 5-diphosphate) and reverse (formation of mevalonate 5-phosphate) reaction. In the next step, mevalonate diphosphate decarboxylase; various abbreviations appear in the literature: MVD (in Fig. 1), also known as MDD, MPD, DPMDC, catalyzes the ATP-dependent decarboxylation of mevalonate 5-diphosphate to form isopentenyl 5-diphosphate (IPP), as indicated in the equation (Fig. 1). This reaction is essential to the MP of polyisoprenoid and sterol synthesis. Activity has been measured in animals, plants, and yeast. Genetic complementation has implied activity in the Staphylococcus aureus and Trypanosoma brucei proteins. Highly purified active enzyme has been prepared from avian, porcine, and rat tissue and in recombinant form from bacteria expressing the yeast, human Trypanosoma brucei, and Staphylococcus aureus proteins. Characterization of the tissue-isolated protein implied the presence of an arginine that influences activity of the avian enzyme and documented the selectivity for divalent cation. The avian enzyme was also used to demonstrate that the transient phosphoryl transfer to the C3 oxygen proceeds with inversion of stereochemistry and, thus, no covalent E–P intermediate forms. Farnesyl transferase (FTase) and geranylgeranyl transferases (GGTase) are two enzymes that carry out the process of prenylation in the cell. This process involves the covalent attachment of hydrophobic molecules (either the C-15 isoprene farnesyl or the C-20 isoprene geranylgeranyl groups) to the C-terminal end of some proteins including the g-subunit of heterotrimeric G proteins, hemeA, nuclear lamins, and small GTP-binding proteins. Prenylation promotes the attachment of these proteins to internal cell membranes by means of a lipid anchors such as palmitate. Such posttranslational modifications and activation GTP-binding proteins Rho, Rac, Rab, Rap, Ras play an important role in many important signaling cascades within the cell. Downstream from FPP the squalene synthase (FDFT1, also known as farnesyl-diphosphate farnesyl transferase 1) catalyzes the first committed step of the specific hepatic cholesterol biosynthesis at the final branch point of the cholesterol biosynthetic pathway, converting farnesyl-pyrophosphate into squalene. Squalene is then converted after a two-step cyclization into lanosterol, which is converted to cholesterol after a series of additional reactions.

448

Mevalonate Pathway

Another downstream branch from FPP is catalyzed by the enzyme geranylgeranyl diphosphate synthase 1 (GGPS1) and leads to the synthesis of GGPP from farnesyl diphosphate and isopentenyl diphosphate. GGPP is an important molecule responsible for the C20-prenylation of proteins and for the regulation of a nuclear hormone receptor (Fig. 1).

Prenylation of Small GTPases and Brain Function As mentioned earlier, FPP and GGPP are substrates for protein prenylation, and small GTP-binding proteins (sGTPases) represent the largest and most extensive characterized group of prenylated proteins. This superfamily is classified according to sequence and function similarity, into Ras, Rho, Rab, Sar1/Arf, and Ran families. The regulation of the activity of sGTPases is somewhat complex, and it has been extensively reviewed in greater detail, nonetheless, in general terms, sGTPases cycle between a guanosine diphosphate (GDP)- and GTP-bound conformation. The GDP-bound conformation is generally considered an inactive state, while the GTP-bound sGTPase activates downstream pathways by binding to specific effectors. sGTPases associate with the cytoplasmic leaflet of cellular membranes, where they perform their respective function. However, sGTPases are not transmembrane proteins, and in order to anchor to membranes they have to be covalently attached to lipid groups, with the exception of the so-called Ran (RAsrelated nuclear protein) also known as GTP-binding nuclear proteins, for which no lipid modification has been reported. All the members of Ras, Rho, and Rab families undergo the lipid posttranslational modification of prenylation as the first committed step toward membrane association, while ARF family members undergo myristoylation. Both FPP and GGPP can be used as moieties for protein prenylation, catalyzed by the enzymes FTase and GGTase-I and -II, respectively. Rho GTPases are essential players in neuronal development and function. Their activity is critical for neuronal cell migration, axon growth and guidance, dendritic arborization, dendritic spine formation and stabilization, growth cone motility and collapse, and synapse formation. Therefore, it is not surprising that studies in which statins or inhibitors of protein prenyltransferases were used, a decrease in neurite outgrowth and dendritic spines density was observed. In fact, inhibiting GGTase-I in vivo induces deficits in memory and learning in mice. However, there are also a number of studies showing that statins can actually increase neurite outgrowth, which might be explained by the different cell models, dosages, and statins that were used. Interestingly, neuronal survival has also been observed to diminish in the presence of statins, which was shown to be independent of a reduction on cholesterol levels, since the reversal of this phenotype was obtained by FPP/GGPP supplementation, but not by cholesterol addition. Studies have shown that increased expression of CYP46A1 (¼ cholesterol 24-hydroxylase) in neurons enhanced prenylation and activation of sGTPases of the Rho and Rab family, and that this effect was dependent on the activation of the MP and GGTase-I activity, which is in accordance to previously mentioned studies. Moreover, it has been reported that this increase of sGTPases prenylation induced by CYP46A1 leads to an increase in neuronal dendritic outgrowth and dendritic protrusion density and elicits an increase of synaptic proteins in crude synaptosomal fractions, further highlighting the importance of how the activation of the MP and sustained production of isoprenoids are essential for neuronal development and function. It is noteworthy that FPP and GGPP levels are much higher in the brain than in other tissues. Accordingly, in mouse brain cytosol, FPP and GGPP synthase activities are higher than those in the corresponding fractions from the liver, perhaps reflecting a higher demand for protein prenylation in the brain. Furthermore, it has been reported that in aged mouse brain, there is a reduction of GGTase-I activity, leading to a reduction of Rho GTPases membrane association, which is suggested to be one of the mechanisms underlying age-related cognitive dysfunction. There is a great lack of studies comparing the levels of nonsteroid products from the MP between neurons and astrocytes. It is noteworthy that CoQ10, which derives from the decaprenoide intermediary, has actually been measured and compared between both cell types (unpublished data reported in a PhD dissertation). Neurons were found to possess CoQ10 levels at around 68 pmol/mg protein, while in astrocytes the levels were significantly lower, about 38 pmol/mg protein. These results suggest that although neurons might have a lower cholesterol synthesis than astrocytes, an opposite profile might exist for isoprenoid synthesis, which supports our hypothesis that neurons undergo a shift from the postsqualene to the nonsterol branch of the MP. Based on those results and data presented in literature we hypothesize that the postsqualene branch and cholesterol synthesis might be downregulated in neurons to redirect the intrinsic MP toward the nonsterol branch, producing important isoprenoid intermediaries for neuronal function, while relying on external supply of cholesterol. Nevertheless, many questions remain unanswered: What are the underlying regulatory mechanisms involved the differential expression patterns of the MP enzymatic machinery in distinct neural cell types? Is the synthesis rate of isoprenoids different in different neural cell types? Is there a shuttle of isoprenoids from astrocytes to neurons? Is there a soluble factor released by astrocytes that signals for the shutdown of the postsqualene pathways in neurons? Are specific neuronal signaling pathways being modulated by isoprenoids levels, in a prenylation independent manner? Is there a specific deregulation of the nonsterol branch in neurons in pathological contexts? In order to answer these questions, the challenges ahead are also dependent on the development of new tools that facilitate imaging and quantification of the prenylome in neurons and astrocytes and to the identification their cognate protein prenyltransferases or unknown prenyl-binding proteins.

Evidence for a Redirection of the MP in Neurons A compelling body of evidence suggests that, although neurons retain an active MP and can synthesize cholesterol, the efficiency of cholesterol synthesis is markedly lower when compared to astrocytes, possibly due to lower expression levels of enzymes belonging

Mevalonate Pathway

449

to the postsqualene branch. There exists a hypothesis indicating that the postsqualene branch and cholesterol synthesis might be downregulated in neurons, and with this downregulation there is a redirection of the intrinsic MP toward the nonsterol branch, while neurons still rely on external supply of cholesterol. This hypothesis seems to be corroborated by data from other laboratories that published a RNA-Seq transcriptome of 7 days postnatal (P7) mice cortical neurons, astrocytes, oligodendrocytes, and vascular cells that corroborated a previous analysis of an array dataset using RNA isolated from forebrain cells of P7 and P16–17 (¼ 16– 17 days old) mice. Based on data present in these two distinct datasets, the expression levels of genes involved in nonsterol, pre, and postsqualene pathways in neurons and astrocytes and also between P7 and P16 neurons were compared. Both datasets suggest that the pre- and postsqualene branches in P7 neurons are substantially downregulated compared to astrocytes (blue and red, respectively). Nonetheless, the genes related to the nonsterol pathway have a similar or a higher-expression level in neurons when compared to astrocytes, with the exception of prenyl (solanesyl) diphosphate synthase, subunit 1 (Pdss1). Furthermore, the array data also enabled to compare the expression levels of genes involved in the MP in P16 neurons when compared to P17 astrocytes. Although there are two genes upregulated in the presqualene pathway (blue), the majority of genes have decreased mRNA levels in neurons when compared to astrocytes. Similar to P7 cells, while P16 neurons exhibit a clear downregulation of postsqualene pathway, they maintain or even increase the mRNA levels of genes belonging to the nonsterol branch, when compared to astrocytes. These expression patterns may indicate that while maintaining the efficiency of the nonsterol branch some what intact or even upregulated, there is a robust downregulation of the pre- and postsqualene pathway in neurons compared to astrocytes, which is a profile that seems to be conserved between these two cell types throughout CNS maturation. The fact that the nonsterol branch, in contrast to the pre- and postsqualene pathways, is not downregulated and might actually be somehow induced in neurons, is in line with the idea that the neuronal MP might be favoring the production of isoprenoids in detriment of cholesterol, which is ought to be supplied by astrocytes. The decrease in mRNA levels from MP-associated genes in the presqualene pathway observed in neurons also favors the nonsterol pathway. Indeed, the affinity of GGPPS for FPP (Km value of 0.6 mM) is much higher than SQS (Km value of  15 mM), thus under limited concentrations of FPP, isoprenoid synthesis will be favored. These kinetic findings together with the fact that during neuronal maturation the mRNA levels of Gggps increase while Fdft1 (the gene that codes for SQS) decrease argues that the nonsterol branch might be boosted in neurons in detriment of the postsqualene pathway. Accordingly, as previously mentioned, sGTPases activity is crucial during neuronal development and has been widely associated with neuronal dendritic development and synaptic regulation. Hence, as neurons mature, the demand for isoprenoids to prenylate sGTPases should increase, which is in agreement with the hypothesis of a redirection of the MP branch toward production of nonsterol intermediaries, while cholesterol needs are met by uptake from astrocytic-derived lipoproteins.

Role of the MP in Cancer Cancer cells reprogram their metabolism to provide energy and the essential building blocks required to maintain their aberrant survival and growth. This reprogramming may occur through either mutations in metabolic enzymes (e.g., isocitrate dehydrogenases (IDHs)) or alterations in cell signaling owing to oncogenic events and/or the remodeled tumor microenvironment. These activated signaling cascades in turn deregulate the expression and/or the activity of enzymes in key metabolic pathways, including the MP. The MP uses acetyl-CoA, NADPH, and ATP to produce sterols and isoprenoids that are essential for tumor growth (Fig. 1). The production of acetyl-CoA occurs following glucose, glutamine, or acetate consumption, which are often increased in cancer cells. NADPH is produced from a variety of sources, including the pentose phosphate pathway, malic enzyme, and IDHs. Therefore, the MP is highly integrated into the overall metabolic network of cancer cells. The transcription of genes encoding MP enzymes is primarily controlled by the SREBP family of basic helix–loop–helix leucine zipper transcription factors. When intracellular sterol levels are high, the SREBPs are maintained in an inactive state at the ER, where some MP enzymes are also localized. In response to sterol deprivation, a feedback response is initiated that leads to the SREBPs, along with their binding partner SREBP cleavageactivating protein (SCAP), dissociating from the insulin-induced genes (INSIGs) and translocating from the ER to the Golgi. At the Golgi, the SREBPs are sequentially cleaved by site-1 protease and site-2 protease, and they translocate to the nucleus where they bind to sterol regulatory elements (SREs) in the promoters of their target genes and activate the transcription of MP genes to restore sterol and isoprenoid levels. The importance of MP metabolites to the survival of cancer cells has been highlighted by recent studies that have identified a large number of MP enzymes as essential for the survival of several cancer cell lines. Additionally, numerous studies have shown that the statin family of drugs, which inhibit the initial flux-controlling enzyme of the MP, HMGCR, decrease growth and increase apoptosis in many cancer types in vitro and in vivo. These observations point to the MP as being a key dependency in tumors, and one that is readily targetable. The MP has been suggested by some studies to be oncogenic. Early work in chronic lymphocytic leukemia showed that mevalonate can stimulate replication in primary leukemic cells. In another study, overexpression of the catalytic domain of HMGCR in primary mouse embryonic fibroblasts cooperated with HRASG12V to promote foci formation, suggesting that HMGCR is a metabolic oncogene. In addition, the direct infusion of MVA into mice harboring breast cancer cell xenografts caused an increase in tumor growth. Data from primary patient samples also suggest a role for the MP in promoting tumorigenesis, with a higher expression of MP genes correlating with poor prognosis in breast cancer. Collectively, this evidence indicates that the MP has a key role in cancer.

450

Mevalonate Pathway

A series of evidences demonstrated that the MP is deregulated in cancer through aberrant cell signaling, which in turn establishes a tumor vulnerability that can be therapeutically targeted to improve outcomes for cancer patients.

MP-Derived Metabolites in Cancer Initially, the regulation and function of the MP and its metabolites were studied in the context of normal and hypercholesterolemic tissues, which led to the Nobel prize-winning discoveries of Bloch and Lynen in 1964, and Brown and Goldstein in 1985. In recent years, the importance of MP-derived metabolites in cancer has become increasingly appreciated (discussed later).

Cholesterol Cholesterol is an important component of most cellular membranes. Highly proliferative cancer cells need to produce membranes rapidly, and an increase in cholesterol synthesis contributes to this process. Cholesterol is also an integral component of lipid rafts, which are necessary to form signaling complexes. The cholesterol content of the ER has recently been linked to the antiviral type I interferon (IFN) response, with low ER cholesterol triggering an IFN response in macrophages that protects mice from viral challenge. Therefore, it is possible that high levels of cholesterol, produced by the MP, could have a role in protecting cancer cells from immune surveillance and various therapies. Cholesterol also serves as the precursor of downstream products, such as steroid hormones and oxysterols: steroid hormones drive the initiation and progression of various cancers, including breast and prostate carcinomas; increased oxysterol production can activate the liver X receptors, which have been proposed to be therapeutic targets in multiple cancer types. Therefore, cancer cells require cholesterol for growth and survival, and decreasing intracellular cholesterol biosynthesis is a promising anticancer strategy.

Isopentenyl-Diphosphate In human cells, the MP is the sole intracellular source of isopentenyl-diphosphate (IPP) (Fig. 1). Aberrant activation of the MP in cancer results in increased intracellular levels of IPP, which has been shown to activate host gd T cells that subsequently kill the IPP-overexpressing cells. These observations led to phase I clinical trials that evaluated the in vivo expansion of gd T cells in response to zoledronate, a bisphosphonate (BP) that inhibits farnesyl diphosphate synthase (FDPS) and leads to the accumulation of IPP, in combination with interleukin-2 (IL-2) treatment in advanced-stage breast cancer and prostate cancer. In both studies, the therapy was well-tolerated and the number of sustained peripheral gd T cells correlated with improved clinical outcome. Future phase II clinical trials will reveal whether combined zoledronate and IL-2 therapy is an effective anticancer strategy.

Farnesyl-Diphosphate and Geranylgeranyl-Diphosphate Farnesyl-diphosphate (FPP) and geranylgeranyl-diphosphate (GGPP) are produced by sequential condensation reactions of dimethylallyl-diphosphate with two or three units of IPP, respectively. FPP and GGPP contain hydrophobic chains that are essential for the isoprenylation of proteins. This posttranslational modification tethers proteins to cell membranes, enabling proper protein localization and function. Most small GTPasesdmany of which are involved in tumorigenesis, such as RAS and RHOdare isoprenylated; inhibition of the MP can reduce the isoprenylation of these small GTPases and can induce the death of some cancer cells. This cell death can be reversed by the addition of GGPP, and sometimes FPP, suggesting that these MP metabolites are essential for tumor cell viability. Evidence suggests that it is unlikely that any one isoprenylated protein can be assigned functional responsibility for this cancer cell dependency on GGPP and FPP; instead, it seems that this is a “class effect,” with the depletion of these isoprenoid pools potentially affecting the many proteins that are isoprenylated. Despite this dependency, directly inhibiting the isoprenylation of proteins using geranylgeranyl transferase inhibitors (GGTIs) or farnesyl transferase inhibitors (FTIs) has not been a successful anticancer strategy to date. The rationale behind these drug development programs was that key isoprenylated oncoproteins, such as RAS, could be targeted. However, the efficacy of FTIs was impeded by alternative isoprenylation using GGPP, and GGTIs have been disappointingly toxic. Further development of next-generation FTIs and GGTIs remains a fairly limited and focused area of research.

Dolichol Dolichol is derived from 18 to 20 IPP molecules and is an essential component of the N-glycosylation of nascent polypeptides in the ER. Protein N-glycosylation is frequently altered in cancer and can contribute to tumor formation, proliferation, and metastasis. Not all N-glycans are associated with tumor progression; the complex branching of N-glycans leads to tumor-suppressive properties in some cancers. Glucose-derived N-acetylglucosamine has recently been shown to be necessary for the N-glycosylation of SCAP before ER-to-Golgi translocation. The SCAP–SREBP complex thus remains inactive in the ER when glucose is absent, even in the presence of low levels of sterols.

Mevalonate Pathway

451

Coenzyme Q Isoprenoids are also used to produce the quinone coenzyme Q (CoQ). The hydrophobic isoprenoid chain localizes CoQ to the inner membrane of the mitochondria, where the quinone group transfers electrons from complex I or II to complex III of the electron transport chain, thus enabling ATP production. Therefore, CoQ is crucial for ATP production in cancer cells that rely on oxidative phosphorylation to produce energy.

Oncogenic Regulation of the MP Intracellular pools of MP metabolites are tightly regulated by modulating the expression and activity of the MP enzymes. MP gene expression is mainly controlled by the SREBP transcription factors (Fig. 2). There are three SREBP proteins, which are transcribed from two genes: SREBP2 is transcribed from the SREBF2 gene and is the main transcription factor for MP-associated genes; SREBP1a and SREBP1c are transcribed from alternative start sites in the SREBF1 gene, with SREBP1a regulating the expression of both MP and fatty acid metabolism genes, and SREBP1c predominantly regulating the expression of fatty acid metabolism genes. Chromatin immunoprecipitation followed by sequencing (ChIP-seq) studies have indicated some overlap in the target genes of each SREBP, including MP genes, indicating some redundancy. Most studies have also shown an overlap in the regulation of the SREBPs; however, the majority of studies limit full characterization to SREBP1, and most do not distinguish between SREBP1a and SREBP1c as available antibodies cannot differentiate between the two. Given the importance of the MP in cancer, a complete characterization of SREBP2 in transformed cells is needed. In recent years, oncogenic and tumor-suppressive pathways have been shown to converge on the MP and its regulatory feedback loop. Cancer cells, with their aberrant growth and metabolism, are thus primed to upregulate the MP to provide essential building blocks for continued proliferation. The integration of cellular signaling from growth factors and essential metabolites, with the regulation of the MP and its SREBP-regulated feedback response, highlights the importance of this pathway in cancer.

Fig. 2 Targets of inhibitors of the MP. Statins inhibit HMGCR, thereby reducing MP metabolites that are also essential for cancer cell growth and survival. This triggers sterol regulatory element-binding protein (SREBP) activation and the transcription of MP genes, thus restoring MP activity. Dipyridamole is one example of an agent that inhibits SREBP cleavage, preventing the restorative feedback response and increasing apoptosis in multiple cancer types. Combining SREBP cleavage inhibitors with statins may increase the therapeutic response compared with the use of statins alone. Dashed boxes represent metabolites or steps that are reduced by the indicated treatments. Bisphosphonates act downstream of statins and inhibit farnesyl pyrophosphate synthase, a key enzyme in the MP, with consecutive decrease of the formation of isoprenoid lipids such as farnesyl pyrophosphate and geranylgeranyl pyrophosphate. This part of the MP is also targeted by farnesyl transferase inhibitors.

452

Mevalonate Pathway

PI3K–AKT The PI3K–AKT signaling pathway is a major regulator of cell survival and proliferation in response to growth factors. It is the single most frequently altered pathway in cancer, and the second most frequently mutated gene is PIK3CA, which encodes PI3K catalytic subunit alpha. Inactivating mutations in the PI3K–AKT pathway negative regulator PTEN and/or the hyperactivity of growth factor receptor tyrosine kinases are also common in cancer. Alterations in the PI3K–AKT pathway generally act to augment signaling and consequently increase the proliferation of cancer cells. PI3K–AKT can activate the MP through various mechanisms. For example, the stimulation of PI3K–AKT signaling by growth factors, such as insulin, platelet-derived growth factor, and vascular endothelial growth factor, can increase the mRNA and protein expression of SREBP1 and SREBP. It should be noted that although PI3K–AKT signaling strongly and consistently increases the mRNA and protein levels of SREBP1a and SREBP1c, its effects on SREBP2 expression are context dependent. AKT has also been suggested to increase the stability of nuclear SREBP1a, SREBP1c, and SREBP2 by preventing their proteasomal degradation mediated by the F-box and WD repeat domain containing 7 (FBXW7) E3 ubiquitin ligase. The importance of this degradation pathway is highlighted by an increase in cholesterol and fatty acid synthesis in FBXW7-deficient cells. The residues that are recognized by FBXW7 are phosphorylated by glycogen synthase kinase-3b; AKT, which inhibits this phosphorylation, may prevent FBXW7-mediated degradation of the SREBPs. Insulin also causes the dissociation of INSIG from SCAP– SREBP1c in a sterol-independent manner, leading to the increased transcription of MP genes. These studies were further validated through genetic approaches, in which SREBP1 and SREBP2 expression and activity were increased with the expression of constitutively active PI3K or AKT, and abrogated by dominant-negative AKT. The increase in lipid and cholesterol production that is mediated by the PI3K–AKT–SREBP axis promotes the proliferation of cancer cells and tumorigenesis in vitro and in vivo. Increased MP activity is inconsequential without the availability of both acetyl-CoA and NADPH, and PI3K–AKT signaling meets this requirement by increasing glucose uptake and the rate of glycolysis in cancer cells. Conversely, inhibition of the MP decreases PI3K activity, possibly through decreased RAS isoprenylation, thus demonstrating a two-way regulatory relationship between PI3K–AKT signaling and the MP.

mTOR Complex 1 Downstream of PI3K–AKT signaling, mTOR complex 1 (mTORC1), acts as a sensor of growth signals (such as insulin) and nutrients (such as amino acids) to regulate cellular growth. mTORC1 is often deregulated in cancer, and this supports aberrant growth. mTORC1 increases mRNA translation by phosphorylating and activating ribosomal S6 kinase 1 (S6K1; also known as RPS6KB1) and repressing the activity of the inhibitor of cap-dependent translation, eukaryotic translation initiation factor 4E-binding protein 1 (4EBP1; also known as EIF4EBP1). SREBPs are major downstream effectors of mTORC1 signaling, as evidenced by increased lipogenesis in response to mTORC1 activation. The observation that SREs are the most common regulatory elements in mTORC1induced genes further strengthens the link between mTORC1 and the SREBPs. This link is also evident in samples from patients with primary breast cancer, as patients with high levels of phosphorylated S6K1 had correspondingly high expression of SREBP target genes, such as fatty acid synthase (FASN), low-density lipoprotein receptor (LDLR), and mevalonate kinase (MVK). This study also compared proteins from tumor samples and adjacent normal breast samples and described an increase in FASN protein levels in the tumors that had higher levels of phosphorylated S6K1. mTORC1 can regulate the SREBP transcription factors at multiple levels although there are some cell- and tissue-type differences. For example, S6K1 has been shown to activate SREBP2 processing and increase the expression of MP genes in a hepatocellular carcinoma (HCC) cell line although the mechanism involved remains unclear. Greater understanding of the role of mTORC1 in SREBP activity came with the development of torins, which are mTOR catalytic site inhibitors. The original allosteric mTOR inhibitor, rapamycin, prevents the phosphorylation of S6K1 but does not inhibit 4EBP1 phosphorylation equally in all systems. By contrast, torins inhibit the phosphorylation of multiple mTOR targets, including S6K1 and 4EBP1. Recent work comparing torin and rapamycin action implicated a role for lipin 1 (LPIN1) in mediating the effects of mTORC1 on the SREBPs. LPIN1 is a nuclear phosphatidic acid phosphatase that is inhibited through direct phosphorylation by mTORC1, independently of S6K1. Active, unphosphorylated LPIN1 indirectly prevents the transcription of SREBP target genes by preventing the SREBPs from binding to chromatin although the mechanism involved remains unclear. A further link between LPIN1 and the MP was uncovered by studies using skeletal muscle, in which statins and LPIN1 were shown to increase autophagy. Given the role of SREBP2 in transcribing numerous autophagy genes, further work is needed to fully understand the interplay between mTORC1, LPIN1, and the SREBPs. The position of the SREBPs as key effectors of mTORC1 signaling presents a potential vulnerability in tumors that have deregulated mTORC1 activity. Previous studies have linked the loss of SREBPs in breast cancer to the induction of ER stress, which induced apoptosis through mTOR. A separate study showed that genetic knockdown of SREBF1 and/or SREBF2 reduced proliferation and increased cell death in mTORC1-activated breast cancer cell lines. The observation that double knockdown of SREBF1 and SREBF2 showed the greatest proapoptotic effect suggests that small-molecule inhibitors that target both SREBP1 and SREBP2 may have the greatest therapeutic benefit.

AMP-Activated Protein Kinase With an opposing role to that of mTORC1, AMP-activated protein kinase (AMPK) acts to dampen anabolic pathways when intracellular ATP levels are low. This role of AMPK as an energy sensor and central regulator of metabolism is crucial in metabolic

Mevalonate Pathway

453

disorders such as type 2 diabetes and cancer. AMPK was discovered through its ability to phosphorylate and reduce the activity of microsomal HMGCR in rat liver extracts. Further studies showed that AMPK phosphorylates Ser872 within the catalytic domain of HMGCR, inhibiting its enzymatic activity in a manner that is independent of its feedback regulation by MP metabolites. The SREBPs are also direct targets of AMPK phosphorylation. Activated AMPK specifically interacts with both the precursor and the nuclear forms of SREBP1c and SREBP2, and phosphorylation by AMPK inhibits SREBP proteolytic processing and transactivation activity. Activation of AMPK in HepG2 liver cancer cells by either polyphenols or metformin stimulates this phosphorylation, which suppresses the accumulation of SREBPs in the nucleus under hyperglycemic and hyperinsulinemic conditions. Moreover, activation of AMPK in the livers of insulin-resistant mice inhibited the transcription of enzymes that are involved in lipid and cholesterol biosynthesis, including the MP enzymes 3-hydroxy-3-methylglutaryl-CoA synthase 1 (HMGCS1) and HMGCR, which consequently resulted in a decrease in hepatic triglyceride and cholesterol levels. AMPK can thus inhibit MP activity both directly via the phosphorylation of HMGCR and indirectly through the phosphorylation and repression of SREBPs. However, the relevance of this regulation in the context of cancer is poorly understood. The MP may also regulate AMPK activity, thereby forming a feedback loop. The tumor suppressor liver kinase B1 (LKB1; also known as STK11), which phosphorylates and activates AMPK, is farnesylated at a highly conserved carboxyterminal CAAX motif. Knock-in mice expressing a mutant form of LKB1, which could not be farnesylated, exhibited reduced membrane-bound LKB1 and impaired AMPK activity. This hints at a negative feedback loop, in which the activation of AMPK in response to decreased cellular energy results in the inhibition of the MP via the phosphorylation of HMGCR and the SREBPs. This in turn reduces the FPP pool within the cell, thereby hindering LKB1 farnesylation and inhibiting AMPK activation.

p53 and RB TP53, which encodes the p53 tumor suppressor, is one of the most frequently altered genes in cancer, and mutations within the coding region of TP53 can confer oncogenic properties to p53. Two gain-of-function mutations (TP53R273H and TP53R280K) enable p53 to functionally interact with nuclear SREBP2 and increase the transcription of MP genes. This MP gene activation was necessary and sufficient for mutant p53 to disrupt normal breast acinar morphology, and mutant TP53 expression in primary breast cancer tissues was correlated with the increased expression of sterol biosynthesis genes. Conversely, wild-type TP53 can reduce lipid synthesis under conditions of glucose starvation by inducing the expression of LPIN1, which, as described earlier, can prevent the association of SREBPs with chromatin. TP53R273H and TP53R280K mutations are also found in tumors from tissues other than the breast, for example, the ovaries, prostate, and lung. The interplay between TP53 and the MP suggests that the MP may be a novel therapeutic target for tumors that harbor these specific p53 gain-of-function mutations. The tumor suppressor protein RB has also been implicated as a regulator of the MP. In a mouse model of adenoma, loss of Rb1 (which encodes RB) enhanced isoprenylation and activation of NRAS. Loss of RB relieved the suppression of the transcription factors E2F1 and E2F3, which were shown to bind and activate the promoters of numerous prenyltransferase genes, FDPS and SREBF1. Moreover, RB prevented the association of SREBP1 and SREBP2 with the FDPS promoter, suggesting that RB negatively regulates the MP at both the transcriptional and the posttranslational levels.

MYC The MYC transcription factor is a potent oncogene that can drive transformation in multiple cancer types. It is deregulated in more than 50% of cancers and can reprogram cancer cell metabolism to enable the proliferation and survival of cancer cells. Like the SREBPs, MYC is a basic helix-loop-helix dimerizing protein and it has been shown to bind to SREBP1 to drive somatic cell reprogramming into induced pluripotent stem cells. Analysis of data from the Encyclopedia of DNA Elements (ENCODE) project also shows that MYC binds to promoters of MP genes in close proximity to SREBP1 and SREBP2 binding regions, suggesting that MYC can contribute to the expression of MP enzymes. As the MP is essential for cancer cells, and because MYC has a major role in metabolic regulation, deregulated MYC may ensure that MP metabolites are not limiting for tumorigenesis. The MP was also shown to be important in a MYC-driven transgenic model of HCC. In that study, atorvastatin reduced tumor initiation and growth, possibly through reduced isoprenylation of the RHO-family GTPase RAC1, leading to the activation of serine/threonine-protein phosphatase 2A, which is a negative regulator of MYC. More recently, Mycþ/ mice (which are haploinsufficient) were shown to have an increased lifespan, which was associated with the decreased expression of MP genes, including Hmgcr and Srebf2. Given the importance of MYC in driving cancer, and the difficulty of targeting it therapeutically, further work is warranted to uncover the relationship between MYC and the MP.

Signaling From the MP Altered metabolism in tumors not only fulfills the energetic and biosynthetic needs of a dividing cell but also produces metabolites that are important for downstream signaling. This is particularly true of the isoprenoid and sterol metabolites produced by the MP, which are also used by cancer cells to modulate multiple downstream signaling pathways that are important for tumor progression.

454

Mevalonate Pathway

Yes-Associated Protein and TAZ It was recently shown that the oncogenes Yes-associated protein (YAP) and transcriptional coactivator with PDZ-binding motif (TAZ; also known as WWTR1) require the MP to be fully functional. YAP and TAZ are transcriptional coactivators that facilitate the transcriptional activation of progrowth genes and the repression of proapoptotic genes. The nuclear localization of YAP and TAZ is negatively regulated, partly by the activation of the tumor-suppressive Hippo signaling pathway. Activation of the Hippo cascade results in the phosphorylation and activation of large tumor suppressor kinase 1 (LATS1) and LATS2, which phosphorylate YAP and TAZ and retain them in the cytoplasm. YAP and TAZ nuclear localization requires the MP, as concurrent knockdown of SREBF1 and SREBF2 reduces nuclear localization of YAP and TAZ. These effects were mimicked by GGTIs (geranylgeranyl transferase inhibitors) and were prevented by a RHOA mutant that does not require geranylgeranylation. This suggests that SREBP-mediated induction of the MP maintains intracellular GGPP pools, which is necessary for RHOA upregulation and for downregulation of RHOB (which is essential for initiating protein degradation and recycling through an endolysosomal pathway), as well as YAP and TAZ nuclear localization. Although some studies showed that MP-mediated YAP and TAZ signaling is independent of LATS1 and LATS2 via RNA interference-knockdown experiments, one study demonstrated that both atorvastatin treatment and GGTI treatment increase the phosphorylation of LATS1 and LATS2, suggesting that geranylgeranylation regulates Hippo signaling. A separate study reported constitutive SREBP activation in the livers of mice with a liver-specific Lats2 deletion, which corresponded to an increase in free cholesterol in the liver and protection from p53-mediated apoptosis. Activation of the MP and activation of YAP and TAZ are correlated with mutant TP53 expression in primary tumors, suggesting a dysfunctional mutant p53–SREBP–YAP– TAZ axis in cancer. Overexpression of TP53R280K in a TP53-null cell line activated YAP and TAZ only when the MP was active, suggesting that the MP is a crucial intermediate in the oncogenic activation of YAP and TAZ by mutant TP53.

Hedgehog Cholesterol has a multifaceted role in the regulation of cell signaling. For example, the Hedgehog (HH) signaling pathway, which has important roles in vertebrate development and tumorigenesis, is regulated by sterols at multiple levels. Cholesterol itself can serve as a substrate for the posttranslational modification of HH ligands, which is required for their proper trafficking. Cholesterol and cholesterol-derived oxysterols can also activate HH signal transduction in medulloblastoma, whereas inhibition of the MP or downstream sterol biosynthesis decreased HH signaling and reduced cell proliferation.

Steroid Hormone Signaling Cholesterol also serves as the precursor of steroid hormones, which drive the initiation and progression of cancers such as hormonedependent breast cancer and prostate cancer. In breast cancer, patients with estrogen receptor-a (ERa)-positive disease are commonly treated with aromatase inhibitors to deprive the tumors of estrogen. Recent work demonstrated that long-term estrogen deprivation of ERa-positive breast cancers leads to the stable epigenetic activation of the MP and cholesterol biosynthesis. This is coupled with an enrichment of SREBP1 and SREBP2 DNA-binding motifs, as determined by DNase I footprinting analyses, suggesting that there is increased SREBP occupancy on open chromatin. The resulting increased levels of 27-hydroxycholesterol were sufficient to activate ER1 (estrogen receptor alpha) signaling in the absence of exogenous estrogen, driving the activation of genes that promote an invasive cell phenotype. Similarly, in prostate cancer, the de novo synthesis of androgens from cholesterol drives androgen receptor activity in castration-resistant disease. This finding, coupled with the observations that SREBP expression is increased in advanced-stage prostate cancer, suggests a role for the MP in prostate cancer progression. These findings warrant further investigation into the utility of inhibitors of the MP and/or SREBPs in the treatment of hormone-driven cancers.

Inhibitors Statins Statins competitively inhibit HMGCR (3-hydroxy-3-methyl-glutaryl-coenzyme A reductase), the first committed enzyme of the MP, blocking the conversion of HMG-CoA to mevalonic acid (Fig. 2). The molecular mechanism of inhibition of HMGR by statins is a catalytic mechanism. The statins molecules occupy the catalytic portion of HMGCR, specifically the binding site of HMG-CoA, thus blocking access of this substrate to the active site. The structures of the catalytic domains of the enzyme in complex with statin molecules have been identified. Efficacy of HMGCR inhibition by various compounds in the family of statin inhibitors is markedly dependent (e.g., nanomolar versus millimolar inhibitor affinity) on whether class I or class II enzyme is used. The inhibitors are characterized by hydroxymethylglutaryl (HMG)-like moieties linked to an extensive hydrophobic scaffold (e.g., a decalin ring in the case of mevastatin and simvastatin). Complexes of human enzyme catalytic domain with a variety of statins have been crystallized, and X-ray structures have been published. The structural results indicated binding of the HMG moiety in the active site pocket where the catalytic glutamate and lysine residues are located. In contrast, the NADPH substrate site is not occupied upon inhibitor binding. The structural results indicating that access to the substrate HMG-CoA is blocked by inhibitor binding are in accord with the observation of competitive inhibition with respect to HMG-CoA. A variety of additional polar interactions with the HMG moiety have also been documented.

Mevalonate Pathway

455

The hydrophobic portion of the inhibitors is bound in a shallow hydrophobic groove of the human protein. A large number of van der Waals contacts between nonpolar amino acids in this groove and the diverse hydrophobic substituents that are a common feature of the various statins are proposed to represent the dominant contribution to high-affinity binding. A structure of lovastatin bound to the class II P. mevalonii enzyme has also been reported. As in the case of the class I enzyme, the structure indicates interactions with residues (e.g., lys, glu) identified in catalysis as well other polar residues in this pocket, with some hydrogen bonds mediated by water molecules. The hydrophobic decalin ring component of the inhibitor blocks a closure of the C-terminal flap domain of the protein, which includes the histidine residue that has been implicated in catalysis. Thus, inhibitor binding both blocks the active site and makes correct orientation of active site amino acids impossible. A large difference between class I and class II enzyme interactions with the statin inhibitor is suggested to involve a pocket formed, in part, by the alpha helix of the P. mevalonii enzyme. The structural results could expedite potential design of HMGCR inhibitors that can discriminate between class I and class II enzymes. Additionally, an approach proposed to increase inhibitor affinity involves derivatization of the parent inhibitory compound to incorporate chemical substituents effective in interaction with the NADPH site. By interrupting cholesterol synthesis in the liver, statins activate the production of microsomal HMGCR and cell surface lowdensity lipoprotein (LDL) receptors. This results in a predictable increased clearance of LDL from the bloodstream and a decrease in blood LDL cholesterol levels that may range from 20% to 55%. Statins have shown strong evidence-based proved capacity of decreasing the cardiovascular morbidity and mortality in both primary and secondary prevention settings. Because of these properties, statins are among the most widely used pharmaceutical agents in the world, their role in the global battle against cardiovascular disease being compared with the crucial role of antibiotics in decreasing the mortality of infections. Subgroup analyses of several large clinical trials have suggested that changes in lipid levels alone may not explain all the beneficial effects of statins in human pathology. Evidence has emerged from both basic research and clinical trials that statins have many cholesterolindependent or so-called pleiotropic effects: improving endothelial function, atherosclerotic plaque-stabilizing effects, antiinflammatory and immunomodulatory effects, antithrombotic properties, effects on bone metabolism, on risk of dementia, induction of apoptosis, and antiproliferative effects. Most of these pleiotropic effects result from inhibition of the synthesis of important isoprenoid intermediates of the MP, such of FPP and GGPP. Mevalonate-derived prenyl groups have been shown to have essential roles in many cellular functions including cell signaling, cell differentiation and proliferation, myelination, cytoskeleton dynamics, and endo/exocytotic transport. In the last decade, substantial progress has been made in understanding the nonlipid-related pharmacological properties of statins and a growing interest in the potential therapeutical implications of these pleiotropic effects of statins has emerged. Recent experimental and clinical data indicate that statins may be of potential therapeutic use in a variety of nonvascular diseases, including autoimmune diseases, multiple sclerosis, rheumatoid arthritis, sepsis, dementia, and different types of cancer. Many studies have shown that statins can directly and specifically inhibit the proliferation of tumor cells. For example, statins promote the apoptosis of cells derived from acute myeloid leukemia (AML), while normal myeloid progenitors do not undergo apoptosis and retain full proliferative potential. This tumor-normal therapeutic index may be due to the altered metabolic reprogramming of AML cells leading to an increased dependence on MP metabolites for growth and survival. The widespread use of statins for cholesterol management also demonstrates that these drugs cause minimal damage to normal cells. The side effects of these drugs are regularly treated by switching to a different statin or potentially by cotreating with CoQ although this cotreatment method is controversial owing to conflicting clinical evidence. The data discussed earlier suggest that statins have a high therapeutic index to target tumors in vivo despite the ubiquitous expression of the MP. This rationale has led to multiple clinical trials investigating the efficacy of various statins as a therapeutic option in a variety of tumor types. Two recent breast cancer window-of-opportunity clinical trials, using atorvastatin and fluvastatin, showed reductions in the Ki67 index in a subset of patients who were administered with cholesterol-management doses of statins between cancer diagnosis and surgery. Statins have also been safely used in combination with other agents to increase efficacy. For example, pravastatin was combined with standard-of-care treatment in HCC and AML, resulting in significantly longer median survival in colorectal cancer and resulting in complete or partial response in 60% of patients with AML. In another study, combining lovastatin with thalidomide and dexamethasone in patients with relapsed or refractory multiple myeloma (MM) led to prolonged overall survival and progression-free survival. Despite evidence of patient response to statins as anticancer agents, many patients remained nonresponsive to statin treatment in other cancer clinical trials. This is consistent with the current paradigm of interpatient tumor heterogeneity. This lack of response might also be expected considering the evidence that we discussed earlier showing that the MP is regulated by many key oncogenic signals. Similar to many anticancer agents, a personalized medicine approach is needed to implement statins, and/or other inhibitors of the MP, as a successful class of cancer therapeutics. To this end, a molecular signature of basal mRNA expression has been developed to predict statin response in breast cancer in vitro, and deregulated MYC expression has been a proposed indicator of statin response in specific tumor types; however, essential follow-up validation is required before these biomarkers can be used clinically. It is currently difficult to predict which cancers will be particularly sensitive to statin therapy. In addition to AML and MM, encouraging results from both clinical trials and epidemiological studies suggest that patients with hormone-dependent cancers, such as breast cancer and prostate cancer, may benefit from the addition of statins to their treatment regimen. This may be partly because the MP end-product cholesterol is the precursor of hormones such as estrogen and androgens, which have a major role in the development of these types of cancers. Colorectal cancer also seems to be particularly responsive to statins, perhaps because of the hepatotropic pharmacology of this family of drugs. Clinical trials are required in these and other cancers to further define the subset of cancers that are particularly statin-sensitive.

456

Mevalonate Pathway

Targeting the SREBP in Combination With Statins Crucial to the regulation of the MP is the tightly controlled, SREBP-mediated feedback mechanism, in which inhibition of the MP results in the activation of the SREBPs and an increase in the expression of MP genes, an effect that may be amplified in cancer cells. SREBP activation also increases the expression of the LDLR, which leads to the increased uptake of exogenous, lipoprotein-derived cholesterol: an effect that has been shown to be important in cancer cells. The SREBPs thus function to replenish MP metabolites, which can dampen the apoptotic response following statin treatment. This would be a classic resistance mechanism, similar to that seen with other anticancer therapeutics such as inhibitors of the BRAF protooncogene in BRAF-mutant melanoma. Cells treated with BRAF inhibitors, such as vemurafenib, can acquire an activating mutation in downstream kinases (e.g., the MEK1 mitogen-activated protein kinase kinase) or can have an increase in expression of receptor tyrosine kinases (e.g., epidermal growth factor receptor), bypassing the need for BRAF activity. These studies demonstrate that inhibiting both the cancer vulnerability and the resistance or feedback mechanism is crucial for maximum efficacy. Therefore, inhibiting the SREBP-regulated feedback response in conjunction with statin therapy could prevent resistance, thereby increasing the efficacy of statins as anticancer agents and the number of responsive patients. Evidence that targeting the SREBPs in combination with statin therapy is a viable strategy has been provided by several studies. One study looking at breast and lung cancer cell lines used a short hairpin RNA screen to uncover genes that, when knocked down, potentiated the proapoptotic effects of statins. The MP genes HMGCS1, GGPS1, SCAP, and SREBF2 all scored highly, adding credence to either inhibiting other enzymes in the MP or inhibiting the SREBP-mediated feedback response in combination with statin therapy. Another study showed that statin-induced SREBP processing can be blocked by another agent that has been approved for a noncancer indication, dipyridamole. Dipyridamole reduced the transcription of SREBP target genes such as HMGCS1 and HMGCR and synergized with statins to increase apoptosis in AML and MM cell lines and patient samples. Other compounds, such as tocotrienols, have also been demonstrated to synergize with statins to induce cancer cell apoptosis, which is an effect that may be associated with their ability to degrade nuclear SREBP2 and inhibit its transcriptional activity. Although several other small molecules, including fatostatin, have been shown to inhibit SREBP processing, their lack of approval for use in patients limits their potential to immediately have an impact on cancer patient care. Therefore, clinical investigation into the utility of combined statins and SREBP inhibitors for the treatment of cancer is currently warranted.

Bisphosphonates BPs are potent inhibitors of osteoclast mediated bone resorption and are currently the most important and effective class of drugs used to treat metabolic bone disease. BPs bind avidly and achieve therapeutic concentration at sites of active bone metabolism in the mineralized bone matrix. By inhibiting bone resorption and inducing osteoclast apoptosis, BPs reduce bone turnover, increase bone mass, and improve bone mineralization. Recent experimental studies have shown that nitrogen-containing BPs may have interesting antitumor properties. As shown in Fig. 2, they inhibit FPP synthase, a key enzyme in the MP, with consecutive decrease of the formation of isoprenoid lipids such as FPP and GGPP. It has been demonstrated that the capacity of inhibiting FPP synthase is correlated to the synthesis from the accumulated isopentenyl diphosphonate of an ATP analogue, named Apppi. This, in turn, prevents the prenylation of a number of small GTPases, such as Rho, Ras, Rac, and Rab, which play a role in malignant transformation of cells by regulating intracellular signaling, cell growth, motility, and invasion. Interestingly, the ability of nitrogencontaining BPs to inhibit FPP synthase appears to be clearly related to the presence of a nitrogen atom at critical positions in the side chain. Consistent with their recently identified biochemical effects on protein prenylation, there is extensive evidence from preclinical research that BPs also exhibit antitumor activity in a variety of human cancers: myeloma, breast cancer, prostate cancer, pancreatic cancer, mesothelioma, mesenchymal tumors, and osteosarcoma. Summarizing, inhibition of the MP is the underlying molecular mechanism of many of these antitumor properties of nitrogen-containing BPs: inhibition of integrin-mediated tumor cells adhesion to bone. BPs act by inhibition of prenylation of small GTPases required for integrin activation–inhibition of cancer cells migration and invasion by inhibition of metalloproteinases and by inhibition of Rho activation by preventing geranylgeranylation. Inhibition of proliferation and induction of apoptosis of cancer cells were observed at high BP concentrations. Although inhibition of the MP was found to be the main molecular mechanism of promoting apoptosis, other mechanisms have also been described, that is, reduction of bcl-2 expression and activation of caspase. In addition, immunomodulatory effects mediated through accumulation of mevalonate metabolites in tumor cells, which stimulate T lymphocytes expressing the T cell receptor, were described.

Other Therapeutic Agents That Target MP Extensive preclinical evidence suggested that manipulation of the MP through inhibition of FTase and GGTase, and, in consequence, inhibition of protein prenylation, can result in alteration of malignant cells’ capacity of proliferation, growth, and migration. FTIs represent a new generation of signal transduction inhibitors, designed to target the critical posttranslational modification of Ras, a protooncogene that is believed to play a critical role in tumor growth or progression and may have prognostic significance.

Mevalonate Pathway

457

Some studies suggest that FTIs inhibit not exclusively Ras proteins but also the farnesylation of additional cellular polypeptides that have not been identified, thereby exerting antitumor effects independent of the presence of activating Ras gene mutations. Lonafarnib SCH-66336 has been the first of these compounds to undergo clinical development. Several other FTIs have entered clinical trials for various cancer indications: examples are tipifarnib (R115777), BMS-214662, and L-778. Although phase I and II studies have shown that FTIs have a significant antitumor activity and an acceptable toxicity profile, the results of phase III trials have been disappointing, showing no overall significant amelioration of survival in solid cancers. The most promising activity to date has been demonstrated in patients with hematological malignancies, in particular AML and myelodysplastic syndrome. GGTIs are a novel class of drugs that are still in the experimental stage. It has been recently demonstrated that when FTase is blocked by FTIs, additional inhibition of geranylation by inhibitors of GGTase can augment antiproliferative properties of FTIs. Treatment of tumor cell lines in culture with varying doses of a GGTI in conjunction with a FTI can inhibit Ki-Ras prenylation and determine an apoptotic response that is greater than the apoptotic response elicited by either agent alone in vitro. However, the tolerability of this protocol in vivo was poor and suggested that the use of this approach in humans should be cautious because of a possible narrow therapeutic index. Recent preclinical research has identified a new inhibitor of protein prenylation, a selective, highly potent, and cell-active GGTase-I inhibitor, GGTI-DU40, and suggested that investigation of GGTIs, as potential new anticancer drugs should continue, in order to define their exact role and toxicity profile. In conclusion, in the last few years, there has been intense interest in understanding the clinical and therapeutical implications of MP, a metabolic pathway which plays a key role in regulation of cellular cholesterol synthesis and in controlling cell proliferation by generating prenyl intermediates. Since the discovery of MK deficiency as the direct biochemical and molecular cause of MVA and hyperimmunoglobulinemia D syndrome, considerable progress has been made in understanding the pathophysiology of these antiinflammatory disorders. However, it is still not known what is the causal relationship between the enzymatic defect and the inflammation and no effective treatment has been established so far. Also, MP is an important, attractive target for many areas of therapeutical research and application, through inhibition of the isoprenylation process of small intracellular proteins. Although extensively studied, many important issues remain unanswered, and this fascinating area of interference between biochemistry, clinical implications, and pharmacology remains a major focus for potential future investigation.

Outlook Understanding tumor metabolism in the context of oncogenic signals has the potential to drive the development of targeted personalized therapies. The various signaling pathways that we describe in this book chapter are important drivers in many cancers, and they all have the ability to deregulate the MP, making these cancers potentially vulnerable to MP inhibition. Whether this occurs in every patient who presents with these lesions remains unclear. More work is needed to understand the extent to which driver mutations increase flux through the MP in patients. Rapidly developing technologies for the comprehensive flux-based analysis of MP metabolites will provide further advances in understanding how the MP receives and responds to oncogenic signals. In patients, it may be more feasible to determine pathway activity by mapping their oncogenic lesions to their sterol feedback response at the protein level (via SREBP localization) or mRNA expression level of MP genes, which may identify patients who will respond to MP inhibition. Designing clinical trials that will identify potential responders before treatment is required to prevent expensive failures of therapies that may still have benefits to a subset of patients. Improving reagents, particularly antibodies to HMGCR and SREBP2, will also aid trial design and interpretation. The essentiality of the MP in many cancers, coupled with affordable and safe drugs that can target this pathway and its feedback response, provides a strong rationale for continuing to explore this key metabolic pathway in cancer and many other diseases.

Acknowledgments Related works in our laboratory are supported in part by the Fonds zur Förderung der wissenschaftlichen Forschung (FWF; The Austrian Science Fund) Project P24370-B19, the WGKK (Social Health Insurance Vienna), and the AUVA (Austrian Social Insurance for Occupational Risks).

Further Reading Buhaescu, I., Izzedine, H., 2007. Mevalonate pathway: a review of clinical and therapeutical implications. Clinical Biochemistry 40 (9–10), 575–584. Miziorko, H.M., 2011. Enzymes of the mevalonate pathway of isoprenoid biosynthesis. Archives of Biochemistry and Biophysics 505 (2), 131–143. Karlic, H., Thaler, R., Gerner, C., Grunt, T., Proestling, K., Haider, F., Varga, F., 2015. Inhibition of the mevalonate pathway affects epigenetic regulation in cancer cells. Cancer Genetics 208 (5), 241–252. Moutinho, M., Nunes, M.J., Rodrigues, E., 2017. The mevalonate pathway in neurons: it’s not just about cholesterol. Experimental Cell Research (Epub ahead of print). Mullen, P.J., Yu, R., Longo, J., Archer, M.C., Penn, L.Z., 2016. The interplay between cell signalling and the mevalonate pathway in cancer. Nature Reviews. Cancer 16 (11), 718–731. Räikkönen, J., Mönkkönen, H., Auriola, S., Mönkkönen, J., 2010. Mevalonate pathway intermediates downregulate zoledronic acid-induced isopentenylpyrophosphate and ATP analog formation in human breast cancer cells. Biochemical Pharmacology 79 (5), 777–783.

Microbiota and Colon Cancer: Orchestrating Neoplasia Through DNA Damage and Immune Dysregulation Alberto Martin and Thergiory Irraza´bal, University of Toronto, Toronto, ON, Canada © 2019 Elsevier Inc. All rights reserved.

Glossary Adenoma Describes a wide range of neoplastic nodules with glandular appearance that arise in the epithelial tissue and show no cytological features of malignancy. Bacteriome Term used to define a bacterial community that inhabit a particular environment and is usually shaped by a symbiotic interaction with its host and other microbes. Dysbiosis Perturbation of the natural balance in the gut microbiota, that is usually associated with pathogenesis. Genotoxin A molecule that cause DNA damage. It could cause DNA mutations (mutagen), trigger cancer (carcinogen), or lead to birth defects (teratogen). Polyp Is a typically benign abnormal growth of diffuse mucosal tissue. Some polyps have a stalk and are called pedunculated, while other polyps are flat and are called sessile. Virome Term used to define a viral community that inhabit a particular environment and is usually shaped by a symbiotic interaction with its host and other microbes.

Nomenclature AOM Azoxymethane APC Adenomatous polyposis coli Apcmin/þ APC multiple intestinal neoplasia ATM Ataxia-telangiectasia mutated BabA Blood group binding adhesin BFT Bacteroides fragilis toxin CAC Colitis-associated colon cancer cagPAI cag pathogenicity island CDT Cytolethal distending toxin CEACAM Carcinoembryonic antigen-related cell adhesion molecule CECs Colon epithelial cells CHK Checkpoint kinase CRC Colorectal cancer DDR DNA damage response DSBs Double strand breaks EPEC Enterotoxigenic Escherichia coli ETBF Enterotoxigenic Bacteroides fragilis Gal-GalNac Galactose-N-acetyl-D-galactosamine H2O2 Hydrogen peroxide HNPCC Hereditary nonpolyposis colorectal cancer HR Homologous recombination IBD Inflammatory bowel disease IFNg Interferon gamma iNOS Inducible nitric oxidase MAP kinase Mitogen-activated protein kinase MMR Mismatch repair MPII Secretory metalloproteinase II MSCRAMM Microbial surface component recognizing adhesive matrix molecules MSI Microsatellite instability NHEJ Nonhomologous end-joining p53 Tumor protein P53 PGE2 Prostaglandin E2

458

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65124-3

Microbiota and Colon Cancer: Orchestrating Neoplasia Through DNA Damage and Immune Dysregulation

459

Rag2 Recombinase-activating gene-2 ROS Reactive oxygen species SMO Spermine oxidase T3SS Type III secretion system T4SS Type IV secretion system Tregs T regulatory cells VacA Vacuolating cytotoxin yH2AX Phosphorylated histone H2A.X

Healthy Human Enteric Microbiota The gut microbiota is a dynamic structure that helps maintain host homeostasis by regulating metabolic and immune processes and establishing a network that keeps pathogenic microbes in line. Microbial composition is shaped by the host immune system, aging processes, dietary metabolites, oxygen levels, in situ microbial interactions, and other environmental factors. In the human colon, the microbiota is composed of archaea, eukarya, viruses, and bacteria. The Methanobrevibacter genus is the most prevalent archaea in healthy individuals and the phyla Ascomycota and Basidiomycota are the most prevalent fungi (Hoffmann et al., 2013). Intestinal bacteria are dominated by an inverse abundance ratio between Bacteroides and Firmicutes (Human Microbiome Project Consortium, 2012; Eckburg et al., 2005). However, the less abundant phylum Proteobacteria, which include pathogens such as Escherichia coli and Helicobacter, is the major source of gene variability between healthy hosts (Bradley and Pollard, 2017). Strict anaerobes such as Bacteroides, Bifidobacterium, Eubacterium, Peptostreptococcus, Atopobium and Fusobacterium constitute the major portion of the microbiota (Tlaskalova-Hogenova et al., 2004), whereas facultative anaerobes such as Enterococci, Streptococci and Enterobacteriaceae represent a minority. Both inflammation and antibiotic treatment can induce an increase in oxygen levels in the colon (Byndloss et al., 2017), which can favor dysbiosis by creating a niche for facultative anaerobes with potential pathogenic activity (Spees et al., 2013; Rivera-Chavez et al., 2016) (reviewed on Rivera-Chavez et al. (2017) and Litvak et al. (2017)).

Human Enteric Microbiota Associated With Carcinogenesis Colorectal cancer (CRC) development is a multistep process that involves accumulation of mutations in colon epithelial cells (CECs), which lead to the formation of aberrant crypts (Irrazabal et al., 2014). Common genetic lesions that lead to colon carcinogenesis include loss of the adenomatous polyposis coli (APC) tumor suppressor gene, or mutations in the b-catenin gene, both of which lead to activation of b-catenin/TCF/LEF-1 signaling that in turn leads to uncontrolled CEC proliferation. In fact, 85% of sporadic CRC carry mutations in the APC gene (Medema and Vermeulen, 2011). These genetic changes are usually followed by mutations in K-Ras, PIK3CA and TP53 genes (Raskov et al., 2014). In addition, mutations or epigenetic silencing that lead to mismatch repair (MMR) deficiency occur in approximately 20% of sporadic CRC (Poulogiannis et al., 2010) and are the most common mutation in patients with hereditary nonpolyposis colorectal cancer (HNPCC) or Lynch syndrome. Recently, several studies have shown that besides these genetics changes, differences in the composition and diversity on the human microbiota are also strongly associated with CRC development. Although findings have shown that CRCs are colonized with specific bacteria (Maddocks et al., 2009; Gagniere et al., 2017; Viljoen et al., 2015; Kostic et al., 2013; Purcell et al., 2017; Biarc et al., 2004), it is not clear whether such bacteria are CRC initiators or passengers (Tjalsma et al., 2012). This distinction is necessary to differentiate causality from association. CRC initiators induce DNA damage either directly through the production of genotoxins, or indirectly, by promoting inflammation that in turn leads to the production of genotoxic agents or by transforming diet-derived metabolites into secondary genotoxic molecules (Belcheva et al., 2015). On the other hand, passenger bacteria exploit the unique microenvironment provided by the tumor to grow and out-compete other microbes. In this case, bacteria could simply be along for the ride, or might promote the growth of the tumor. Current evidence supporting bacterial causality in human CRC has been challenging, which is mainly due to three factors: bacterial genetic plasticity, host genetic variability, and the prolonged period between the initiation of carcinogenesis and the detection of the cancerous lesion in the clinic. First, the microbiota is a dynamic structure, the composition of which is affected by many factors. Due to the high mutation rates in viruses (Minot et al., 2013), the constant horizontal gene transfer between unrelated bacteria (Lloyd-Price et al., 2016), and the very long times that the microbiota develop and adapt to each other in an adult individual, it is likely that each person will ultimately possess a unique virome, composed primarily of phages, and a unique bacteriome. Hence, trying to precisely implicate a single bacterium at a determinate point to CRC initiation can be likewise compared to trying to measure with absolute precision the position and momentum of a quantum particle, as outlined by Heisenberg’s uncertainty principle. Second, most of the potentially carcinogenic bacteria can colonize humans asymptomatically, and are able to induce or promote carcinogenesis only after chronic infection in genetically susceptible hosts. Hence, such bacteria might be carcinogenic only under

460

Microbiota and Colon Cancer: Orchestrating Neoplasia Through DNA Damage and Immune Dysregulation

certain conditions, and defining these conditions might prove to be challenging. Furthermore, considering that not all inflammatory processes are equally detrimental to the mucosa, a detailed characterization of the host inflammatory state is critical, since gross histological scores minimize the potential relevance of unique inflammatory responses. Third, although a single acute exposure of a specific genotoxic bacterium might initiate genetic defects that facilitate cancer initiation, the bacterial pathogen that initiated the oncogenic process might have disappeared by the time the patient presents with an adenoma or CRC at the clinic. In this regard, more research should focus on how a single challenge with pathogenic bacteria could potentially lead to chronic dysbiosis, and/or intestinal disease. Despite these caveats, there is nevertheless strong evidence for the involvement of specific microbes in colon cancer, and these are described below.

Helicobacter pylori Helicobacter pylori (H. pylori) is a Gram-negative human pathogen found in approximately 50% of the population (Moayyedi and Hunt, 2004). This pathogen promotes chronic inflammation in the upper gastrointestinal tract and is a known inducer of gastric cancer. In fact, H. pylori is the only bacteria classified as a Group I carcinogen by the International Agency for Research on Cancer (Anon, 1994). This link has led to extensive research of a possible role of H. pylori in other gastrointestinal cancers, such as CRC. Although H. pylori has been detected in the human colon (Keenan et al., 2010) and in human colonic tumors (Papastergiou et al., 2016), there is no clear mechanism by which this pathogen can initiate CRC. Furthermore, the association between H. pylori infection and CRC is conflicting (Breuer-Katschinski et al., 1999; Fujimori et al., 2005; Moss et al., 1995; Shmuely et al., 2001). Recently, an epidemiologic study in the United States showed that the ethnic distribution of sessile serrated polyps is inversely associated with H. pylori prevalence (Sonnenberg et al., 2017). In contrast, a study focused on the Korean population showed that the rate of H. pylori infection was higher in patients with adenoma compared to polyp-free controls (Nam et al., 2017). In addition, while some studies have shown that the presence of anti-H. pylori IgG antibodies is positively associated with advanced colon cancer (Lee et al., 2016), other studies have not found such associations (Blase et al., 2016). Although, none of these studies have demonstrated that this pathogen could directly induce colonic carcinogenesis, given the known mechanism by which H. pylori induces gastric cancer, it is reasonable to hypothesize that H. pylori might directly promote CRC. In this section, we will discuss the mechanisms by which H. pylori induce gastric cancer and potentially CRC.

H. pylori Virulence Factors H. pylori virulence is related to the presence of a cag pathogenicity island (cagPAI) locus, which encodes for a type IV secretion system (T4SS) and the CagA virulence factor. T4SS binds to b integrins and injects CagA into host epithelial cells (Fig. 1a). T4SS also delivers peptidoglycan into gastric epithelial cells, which is recognized by Nod1 receptors, leading to IL-8 induction (Fig. 1b) (Viala et al., 2004). After CagA is injected into the epithelial cells, this oncoprotein hijacks key signaling pathways in the gastric cell, such as NFkb, MAP kinase and Jak/STAT pathways. These pathways control fundamental processes in gastric cells, such as proliferation, inflammation, cell cycle, and apoptosis (reviewed on Tegtmeyer et al. (2017)) leading to deregulation in cell homeostasis. Another virulent factor produced by H. pylori is the vacuolating cytotoxin VacA, which is the only known H. pylori protein that targets the mitochondria. VacA is internalized into gastric epithelial cells in endosomes, a process that alters intracellular trafficking and inhibits antigen presentation (Molinari et al., 1998). Internalized VacA localizes to the mitochondrial inner membrane where it forms anion-conductive channels that reduce the mitochondrial membrane potential, resulting in a lower production of ATP, cytochrome c release and apoptosis (Fig. 1c) (McClain et al., 2017). VacA channel-forming activity also leads to autophagy, which seems to be a host defense mechanism to reduce cellular damage (Terebiznik et al., 2009). On the other hand, VacA can trigger reactive oxygen species (ROS)-induced autophagy by decreasing intracellular levels of glutathione, one of the main antioxidants in the cell (Tsugawa et al., 2012). Additionally, VacA can inhibit the function and proliferation of several adaptive and innate immune cells (McClain et al., 2017). VacA activity differs between H. pylori strains due to gene variability (Blaser, 2002). Nevertheless, H. pylori strains that possess a functional T4SS, a CagA gene, and a cytotoxic VacA gene, are positively associated with induction of gastric tumor lesions (Blaser, 2002). Once H. pylori reaches the stomach, it adheres to gastric epithelial cells by binding to some members of the CEACAM (carcinoembryonic antigen-related cell adhesion molecule) family expressed by epithelial cells (Koniger et al., 2016). Curiously, while CEACAM1 is a recognized tumor marker in the colon, this protein is downregulated during colon carcinogenesis (Neumaier et al., 1993), and its deficiency contributes to colon carcinogenesis in mice (Leung et al., 2006). These data suggest that the environment generated after tumor formation in the colon do not favor H. pylori colonization, but doesn’t rule out such a role for H. pylori in tumor initiation. Another protein that facilitate H. pylori adherence to gastric epithelial cells is the blood group binding adhesin (BabA), which binds to ABO/Leb blood group antigens and fucosylated carbohydrates, such as the ones present in mucus. Interestingly, BabA binding to epithelial cells is reduced by low pH, and increased in neutral environments (Bugaytsova et al., 2017). Since the acidic gastric content is neutralized in the intestine, H. pylori could potentially bind mucus in the small and large intestine through BabAmediated interactions. Furthermore, it is unknown whether an increase in mucus production, which is a phenotype of colitis and colitis-associated colon cancer (CAC), could affect H. pylori potential for intestinal colonization.

Microbiota and Colon Cancer: Orchestrating Neoplasia Through DNA Damage and Immune Dysregulation

461

Fig. 1 (a) H. pylori injects the toxin CagA into the gastric epithelial cell using a type IV secretion system. CagA induces the expression of spermine oxidase (SMO), which produces H2O2 damaging the DNA. Notice that the enterotoxigenic B. fragilis toxin also induces SMO and DNA damage (b) H. pylori also delivers peptidoglycan into gastric epithelial cells, which is recognized by Nod1 receptors, leading to inflammation. (c) H. pylori produces VacA, which targets the mitochondria. VacA is internalized in endosomes, and localizes to the mitochondrial inner membrane where it forms pores that reduce the mitochondrial membrane potential, resulting in a lower production of ATP, and apoptosis. VacA also leads to autophagy as a response to antioxidant depletion.

Mechanisms by Which H. pylori Could Induce Colon Carcinogenesis After H. pylori colonizes the gastric epithelial cells, it induces several physiological changes that have been associated with the induction of either gastritis or gastric cancer. These changes include: hypergastrinemia, inflammation, destruction of gastric cells, hypochlorhydria, and DNA damage, all of which could potentially promote colon carcinogenesis.

Hypergastrinemia Chronic infection with H. pylori can increase the release of gastrin, which is a known inducer of hyperproliferation in human CECs (Renga et al., 1997); however, the mitogenic effects of gastrin are not related to the prevalence of adenomas in the colon of patients presenting hypergastrinemia (Sobhani et al., 1993). Since cell hyperproliferation is not sufficient for tumor formation, it is possible that H. pylori promotes neoplasia initiation under certain circumstances, such as in DNA repair-deficient hosts. Furthermore, given that mutations on DNA damage repair proteins appear early during colon carcinogenesis, it is possible that under these circumstances H. pylori accelerates cancer development.

Inflammation Chronic inflammation induced by H. pylori leads to destruction of gastric glandular cells, a process that can facilitate microbial dysbiosis across the gastrointestinal tract (Kanno et al., 2009). Dysbiosis, in turn, could generate an environment suitable for overgrowth of other cancer-promoting pathogens and/or exacerbate inflammation. Chronic inflammation can also favor cell transformation by silencing some genes through epigenetic modifications, or by leading to the production of ROS which in turn induces DNA damage.

Hypochlorhydria H. pylori also cause hypochlorhydria, which can cause a decrease in stomach protein hydrolysis and an increase in colonic protein fermentation (Cater 2nd, 1992). An increase in colonic protein availability could potentially provide an advantage to pathogenic bacteria that can ferment proteins and/or use nitrogen as a source of energy (Byndloss et al., 2017; Spees et al., 2013; Winter et al., 2013); under these circumstances H. pylori could act as a passenger that facilitate colonic tumor development.

DNA damage H. pylori might promote DNA damage by at least three mechanisms: increasing ROS production (Chaturvedi et al., 2011), inducing mutations in mitochondrial and nuclear DNA, and reducing DNA repair (Machado et al., 2009). One of the main factors related to H. pylori-induced DNA damage is the toxin CagA. This virulence factor could potentially induce CRC by deregulating Wnt (Yong et al., 2016; Gnad et al., 2010) and p53 pathways (Buti et al., 2011; Wei et al., 2010), by inducing the expression of inflammatory mediators that are known CRC promoters, such as IL-8 (Ferreira et al., 2016; Chang et al., 2015) or by inducing DNA damage

462

Microbiota and Colon Cancer: Orchestrating Neoplasia Through DNA Damage and Immune Dysregulation

(Chaturvedi et al., 2011; Xu et al., 2004). CagA positive H. pylori infection of gastric epithelial cells induces DNA double-strand breaks (DSBs) and telomere shortening. Specifically, H. pylori strains that express the oncogenic protein CagA induce upregulation of spermine oxidase (SMO) (Chaturvedi et al., 2011). SMO produces hydrogen peroxide (H2O2) as a by-product, leading to oxidative stress, DNA damage, and apoptosis of gastric epithelial cells (Fig. 1a) (Chaturvedi et al., 2011). Surprisingly, after H. pylori infection, a subset of gastric epithelial cells with high levels of DNA damage do not undergo apoptosis, a phenomenon that has been associated with the activation of the EGF receptor by H. pylori (Chaturvedi et al., 2014). Another mechanism that could increase survival of cells with high levels of DNA damage could take place in cells that harbor mutations in the DNA damage response (DDR). When the DDR is not active, cell cycle checkpoints are not activated, and cells with high levels of damaged DNA will proliferate. For example, ATM (ataxia-telangiectasia mutated)-deficient cells proliferate after irradiation due to its inability to activate the cell cycle checkpoints in response to double-stranded DNA breaks (Painter and Young, 1980; Falck et al., 2001). Although CagA could induce CRC by several mechanisms, and some studies have shown that CagA seropositivity in patients infected with H. pylori is associated with a higher incidence of CRC (Shmuely et al., 2001), other reports have found no such correlations (Papastergiou et al., 2016; Chen et al., 2016). In the same way, it has also been shown that seropositivity against the toxin VacA is strongly associated with CRC risk (Epplein et al., 2013), but although both CagA (Chaturvedi et al., 2011) and VacA (Tsugawa et al., 2012) induce ROS and DNA damage in gastric cells, no reports have shown a direct effect of these toxins on CECs. In conclusion, although it is possible that H. pylori could induce CRC, more research is required with specific focus on specific susceptible subjects. For example, H. pylori can directly induce DNA damage, and the potential role of H. pylori on DNA repairdeficient subjects have not been fully studied.

Helicobacter hepaticus Other Helicobacter species, such as Helicobacter hepaticus (H. hepaticus), have also been associated with carcinogenesis. In 1992, H. hepaticus was shown to cause hepatocellular and hepatocholangiolar adenomas and carcinomas in genetically susceptible mice (Falsafi and Mahboubi, 2013). Subsequent work showed that H. hepaticus infection induced colitis and CRC in immunodeficient mice (Erdman et al., 2003, 2009; Ge et al., 2017; Nagamine et al., 2008a, 2008b; Wang et al., 2017).

Mechanisms by Which H. hepaticus Could Induce Colon Carcinogenesis Although H. hepaticus has been found in patients with chronic liver disease (Nilsson et al., 2000; Yang et al., 2013; Hamada et al., 2009), there is no data that demonstrate that H. hepaticus colonize the human intestinal tract and therefore a direct causality on inflammatory bowel disease (IBD) or CRC in humans is nonexistent. Nevertheless, current research suggests that H. hepaticus can induce CRC in mice by at least two mechanisms: inflammation and DNA damage.

Inflammation Lymphocyte homeostasis directly affects the progress of intestinal disease during H. hepaticus infection; SMAD3-deficient mice develop colitis after infection with H. hepaticus, a process that is dependent on the activation and dysregulation of effector lymphocytes in the gut (McCaskey et al., 2012). Although, one might expect that lymphocyte-deficient mice will not develop colitis after infection with H. hepaticus, recombinase-activating gene-2-deficient (Rag2/) mice inoculated with H. hepaticus develop colitis and colonic polyps, which is attenuated by providing mice with T regulatory cells (Tregs) (Erdman et al., 2003, 2009). In the same line, mice deficient in the immunoregulatory molecule IL-10, also develop colitis after infection with H. hepaticus, and it has been suggested that monocyte recruitment into the lamina propria lead to CAC (Bain et al., 2018). While, the pathogenesis of H. hepaticus was thought to be due to its capacity to activate inducible nitric oxidase (iNOS) in macrophages, recent data suggest that H. hepaticus induce iNOS directly on the apical side of cecal epithelial cells (Fig. 2a). In Rag2/ mice, H. hepaticus-mediated nitric oxide production leads to cecal epithelial cell hyperproliferation and DNA damage; phenotypes that are reverted after IL-22 depletion. H. hepaticus-mediated IL-22 induction was not only associated with nitric oxide production, but also with dysbiosis (Wang et al., 2017).

DNA damage H. hepaticus also produces the cytolethal distending toxin (CDT), which is formed by three subunits encoded by the genes cdtA, cdtB and cdtC. The subunit CdtA allows the toxin to attach to the cell membrane, while CdtC promotes the translocation of the DNase CdtB into the nucleus (Fig. 2b) (Avenaud et al., 2004). Once this toxin is translocated into the nucleus, it catalyzes the formation of DSBs during S-phase of the cell cycle (Fedor et al., 2013), inducing DNA damage (Ge et al., 2017) and G2/M cell cycle arrest (Young et al., 2000). The mechanism by which CDT induces colitis and CRC has been extensively investigated in the context of other pathogens that can produce this toxin, such as enterotoxigenic Escherichia coli (see below), however the mechanisms by which H. hepaticus CDT induces disease in vivo has not been fully investigated. In vitro studies have shown that the DNA repair pathways activated by H. hepaticus CDT are very similar to those activated after irradiation (IR)-induced DNA damage. CDT activates both ATM-CHK2 and ATR-CHK1 pathways, and leads to the formation of persistent yH2AX foci in human fibroblasts (Fahrer et al., 2014) and colon epithelial cells from Rag2-deficient mice (Ge et al., 2017). When CDT-induced DNA damage is detected, cells undergo G2/M cell cycle arrest (Young et al., 2000), which is consistent with the protection against CDT-induced DSBs conferred

Microbiota and Colon Cancer: Orchestrating Neoplasia Through DNA Damage and Immune Dysregulation

463

Fig. 2 (a) H. hepaticus induce IL-22 production by immune cells, which triggers iNOS expression on the apical side of the epithelial cells and nitric oxide production. Nitric oxide leads to cell hyperproliferation, DNA damage and dysbiosis. (b) H. hepaticus and EPEC produce the cytolethal distending toxin (CDT), which induces the formation of DSBs. (c) S. bovis can directly induce hyperproliferation of CECs and increase release of CXC chemokines, IL-8 and prostaglandins E2 (PGE2). PGE2 can activate the transcription of genes that regulate proliferation in the colon, such as the Wnt-pathway. (d) ETBF toxin (BFT) cleaves E-cadherin increasing intestinal permeability, and leading to the translocation of b-catenin into the nucleus, which induces proliferation. (e) After ETBF infection, Tregs are recruited into the lamina propria, creating a microenvironment that facilitates Th17 polarization, which promotes IL-17 mediated carcinogenesis. (f) F. nucleatum’s FadA adhesin binds to E-cadherin, which increases free b-catenin, thereby activating the Wnt-pathway. (g) F. nucleatum also activates TLR4, which induces the expression of miR21a leading to the activation of the Wnt-pathway. (h) EPEC secretes EspF, which targets the mitochondria leading to the downregulation of the MMR activity.

by both homologous recombination (HR) and nonhomologous end-joining (NHEJ) in human fibroblasts (Fahrer et al., 2014; Bezine et al., 2016). Interestingly, since it is known that the MMR proteins MLH1 and MSH2 are necessary for DNA repair and the activation of the ATM-CHK2 pathway (Brown et al., 2003), it is possible that patients with MMR-deficiencies could be especially susceptible to CDT-induced DNA damage, although this would need to be investigated. In conclusion, while H. hepaticus is clearly linked to CRC development in mice, the mechanism by which it does so in vivo is lacking. Other pathogens that produce CDT and have been associated with induction of CRC in humans, do not colonize mice efficiently, and therefore, although H. hepaticus might not colonize the human colon, it easily colonizes the mouse intestine without the need of antibiotic pretreatment, providing an excellent model to study the effects of CDT on colitis and CRC. Furthermore, different pathogens could lead to distinct immune responses resulting in divergent phenotypes. Therefore, one cannot assume that the effect of one toxin produced by a specific pathogen would be the same when the toxin is produced by another distant pathogen.

464

Microbiota and Colon Cancer: Orchestrating Neoplasia Through DNA Damage and Immune Dysregulation

Streptococcus bovis Biotype I Streptococcus bovis (S. bovis) is a gram-positive commensal bacterium that belongs to the Firmicutes phylum, and is found in 2.5%– 15% of healthy humans (Hooper and Gordon, 2001). Surprisingly, when this opportunistic pathogen causes endocarditis, its infection is linked to CRC (McCoy and Mason 3rd, 1951; Hoppes and Lerner, 1974; Klein et al., 1977). Specifically, there is a striking association between endocarditis induced by S. bovis biotype I (S. gallolyticus subsp. gallolyticus, referred after as S. bovis) and CRC (Ruoff et al., 1989; Corredoira et al., 2005). In fact, both ulcerative carcinomas and precancerous polyps are associated with this microbe (Burns et al., 1985; Tjalsma et al., 2006). However, it is still under debate whether S. bovis infection can directly induce CRC.

S. bovis Adherence and Colonization of Epithelial Cells Although S. bovis usually do not infect intestinal tissue, some reports suggest that ulceration in colorectal carcinomas facilitate the entry of S. bovis into the blood stream leading to bacteremia and endocarditis. S. bovis express several surface proteins that belong to the Microbial Surface Component Recognizing Adhesive Matrix Molecules (MSCRAMM) family (Sillanpaa et al., 2009). These proteins can bind to collagen-rich surfaces, which are found in both dysplastic colonic epithelium and damaged cardiac tissue. In vitro studies suggest that after S. bovis has attached to the epithelium, these bacteria can cross the epithelial layer via a paracellular mechanism, without inducing IL-1b or IL-8 responses that would otherwise clear the infection (Boleij et al., 2011). In contrast, in vivo studies have found that S. bovis infection lead to inflammation of the colonic mucosa (Ellmerich et al., 2000), which is associated with cancer development (Biarc et al., 2004).

Mechanisms by Which S. bovis Could Induce Colon Carcinogenesis Animal models have shown that S. bovis can directly induce cellular transformation in vivo. Rats treated with the carcinogen azoxymethane (AOM) and either S. bovis or bacterial antigens extracted from its cell wall developed preneoplastic lesions characterized by hyperproliferation of CECs and increased release of CXC chemokines, IL-8 and prostaglandins E2 (PGE2) in the colonic epithelium (Biarc et al., 2004; Ellmerich et al., 2000). PGE2 can activate the transcription of genes that are usually under the control of the Wntpathway (Fig. 2c) (Shao et al., 2005), which when dysregulated can enhance tumor formation. Therefore, S. bovis-induced PGE2 could be a link between inflammation and carcinogenesis.

Potential Role of S. bovis in Human CRC A causative association between S. bovis infection and CRC seems to be restricted to certain scenarios such as when the host is chronically infected and the colonic epithelium is inflamed, the latter of which could lead to an increased rate of genetic mutations and/or hyperproliferation. At least two studies have found that IBD patients have a trend toward an increase in the frequency of fecal S. bovis, results that were likely not significant due to the size of the sample (Klein et al., 1977; Dubrow et al., 1991; Moshkowitz et al., 1992). Nevertheless, further work is required to establish S. bovis in CRC in both humans and colitis-prone animal models. In conclusion, it is not clear whether S. bovis is a CRC initiator or passenger, however given that this bacterium can induce inflammation and cell transformation in vivo, it is reasonable to hypothesize that S. bovis could generate an environment that can exacerbate colon carcinogenesis.

Enterotoxigenic Bacteroides fragilis Bacteroides are one of the major phylum found in the colon, and compromise several symbionts, such as the human anaerobe Bacteroides fragilis, which constitute 1%–2% of cultured fecal bacteria. The presence of a 6 kb pathogenicity island that encodes the secretory metalloproteinase II (MPII) and one of the three homologous toxins named B. fragilis toxins (BFT), distinguishes two classes of B. fragilis. Strains that secrete BFT are called enterotoxigenic Bacteroides fragilis (ETBF) while strains that do not produce this toxin are called nontoxigenic B. fragilis (Sears et al., 2014). Although, ETBF can asymptomatically colonize the human colon (Sears, 2009), several reports linked ETBF infection to acute diarrhea in young children (Sack et al., 1992) and CRC (Viljoen et al., 2015; Purcell et al., 2017; Zhou et al., 2016; Wu et al., 2009; Goodwin et al., 2011; Housseau et al., 2016; Geis et al., 2015), which will be discussed in this section.

Mechanisms by Which B. fragilis Toxin Could Induce Colon Carcinogenesis The mechanisms of action of BFT are not fully understood, however current research suggests that BFT can cleave E-cadherin, increase intestinal permeability, trigger inflammation and DNA damage, and induce morphological changes and cell proliferation in intestinal epithelial cells.

Microbiota and Colon Cancer: Orchestrating Neoplasia Through DNA Damage and Immune Dysregulation

465

E-cadherin cleavage The adhesion molecule E-cadherin, which usually enables cell-to-cell contacts in the epithelium, is a target for both MPII and BFT. While MPII binds to E-cadherin (Remacle et al., 2014), BFT cleaves it thereby weakening tight junctions, and increasing cell permeability, which lead to a series of morphological changes in human intestinal cells (Wu et al., 1998). E-cadherin is not only an adhesion molecule, its cytoplasmic domain associates with b-catenin. Binding of E-cadherin can lead to the translocation of b-catenin into the nucleus and transcriptional activation of several genes that control cell proliferation (Fig. 2d). Accordingly, when human CECs (HT29/C1) are treated with BFT, cleavage of membrane associated-E-cadherin leads to nuclear localization of b-catenin, and transcriptional upregulation of c-myc, triggering cell proliferation (Wu et al., 2003). Some studies have suggested that BFT induces IL-8 secretion by cleaving E-cadherin in human intestinal cells. In fact, E-cadherin deficient cells do not express IL-8 after exposure to BFT. However, disruption of cell-to-cell interactions after EDTA-treatment is enough to induce IL-8 production by HT29/C1 cells. Furthermore b-catenin activation alone can induce IL-8 secretion, suggesting that E-cadherin cleavage is not required to induce IL-8 secretion (Hwang et al., 2013).

Inflammation ETBF has also been shown to trigger either colitis or CAC depending on the host’s T-cell polarization after infection. For example, Apcmin/þ (multiple intestinal neoplasia) mice infected with ETBF develop colonic neoplasia via activation of inflammatory Th17 cells (Wu et al., 2009). After ETBF infection, anti-inflammatory Tregs are recruited into the lamina propria, limiting IL-2 availability, and creating a microenvironment that facilitates pathogenic Th17 polarization, which promotes IL-17 mediated carcinogenesis (Fig. 2e) (Housseau et al., 2016; Geis et al., 2015). In agreement with this finding, ETBF-infected Treg deficient mice do not develop CAC, however in this scenario IL-2 is available for Th1 cell polarization, which produce interferon gamma (IFNg) inducing colitis but not CRC. In this case, EBTF infection could result in two different inflammatory pathways, which depend on the immune status of the host. These findings highlight the importance of characterizing the nature of the inflammatory response after microbial infection (Irrazabal and Martin, 2015), information that is lacking with other CRC-associated bacteria.

DNA damage BFT has also been shown to increase ROS production by inducing SMO, similar to that observed with the H. pylori toxin CagA, which in turn can lead to DNA damage (Fig. 2d). BFT upregulates the enzyme SMO in CEC cell lines (i.e., HT29/c1) and colonic tissue of C57BL/6 mice (Goodwin et al., 2011). SMO is a polyamine catabolic enzyme that leads to increased ROS production after inflammatory stimuli. BFT-mediated SMO-induction increases the levels of y-H2AX in colon cancer cell lines. On the other hand, chronic inflammation and CEC hyperproliferation in ETBF-infected mice is reduced when mice are treated with a SMO-inhibitor (Goodwin et al., 2011). In fact, colonic polyps can be reduced with administration of a SMO-inhibitor in BFT-infected Apcmin/þ mice.

Evidence Supporting a Role for ETBF in Human Colon Carcinogenesis Given the evidence that ETBF could induce CRC in mice, research has now focused on studying the prevalence and association of ETBF with colon carcinogenic lesions in humans. Indeed, the presence and abundance of the BFT gene in mucosal tissue from patients undergoing colonoscopy is positively associated with early-stage tumoral lesions (Purcell et al., 2017). In addition, studies focused on different populations have found that ETBF is significantly increased in tumoral tissue compared to both adjacent normal tissue and healthy controls (Viljoen et al., 2015; Zhou et al., 2016). In conclusion, extensive research supports a role of ETBF in CAC induction in genetically susceptible mouse models. ETBF is also increased in tumoral tissue in human subjects. However, there is not enough evidence to conclude that ETBF is linked to tumor initiation in healthy subjects. ETBF is found in normal tissue and control subjects, suggesting that ETBF might have pathological effects only under certain conditions.

Fusobacterium nucleatum Fusobacterium nucleatum (F. nucleatum) is an invasive, adherent and proinflammatory gram-negative anaerobe that is normally found in the oral cavity, and rarely detected in the gut of healthy humans. However, F. nucleatum can be isolated from the intestines of individuals suffering with IBD (Strauss et al., 2011) which has stimulated further research into F. nucleatum, inflammatory intestinal disease, and CAC.

Mechanisms by Which F. nucleatum Could Induce Colon Carcinogenesis F. nucleatum adheres to CECs, which has been reported to be mediated by binding of the fusobacterial protein Fap2 to Gal-GalNAc lectin that is expressed on CECs, and overexpressed in CRC cells (Abed et al., 2016). In fact, binding of Fap2-deficient F. nucleatum to CRC cells that express Gal-GalNAc is reduced (Abed et al., 2016), suggesting that colonic tumor microenvironment favors F. nucleatum attachment to epithelial cells. It has been proposed that F. nucleatum can induce CRC mainly by inducing CEC proliferation and inhibiting inflammatory cells.

466

Microbiota and Colon Cancer: Orchestrating Neoplasia Through DNA Damage and Immune Dysregulation

Proliferation inducer Cancer cell lines (Caco-2 and SW480) infected with an invasive F. nucleatum strain extracted from human CRC tissue have increased protein levels of activated b-catenin and its targets c-myc and cyclin D1 (Chen et al., 2017) suggesting that invasive F. nucleatum can directly induce hyperproliferation of CRC cells. There are a number of reports that suggest that the effects of F. nucleatum on proliferation could be mediated by the action of the adhesin FadA and through TLR4 activation. First, F. nucleatum’s FadA adhesin binds to E-cadherin, which increases free b-catenin, thereby activating the Wnt-pathway and inducing cell proliferation (Fig. 2f) (Rubinstein et al., 2013). This adhesin could potentially be used as a diagnostic marker for CRC since the levels of the fadA gene in the colon of patients with adenomas are > 10–100 times higher compared to controls. Second, inhibition of the Tolllike receptor 4 in CRC cells decreases F. nucleatum-induced activation of the b-catenin pathway (Chen et al., 2017). Interestingly, microRNA (miRNA) 21, which is expressed under the control of TLR4, can also affect CEC proliferation since inhibitors of miR21 reduce F. nucleatum-induced proliferation and invasion in cultured cells (Fig. 2g) (Yang et al., 2017). Indeed, F. nucleatum-infected miR21a-deficient mice have less tumors and increased survival compared to control mice. In addition, it has also been reported that elevated levels of F. nucleatum DNA in human tumors correlate with high MiR21 expression and worse clinical outcome (Yang et al., 2017).

Immunosuppresor Besides being necessary for adherence, F. nucleatum Fap2 protein can inhibit specific immune responses. Fap2 interacts with the inhibitory receptor TIGIT, which is expressed in all human natural killer (NK) cells and on some T-cell populations, and therefore F. nucleatum can inhibit the activity of NK cells and tumor infiltrating lymphocytes (Gur et al., 2015). In fact, there is an inverse correlation between CRCs associated with F. nucleatum and CD3þ T-cell density (Mima et al., 2015), suggesting that this bacterium could promote CRC by downregulating antitumoral T-cell cytotoxicity. In addition, F. nucleatum-induced CRCs in ApcMin/þ mice have an elevated number of infiltrating myeloid-derived suppressor cells, which have immunosuppressive activity and allow tumor growth (Kostic et al., 2013). APCmin/þ mice gavaged with F. nucleatum develop more tumors in the colon (Kostic et al., 2013; Yang et al., 2017), and have higher mortality than vehicle treated controls (Yang et al., 2017). Tumors from F. nucleatum-treated APCmin/þ mice show a proinflammatory profile similar to human Fusobacterium-positive tumors, however F. nucleatum does not worsen colitis and CAC in mice (Kostic et al., 2013), suggesting that this bacterium can induce an antiinflammatory environment that facilitates tumor progression but not tumor initiation. Moreover, F. nucleatum produce butyrate, and butyrate can induce Treg differentiation (Furusawa et al., 2013), which could facilitate tumor growth by hampering antitumoral immune responses.

Evidence Supporting a Role of F. nucleatum in Human Colon Carcinogenesis F. nucleatum is significantly increased in tumors compared to adjacent normal tissue and tissue from healthy controls (Viljoen et al., 2015; Zhou et al., 2016; Castellarin et al., 2012; Kostic et al., 2012). The presence of F. nucleatum is also significantly associated with microsatellite instability (MSI) high tumors (Nosho et al., 2016). In addition, F. nucleatum strains extracted from inflamed mucosa of IBD patients are more invasive than strains extracted from control patients (Strauss et al., 2011), and invasive F. nucleatum infection is positively associated with b-catenin activation in CRC tissues (Chen et al., 2017). Overall there is substantial evidence supporting a positive association between F. nucleatum infection and CRC in humans. However, in contrast to other bacteria associated to CRC, F. nucleatum seems to have antiinflammatory capacities that facilitate tumor escape from immune surveillance, and therefore might promote tumor progression rather than initiate tumors.

Escherichia coli Escherichia coli (E. coli) are commensal gram-negative bacteria found in normal gut microbiota and rarely cause disease. However, a subgroup of these bacteria, named enteropathogenic E. coli (EPEC), can attach to the intestinal mucosa and produce toxins that induce bowel disease in humans.

Mechanisms by Which EPEC Could Induce Colon Carcinogenesis Some reports have found that EPEC incidence is significantly higher in CRC patients than in healthy controls (Magdy et al., 2015), prompting research in search for a possible link between EPEC and CRC. Current research indicates that inflammation might favor EPEC carcinogenesis, which could potentially be induced by two mechanisms: DNA damage, and MMR downregulation.

DNA damage EPEC uses a type 3 secretion system (T3SS) to inject various effector proteins into the host cells, and it also express the genotoxins colibactin and CDT (Fig. 2b, see H. hepaticus). Colibactin is encoded within the polyketide synthase (pks) gene island, and crosslinks DNA (Vizcaino and Crawford, 2015), promoting genomic instability in vitro (Cuevas-Ramos et al., 2010), and DSBs in mouse enterocytes (Cuevas-Ramos et al., 2010; Arthur et al., 2012). In a colitis AOM/IL-10/ model of CRC, infection with pksþ E. coli increased tumor counts (Arthur et al., 2012). This finding was recapitulated in an AOM/DSS treated mice models (Cuevas-Ramos et al., 2010; Cougnoux et al., 2014; Dalmasso et al., 2014). Additionally, inflammation seems to accelerate pksþ E. coli-induced carcinogenesis: AOM-treated IL-10/ mice infected with pksþ E. coli exhibit significantly more tumors than

Microbiota and Colon Cancer: Orchestrating Neoplasia Through DNA Damage and Immune Dysregulation

467

lymphocyte-deficient AOM-treated Rag2/ IL-10/ mice infected with pksþ E. coli (Arthur et al., 2014). The E. coli pks island is one of the top five operons upregulated in AOM-treated IL-10/ mice compared to untreated mice (Arthur et al., 2014), suggesting that during carcinogenesis, the inflamed intestinal environment induce pks transcription hastening carcinogenesis in infected mice. The EPEC genotoxin CDT also induces DSBs in cultured cells, which leads to ATM-mediated induction of G2/M cell cycle arrest (Fedor et al., 2013). Moreover, CDT accelerates growth in p53-deficient but not in KRAS-deficient human CECs (Graillot et al., 2016), suggesting that CDT genotoxicity could be exacerbated in DNA repair-defective cells.

MMR downregulation EPEC has been found to downregulate the expression of the MMR proteins MSH2 and MLH1 in cultured CRC cell lines (Fig. 2h) (Maddocks et al., 2009). MMR activity downregulation is mediated by EPEC secretion of the protein EspF, which targets the mitochondria (Maddocks et al., 2013). Interestingly, H. pylori VacA also targets the mitochondria and reduce the MMR function, suggesting that these toxins may disrupt similar mitochondrial pathways.

Evidence Supporting a Role of EPEC on Human Colon Carcinogenesis Sixty seven percentage of CRC patients harbor pks þ E. coli compared to 21% in normal controls (Arthur et al., 2012). Furthermore, although there is a significant increase in total E. coli colonization in human MSI tumors, colibactin-producing E. coli have been found to be enriched in microsatellite stable (MSS) tumors (Gagniere et al., 2017). In contrast, another report found that there is no significant difference in the concentration of pks-positive bacterial DNA between CRC, adenoma, and control patient groups (Shimpoh et al., 2017). In summary, EPEC can induce CRC in inflammatory mouse models. In addition, EPEC encode two genotoxins that control the cell cycle and induce DNA damage. Moreover, current data indicate that there is a positive association between EPEC infection and CRC in humans (Magdy et al., 2015; Arthur et al., 2012), supporting a role for EPEC in CRC initiation in humans.

Conclusion It is evident that most bacteria that have been linked to CRC seem to exploit similar mechanisms to induce carcinogenesis: DNA damage, immune dysregulation, and hyperproliferation (Fig. 2). All the bacteria previously described, with exception of S. bovis and F. nucleatum, have the capacity to induce DNA damage (Chaturvedi et al., 2011; Xu et al., 2004; Wang et al., 2017; Goodwin et al., 2011; Cuevas-Ramos et al., 2010). Remarkably, as with irradiation-induced DNA damage, most bacteria seem to trigger ROS production, which in turn leads to DSB formation. Oxidative stress induces DNA oxidation which is usually repaired by the MMR and BER (base excision repair) pathways (Russo et al., 2007). Therefore, DNA repair-deficient subjects might be more susceptible to the effects of ROS. Furthermore, it has been implied that E. coli can create the perfect storm for colon carcinogenesis by both inducing DNA damage and downregulating DNA damage repair proteins (Maddocks et al., 2013). Immune dysregulation is another factor exploited by many microbes. S. bovis induce an inflammatory response that facilitate uncontrolled cell proliferation, while carcinogenesis induced by H. pylori, H. hepaticus, ETBF, and pksþ E. coli is facilitated by inflammation. Furthermore, ETBF infection in mice could lead to either colitis or CAC depending on the inflammatory response mounted by the host (Geis et al., 2015). In contrast, F. nucleatum seems to induce an immunosuppressive environment that facilitate tumor progression (Mima et al., 2015). Many microbes implicated in CRC also induce CEC hyperproliferation. The most common CRC-associated mutations are in genes that control CEC proliferation (i.e., APC and b-catenin), and some microbes seem to exploit this pathway to induce proliferation of CECs. On the other hand, hyperproliferation induced by microbes could be a consequence of deficiencies in DDR in the host, since genetic mutations induced by microbes would not lead to the activation of cell cycle checkpoints, allowing damaged DNA to be propagated to daughter cells, facilitating carcinogenesis. Finally, current research suggests that H. pylori is the only bacterium that can induce carcinogenesis in any host, irrespective of genetic background, while most of the aforementioned bacteria cannot induce CEC transformation in healthy subjects and need a “first hit,” such as a specific genetic deficiency, that sensitizes the host to the oncogenic effects of the microbe. Indeed, this pathogen þ susceptibility gene formula has been previously proposed for IBD (Cadwell et al., 2010), and seems applicable to many cases of microbial-induced CRC or CAC.

Acknowledgments We would like to thank the Martin lab for helpful comments, and to Rodrigo Irrazábal for the illustrations (Figs. 1 and 2). This work is funded by the Canadian Cancer Society (grant 703185) and Canadian Institute of Health Research (grant 144628).

See also: Carcinogenesis: Role of Reactive Oxygen and Nitrogen Species. Colorectal Cancer: Pathology and Genetics.

468

Microbiota and Colon Cancer: Orchestrating Neoplasia Through DNA Damage and Immune Dysregulation

References Abed, J., Emgard, J.E., Zamir, G., Faroja, M., Almogy, G., Grenov, A., et al., 2016. Fap2 mediates fusobacterium nucleatum colorectal adenocarcinoma enrichment by binding to tumor-expressed Gal-GalNAc. Cell Host & Microbe 20 (2), 215–225. Anon, 1994. Infection with Helicobacter pylori. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans 61, 177–240. Arthur, J.C., Perez-Chanona, E., Muhlbauer, M., Tomkovich, S., Uronis, J.M., Fan, T.J., et al., 2012. Intestinal inflammation targets cancer-inducing activity of the microbiota. Science 338 (6103), 120–123. Arthur, J.C., Gharaibeh, R.Z., Muhlbauer, M., Perez-Chanona, E., Uronis, J.M., McCafferty, J., et al., 2014. Microbial genomic analysis reveals the essential role of inflammation in bacteria-induced colorectal cancer. Nature Communications 5, 4724. Avenaud, P., Castroviejo, M., Claret, S., Rosenbaum, J., Megraud, F., Menard, A., 2004. Expression and activity of the cytolethal distending toxin of Helicobacter hepaticus. Biochemical and Biophysical Research Communications 318 (3), 739–745. Bain CC, Oliphant CJ, Thomson CA, Kullberg MC, Mowat AM. Pro-inflammatory role of monocyte-derived CX3CR1int macrophages in Helicobacter hepaticus-induced colitis. Infection and Immunity 2018, 86(2). pii: e00579-17. Belcheva, A., Irrazabal, T., Martin, A., 2015. Gut microbial metabolism and colon cancer: Can manipulations of the microbiota be useful in the management of gastrointestinal health? BioEssays 37 (4), 403–412. Bezine, E., Malaise, Y., Loeuillet, A., Chevalier, M., Boutet-Robinet, E., Salles, B., et al., 2016. Cell resistance to the Cytolethal distending toxin involves an association of DNA repair mechanisms. Scientific Reports 6, 36022. Biarc, J., Nguyen, I.S., Pini, A., Gosse, F., Richert, S., Thierse, D., et al., 2004. Carcinogenic properties of proteins with pro-inflammatory activity from Streptococcus infantarius (formerly S. bovis). Carcinogenesis 25 (8), 1477–1484. Blase, J.L., Campbell, P.T., Gapstur, S.M., Pawlita, M., Michel, A., Waterboer, T., et al., 2016. Prediagnostic Helicobacter pylori antibodies and colorectal cancer risk in an elderly, Caucasian population. Helicobacter 21 (6), 488–492. Boleij, A., Muytjens, C.M., Bukhari, S.I., Cayet, N., Glaser, P., Hermans, P.W., et al., 2011. Novel clues on the specific association of Streptococcus gallolyticus subsp gallolyticus with colorectal cancer. The Journal of Infectious Diseases 203 (8), 1101–1109. Bradley, P.H., Pollard, K.S., 2017. Proteobacteria explain significant functional variability in the human gut microbiome. Microbiome 5 (1), 36. Breuer-Katschinski, B., Nemes, K., Marr, A., Rump, B., Leiendecker, B., Breuer, N., et al., 1999. Helicobacter pylori and the risk of colonic adenomas. Colorectal Adenoma Study Group. Digestion 60 (3), 210–215. Brown, K.D., Rathi, A., Kamath, R., Beardsley, D.I., Zhan, Q., Mannino, J.L., et al., 2003. The mismatch repair system is required for S-phase checkpoint activation. Nature Genetics 33 (1), 80–84. Bugaytsova, J.A., Bjornham, O., Chernov, Y.A., Gideonsson, P., Henriksson, S., Mendez, M., et al., 2017. Helicobacter pylori adapts to chronic infection and gastric disease via pHresponsive BabA-mediated adherence. Cell Host & Microbe 21 (3), 376–389. Burns, C.A., McCaughey, R., Lauter, C.B., 1985. The association of Streptococcus bovis fecal carriage and colon neoplasia: Possible relationship with polyps and their premalignant potential. The American Journal of Gastroenterology 80 (1), 42–46. Buti, L., Spooner, E., Van der Veen, A.G., Rappuoli, R., Covacci, A., Ploegh, H.L., 2011. Helicobacter pylori cytotoxin-associated gene a (CagA) subverts the apoptosis-stimulating protein of p53 (ASPP2) tumor suppressor pathway of the host. Proceedings of the National Academy of Sciences of the United States of America 108 (22), 9238–9243. Byndloss, M.X., Olsan, E.E., Rivera-Chavez, F., Tiffany, C.R., Cevallos, S.A., Lokken, K.L., et al., 2017. Microbiota-activated PPAR-gamma signaling inhibits dysbiotic Enterobacteriaceae expansion. Science 357 (6351), 570–575. Cadwell, K., Patel, K.K., Maloney, N.S., Liu, T.C., Ng, A.C., Storer, C.E., et al., 2010. Virus-plus-susceptibility gene interaction determines Crohn’s disease gene Atg16L1 phenotypes in intestine. Cell 141 (7), 1135–1145. Castellarin, M., Warren, R.L., Freeman, J.D., Dreolini, L., Krzywinski, M., Strauss, J., et al., 2012. Fusobacterium nucleatum infection is prevalent in human colorectal carcinoma. Genome Research 22 (2), 299–306. Cater 2nd, R.E., 1992. The clinical importance of hypochlorhydria (a consequence of chronic Helicobacter infection): Its possible etiological role in mineral and amino acid malabsorption, depression, and other syndromes. Medical Hypotheses 39 (4), 375–383. Chang, W.L., Yang, H.B., Cheng, H.C., Yeh, Y.C., Kao, C.Y., Wu, J.J., et al., 2015. Intracellular osteopontin induced by CagA-positive Helicobacter pylori promotes beta-catenin accumulation and interleukin-8 secretion in gastric epithelial cells. Helicobacter 20 (6), 476–484. Chaturvedi, R., Asim, M., Romero-Gallo, J., Barry, D.P., Hoge, S., de Sablet, T., et al., 2011. Spermine oxidase mediates the gastric cancer risk associated with Helicobacter pylori CagA. Gastroenterology 141 (5), 1696-708 e1-2. Chaturvedi, R., Asim, M., Piazuelo, M.B., Yan, F., Barry, D.P., Sierra, J.C., et al., 2014. Activation of EGFR and ERBB2 by Helicobacter pylori results in survival of gastric epithelial cells with DNA damage. Gastroenterology 146 (7), 1739-51 e14. Chen, X.Z., Schottker, B., Castro, F.A., Chen, H., Zhang, Y., Holleczek, B., et al., 2016. Association of Helicobacter pylori infection and chronic atrophic gastritis with risk of colonic, pancreatic and gastric cancer: A ten-year follow-up of the ESTHER cohort study. Oncotarget 7 (13), 17182–17193. Chen, Y., Peng, Y., Yu, J., Chen, T., Wu, Y., Shi, L., et al., 2017. Invasive Fusobacterium nucleatum activates beta-catenin signaling in colorectal cancer via a TLR4/P-PAK1 cascade. Oncotarget 8 (19), 31802–31814. Corredoira, J.C., Alonso, M.P., Garcia, J.F., Casariego, E., Coira, A., Rodriguez, A., et al., 2005. Clinical characteristics and significance of Streptococcus salivarius bacteremia and Streptococcus bovis bacteremia: A prospective 16-year study. European Journal of Clinical Microbiology & Infectious Diseases 24 (4), 250–255. Cougnoux, A., Dalmasso, G., Martinez, R., Buc, E., Delmas, J., Gibold, L., et al., 2014. Bacterial genotoxin colibactin promotes colon tumour growth by inducing a senescenceassociated secretory phenotype. Gut 63 (12), 1932–1942. Cuevas-Ramos, G., Petit, C.R., Marcq, I., Boury, M., Oswald, E., Nougayrede, J.P., 2010. Escherichia coli induces DNA damage in vivo and triggers genomic instability in mammalian cells. Proceedings of the National Academy of Sciences of the United States of America 107 (25), 11537–11542. Dalmasso, G., Cougnoux, A., Delmas, J., Darfeuille-Michaud, A., Bonnet, R., 2014. The bacterial genotoxin colibactin promotes colon tumor growth by modifying the tumor microenvironment. Gut Microbes 5 (5), 675–680. Dubrow, R., Edberg, S., Wikfors, E., Callan, D., Troncale, F., Vender, R., et al., 1991. Fecal carriage of Streptococcus bovis and colorectal adenomas. Gastroenterology 101 (3), 721–725. Eckburg, P.B., Bik, E.M., Bernstein, C.N., Purdom, E., Dethlefsen, L., Sargent, M., et al., 2005. Diversity of the human intestinal microbial flora. Science 308 (5728), 1635–1638. Ellmerich, S., Scholler, M., Duranton, B., Gosse, F., Galluser, M., Klein, J.P., et al., 2000. Promotion of intestinal carcinogenesis by Streptococcus bovis. Carcinogenesis 21 (4), 753–756. Epplein, M., Pawlita, M., Michel, A., Peek Jr., R.M., Cai, Q., Blot, W.J., 2013. Helicobacter pylori protein-specific antibodies and risk of colorectal cancer. Cancer Epidemiology, Biomarkers & Prevention 22 (11), 1964–1974. Erdman, S.E., Rao, V.P., Poutahidis, T., Ihrig, M.M., Ge, Z., Feng, Y., et al., 2003. CD4(þ)CD25(þ) regulatory lymphocytes require interleukin 10 to interrupt colon carcinogenesis in mice. Cancer Research 63 (18), 6042–6050. Erdman, S.E., Rao, V.P., Poutahidis, T., Rogers, A.B., Taylor, C.L., Jackson, E.A., et al., 2009. Nitric oxide and TNF-alpha trigger colonic inflammation and carcinogenesis in Helicobacter hepaticus-infected, Rag2-deficient mice. Proceedings of the National Academy of Sciences of the United States of America 106 (4), 1027–1032.

Microbiota and Colon Cancer: Orchestrating Neoplasia Through DNA Damage and Immune Dysregulation

469

Fahrer, J., Huelsenbeck, J., Jaurich, H., Dorsam, B., Frisan, T., Eich, M., et al., 2014. Cytolethal distending toxin (CDT) is a radiomimetic agent and induces persistent levels of DNA double-strand breaks in human fibroblasts. DNA Repair (Amst) 18, 31–43. Falck, J., Mailand, N., Syljuasen, R.G., Bartek, J., Lukas, J., 2001. The ATM-Chk2-Cdc25A checkpoint pathway guards against radioresistant DNA synthesis. Nature 410 (6830), 842–847. Falsafi, T., Mahboubi, M., 2013. Helicobacter hepaticus, a new pathogenic species of the Helicobacter genus: Similarities and differences with H. pylori. Iranian Journal of Microbiology 5 (3), 185–194. Fedor, Y., Vignard, J., Nicolau-Travers, M.L., Boutet-Robinet, E., Watrin, C., Salles, B., et al., 2013. From single-strand breaks to double-strand breaks during S-phase: A new mode of action of the Escherichia coli cytolethal distending toxin. Cellular Microbiology 15 (1), 1–15. Ferreira, R.M., Pinto-Ribeiro, I., Wen, X., Marcos-Pinto, R., Dinis-Ribeiro, M., Carneiro, F., et al., 2016. Helicobacter pylori cagA promoter region sequences influence CagA expression and interleukin 8 secretion. The Journal of Infectious Diseases 213 (4), 669–673. Fujimori, S., Kishida, T., Kobayashi, T., Sekita, Y., Seo, T., Nagata, K., et al., 2005. Helicobacter pylori infection increases the risk of colorectal adenoma and adenocarcinoma, especially in women. Journal of Gastroenterology 40 (9), 887–893. Furusawa, Y., Obata, Y., Fukuda, S., Endo, T.A., Nakato, G., Takahashi, D., et al., 2013. Commensal microbe-derived butyrate induces the differentiation of colonic regulatory T cells. Nature 504, 446–450. Gagniere, J., Bonnin, V., Jarrousse, A.S., Cardamone, E., Agus, A., Uhrhammer, N., et al., 2017. Interactions between microsatellite instability and human gut colonization by Escherichia coli in colorectal cancer. Clinical Science (London, England) 131 (6), 471–485. Ge, Z., Feng, Y., Ge, L., Parry, N., Muthupalani, S., Fox, J.G., 2017. Helicobacter hepaticus cytolethal distending toxin promotes intestinal carcinogenesis in 129Rag2-deficient mice. Cellular Microbiology 19 (7). Geis, A.L., Fan, H., Wu, X., Wu, S., Huso, D.L., Wolfe, J.L., et al., 2015. Regulatory T-cell response to enterotoxigenic Bacteroides fragilis colonization triggers IL17-dependent colon carcinogenesis. Cancer Discovery 5 (10), 1098–1109. Gnad, T., Feoktistova, M., Leverkus, M., Lendeckel, U., Naumann, M., 2010. Helicobacter pylori-induced activation of beta-catenin involves low density lipoprotein receptor-related protein 6 and Dishevelled. Molecular Cancer 9, 31. Goodwin, A.C., Destefano Shields, C.E., Wu, S., Huso, D.L., Wu, X., Murray-Stewart, T.R., et al., 2011. Polyamine catabolism contributes to enterotoxigenic Bacteroides fragilisinduced colon tumorigenesis. Proceedings of the National Academy of Sciences of the United States of America 108 (37), 15354–15359. Graillot, V., Dormoy, I., Dupuy, J., Shay, J.W., Huc, L., Mirey, G., et al., 2016. Genotoxicity of cytolethal distending toxin (CDT) on isogenic human colorectal cell lines: Potential promoting effects for colorectal carcinogenesis. Frontiers in Cellular and Infection Microbiology 6, 34. Gur, C., Ibrahim, Y., Isaacson, B., Yamin, R., Abed, J., Gamliel, M., et al., 2015. Binding of the Fap2 protein of Fusobacterium nucleatum to human inhibitory receptor TIGIT protects tumors from immune cell attack. Immunity 42 (2), 344–355. Hamada, T., Yokota, K., Ayada, K., Hirai, K., Kamada, T., Haruma, K., et al., 2009. Detection of Helicobacter hepaticus in human bile samples of patients with biliary disease. Helicobacter 14 (6), 545–551. Hoffmann, C., Dollive, S., Grunberg, S., Chen, J., Li, H., Wu, G.D., et al., 2013. Archaea and fungi of the human gut microbiome: Correlations with diet and bacterial residents. PLoS One 8 (6), e66019. Hooper, L.V., Gordon, J.I., 2001. Commensal host–bacterial relationships in the gut. Science 292 (5519), 1115–1118. Hoppes, W.L., Lerner, P.I., 1974. Nonenterococcal group-D streptococcal endocarditis caused by Streptococcus bovis. Annals of Internal Medicine 81 (5), 588–593. Housseau, F., Wu, S., Wick, E.C., Fan, H., Wu, X., Llosa, N.J., et al., 2016. Redundant innate and adaptive sources of IL17 production drive colon tumorigenesis. Cancer Research 76 (8), 2115–2124. Human Microbiome Project Consortium, 2012. Structure, function and diversity of the healthy human microbiome. Nature 486 (7402), 207–214. Hwang, S., Gwon, S.Y., Kim, M.S., Lee, S., Rhee, K.J., 2013. Bacteroides fragilis toxin induces IL-8 secretion in HT29/C1 cells through disruption of E-cadherin junctions. Immune Network 13 (5), 213–217. Irrazabal, T., Martin, A., 2015. T regulatory cells gone bad: An oncogenic immune response against enterotoxigenic B. fragilis infection leads to colon cancer. Cancer Discovery 5 (10), 1021–1023. Irrazabal, T., Belcheva, A., Girardin, S.E., Martin, A., Philpott, D.J., 2014. The multifaceted role of the intestinal microbiota in colon cancer. Molecular Cell 54 (2), 309–320. Kanno, T., Matsuki, T., Oka, M., Utsunomiya, H., Inada, K., Magari, H., et al., 2009. Gastric acid reduction leads to an alteration in lower intestinal microflora. Biochemical and Biophysical Research Communications 381 (4), 666–670. Keenan, J.I., Beaugie, C.R., Jasmann, B., Potter, H.C., Collett, J.A., Frizelle, F.A., 2010. Helicobacter species in the human colon. Colorectal Disease 12 (1), 48–53. Klein, R.S., Recco, R.A., Catalano, M.T., Edberg, S.C., Casey, J.I., Steigbigel, N.H., 1977. Association of Streptococcus bovis with carcinoma of the colon. The New England Journal of Medicine 297 (15), 800–802. Koniger, V., Holsten, L., Harrison, U., Busch, B., Loell, E., Zhao, Q., et al., 2016. Helicobacter pylori exploits human CEACAMs via HopQ for adherence and translocation of CagA. Nature Microbiology 2, 16188. Kostic, A.D., Gevers, D., Pedamallu, C.S., Michaud, M., Duke, F., Earl, A.M., et al., 2012. Genomic analysis identifies association of Fusobacterium with colorectal carcinoma. Genome Research 22 (2), 292–298. Kostic, A.D., Chun, E., Robertson, L., Glickman, J.N., Gallini, C.A., Michaud, M., et al., 2013. Fusobacterium nucleatum potentiates intestinal tumorigenesis and modulates the tumor-immune microenvironment. Cell Host & Microbe 14 (2), 207–215. Lee, J.Y., Park, H.W., Choi, J.Y., Lee, J.S., Koo, J.E., Chung, E.J., et al., 2016. Helicobacter pylori infection with atrophic gastritis is an independent risk factor for advanced colonic neoplasm. Gut Liver 10 (6), 902–909. Leung, N., Turbide, C., Olson, M., Marcus, V., Jothy, S., Beauchemin, N., 2006. Deletion of the carcinoembryonic antigen-related cell adhesion molecule 1 (Ceacam1) gene contributes to colon tumor progression in a murine model of carcinogenesis. Oncogene 25 (40), 5527–5536. Litvak, Y., Byndloss, M.X., Tsolis, R.M., Baumler, A.J., 2017. Dysbiotic Proteobacteria expansion: A microbial signature of epithelial dysfunction. Current Opinion in Microbiology 39, 1–6. Lloyd-Price, J., Abu-Ali, G., Huttenhower, C., 2016. The healthy human microbiome. Genome Medicine 8 (1), 51. Machado, A.M., Figueiredo, C., Touati, E., Maximo, V., Sousa, S., Michel, V., et al., 2009. Helicobacter pylori infection induces genetic instability of nuclear and mitochondrial DNA in gastric cells. Clinical Cancer Research 15 (9), 2995–3002. Maddocks, O.D., Short, A.J., Donnenberg, M.S., Bader, S., Harrison, D.J., 2009. Attaching and effacing Escherichia coli downregulate DNA mismatch repair protein in vitro and are associated with colorectal adenocarcinomas in humans. PLoS One 4 (5), e5517. Maddocks, O.D., Scanlon, K.M., Donnenberg, M.S., 2013. An Escherichia coli effector protein promotes host mutation via depletion of DNA mismatch repair proteins. MBio 4 (3), e00152-13. Magdy, A., Elhadidy, M., Abd Ellatif, M.E., El Nakeeb, A., Abdallah, E., Thabet, W., et al., 2015. Enteropathogenic Escherichia coli (EPEC): Does it have a role in colorectal tumourigenesis? A Prospective Cohort Study. International Journal of Surgery 18, 169–173. McCaskey, S.J., Rondini, E.A., Clinthorne, J.F., Langohr, I.M., Gardner, E.M., Fenton, J.I., 2012. Increased presence of effector lymphocytes during Helicobacter hepaticus-induced colitis. World Journal of Gastroenterology 18 (13), 1459–1469. McClain, M.S., Beckett, A.C., Cover, T.L., 2017. Helicobacter pylori vacuolating toxin and gastric cancer. Toxins (Basel) 9 (10). McCoy, C.W., Mason 3rd., J.M., 1951. Enterococcal endocarditis associated with carcinoma of the sigmoid; report of a case. Journal of the Medical Association of the State of Alabama 21 (6), 162–166.

470

Microbiota and Colon Cancer: Orchestrating Neoplasia Through DNA Damage and Immune Dysregulation

Medema, J.P., Vermeulen, L., 2011. Microenvironmental regulation of stem cells in intestinal homeostasis and cancer. Nature 474 (7351), 318–326. Mima, K., Sukawa, Y., Nishihara, R., Qian, Z.R., Yamauchi, M., Inamura, K., et al., 2015. Fusobacterium nucleatum and T cells in colorectal carcinoma. JAMA Oncology 1 (5), 653–661. Minot, S., Bryson, A., Chehoud, C., Wu, G.D., Lewis, J.D., Bushman, F.D., 2013. Rapid evolution of the human gut virome. Proceedings of the National Academy of Sciences of the United States of America 110 (30), 12450–12455. Moayyedi, P., Hunt, R.H., 2004. Helicobacter pylori public health implications. Helicobacter 9 (Suppl. 1), 67–72. Molinari, M., Salio, M., Galli, C., Norais, N., Rappuoli, R., Lanzavecchia, A., et al., 1998. Selective inhibition of ii-dependent antigen presentation by Helicobacter pylori toxin VacA. The Journal of Experimental Medicine 187 (1), 135–140. Moshkowitz, M., Arber, N., Halpern, Z., Gilat, T., 1992. Fecal carriage of Streptococcus bovis and inflammatory bowel disease. Gastroenterology 102 (5), 1829. Moss, S.F., Neugut, A.I., Garbowski, G.C., Wang, S., Treat, M.R., Forde, K.A., 1995. Helicobacter pylori seroprevalence and colorectal neoplasia: Evidence against an association. Journal of the National Cancer Institute 87 (10), 762–763. Nagamine, C.M., Rogers, A.B., Fox, J.G., Schauer, D.B., 2008a. Helicobacter hepaticus promotes azoxymethane-initiated colon tumorigenesis in BALB/c-IL10-deficient mice. International Journal of Cancer 122 (4), 832–838. Nagamine, C.M., Sohn, J.J., Rickman, B.H., Rogers, A.B., Fox, J.G., Schauer, D.B., 2008b. Helicobacter hepaticus infection promotes colon tumorigenesis in the BALB/c-Rag2(/ ) Apc(Min/þ) mouse. Infection and Immunity 76 (6), 2758–2766. Nam, J.H., Hong, C.W., Kim, B.C., Shin, A., Ryu, K.H., Park, B.J., et al., 2017. Helicobacter pylori infection is an independent risk factor for colonic adenomatous neoplasms. Cancer Causes & Control 28 (2), 107–115. Neumaier, M., Paululat, S., Chan, A., Matthaes, P., Wagener, C., 1993. Biliary glycoprotein, a potential human cell adhesion molecule, is down-regulated in colorectal carcinomas. Proceedings of the National Academy of Sciences of the United States of America 90 (22), 10744–10748. Nilsson, I., Lindgren, S., Eriksson, S., Wadstrom, T., 2000. Serum antibodies to Helicobacter hepaticus and Helicobacter pylori in patients with chronic liver disease. Gut 46 (3), 410–414. Nosho, K., Sukawa, Y., Adachi, Y., Ito, M., Mitsuhashi, K., Kurihara, H., et al., 2016. Association of Fusobacterium nucleatum with immunity and molecular alterations in colorectal cancer. World Journal of Gastroenterology 22 (2), 557–566. Painter, R.B., Young, B.R., 1980. Radiosensitivity in ataxia-telangiectasia: A new explanation. Proceedings of the National Academy of Sciences of the United States of America 77 (12), 7315–7317. Papastergiou, V., Karatapanis, S., Georgopoulos, S.D., 2016. Helicobacter pylori and colorectal neoplasia: Is there a causal link? World Journal of Gastroenterology 22 (2), 649–658. Peek Jr., R.M., Blaser, M.J., 2002. Helicobacter pylori and gastrointestinal tract adenocarcinomas. Nature Reviews Cancer 2 (1), 28–37. Poulogiannis, G., Frayling, I.M., Arends, M.J., 2010. DNA mismatch repair deficiency in sporadic colorectal cancer and Lynch syndrome. Histopathology 56 (2), 167–179. Purcell, R.V., Pearson, J., Aitchison, A., Dixon, L., Frizelle, F.A., Keenan, J.I., 2017. Colonization with enterotoxigenic Bacteroides fragilis is associated with early-stage colorectal neoplasia. PLoS One 12 (2), e0171602. Raskov, H., Pommergaard, H.C., Burcharth, J., Rosenberg, J., 2014. Colorectal carcinogenesisdUpdate and perspectives. World Journal of Gastroenterology 20 (48), 18151– 18164. Remacle, A.G., Shiryaev, S.A., Strongin, A.Y., 2014. Distinct interactions with cellular E-cadherin of the two virulent metalloproteinases encoded by a Bacteroides fragilis pathogenicity island. PLoS One 9 (11), e113896. Renga, M., Brandi, G., Paganelli, G.M., Calabrese, C., Papa, S., Tosti, A., et al., 1997. Rectal cell proliferation and colon cancer risk in patients with hypergastrinaemia. Gut 41 (3), 330–332. Rivera-Chavez, F., Zhang, L.F., Faber, F., Lopez, C.A., Byndloss, M.X., Olsan, E.E., et al., 2016. Depletion of butyrate-producing clostridia from the gut microbiota drives an aerobic luminal expansion of Salmonella. Cell Host & Microbe 19 (4), 443–454. Rivera-Chavez, F., Lopez, C.A., Baumler, A.J., 2017. Oxygen as a driver of gut dysbiosis. Free Radical Biology & Medicine 105, 93–101. Rubinstein, M.R., Wang, X., Liu, W., Hao, Y., Cai, G., Han, Y.W., 2013. Fusobacterium nucleatum promotes colorectal carcinogenesis by modulating E-cadherin/beta-catenin signaling via its FadA adhesin. Cell Host & Microbe 14 (2), 195–206. Ruoff, K.L., Miller, S.I., Garner, C.V., Ferraro, M.J., Calderwood, S.B., 1989. Bacteremia with Streptococcus bovis and Streptococcus salivarius: Clinical correlates of more accurate identification of isolates. Journal of Clinical Microbiology 27 (2), 305–308. Russo, M.T., De Luca, G., Degan, P., Bignami, M., 2007. Different DNA repair strategies to combat the threat from 8-oxoguanine. Mutation Research 614 (1–2), 69–76. Sack, R.B., Myers, L.L., Almeido-Hill, J., Shoop, D.S., Bradbury, W.C., Reid, R., et al., 1992. Enterotoxigenic Bacteroides fragilis: Epidemiologic studies of its role as a human diarrhoeal pathogen. Journal of Diarrhoeal Diseases Research 10 (1), 4–9. Sears, C.L., 2009. Enterotoxigenic Bacteroides fragilis: A rogue among symbiotes. Clinical Microbiology Reviews 22 (2), 349–369. Table of Contents. Sears, C.L., Geis, A.L., Housseau, F., 2014. Bacteroides fragilis subverts mucosal biology: From symbiont to colon carcinogenesis. The Journal of Clinical Investigation 124 (10), 4166–4172. Shao, J., Jung, C., Liu, C., Sheng, H., 2005. Prostaglandin E2 stimulates the beta-catenin/T cell factor-dependent transcription in colon cancer. The Journal of Biological Chemistry 280 (28), 26565–26572. Shimpoh, T., Hirata, Y., Ihara, S., Suzuki, N., Kinoshita, H., Hayakawa, Y., et al., 2017. Prevalence of pks-positive Escherichia coli in Japanese patients with or without colorectal cancer. Gut Pathogens 9, 35. Shmuely, H., Passaro, D., Figer, A., Niv, Y., Pitlik, S., Samra, Z., et al., 2001. Relationship between Helicobacter pylori CagA status and colorectal cancer. The American Journal of Gastroenterology 96 (12), 3406–3410. Sillanpaa, J., Nallapareddy, S.R., Qin, X., Singh, K.V., Muzny, D.M., Kovar, C.L., et al., 2009. A collagen-binding adhesin, Acb, and ten other putative MSCRAMM and pilus family proteins of Streptococcus gallolyticus subsp. gallolyticus (Streptococcus bovis group, biotype I). Journal of Bacteriology 191 (21), 6643–6653. Sobhani, I., Lehy, T., Laurent-Puig, P., Cadiot, G., Ruszniewski, P., Mignon, M., 1993. Chronic endogenous hypergastrinemia in humans: Evidence for a mitogenic effect on the colonic mucosa. Gastroenterology 105 (1), 22–30. Sonnenberg, A., Turner, K.O., Genta, R.M., 2017. The ethnic distribution of sessile serrated polyps in the United States is inversely associated with Helicobacter pylori prevalence. Colorectal Disease 19 (11), 996–1002. Spees, A.M., Wangdi, T., Lopez, C.A., Kingsbury, D.D., Xavier, M.N., Winter, S.E., et al., 2013. Streptomycin-induced inflammation enhances Escherichia coli gut colonization through nitrate respiration. MBio 4 (4). Strauss, J., Kaplan, G.G., Beck, P.L., Rioux, K., Panaccione, R., Devinney, R., et al., 2011. Invasive potential of gut mucosa-derived Fusobacterium nucleatum positively correlates with IBD status of the host. Inflammatory Bowel Diseases 17 (9), 1971–1978. Tegtmeyer, N., Neddermann, M., Asche, C.I., Backert, S., 2017. Subversion of host kinases: A key network in cellular signaling hijacked by Helicobacter pylori CagA. Molecular Microbiology 105 (3), 358–372. Terebiznik, M.R., Raju, D., Vazquez, C.L., Torbricki, K., Kulkarni, R., Blanke, S.R., et al., 2009. Effect of Helicobacter pylori’s vacuolating cytotoxin on the autophagy pathway in gastric epithelial cells. Autophagy 5 (3), 370–379. Tjalsma, H., Scholler-Guinard, M., Lasonder, E., Ruers, T.J., Willems, H.L., Swinkels, D.W., 2006. Profiling the humoral immune response in colon cancer patients: Diagnostic antigens from Streptococcus bovis. International Journal of Cancer 119 (9), 2127–2135. Tjalsma, H., Boleij, A., Marchesi, J.R., Dutilh, B.E., 2012. A bacterial driver-passenger model for colorectal cancer: Beyond the usual suspects. Nature Reviews. Microbiology 10 (8), 575–582.

Microbiota and Colon Cancer: Orchestrating Neoplasia Through DNA Damage and Immune Dysregulation

471

Tlaskalova-Hogenova, H., Stepankova, R., Hudcovic, T., Tuckova, L., Cukrowska, B., Lodinova-Zadnikova, R., et al., 2004. Commensal bacteria (normal microflora), mucosal immunity and chronic inflammatory and autoimmune diseases. Immunology Letters 93 (2–3), 97–108. Tsugawa, H., Suzuki, H., Saya, H., Hatakeyama, M., Hirayama, T., Hirata, K., et al., 2012. Reactive oxygen species-induced autophagic degradation of Helicobacter pylori CagA is specifically suppressed in cancer stem-like cells. Cell Host & Microbe 12 (6), 764–777. Viala, J., Chaput, C., Boneca, I.G., Cardona, A., Girardin, S.E., Moran, A.P., et al., 2004. Nod1 responds to peptidoglycan delivered by the Helicobacter pylori cag pathogenicity island. Nature Immunology 5 (11), 1166–1174. Viljoen, K.S., Dakshinamurthy, A., Goldberg, P., Blackburn, J.M., 2015. Quantitative profiling of colorectal cancer-associated bacteria reveals associations between Fusobacterium spp., enterotoxigenic Bacteroides fragilis (ETBF) and clinicopathological features of colorectal cancer. PLoS One 10 (3), e0119462. Vizcaino, M.I., Crawford, J.M., 2015. The colibactin warhead crosslinks DNA. Nature Chemistry 7 (5), 411–417. Wang, C., Gong, G., Sheh, A., Muthupalani, S., Bryant, E.M., Puglisi, D.A., et al., 2017. Interleukin-22 drives nitric oxide-dependent DNA damage and dysplasia in a murine model of colitis-associated cancer. Mucosal Immunology 10 (6), 1504–1517. Wei, J., Nagy, T.A., Vilgelm, A., Zaika, E., Ogden, S.R., Romero-Gallo, J., et al., 2010. Regulation of p53 tumor suppressor by Helicobacter pylori in gastric epithelial cells. Gastroenterology 139 (4), 1333–1343. Winter, S.E., Winter, M.G., Xavier, M.N., Thiennimitr, P., Poon, V., Keestra, A.M., et al., 2013. Host-derived nitrate boosts growth of E. coli in the inflamed gut. Science 339 (6120), 708–711. Wu, S., Lim, K.C., Huang, J., Saidi, R.F., Sears, C.L., 1998. Bacteroides fragilis enterotoxin cleaves the zonula adherens protein, E-cadherin. Proceedings of the National Academy of Sciences of the United States of America 95 (25), 14979–14984. Wu, S., Morin, P.J., Maouyo, D., Sears, C.L., 2003. Bacteroides fragilis enterotoxin induces c-Myc expression and cellular proliferation. Gastroenterology 124 (2), 392–400. Wu, S., Rhee, K.J., Albesiano, E., Rabizadeh, S., Wu, X., Yen, H.R., et al., 2009. A human colonic commensal promotes colon tumorigenesis via activation of T helper type 17 T cell responses. Nature Medicine 15 (9), 1016–1022. Xu, H., Chaturvedi, R., Cheng, Y., Bussiere, F.I., Asim, M., Yao, M.D., et al., 2004. Spermine oxidation induced by Helicobacter pylori results in apoptosis and DNA damage: Implications for gastric carcinogenesis. Cancer Research 64 (23), 8521–8525. Yang, J., Ji, S., Zhang, Y., Wang, J., 2013. Helicobacter hepaticus infection in primary hepatocellular carcinoma tissue. Singapore Medical Journal 54 (8), 451–457. Yang, Y., Weng, W., Peng, J., Hong, L., Yang, L., Toiyama, Y., et al., 2017. Fusobacterium nucleatum increases proliferation of colorectal cancer cells and tumor development in mice by activating toll-like receptor 4 signaling to nuclear factor-kappab, and up-regulating expression of MicroRNA-21. Gastroenterology 152 (4), 851–866.e24. Yong, X., Tang, B., Xiao, Y.F., Xie, R., Qin, Y., Luo, G., et al., 2016. Helicobacter pylori upregulates Nanog and Oct4 via Wnt/beta-catenin signaling pathway to promote cancer stem cell-like properties in human gastric cancer. Cancer Letters 374 (2), 292–303. Young, V.B., Knox, K.A., Schauer, D.B., 2000. Cytolethal distending toxin sequence and activity in the enterohepatic pathogen Helicobacter hepaticus. Infection and Immunity 68 (1), 184–191. Zhou, Y., He, H., Xu, H., Li, Y., Li, Z., Du, Y., et al., 2016. Association of oncogenic bacteria with colorectal cancer in South China. Oncotarget 7 (49), 80794–80802.

Mitogen-Activated Protein Kinases (MAPK) in Cancer Ana Cuenda, Centro Nacional de Biotecnología/CSIC (CNB-CSIC), Madrid, Spain © 2019 Elsevier Inc. All rights reserved.

Glossary Apoptosis Programmed cell death. Cell signaling Ability of the cell to perceive changes in the environment and to produce a correct response. Differentiation Process by which a less specialized cell evolves to a more specialized cell type. Kinase Enzyme that catalyzes the transfer of a phosphate group (phosphorylation) from a phosphate donating molecule to specific substrates. Oncogene Gene that encodes a protein that promote tumor development. Phosphatase Enzyme that catalyzes the removal of a phosphate group from its substrate. Proliferation Process by which the cell number increases. It is the balance between cell division and cell loss. Scaffold protein Protein that bring together two or more proteins in a relatively stable configuration. Tumorigenesis The process of tumor development. Tumor suppressor gene Gene that encodes a protein that suppresses the development of tumor.

Introduction Mitogen-activated protein kinases (MAPKs) are a family of kinases, which are ubiquitous and highly conserved in all eukaryotes. They belong to the CMGC group of protein kinases that also includes glycogen synthase kinases (GSK), cyclin-dependent kinases (CDKs) and CDK-like kinases. The different members of the MAPK family share a number of common structural and regulatory features. MAPKs exhibit more than 40% amino acid identity across the kinase domain, but also have unique characteristics. All MAPKs are integrated in signaling modules called MAPK pathways. Several molecular intermediaries compose MAPK pathways: a MAPK, a MAPK activator (MAPK kinase, MKK, MAP2K or MEK), and a MKK activator (MKK kinase (MKKK, MAP3K or MEKK)) that become activated sequentially in response to vast amount of intra- and extracellular stimuli (Fig. 1). The different MAPK pathways are involved in the transduction of extracellular changes into coordinated and integrated adaptive intracellular responses. They, therefore, control many cellular functions including gene transcription, protein biosynthesis, proliferation, cellular differentiation, metabolism rate, apoptosis, migration or cellular interaction among others. MAPKs are activated by dual phosphorylation of Thr and Tyr residues, in a conserved Thr-X-Tyr motif in the activation loop. This phosphorylation is catalyzed by the dual-specificity kinases MEKs. MAPK phosphatases reverse this phosphorylation and return the MAPK to their inactive state. Further MAPK regulation is achieved by scaffold proteins, which are usually specific for each of the MAPK pathways, and collect specific MEKK–MEK–MAPK core into organized signaling modules. Scaffold proteins are spatial regulators of MAPK signals, and depending on the subcellular localization from which the activating signals arise, defined scaffolds determine which substrates are phosphorylated (Fig. 2). After activation MAPKs phosphorylate specific Ser and Thr (followed by Pro) residues of target substrates, which include many transcription factors, and other kinases and proteins. MEKs are highly selective in phosphorylating specific MAPKs, and they are in turn activated upon phosphorylation of Ser/Thr residues in a conserved motif catalyzed by MEKKs. MEKKs are the first component activated in the MAPK signaling module (Fig. 1) and have distinct motifs in their sequences that confer selectivity to their activation in response to different stimuli; however, their regulation by a range of multiple upstream components is still poorly understood. MAPKs have been extensively studied and, in mammals, subdivided mainly into four subfamilies, also known as “conventional MAPK” subfamilies: extracellular signal-regulated protein kinase 1/2 (ERK1/2), c-Jun N-terminal kinases 1–3 (JNK1, JNK2 and JNK3), p38 MAPKs (p38a, p38b, p38g and p38d) and ERK5 (also known as BMK) (Fig. 1). There are other MAPKs so called “atypical MAPK” that include ERK3/ERK4, Nemo-like kinase (NLK) and ERK7, which are not activated by MEKs and their biological function remains largely unknown.

ERK1/2 Pathway ERK1 and ERK2 (ERK1/2) were the first MAPKs to be identified and are encoded by two genes MAPK3 and MAPK1, respectively. Both are ubiquitously expressed, although their relative levels varies in different tissues. ERK1/2 are activated by a wide range of growth factors and mitogens, but also by cytokines, certain stresses, transforming agents and G protein couple-receptors among others. ERK1/2 activation is implicated in many cellular processes such as cell motility, proliferation, survival or differentiation. As all conventional MAPKs, ERK1/2 require dual phosphorylation on Thr185 and Tyr187 (for ERK2), in a Thr-Glu-Tyr motif in

472

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.64980-2

Mitogen-Activated Protein Kinases (MAPK) in Cancer

MAP3K

MEKK1-3 TPL2 A-/B-/C-Raf

MEKK1-4 MLK, TAK1 ASK, TAO TPL2

MEKK1-4 MLKs ASKs

MEKK2/3

MKK1/2

MKK6/3

MKK7/4

MKK5

JNK1/2/3

ERK5

473

P P

MAP2K P

P

MAPK

ERK1/2

p38

p38

p38

p38

P

Substrate

Transcription factors

MSK1/2 MNK1/2

MK2 & 3

Transcription factors

hDlg, PSD95 1- Syntrophin PTPH1, MyoD HSF1, Tau

PKD1, Tau eEF2K, p62

Transcription Transcription factors factors

Fig. 1 Schematic representation of conventional MAPK pathways in mammals. ASK, apoptosis signal-regulating kinase; MLK, mixed lineage kinase; TAK1, transforming growth factor beta-activated kinase-1; TAO, thousand and one amino acid kinase. See text for more details.

Stimuli

Scaffold protein

MEKK MEK

MAPK

Response

Fig. 2 MAPKs scaffold proteins. The figure shows some of the most studied scaffold proteins in the different MAPK pathways. Scaffold proteins bring together two or more proteins (from specific MAPK signaling pathways) in a relatively stable configuration and facilitate their activation in response to different stimuli to trigger the cellular response. KSR, kinase suppressor of Ras; IQGAP, IQ-motif-containing GTPase-activating protein; JIP, JNK interacting partner; POSH, plenty of SH3 domain; OSM, osmosensing scaffold for MEKK3.

474

Mitogen-Activated Protein Kinases (MAPK) in Cancer

the activation loop of subdomain VIII to be activated. More than 150 substrates for ERK1/2 have been described so far, and among them, ERK1/2 phosphorylate the transcription factors c-Myc, c-Fos, Elk1 and Ets, and also the protein kinases p90 ribosomal S6 kinase (RSK), mitogen- and stress-activated kinase 1 and 2 (MSK1/2) and the MAPK interacting kinase 1 and 2 (MNK1/2). ERK1/2 phosphorylation is catalyzed by two MEKs (MEK1 and MEK2), which are in turn activated mainly by the MAP3Ks Raf (Fig. 1), although in certain cell types and in response to specific stimuli MEK1/2 can also be phosphorylated by other MAP3K such as TPL2 (tumor progression locus 2; also known as cancer Osaka thyroid, COT). There are three Raf family members: ARaf, BRaf and CRaf, also called Raf-1, which is the most studied. Raf-1 is activated by a combination of cellular localization, phosphorylation and binding to Ras. Scaffold proteins also play a crucial role as ERK1/2 regulators, maintaining the pathway integrity and efficiency. At least 15 proteins have been identified as scaffolds for ERK1/2 pathway, which fine-tune signal amplitude and duration, and provide signal fidelity by isolating these complexes from external interferences. Scaffold proteins of the ERK1/2 pathway include kinase suppressor of Ras (KRS), MEK partner 1 (MP1), b-arrestins and IQGAP (Fig. 2). ERK1/2 pathway regulates processes that are central for cellular transformation; for example, they are implicated in cell-cycle progression by controlling the expression of cyclin D1 and cyclin-dependent kinase (CDK) inhibitors p21 and p27. The intensity and the duration of ERK1/2 signaling determine whether cell undergoes differentiation, proliferation or cell-cycle arrest. In addition, ERK1/2 signaling controls cell survival and apoptosis by regulating both the expression and the activity of proapoptotic proteins. ERK1/2 signaling is upregulated in cancer cells and in human tumors, which has led to the development of inhibitors of this pathway for cancer therapy. There is a broad spectrum of human tumors known to display aberrant ERK1/2 pathway activation caused by activating mutations in Ras (KRas, NRas and HRas) and Raf (BRaf). Among these tumors are: breast, colon, pancreatic, ovarian, prostate, non-small-cell lung and papillary thyroid cancer, as well as melanoma, glioma, acute myeloid leukemia or acute lymphocytic leukemia. Therefore, there is a need for exploring the potential of ERK1/2-pathway inhibitors as anticancer agents. Ras, Raf and MEK are better therapeutic targets than ERK1/2, since they are proteins that posses unique structural features. Very potent small-molecule inhibitors have been generated for Ras, Raf and MEK, and have been tested in several clinical trials, being so far Raf and MEK inhibitors more effective than Ras inhibitors. The treatment of cutaneous melanoma is one example in which MAPK pathway inhibitors have been used. The development of MEK inhibitors PD908059 and U0126 demonstrate the role of MEKERK1/2 pathway in human melanoma cell proliferation, survival and invasion. Approximately 50% and 28% of melanoma patients display mutations in BRaf (BRafV600E is the most predominant) and NRas, respectively. The Braf inhibitors vemurafenib (PLX4032) and dabrafenib (GSK2118436) have been used in the treatment of BRafV600E mutant melanoma patients, showing very good clinical responses in phase I to III trials (overall response of 80%). However, the response to the inhibitors is transient in most patients due to the development of drug-resistance caused by various mechanisms, most of them involving re-wiring of the MAPK pathway. In addition, approximately 30% of the patients treated with the BRaf inhibitors developed Ras-driven secondary cancers such as squamous cell carcinomas, colon cancer or leukemia, probably due to CRaf activation. This has prompted to the development of combination therapies using both BRaf and MEK inhibitors. The combination of Raf and MEK inhibitors in melanoma therapies reduces the development of secondary cancers and prolongs the response, but unfortunately patients still develop resistance. Nonetheless, therapies with BRaf and MEK inhibitors, such as trametinib/dabrafenib and cobimetinib/vemurafenib are currently accepted for the treatment of BRaf-mutant advanced melanomas. Additionally, alternative strategies are being tested nowadays, and drugs targeting ERK, RTK or BRaf and CRaf have been described.

JNK Pathway There are three JNKs encoded by different genes JNK1 (MAPK8), JNK2 (MAPK9) and JNK3 (MAPK10). Additionally, the JNKs are alternatively spliced giving rise to at least 10 different JNK isoforms. JNK1 and JNK2 are expressed ubiquitously, whereas JNK3 is mainly expressed in the nervous system. JNKs are activated by dual phosphorylation of Thr183 and Tyr185 residues in the Thr-ProTyr motif of the activation loop. This phosphorylation is catalyzed by the MAP2Ks, MKK4 and MKK7. The main targets of JNK are members of the activator protein 1 (AP1) group of transcription factors, such the transcription factors c-Jun, JunB, JunD and related proteins as Fos. JNK phosphorylates the N-terminal part of these molecules causing their activation and therefore promoting gene transcription. Other JNK targets are proteins involved in the control of the cell cycle, survival, cell death, migration, invasion and gene transcription. As ERK1/2, JNKs are also activated by mitogens, although they are preferentially and strongly activated by environmental stresses, toxins, translational inhibitor (cycloheximide and anisomycin), pathogen-associated molecular patterns (PAMPs) and damage-associated molecular patterns (DAMPs), and by proinflammatory cytokines. As all MAP2Ks, MKK4 and MKK7 are regulated by Ser/Thr phosphorylation catalyzed by a variety of MEKK depending on the stimulus and on the cell type (Fig. 1). The specificity of JNK activation is achieved in part by scaffold protein such as the JNK interacting protein (JIP) family or the plenty of SH3 domains (POSH) family (Fig. 2). It is well established that JNKs are involved in cell proliferation, survival, cell death and differentiation. However, their specific effect and degree of implication in those processes depends on the JNK family member, the cellular context, the stimuli, or on the strength and duration of the JNK activation. Therefore, to define the implication of the JNKs in cancer development is complex. It is well documented that JNK pathways are implicated in both pro- and antioncogenic processes. The oncogenic function of the JNKs is mainly based on their ability to activate AP1 by phosphorylating and stabilizing c-Jun, and their role as negative regulators of the tumor suppressor p53. On the other hand, tumor-suppressive functions are probably related to their proapoptotic activity. The role of JNK in cancer development is being investigated using mouse models. Studies using JNK1-null mice and JNK2-null mice have

Mitogen-Activated Protein Kinases (MAPK) in Cancer

475

further demonstrated that JNK can cause negative and positive effects in tumor development. For example, the deletion of JNK1 and JNK2 has opposite effects in the development of chemical-induced skin cancer. JNK2 deletion increases skin cancer, whereas the lack of JNK1 decreases the formation of skin tumors. Moreover, chemical-induced hepatocellular carcinoma (HCC) and Bcr-Abl-induced lymphoma are suppressed by JNK1 deletion, but not by the lack of JNK2. In contrast, studies of glioblastoma, prostate cancer and lung carcinoma cell lines have identified important roles for JNK2, but not for JNK1 in oncogenic processes. These data confirm that JNK pathways play a role in cancer development, and that the relative roles of each JNK isoform depend on type of cancer. One clear example of the implication of the JNK pathway in cancer development is the HCC, which is the third most common cause of death from cancer in the world. Studies using knockout mice have demonstrated that JNK1 deletion, but not JNK2 deletion, decreases the development of chemical-induced HCC with the compound N-nitrosodiethylamine (DEN). In the liver, DEN causes hepatocyte death followed by compensatory regeneration, and maintained administration of DEN leads to the development of HCC. Compared to wild-type mice, JNK1-null mice exhibit decreased liver cell proliferation and tumor formation after DEN treatment, which is caused by a reduction in the expression of c-Myc and an increase in the levels of the cyclin-dependent kinase (CDK) inhibitor p21. These effects have also been found when JNK1 is depleted in human HCC cell lines. Thus JNK1 is central for HCC tumorigenesis. Pharmacological inhibition of JNK reduced HCC development in both DEN-induced liver cancer in mice and in xenografted human HCC cells, indicating that JNK1 is a promising therapeutic target for the HCC treatment. The contribution of different JNKs to skin cancer development has also been described using the DMBA (7,12-dimethylbenz(a) anthracene)/TPA (12-O-tetradecanoylphorbol-13-acetate) model. In this model, JNK1 acts as a tumor suppressor as JNK1-null mice are more susceptible to skin tumor development than wild-type mice, whereas in JNK2-null mice the development of skin tumors is suppress, indicating an oncogenic role for JNK2 in this particular cancer model. The molecular mechanisms by which JNKs regulate skin cancer development are still under study. Nonetheless, it has been suggested that the altered AP1 transcription factor activity, together with the differential regulation of essential signaling pathways such as ERK1/2 or Akt, could account for the specific functions of JNK1 and JNK2 in skin cancer development. Akt pathway is a signal transduction pathway regulated by phosphoinositide 3kinase and the mTOR (mammalian target of rapamycin) that promotes survival and growth in response to extracellular stimuli. JNKs have also been suggested to be targets for antimelanoma therapy as they regulate the survival of melanoma cells. Cutaneous melanoma is an extremely aggressive tumor with a high tendency to relapse and high metastasis rate, its represent only 5% of skin cancers, but it causes approximately the 75% of skin cancer deaths. In melanoma JNKs are frequently active; however their role is not clear as they can act as either oncogenes or tumor suppressors, depending on the upstream or the downstream JNK pathway components presents in a given cell. JNK inhibition causes cell-cycle arrest or apoptosis depending on the cell line. For example, in the human melanoma WM983B cells JNK inhibition causes apoptosis mediated by p53, Bad (Bcl-2-associated death promoter) and Bax (Bcl-2-associated X protein), whereas in 1205Lu cells causes the arrest of the cell cycle mediated by p21. Additionally, somatic mutations in the JNK pathway (JNK1, JNK2, JNK3 and their upstream activators MKK4 and MKK7) have been identified in large-scale sequencing analyses of protein kinases in human tumors suggesting their involvement in cancer development. These mutations could cause either the loss of activity or the activation of the kinase; thus, the loss of JNK3 function promotes tumor formation, and mutations in the MAPK10 gene were found in
53% of human brain tumor examined. Also, sustained activation of JNK1 is associated with altered histone H3 methylation in human HCC, and several members of the JNK pathway are upregulated in prostate cancer. The role of JNKs in cell proliferation and apoptosis is being used for the development of therapeutic inhibitors. Several peptide inhibitors have been generated, such as JNKI1, which has been successfully used in mouse models for HCC; or BI-78D3, which inhibits JNK activity by interfering with the binding to JIP1 scaffold. The application of JNK inhibitors has been proven to work in human cells and in vivo in animal models; however, JNK pathway has not been translated into clinical use.

p38 MAPK Pathways There are four p38MAPKs encoded by different genes: p38a (MAPK14), p38b (MAPK11), p38g (MAPK12) and p38d (MAPK13). Alternative spliced variants of p38a has been identified: CSBP1 (cytokine suppressive antiinflammatory drug-binding protein), Mxi2 (Max-interacting protein) or Exip (for exon skip), although their physiological functions have been poorly studied. Most of the literature on p38MAPK refers to p38a. This was the first p38MAPK identified, and therefore is the best characterized. Based on sequence homology, substrate specificities, and sensitivity to chemical inhibitors, the p38MAPKs can be further divided into two subgroup: p38a/p38b and p38g/p38d. p38g and p38d are also known as alternative p38MAPKs. Although p38MAPKs are widely expressed, their expression pattern varies in tissues. For instance, p38a is ubiquitously expressed in all cell types and tissues; p38b is highly expressed in the brain, thymus and spleen; p38g is very abundant in skeletal muscle; and p38d levels are high in pancreas, intestine, adrenal gland, kidney and heart. The p38MAPKs are strongly activated by a wide variety of environmental and cellular stresses and by inflammatory cytokines, but they are poorly activated by growth factors. p38MAPKs and JNKs are activated by the same stimuli, and together are also called stress-activated protein kinase pathways. All p38MAPKs are activated by dual phosphorylation on Thr180 and Tyr182 of the TheGly-Tyr activation motif mediated by the MAP2Ks MKK3 and MKK6, and in the case of p38a also by MKK4. p38a may also be activated by autophosphorylation in cardiomyocytes in ischemic heart, and in T lymphocytes stimulated through the T-cell antigen receptor (TCR). The activation of MKK3 and MKK6 is by phosphorylation of Ser/Thr residues mediated by different MEKKs, which are stimulus and cell type specific (Fig. 1).

476

Mitogen-Activated Protein Kinases (MAPK) in Cancer

The use of kinase inhibitors that specifically inhibit p38a and p38b, but do not block p38g and p38d activity, and the genetic deletion of specific p38MAPK isoforms has showed that they have some overlapping substrates and functional redundancy. However, there are particular proteins that are better substrates for p38a/p38b than for p38g/p38d. More than 100 proteins can be phosphorylated by p38a/p38b and a significant proportion of them is involved in the regulation of gene expression such as the transcription factors ATF2 (activating transcription factor 2), MEF2 (myocyte enhancer factor 2) or ELK1 (Ets like gene 1). Other p38a/p38b substrates are protein kinases such as MAPK-activated protein kinase 2 (MK2) or MK3; MSK1/2 or MNK1/2 (Fig. 1), which control numerous cellular processes. p38d phosphorylates proteins such as the eukaryotic elongation factor 2 kinase (eEF2K), the protein kinase D1 (PKD1) or the signal adaptor p62; and p38g the transcription factor MyoD. p38g associates with PDZ-domain containing proteins, such as a1-syntrophin, SAP (synapse-associated protein) 90/PSD (postsynapse density) 95, hDlg (human disk large also known as SAP97) and the protein tyrosine phosphatase PTPH1 and under stress conditions it is able to phosphorylate them and modulate their activity. p38g is unique among the MAPKs and contains a short sequence at its C-terminal that associates with PDZ-domains, whose function is facilitate protein–protein interaction (Fig. 3). Also, p38g and p38d are the principal p38 MAPKs phosphorylating the neuronal microtubule-associated protein Tau and the heat shock factor 1 (HSF-1). p38 MAPK pathway are involved in processes that promote cell transformation and tumor development. They control the cell cycle, cell growth, apoptosis, angiogenesis, tissue invasion and metastasis. p38 MAPK pathway regulates the checkpoint controls and cell cycle at G0, G1/S and G2/M transitions. Depending on the cell type, p38a can either induce progression or inhibition at G1/S transition by differential regulation of specific cyclin levels (cyclin A or D1) as well as by phosphorylation of the retinoma protein (which is a hallmark of G1/S progression), and by phosphorylation of the p53 tumor suppressor and the upregulation of p16 tumor suppressor. On the other hand, activation of p38 MAPK in mammalian cells in response to various environmental stresses initiates G2/M checkpoint, which induces cell-cycle arrest and facilitate DNA repair. DNA damage checkpoints function as surveillance

Fig. 3 Schematic representation of the structures of MAPKs. All MAPKs contain a kinase domain, and N- and C-terminal regions of different lengths. Additionally, some MAPK have specific domains/regions/sequences: NLS is a nuclear localization sequence; PR1 and PR2 are Pro rich regions; in p38g there is a PDZ binding sequence; in ERK3 and ERK4 are C-terminal conserved region; AQHr is an Ala, Glu and His rich domain.

Mitogen-Activated Protein Kinases (MAPK) in Cancer

477

mechanisms during cell division to ensure that each step is completed properly, thus maintaining genetic integrity. Most of the work published on cell-cycle regulation by p38 MAPK pathway has been focused on studying the role of the isoforms p38a and p38b. In the case of p38g, one report indicates its implication in G2 arrest after ionizing radiation. Although p38a is normally associated with negative regulation of proliferation, there are reports indicating that this kinase can also positively control cellular proliferation in several cancer cell lines. These opposite effects could be due to the strength of the p38 MAPK activation and on the crosstalk with different signaling pathways. p38a is also implicated in the induction of apoptosis by modulating survival pathways and pro- and antiapoptotic proteins; whereas p38b seems to have antiapoptotic functions. Of particular interest is the key role of p38a in preventing tumorigenesis by promoting growth arrest and apoptosis in response to reactive oxygen species (ROS) in immortalized cells. In contrast, p38a can also facilitate the survival and proliferation of tumor cells supporting tumor growth and metastasis; it regulates angiogenesis and cell invasion by controlling the production, secretion and signaling of angiogenic factors such as VEGF (vascular endothelial growth factor) or MMPs (matrix metalloproteinases). Increased levels of phosphorylated p38a have been correlated with malignancy and poor prognosis in several types of cancers, such as breast carcinoma, glioma, head and neck squamous cell carcinoma, as well as in follicular, lung and thyroid carcinoma. In addition, p38a activation helps tumor cells to survive chemotherapeutic treatments, for example, the tamoxifen resistance in breast cancer patients. The use of: p38a-deficient mice, p38a-inhibitors (such as SB203580, LY2228820 or PH797804), and the deletion of p38a-regulators (such as MKK6/MKK3 or the phosphatase Wip1), have provided evidence of the tumor suppressor role of p38a in mammary epithelial cells, lung, liver and colon. Studies using mice deficient in Wip1, which is a phosphatase that targets p38a, show reduced breast tumorigenesis upon expression of the oncogenes Erbb2 or HRas; this correlates with higher p38 MAPK phosphorylation. Contrary, pharmacological inhibition of p38a using the compounds LY2228820 and PH797804 impairs tumor growth in xenografts of human breast cancer cell lines and in mice with breast tumors induced by polyoma middle T (PyMT) expression. p38a-deficient mice are more sensitized to KrasG12V-induced lung tumorigenesis, and the specific deletion of p38a in hepatocytes facilitates the DEN-induced HCC by activating the JNK pathway and enhancing the accumulation of ROS. Not only p38a but also other isoforms could mediate tumorigenesis. Studies in cancer patient samples, human cancer cell lines, cells from knockout mice, and in mouse cancer models indicate that p38g and p38d have both pro- and antioncogenic roles in tumor development. Thus, the tumor suppressor role of p38g and p38d has been shown in mouse embryonic fibroblasts, in which the deficiency in these kinases increases cell migration and MMP-2 secretion, and impairs cell contact inhibition. Additionally, the loss of p38g in KRas-transformed fibroblasts leads to increased cell proliferation and tumorigenesis, both in vitro and in vivo. In contrast, it has been described that p38d is overexpressed in cholangiocarcinoma, invasive human cutaneous squamous cell carcinoma (SCC), head and neck SCC, and in prostate cancer, whereas p38g is overexpressed in triple-negative breast cancer cells, supporting the potential pro-oncogenic role of these kinases. This was confirmed in the two-step 7,12dimethylbenz(a)anthracene (DMBA)/TPA chemical skin carcinogenesis model, in which p38g and p38d deletion blocks skin tumor development, probably by enabling formation of a proinflammatory microenvironment that fosters epidermal hyperproliferation and tumorigenesis. TPA-induced epidermal hyperproliferation and inflammation is reduced in p38g/p38d-deficient mice. Inflammation has long been associated with cancer development. During tumorigenesis, innate and adaptive immune cells infiltrate the tumor microenvironment and regulate tumor cell fate either directly or via the production of inflammatory molecules. p38a, p38g and p38d control multiple functions of the immune cells and also the production of cytokines and chemokines, which may either promote or suppress tumor growth. Colitis-associated cancer (CAC) is a clear example of tumor development associated to inflammation. CAC is a colon cancer subtype associated with inflammatory bowel disease (IBD) such as ulcerative colitis or Crohn’s disease. IBD confer an increased risk for colorectal cancer, one of the most common fatal malignancies worldwide. Patients with long-standing ulcerative colitis are particularly prone to CAC, the major cause of death in these patients. It has been shown that p38a controls colon homeostasis and that the loss of p38a in intestinal epithelial cells (IEC) increases proliferation, reduces the number of mucus-producing goblet cells and affects epithelial barrier function by altering tight junction assembly. The intestinal epithelia serve as a barrier that protects the intestinal tract against luminal invading pathogens and ingested toxin, which can promote inflammatory responses. p38a in IEC also controls the expression of chemokines that are essential for the recruitment of immune cells, which protect against mucosal infection and dextran sodium sulfate (DSS)-induced colitis. As a result, mice that do not express p38a in IEC are more susceptible to colitis-associated colon tumorigenesis in an azoxymethane (AOM)/DSS colitis-associated colon cancer model. In contrast, the downregulation of p38a in colon tumor cells reduces tumor number in mice. Pharmacological inhibition of p38a using the compounds PH797804 or SB202190 also reduces colon tumor growth either in AOM/DSS model, in xenografts of human colon cancer cell lines or in mice that express APCmin. These studies suggest a dual role for p38a in IEC, supporting colon tumor progression and suppressing the initiation of inflammation-associated colon tumor. p38g and p38d also regulate inflammatory signaling that promote colon tumorigenesis in an AOM/DSS mouse model of CAC. Mice that do not express p38g and p38d are less susceptible to colitis-associated colon tumorigenesis compared to wild-type mice. These mice show a decreased number of colon tumors, which correlates with reduced inflammation, cytokine and chemokine production and inflammatory cell infiltration in the colon. Experiments using bone marrow transplants demonstrated that p38g and p38d in hematopoietic cells are essential for CAC development. Other study nonetheless shows that p38g in IEC is also important for CAC development, as colon tumor formation is reduced in mice with IEC-specific p38g deletion. These studies highlight the importance of p38g and p38d in cancer, although their specific functions in different cells are yet to be determined.

478

Mitogen-Activated Protein Kinases (MAPK) in Cancer

ERK5 Pathway Extracellular signal-regulated kinase (ERK) 5, also known as Big MAPK (BMK), is twice the size of other MAPKs and is encoded by the MAPK7 gene. Unlike the classical MAPKs, ERK5 has two distinct functional domains, a catalytic N-terminal domain, which is very similar to that of ERK1/2 and is responsible for conventional protein kinase activity; and a unique C-terminal half with no homology to any other protein, and responsible for ERK5 transcriptional transactivation activity. C-terminal domain contains a transcriptional transactivation domain, two proline-rich regions and a nuclear localization sequence (NLS) (Fig. 3). Interaction between the C- and N-terminal domains has an important role in regulating ERK5 localization and activity. ERK5 is activated selectively by dual phosphorylation on Thr218 and Tyr220 in the TEY motif by its only upstream kinase, MEK5 (Fig. 1). ERK5 is activated in cells by oxidative and osmotic stress, but also by several growth factors and cytokines. Some downstream targets of ERK5 have been identified, including the myocyte enhancer factor-2 (MEF2) transcription factor family members MEF2A, MEF2C and MEF2D, and other transcription factors such as SAP1 (switch-activating protein 1), c-Myc and cAMP response element-binding (CREB). ERK5 plays a central role in cellular survival, differentiation and proliferation, and also in cell-cycle progression. Additionally, in the last decade the ERK5 signaling has been proposed to play a central role in the regulation of tumor angiogenesis, the development and aggressiveness of several types of cancer and in the function of a wide variety of oncogenes. Thus, the ERK5 pathway has been associated with advanced-stages of prostate adenocarcinomas in patients. Prostate cancer (PCa) is the second-leading diagnosed cancer in men in the world. ERK5 and MEK5 are over expressed in prostate cancer cells, which correlate with aggressiveness and more metastasis. For example, in the prostate cancer cell line PC-3 MEK5/ERK5 expression increases the proliferation of these cells by inducing DNA replication, whereas the silencing of ERK5 inhibits PC-3 proliferation and invasion. In addition, ERK5 has been suggested as a therapeutic target in patients with FOXF1-positive PCa. FOXF1 (Forkhead box F1) is a transcription factor that normal prostate tissue does not express, and it has been proposed that FOXF1 regulates ERK5 phosphorylation by increasing the expression of ERK5 upstream activators. Using murine orthotopic models of PCa, it has been shown that overexpression of FOXF1 in the Myc-CaP and TRAMP PCa cell lines induces tumor and metastasis progression, which is associated with increased ERK5 phosphorylation. Also, the silencing of ERK5 in Myc-CaP and TRAMP cells decreased proliferation and metastasis. On the other hand, the implication of ERK5 pathway in metastasis is also supported by its involvement in the regulation of key proteins and pathways, such as the MMP or the regulation of integrins/FAK signaling, that are implicated in invasion and metastasis. Deletion of ERK5 in PTEN-null mice increases T-cell infiltration in PCa tumors compared to single PTEN-deficient mice. The combined loss of PTEN and ERK5 reduce PCa tumor size and cell proliferation, and increased the expression of T-cell chemoattractants CCL5 and CXCL10, and the infiltration of CD4 þ T cells in the tumors, which indicate that ERK5 can be a good target for PCa immunotherapy. PTEN (from phosphatase and tensin homolog) is a phosphatase that regulates intracellular levels of phosphatidylinositol3,4,5-trisphosphate in cells. Additional in vivo (using mouse models) and clinical data indicate that abnormal ERK5 signaling is also implicated in many others type of cancers such as colorectal cancer, esophageal cancer, osteosarcoma, pancreatic, neuroblastoma, bladder, lung, HCC, leukemia or in triple-negative breast cancer where ERK5 is considered a potential therapeutic target. ERK5 silencing and ERK5 kinase inhibitors, such as the compounds TG02 or XMD8-92, block cell proliferation of different cancer cell lines and tumor growth in xenografts models. For example, it has been shown that the treatment with XMD8-92 blocks the growth of skin tumor, lung carcinoma, pancreatic ductal adenocarcinoma, HCC or neuroblastoma in xenograft models; whereas in triple-negative breast cancer, the administration of XMD8-92 increase the sensitization of cancer cells to chemotherapy agents, such as doxorubicin, cisplatin, etoposide or bortezomib among others, potentiating their antitumor activity in vitro and in vivo.

Atypical MAPK In addition to the “conventional MAPKs” (ERK1/2, JNK, p38MAPK and ERK5), there are “atypical MAPKs,” which are not activated by MEKs. Atypical MAPKs include ERK3/ERK4, Nemo-like kinase (NLK) and ERK7 (also known as ERK8).

ERK3/ERK4 ERK3 and ERK4 are MAPKs encoded by the genes MAPK6 and MAPK4, respectively. ERK3 is ubiquitously expressed, although its expression is higher in brain, skeletal muscle and the gastrointestinal tract; whereas ERK4 expression is more restricted to colon, heart, kidney, eye, lung, placenta, prostate, skin, ovary, pancreas and brain (where its expression is higher). ERK3/ERK4 have very similar structure and their activation loops contain a single phospho-acceptor residue in the Ser-Glu-Gly motif that lacks the phospho-acceptor residue Tyr. Therefore, ERK3/ERK4 are very poor MEKs substrates. The identity of ERK3/ERK4 upstream activator kinase is unclear, although there are some reports identifying the group 1 of p21 activated kinases (PAKs) as ERK3/ERK4 activation-loop kinases. Ser189 residue in ERK3 and Ser186 in ERK4 are constitutively phosphorylated in vivo in growing cells, and so far no stimuli have been found to promote ERK3/ERK4 phosphorylation or activation in vivo. The only known ERK3/ERK4 substrate is the MAPK-activated protein kinase MK5.

Mitogen-Activated Protein Kinases (MAPK) in Cancer

479

ERK3 is a very unstable protein, with a half-life of 30–45 min depending on the cellular context, contrary to ERK4, which is highly stable. The biological role of ERK4 is largely unknown; however, several studies show that ERK3 participates in a number of biological processes. It has been suggested that ERK3 is a negative regulator cell proliferation and a positive regulator of cell differentiation. Is possible that ERK3 is implicated in oncogenic processes since its expression is induces in Raji lymphoma cells or in squamous cell carcinoma after stimuli that inhibit proliferation. Additionally, ERK3 phosphorylation and protein stability is regulated by CDK1 and the phosphatase Cdc14, indicating that ERK3 has a role in cell-cycle progression through mitosis. Also, ERK3 overexpression in various cell lines was found to inhibit the progression through S phase. Unfortunately, mice lacking ERK3 are not viable, and conditional knockout mouse models will be required to perform studies of the implication of this kinase (and of ERK4) in tumor development.

Nemo-Like Kinase (NLK) The NLK kinase was originally identified as a vertebrate ortholog of Drosophila nemo, its catalytic domain shows a 45% amino acid identity with ERK2. NLK is ubiquitously expressed, but its expression is higher in brain and lymphoid organs. NLK lacks the phospho-acceptor residue Tyr in the Thr-Gln-Glu motif in the activation loop, and possesses unique N- and C-terminal extensions, that may contribute to the interaction with its substrates (Fig. 3). The exact MEKs that directly phosphorylate NLK are still unknown, but a potential kinase that could lead to NLK activation is the kinase homeodomain-interacting protein kinase 2 (HIPK2). Also, NLK could autophosphorylate itself in vitro. In cells, NLK can be activated by stimuli of the Wnt signaling pathway, and also by interleukin-6 (IL-6), granulocyte colony-stimulating factor (G-CSF) and transforming growth factor b (TGF-b). Transcription factors of the T-cell factor/lymphoid enhancer factor (TCF/LEF) family and STAT3 are NLK substrates. Increasing evidences show that NLK can either act as an oncogene or as a tumor suppressor depending on the cell context. NLK has been described to be important for tumor cell proliferation, migration and apoptosis. Downregulation of NLK decreases the proliferation of gallbladder cancer cells and of HCC cells, by inducing G1 arrest. Contrary, NLK induction inhibits cell proliferation and increases apoptosis in human colon carcinoma cells, prostate and breast cancer cell, and in glioblastoma cells. In addition, altered expression of NLK has been found in different types of cancers. For example, the expression of NLK is increased in cervical squamous cell carcinoma, HCC or gallbladder; and decreased in prostate, ovarian, lung, breast cancer or glioma. Also, it has been suggested that NLK expression is associated with the risk of invasive epithelial ovarian cancer. Further studies are needed for the elucidation of molecular mechanisms mediated by NLK in tumor development.

ERK7 ERK7 (encoded by the gene MAPK15) was cloned from a rat brain cDNA library, and some years later the human kinase was identified and called ERK8. Surprisingly, ERK7 and ERK8 only show approximately 69% amino acid identity, which is lower than the typical identity found between other human and rodents MAPK orthologs. From now, I will refer to the human kinase as human ERK7, which is predominantly expressed in lung and kidney. ERK7 activation loop contains the Thr-Glu-Tyr motif, being Thr and Tyr residues constitutively phosphorylated, which indicates that a MEK is involved in its activation. However, when inactive ERK7 mutant is expressed in cells, this is not phosphorylated in the Thr-Glu-Tyr motif suggesting that ERK7 phosphorylation is regulated by autophosphorylation. Human ERK7 can also be regulated by stimuli such as serum, amino acid starvation or hydrogen peroxide. So far no physiological substrates have been identified for ERK7. Although the implication of ERK7 in cancer development has not been described, it is known that this kinase plays a role in cell proliferation, autophagy and in the response to glucocorticoids and estrogens.

Conclusions Members of the MAPK family function in a cell context- and cell type-specific manner to integrate signal that affect numerous processes that are essential in tumorigenesis. The involvement of MAPK family members in cancer development has supported the potentially beneficial role of MAPK inhibitors with current cancer therapy. In addition, given the redundancy in MAPK signaling pathways and the adaptive capacity of cancer cells, drug combinations are increasingly being investigated for therapy. For instance, the incorporation of several Raf/MEK inhibitors in combination therapies are currently being tested in clinical trials. The toxic effect that MAPK inhibitors could have in normal proliferating cells is a concern, so the development of small-molecule inhibitors with sufficient potency and selectivity to inhibit MAPK signaling pathway specifically in cancer cells is a real challenge. Studies in mouse models have been essential to better understand how MAPKs control cancer development. The development of new MAPK mouse models, structural analysis and mutational analysis by next generation genomic sequencing are expected to provide new strategies for the design of improved therapeutic approaches.

See also: Mutational Signatures and the Etiology of Human Cancers.

480

Mitogen-Activated Protein Kinases (MAPK) in Cancer

Further Reading Cargnello, M., Roux, P.P., 2011. Activation and function of the MAPKs and their substrates, the MAPK-activated protein kinases. Microbiology and Molecular Biology Reviews 75, 50–83. Casar, B., Crespo, P., 2016. ERK signals: Scaffolding scaffolds? Frontiers in Cell and Development Biology 4, 49. Chen, F., Castranova, V., 2009. Beyond apoptosis of JNK1 in liver cancer. Cell Cycle 8, 1145–1147. Chen, Z., Gibson, T.B., Robinson, F., Silvestro, L., Pearson, G., Xu, B., Wright, A., Vanderbilt, C., Cobb, M.H., 2001. MAP kinases. Chemical Reviews 101, 2449–2476. Coulombe, P., Meloche, S., 2007. Atypical mitogen-activated protein kinases: Structure, regulation and functions. Biochimica et Biophysica Acta 1773, 1376–1387. Cuenda, A., Rousseau, S., 2007. p38 MAP-kinases pathway regulation, function and role in human diseases. Biochimica et Biophysica Acta 1773, 1358–1375. Cuenda, A., Sanz-Ezquerro, J.J., 2017. p38gamma and p38delta: From spectators to key physiological players. Trends in Biochemical Sciences 42, 431–442. Escos, A., Risco, A., Alsina-Beauchamp, D., Cuenda, A., 2016. p38gamma and p38delta mitogen activated protein kinases (MAPKs), new stars in the MAPK galaxy. Frontiers in Cell and Development Biology 4, 31. Garcia-Cano, J., Roche, O., Cimas, F.J., Pascual-Serra, R., Ortega-Muelas, M., Fernandez-Aroca, D.M., Sanchez-Prieto, R., 2016. p38MAPK and chemotherapy: We always need to hear both sides of the story. Frontiers in Cell and Development Biology 4, 69. Gomez, N., Erazo, T., Lizcano, J.M., 2016. ERK5 and cell proliferation: Nuclear localization is what matters. Frontiers in Cell and Development Biology 4, 105. Huang, Y., Yang, Y., He, Y., Li, J., 2015. The emerging role of Nemo-like kinase (NLK) in the regulation of cancers. Tumour Biology 36, 9147–9152. Igea, A., Nebreda, A.R., 2015. The stress kinase p38alpha as a target for cancer therapy. Cancer Research 75, 3997–4002. Imperial, R., Toor, O.M., Hussain, A., Subramanian, J., Masood, A., 2017. Comprehensive pancancer genomic analysis reveals (RTK)-RAS-RAF-MEK as a key dysregulated pathway in cancer: Its clinical implications. Seminars in Cancer Biology. https://doi.org/10.1016/j.semcancer.2017.11.016. Inesta-Vaquera, F.A., Campbell, D.G., Tournier, C., Gomez, N., Lizcano, J.M., Cuenda, A., 2010. Alternative ERK5 regulation by phosphorylation during the cell cycle. Cellular Signalling 22, 1829–1837. Kidger, A.M., Keyse, S.M., 2016. The regulation of oncogenic Ras/ERK signalling by dual-specificity mitogen activated protein kinase phosphatases (MKPs). Seminars in Cell & Developmental Biology 50, 125–132. Kyriakis, J.M., Avruch, J., 2012. Mammalian MAPK signal transduction pathways activated by stress and inflammation: A 10-year update. Physiological Reviews 92, 689–737. Li, S., Jang, H., Zhang, J., Nussinov, R., 2018. Raf-1 cysteine-rich domain increases the affinity of K-Ras/Raf at the membrane, promoting MAPK signaling. Structure 26 (513– 525), e512. Loveridge, C.J., Mui, E.J., Patel, R., Tan, E.H., Ahmad, I., Welsh, M., Galbraith, J., Hedley, A., Nixon, C., Blyth, K., et al., 2017. Increased T-cell infiltration elicited by Erk5 deletion in a Pten-deficient mouse model of prostate carcinogenesis. Cancer Research 77, 3158–3168. Manning, G., Whyte, D.B., Martinez, R., Hunter, T., Sudarsanam, S., 2002. The protein kinase complement of the human genome. Science 298, 1912–1934. McKay, M.M., Morrison, D.K., 2007. Integrating signals from RTKs to ERK/MAPK. Oncogene 26, 3113–3121. Pearson, G., Robinson, F., Beers Gibson, T., Xu, B.E., Karandikar, M., Berman, K., Cobb, M.H., 2001. Mitogen-activated protein (MAP) kinase pathways: Regulation and physiological functions. Endocrine Reviews 22, 153–183. Perander, M., Keyse, S.M., Seternes, O.M., 2016. New insights into the activation, interaction partners and possible functions of MK5/PRAK. Frontiers in Bioscience (Landmark Edition) 21, 374–384. Raman, M., Chen, W., Cobb, M.H., 2007. Differential regulation and properties of MAPKs. Oncogene 26, 3100–3112. Risco, A., Cuenda, A., 2012. New insights into the p38gamma and p38delta MAPK pathways. Journal of Signal Transduction 2012, 520289. Sebolt-Leopold, J.S., English, J.M., 2006. Mechanisms of drug inhibition of signalling molecules. Nature 441, 457–462. Sebolt-Leopold, J.S., Herrera, R., 2004. Targeting the mitogen-activated protein kinase cascade to treat cancer. Nature Reviews. Cancer 4, 937–947. Simoes, A.E., Rodrigues, C.M., Borralho, P.M., 2016. The MEK5/ERK5 signalling pathway in cancer: A promising novel therapeutic target. Drug Discovery Today 21, 1654–1663. Uzdensky, A.B., Demyanenko, S.V., Bibov, M.Y., 2013. Signal transduction in human cutaneous melanoma and target drugs. Current Cancer Drug Targets 13, 843–866. Wagner, E.F., Nebreda, A.R., 2009. Signal integration by JNK and p38 MAPK pathways in cancer development. Nature Reviews. Cancer 9, 537–549. Wellbrock, C., Arozarena, I., 2016. The complexity of the ERK/MAP-kinase pathway and the treatment of melanoma skin cancer. Frontiers in Cell and Development Biology 4, 33.

Mod Squad: Altered Histone Modifications in Cancer Matthew Murtha, Bellvitge Biomedical Biomedical Research Institute (IDIBELL), Barcelona, Catalonia, Spain Manel Esteller, Bellvitge Biomedical Biomedical Research Institute (IDIBELL), Barcelona, Catalonia, Spain; University of Barcelona, Barcelona, Catalonia, Spain; and Catalan Institution for Research and Advanced Studies (ICREA), Barcelona, Catalonia, Spain © 2019 Elsevier Inc. All rights reserved.

Introduction It is now understood that “An epigenetic trait is a stably heritable phenotype resulting from changes in a chromosome without alterations in the DNA sequence” or in other words epigenetics is the heritable molecular determinants of a trait that are independent of DNA sequence. The principle mechanisms by which this is achieved are through DNA methylation, covalent histone tail modifications, and noncoding RNAs. These epigenetic modifications work in concert with the genetic sequence to determine cellular physiology and identity in human cells. It is now well understood that perturbations to these marks can lead to pathologies such as cancer. Each class of epigenetic mark, for example, DNA methylation, represents large fields of research. Here we will focus solely on the role of covalent modifications of histone tails to the appearance and progression of cancers. Excellent reviews for DNA methylation and noncoding RNAs can be found elsewhere. The DNA molecules that comprise the genome are packaged within a cell by winding about octomers of histone proteins to form nucleosomes, the building block of chromatin. The histone proteins H2A, H2B, H3, and H4 are small positively charged proteins with exposed N-terminal tails that dimerize to form the core of the nucleosome particle. Approximately 147 bp of superhelicial DNA wraps around each octomer to form the core particle. The exposed N-terminal tails are the substrate for posttranslational modifications and are the basis for the “histone code,” the molecular logic behind chromatin-based inheritance. The principle histone modifications are acetylation, methylation, phosphorylation, ubiquitination, and SUMOylation and lead to chromatin compaction and transcriptional expression or repression. Notable examples are listed in Table 1. Canonical nomenclature for the modification is to list the histone protein number, then the modified amino acid and its position in the protein, and finally the type of modification. For example, H3K27me3 represents trimethylation of a lysine at the 27th residue in histone H3. Histone modifications are synthesized by an array of enzymes such as histone deacetylases and methyl transferases. These enzymes incorporate or remove the modifications or “marks” and are the target of a number of oncogenic mutations. While there are a large number of characterized chromatin marks, and advances in proteomics promises the discovery of more to come, a few classical marks such as H3K9me3 are well studied. While the functions of the best-studied marks are remarkably context dependent certain patterns have emerged, such as H3K9me3 being associated to chromatin compaction and transcriptional repression, and canonical marks are summarized in Fig. 1. Most tumors harbor both genetic and epigenetic defects with complex interactions between the two. Indeed, efforts to sequence the human cancer genome found some of the most frequent mutations were in genes encoding the subunits of protein complexes that read and write histone modifications like ATP-dependent remodeling complexes SWI/SNF and BAF and suggest that more than 20% of human cancers bear one of these remodeling mutations. In response there have been a number of international epigenome mapping projects such as the NIH’s Roadmap Epigenomics Mapping Consortium, International Human Epigenome Consortium, The Cancer Genome Atlas Network, BLUEPRINT, and the International Cancer Genome Consortium. The result of these consortiums has been an ability to subclassify tumors based on their molecular etiology rather than clinical features. This ability allows for the advancement of precision medicine and patient specific treatments and will be discussed further at the end of the article. Cancer is principally unrestrained cell division and is typically thought to be caused by dysregulation of gene expression. While transcription factors such as p53 or MYC are essential to tumor formation, they represent difficult drug targets as protein:protein and protein:DNA interactions are difficult to target with any specificity. Epigenetic proteins represent much better therapeutic targets given than many are enzymes. It should be noted that epigenome targeting drugs such as HDACis (histone deacetylase inhibitors) have been FDA approved and in use for years. Here we will describe how disruptions in histone modifications caused by HDACs, BETs, and methyltransferases lead to cancer. Table 1

List of important histone modifications and their impact on transcription

Modification

Position

Transcriptional effect

Associated enzymes

Methylation

H3K4 H3K9 H3K37 H3K36 H3K79 H3R2, H3R17, H3R26 H3R8 H2AR3 H4K20 H3K9, H3K14, H3K18, H3K23, H3K27 H4K5, H4K8, H4K12, H4K16

Activation Repression Repression Activation Activation Activation Repression Activation Repression Activation Activation

MLL1-5, SETD1 SUV39H1, G9A, SETDB1 EZH1,2 SETD2, ASH1L DOT1L PRMT4 PRMT5 PRMT1 PRMT5 SETD8, SUV4 KAT2A, KAT2B, p300, CBP KAT5, KAT7, p300, CBP

Acetylation

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65061-4

481

482

Mod Squad: Altered Histone Modifications in Cancer

Fig. 1 Illustration of the nucleosome with selected important histone modifications. Schematic of the nucleosome with four pairs of histone proteins and important N-tail histone modifications. Methylation events are represented with a red circle, acetylation by a green triangle.

Altered Histones in Cancer A hallmark study of altered histone modifications in cancer examined posttranslational modifications to histone H4 in a comprehensive panel of normal tissues, cancer cell lines and primary tumors. The study found a loss of monoacetylated and trimethylated forms of histone H4 characterized cancer cells. These changes appeared early and accumulated during the tumorigenic process, and were also observed in a mouse model of skin carcinogenesis. The losses occurred predominantly at H4K16Ac and H4K20me3 and were associated with the hypomethylation of DNA repetitive sequences, a well-known characteristic of cancer cells. This study was the first to suggest that the global loss of monoacetylation and trimethylation of histone H4 is a common hallmark of human tumor cells and alerted the field to the importance of altered histone modifications in cancer.

Chromatin Remodelers as Tumor Suppressors Chromatin-remodeling complexes are large multisubunit complexes that mediate the compaction and expansion of chromatin in the nucleus while allowing for precise gene expression, DNA repair, and replication. The mammalian BAF (Brg/Brahma associated

Mod Squad: Altered Histone Modifications in Cancer

483

factors) or SWI/SNF complex can be imagined as a single unit comprised of 15 subunits encoded by 29 genes. A mutation to one of these genes is found in at least 20% of cancers. The genes were first identified and characterized in yeast studies where they were discovered in screens attempting to identify signal transduction pathway members. Their role in chromatin compaction was confirmed by phenotypic rescue with secondary mutations in histone proteins, demonstrating that they remodeled chromatin rather than participated in signal transduction. In human cells they appear to primarily act to oppose the effects of the PRC2 complex, which is best known for its role in depositing the H3K27me3 mark. There are many lines of evidence linking mutations to the BAF complex to cancer. Initial observations of malignant cell lines found that these lines often had deletions or mutations to critical subunits of the complex, suggesting a role as a tumor suppressor. Further study revealed subunits were able to bind RB protein and repress E2F function. Their role as tumor suppressor was confirmed through a study of malignant rhabdoid tumor (MRT), a rare cancer that occurs in children. Versteege et al. found that MRT tumors feature uniform loss of BAF47 (i.e., INI1, hSNF5, SMARCB1), typically from a germline loss of one allele followed by the loss of a second allele in the tumor tissue. This indicated that BAF47 mutations act as classic tumor suppressors, resembling the genetics of retinoblastoma. Further studies found cancer cell lines that featured inactivation of both Brg and Brm subunits, a finding that was replicated in several breast cancer cell lines and led to the common view that tumor suppression likely required at least one functional allele of either subunit. The mechanisms by which mutations to chromatin remodelers caused cancer remained mysterious until the advent of genomewide mapping of chromatin marks. The advent of genome sequencing allowed the sensitive detection of mutations, however this revealed that loss of both alleles of particular subunits were not necessary. For example, initial reports of ovarian cancers revealed that while 57% of ovarian clear cell carcinoma and 46% of endometriosis-associated ovarian carcinoma contained inactivating mutations to BAF250A (ARID1A), nearly all of the cases the mutations were in only a single allele. This suggested that at least in the case of ovarian cancers, BAF250A operated as a dominant tumor suppressor subunit, a finding that was confirmed by repeated studies. Further confounding mechanistic insight was the observation that BAF250 is not necessary for chromatin remodeling, at least not with respect to remodeling events that can be observed from in vitro nucleosomal template experiments. In addition, many subunits do not appear to be involved in tumor suppression as they are not observed to be mutated at particularly high rates in any known sequenced tumor. So how can the BAF complex be both extremely sensitive to the loss of a given allele while at the same time having neither its function nor other subunits altered? Standford biologist Gerald Crabtree observed that this places three constraints on the mechanism for BAF tumor suppression. Firstly, it is dosage-sensitive, as evidenced by the observed genetic dominance. Secondly, the fundamental mechanism cannot be detected in an in vitro chromatin-remodeling assay. Lastly, the tumor-suppression function manifests itself in a highly tissue specific manner. A clear example of this is seen within small cell cancers: 90% of small cell ovarian cancers feature mutations to the ATPase BRG (SMARCA4) while less than 5% of small cell lung cancers carry mutations to the gene. So how do mutations in BAF complex subunits mediate oncogenesis? BAF complexes target repressive Polycomb targets and work to oppose transcriptional repression. Mutations apparently cause mis-targeting and improper gene regulation. For example, mutation of BAF subunit ATPase Brg1 (Smarca4) in embryonic stem cells causes accumulation of H3K27me3 and the subsequent repression of genes underlying the mark. A similar mechanism, the inability for the BAF complex to properly regulate the placement and function of the PRC (Polycomb repressive complex) is at work in cancer. For example, tissue specific deletion of BAF47 in mice leads to rapid rise of T cell lymphoma. In these mutant cells, BAF deletion causes improper removal of Polycomb from the Ink4a (Cdkn2a) locus, which results in an aberrant accumulation of H3K27me3 and repression of Ink4a. Thus, altered modification of H3K27 causes the repression of a potent tumor suppressor and proliferation regulator. A similar mechanism is thought to be at play in MRTs albeit with different tissue specificity. That MRTs have the lowest mutational burden observed in any tumor suggests the tumors are induced into oncogenesis solely through changes to histone marks and epigenetic regulation. Another beautiful illustration of altered histone modifications driving cancer progression is seen in human synovial sarcoma, a soft-tissue tumor featuring a precise translocation of the BAF SS18 (C90) subunit. This translocation fuses 78 amino acids of the C terminus of SSX to the C terminus of the SS18 subunit to create a well-studied oncogenic fusion protein termed SS18-SSX. This fusion protein outcompetes wild type SS18 in BAF complexes and results in the mistargeting of BAF47 and SS18-SSX complexes to loci containing oncogenic genes like SOX2 and PAX6. This retargeting results in the loss of PRC complexes at these loci and the subsequent loss of H3K27me3, thus allowing the expression of the oncogenes. Not all BAF driven tumors operate through these mechanisms and even in the case of MRT and synovial tumors where BAF47 is mistargeted, the molecular mechanism is very distinct. These examples however clearly show the importance of correctly modified histones and the ability of epigenetic changes to induce cancer.

Interpreting Histone Acetylation ε-N-acetylation of lysine residues in the N-terminal tails of histone proteins is one of the predominant and best-studied histone modifications. Acetylation is regulated by histone acetyltransferases (HATs) which act as “writers” adding acetyl groups to the histone tail, while histone deacetylases (HDACs) act as “erasers” and remove the mark. There are many examples of these enzymes suffering inactivating mutations that drive cancer. For example, mutations in CBP (KAT3A) and p300 (KAT3B) are found in B cell lymphomas, monoallelic loss of KAT5 in observed in human lymphomas, breast and head and neck cancers and homozygous

484

Mod Squad: Altered Histone Modifications in Cancer

deletion of KAT6B in small cell lung cancer. Additionally, HDAC deregulation results in the silencing of tumor-suppressor genes or overexpression of oncogenes. For example, HDAC1, HDAC3 and HDAC6 are overexpressed in tumors. These studies have provided the basis for the development of HAT11 and HDAC inhibitors, some of which had already proved successful in clinical oncology and will be discussed further below. Bromodomains (BRDs) are “readers” of acetyl marks in histone tails, targeting chromatin-modifying enzymes and other protein machinery to specific sites in the chromatin, thus regulating gene transcription. Bromodomains are a family of evolutionarily conserved motifs identified for in the brahma gene of Drosophila melanogaster. BRDs bind the acetylated lysines in histone tails allowing recruitment of other chromatin factors and transcriptional machinery. A total of 61 bromodomains were found in 46 different proteins of the human proteome. The bromodomain and extra-terminal (BET) family has been thoroughly investigated and holds great potential as a chemotherapeutic target in cancer therapy. The BET family is comprised of ubiquitously expressed BRD2, BRD3, BRD4, and BRDT, and BRDT, which is only expressed in testis. BET proteins are characterized by their two bromodomains in tandem (BD1 and BD2) in the N-terminal and a C-terminal extra-terminal (ET) domain. It was elucidated some years ago that BRD4 and BRD2 play an important role in transcription elongation by recruiting the positive transcription elongation factor complex (P-TEFb) to acetylated chromatin. P-TEFb is composed of cyclin-dependent kinase-9 (CDK9) and its activator, cyclin T and mediates phosphorylation of the C-terminal repeat domain (CTD) of RNA polymerase II (RNAP II) to ensure efficient transcription. BRD4 recruits P-TEFb to acetylated chromatin through its BRD, allowing activation of the CDK9 kinase subunit, phosphorylation of RNAPII and efficient transcript elongation. BRD4 also controls the release of active P-TEFb from its inactive complex with HEXIM1 protein and 7S snRNA. Thus, correct histone acetylation is essential for the proper recruitment of P-TEFb by BRD4, which is then critical for proper transcriptional initiation and elongation of genes controlling cell proliferation. BET proteins and their interactions with acetylated histone tails are now strongly implicated in cancer, not the least because BETs have been show to directly regulate the expression of oncogenes such as c-MYC. Indeed, inhibition of BET protein binding at the MYC locus with BET inhibitors leads to a reduction in cell proliferation. The BET family can function as cell cycle regulators as well by controlling the expression of genes required for M to early G1 phase transition. BRD2 recruits the key transcriptional cell cycle-regulatory genes E2F1 and E2F2 and acts as a scaffold on which they assemble. BRD4 regulates gene transcription globally, so its inhibition would be expected to cause downregulation of activity throughout the genome. Surprisingly, the impact of BRD4 inhibition is modest and somewhat specific as only a few hundred genes are downregulated. However, most of these genes are very important in tumorigenesis. The molecular basis for this surprising discrepancy is explained by the observation that BRD4 has a strong preference for binding enhancers and super-enhancers in key genes of hematological and solid tumors, such as MYC. Brd4 acts as a general regulator that couples the acetylation state of chromatin with Pol II elongation. Midline carcinoma is generated when BRD4 is mutated by chromosomal translocation to form in-frame fusions with the NUT gene. The resulting Brd4-NUT oncoprotein is an aberrant transcriptional regulator.

Targeting BET Bromodomains With Small-Molecules for Therapy Mediating the interaction between histone modifications and transcription with enzymatic activity make BET a desirable target for small molecule therapies. Indeed, the field is proceeding rapidly as in 2010 two independent groups reported the antitumor efficacy of small-molecule inhibitors of BET bromodomains, called JQ1 and I-BET, which have a similar chemical structure and mode of target inhibition. JQ1 and I-BET are notable for their high affinity for bromodomains of the BET family (BD1 and BD2 of Brd2/ Brd3/Brd4/Brdt) over other bromodomain subfamilies. BET inhibitors also have suitable pharmacokinetics for in vivo application, which has enabled a rapid evaluation of their potential therapeutic activity in various disease models. BETs can become oncogenes upon translocational mutation. For example, BRD4 can be mutated by chromosomal translocation to form in-frame fusions with the NUT gene to initiate an aggressive cancer called midline carcinoma. The resulting Brd4-NUT oncoprotein is an aberrant transcriptional regulator that relies on the bromodomains of Brd4 for its oncogenic function. As such, this malignancy provides a clear initial test case for the evaluation of therapeutic benefit of BET inhibitors. In midline carcinoma cell lines, JQ1 treatment led to the release of Brd4-NUT from chromatin and triggered terminal squamous cell differentiation and apoptosis. In patient-derived xenograft models of midline carcinoma, it was shown that daily exposure to JQ1 extended the survival of cancer-bearing mice at doses that had minimal toxicity to normal tissues. This pivotal study provided the first demonstration of efficacy for a BET inhibitor in a preclinical cancer model.

HDACi Demonstrate Utility of Altered Histone Modifications to Fight Cancer HDAC inhibitors (HDACi) are a class of small molecules that interfere deacetylation and were first approved for the treatment of cutaneous T-cell lymphoma in 2006. HDACi’s now have shown efficacy, either alone or in combination, in multiple myeloma, acute myelogenous leukemia and myelodysplastic syndrome. Treating solid tumors with HDACi has been challenging, although some efficacy was recently reported in non-small cell lung cancer when combined with a second epigenetic therapy.

Mod Squad: Altered Histone Modifications in Cancer

485

Mapping the Histone Modifications of the Epigenome The advent of whole genome and high-throughput sequencing technologies has enabled fast, accurate genome-wide maps of histone modifications at rapidly diminishing costs. This has led to international epigenome mapping projects such as the NIH’s Roadmap Epigenomics Mapping Consortium, International Human Epigenome Consortium, The Cancer Genome Atlas Network, BLUEPRINT, and the International Cancer Genome Consortium. These projects aim to map the epigenomes of healthy and cancer cells and have led to a surplus of data and new insights. Most strikingly, these epigenomic maps can be used to subclassify tumor samples allowing for precision and personalized medicine. There are myriad examples to choose from. H3K4 acetylation, H3K9 acetylation and H3K27 methylation can be used to identify breast tumor molecular subtypes. Indeed, most every major tumor type has now been profiled and study epigenetically, see for example malignant melanoma.

Targeting Altered Histone Modifications for Therapy Drugs targeting the epigenome are now being extensively developed given their promise for halting tumor progression and preventing chemoresistance. They are considered broadly to act either in a broad pan-genome manner, termed genome medicine, or are developed to treat specific tumor subsets for precision or personalized medicine. HDACi were initially discovered based on drug screens for differentiation inducers in leukemias. There are now several prominent HDACi used in the clinic: for example, vorinostat (Zolinza; Merck & Co.), belinostat (Beleodaq; Spectrum Pharmaceuticals) and romidepsin (Istodax; Celgene) have all been approved for the treatment of cutaneous or peripheral T cell lymphomas, and panobinostat (Farydak; Novartis) was recently approved for the treatment of drug-resistant multiple myeloma when used in combination with the proteasome inhibitor bortezomib (Velcade; Millennium Pharmaceuticals). Notable examples of drugs targeting histone modification pathways are listed in Table 2. HDACi efficacy has been examined in other malignancies have shown limited single agent activity. iBETs, which reversibly bind to the bromodomains of BET proteins, are another prominent class of genome medicines that have shown promise. Most iBETs target BRD4, which is translocated in some cancers and is key in correct interpretation of histone acetylation (see above). There are several notable examples of targeted therapies showing efficacy. Specific genetic defects in epigenetic pathways can be targeted using small molecules. One such example is apparent given mutation in the H3K27 histone N-methyltransferase EZH2 which is activated by mutations in lymphomas. Researchers found the application of a selective EZH2 inhibitor induced the specific death of cells harboring the EZH2 mutation in cell lines. Synthetic lethality approaches can also be used to achieve subtype specificity for targeted therapy and precision medicine. For instance, inhibitors of H3K79 N-methyltransferase DOT1L have shown effective in vitro study of leukemias with activation of MLL (mixed-lineage leukemia; also known as histone-lysine N-methyltransferase 2A (KMT2A)), whereas a drug targeting lysine-specific histone demethylase 1 (LSD1; also known as KDM1A) is active in vitro only in malignancies with specific DNA methylation patterns. Preclinical studies have shown that these targeted therapies have particular efficacies in specific patient subsets, which is to be expected from targeted therapies and is different from broad genomic medicine approaches. Indeed, in the field of DNA methylation, epigenetics can aid in assigning the correct treatment regimen to Cancers of Unknown Origen suggesting that accurate epigenome profiling of histone modifications may also aid in developing precision medicine.

Table 2

Selection of drugs targeting regulators of histone modifications currently FDA approved or in advanced clinical trials

Inhibitor class

Mechanism

Drug

Target

Cancer

HDACi

Inhibition of histone deacetylation Inhibition of histone deacetylation Inhibition of histone deacetylation Inhibition of histone deacetylation Inhibition of histone deacetylation Inhibition of histone deacetylation Inhibition of histone deacetylation Inhibition of histone deacetylation Inhibition of histone deacetylation Inhibition of BET binding to acetylated histones Inhibition of BET binding to acetylated histones Inhibition of BET binding to acetylated histones Inhibition of BET binding to acetylated histones Inhibition of H3K27 methylation Inhibition of H3K27 methylation

Belinostat Panobinostat Romidepsin VOrinostat Entinostat Givinostat Mocetinostat Resminostat Ricolinostat OTX015 INCB054329 BMS-986158 GSK525762 CPI-1205 Tazemetostat

Class I & II HDACs Class I, II, & IV Class I Class I, II, & IV Class I Class I & II Class I HDAC1,3,6 HDAC6 Pan-BET Pan-BET Pan-BET Pan-BET EZH2 EZH2

Peripheral T-cell lymphoma Multiple myeloma Cutaneous T-cell lymphoma Cutaneous T-cell lymphoma Breast cancer Hematological malignancies Hematological malignancies Hepatocellular carcinoma Solid tumors and hematological malignancies Hematological malignancies Leukemias and solid tumors Solid tumors Solid tumors and hematological malignancies Lymphoma Lymphomas & sarcomas

iBET

EZH2 inhibitor

486

Mod Squad: Altered Histone Modifications in Cancer

Concluding Remarks As our understanding of the interactions between histone modifications and cell identity become clearer so too does our appreciation for the ways that altered histone modifications can lead to cancer. It stands to reason that deeper study of these covalent modifications will shed light on to the basic biology of how epigenetics can govern cellular functions in addition to the discovery of therapeutically relevant targets for chemotherapy development.

Acknowledgments The authors would like to acknowledge graphic artist Vanessa Korbackova who helped create Fig. 1.

See also: Environmental Exposures and Epigenetic Perturbations. Mutations in Chromatin Remodeling Factors. Mutations in Histone Lysine Methyltransferases and Demethylases.

Further Reading Bae, J.-B., 2013. Perspectives of international human epigenome consortium. Genomics & Informatics 11, 7–14. Berger, S.L., Kouzarides, T., Shiekhattar, R., Shilatifard, A., 2009. An operational definition of epigenetics. Genes & Development 23, 781–783. Bernstein, B.E., Stamatoyannopoulos, J.A., Costello, J.F., Ren, B., Milosavljevic, A., Meissner, A., Kellis, M., Marra, M.A., Beaudet, A.L., Ecker, J.R., et al., 2010. The NIH roadmap epigenomics mapping consortium. Nature Biotechnology 28, 1045–1048. Bolden, J.E., Peart, M.J., Johnstone, R.W., 2006. Anticancer activities of histone deacetylase inhibitors. Nature Reviews. Drug Discovery 5, 769–784. Cancer Genome Atlas Research Network, Weinstein, J.N., Collisson, E.A., Mills, G.B., Shaw, K.R.M., Ozenberger, B.A., Ellrott, K., Shmulevich, I., Sander, C., Stuart, J.M., 2013b. The Cancer Genome Atlas Pan-Cancer Analysis project. Nature Genetics 45, 1113–1120. Chadwick, L.H., 2012. The NIH roadmap epigenomics program data resource. Epigenomics 4, 317–324. Costa, A.L., Lobo, J., Jerónimo, C., Henrique, R., 2017. The epigenetics of testicular germ cell tumors: Looking for novel disease biomarkers. Epigenomics 9, 155–169. Delmore, J.E., Issa, G.C., Lemieux, M.E., Rahl, P.B., Shi, J., Jacobs, H.M., Kastritis, E., Gilpatrick, T., Paranal, R.M., Qi, J., et al., 2011. BET bromodomain inhibition as a therapeutic strategy to target c-Myc. Cell 146, 904–917. Esteller, M., 2011a. Epigenetic changes in cancer. Biology of Reproduction 3, 9. Esteller, M., 2011b. Non-coding RNAs in human disease. Nature Reviews. Genetics 12, 861–874. Esteller, M., 2017. Epigenetic drugs: More than meets the eye. Epigenetics 12, 307. Fernández, J.M., de la Torre, V., Richardson, D., Royo, R., Puiggròs, M., Moncunill, V., Fragkogianni, S., Clarke, L., Consortium, B.L.U.E.P.R.I.N.T., Flicek, P., et al., 2016. The BLUEPRINT data analysis portal. Cell Systems 3, 491-495.e495. Filippakopoulos, P., Qi, J., Picaud, S., Shen, Y., Smith, W.B., Fedorov, O., Morse, E.M., Keates, T., Hickman, T.T., Felletar, I., et al., 2010. Selective inhibition of BET bromodomains. Nature 468, 1067–1073. Fraga, M.F., Ballestar, E., Villar-Garea, A., Boix-Chornet, M., Espada, J., Schotta, G., Bonaldi, T., Haydon, C., Ropero, S., Petrie, K., et al., 2005. Loss of acetylation at Lys16 and trimethylation at Lys20 of histone H4 is a common hallmark of human cancer. Nature Genetics 37, 391–400. Jones, P.A., Issa, J.-P.J., Baylin, S., 2016. Targeting the cancer epigenome for therapy. Nature Reviews. Genetics 17, 630–641. Kadoch, C., Crabtree, G.R., 2015. Mammalian SWI/SNF chromatin remodeling complexes and cancer: Mechanistic insights gained from human genomics. Science Advances 1, e1500447. Kulis, M., Esteller, M., 2010. DNA methylation and cancer. Advances in Genetics 70, 27–56. Mohammad, H.P., Smitheman, K.N., Kamat, C.D., Soong, D., Federowicz, K.E., Van Aller, G.S., Schneck, J.L., Carson, J.D., Liu, Y., Butticello, M., et al., 2015. A DNA hypomethylation signature predicts antitumor activity of LSD1 inhibitors in SCLC. Cancer Cell 28, 57–69. Moran, S., Martínez-Cardús, A., Sayols, S., Musulen, E., Balañá, C., Estival-Gonzalez, A., Moutinho, C., Heyn, H., Diaz-Lagares, A., de Moura, M.C., et al., 2016. Epigenetic profiling to classify cancer of unknown primary: A multicentre, retrospective analysis. The Lancet Oncology 17, 1386–1395. Moran, S., Martínez-Cardús, A., Boussios, S., Esteller, M., 2017b. Precision medicine based on epigenomics: The paradigm of carcinoma of unknown primary. Nature Reviews. Clinical Oncology 14, 682–694. Pérez-Salvia, M., Simó-Riudalbas, L., Llinàs-Arias, P., Roa, L., Setien, F., Soler, M., de Moura, M.C., Bradner, J.E., González-Suárez, E., Moutinho, C., et al., 2017. Bromodomain inhibition shows antitumoral activity in mice and human luminal breast cancer. Oncotarget 8, 51621–51629. San-Miguel, J.F., Hungria, V.T.M., Yoon, S.-S., Beksac, M., Dimopoulos, M.A., Elghandour, A., Jedrzejczak, W.W., Günther, A., Nakorn, T.N., Siritanaratkul, N., et al., 2014. Panobinostat plus bortezomib and dexamethasone versus placebo plus bortezomib and dexamethasone in patients with relapsed or relapsed and refractory multiple myeloma: A multicentre, randomised, double-blind phase 3 trial. The Lancet Oncology 15, 1195–1206. Shi, J., Vakoc, C.R., 2014. The mechanisms behind the therapeutic activity of BET bromodomain inhibition. Molecular Cell 54, 728–736. Simó-Riudalbas, L., Esteller, M., 2014. Cancer genomics identifies disrupted epigenetic genes. Human Genetics 133, 713–725. Stunnenberg, H.G., International Human Epigenome Consortium, Hirst, M., 2016. The international human epigenome consortium: A blueprint for scientific collaboration and discovery. Cell 167, 1897. Yang, Z., Yik, J.H.N., Chen, R., He, N., Jang, M.K., Ozato, K., Zhou, Q., 2005. Recruitment of P-TEFb for stimulation of transcriptional elongation by the bromodomain protein Brd4. Molecular Cell 19, 535–545. Yardley, D.A., Ismail-Khan, R.R., Melichar, B., Lichinitser, M., Munster, P.N., Klein, P.M., Cruickshank, S., Miller, K.D., Lee, M.J., Trepel, J.B., 2013. Randomized phase II, doubleblind, placebo-controlled study of exemestane with or without entinostat in postmenopausal women with locally recurrent or metastatic estrogen receptor-positive breast cancer progressing on treatment with a nonsteroidal aromatase inhibitor. Journal of Clinical Oncology 31, 2128–2135. Zhang, Y., Lv, J., Liu, H., Zhu, J., Su, J., Wu, Q., Qi, Y., Wang, F., Li, X., 2010. HHMD: The human histone modification database. Nucleic Acids Research 38, D149–D154. Zöllner, S.K., Rössig, C., Toretsky, J.A., 2015. Synovial sarcoma is a gateway to the role of chromatin remodeling in cancer. Cancer Metastasis Reviews 34, 417–428.

Molecular Epidemiology and Cancer Riskq Paulina Gomez-Rubio and Evangelina Lo´pez de Maturana, Spanish National Cancer Research Center (CNIO), Madrid, Spain; and Biomedical Research Networking Centre in Oncology, Madrid, Spain © 2019 Elsevier Inc. All rights reserved.

Glossary Biomarker Any biological indicator that serves as an objective measure of physiological or pathogenic processes used to detect exposure to carcinogens and assess carcinogenic hazard. Causal inference The process of deciding whether existing evidence is sufficient to conclude there is a causal relationship between a perturbation and an effect. Exposome All environmental (nongenetic) exposures in a lifetime intended to explore the role of the environment in human health. Molecular epidemiology A scientific field that seeks to study the contribution of risk factors, identified through biological indicators (i.e., biomarkers), to the etiology of pathological processes. Omics The field that seeks to characterize and quantify large-scale data considered collectively from different biological fields such as genomics or proteomics. Primary risk prevention Intervention applied before evidence of disease with the aim of preventing the onset of disease. Reverse causation Observed when the outcome of interest precedes the exposure instead of the common presumption that exposure occurs before the outcome of interest. Subphenotypes A subset of the phenotype with characteristics specific to a part of the population.

Introduction Cancer continues to be a dire predicament worldwide. Established as the second leading cause of death after cardiovascular diseases in 2015; cancer represents a substantial personal and social burden with serious impact in both economy and quality of life (Fitzmaurice et al., 2017). Classic epidemiology has greatly contributed to the study of cancer risk. Illustrative of this is that most of what we initially learned from smoking and lung cancer resulted from traditional epidemiologic studies. During the early 1950’s evidence based on questionnaire information started to surface linking smoking with lung cancer. The consistent evidence mounting over the years by different studies and the replicability of results across investigations made causal inference conceivable (Samet, 2016). More recently, molecular epidemiology approaches have enabled to further explore and understand carcinogenic mechanisms of tobacco compounds through the identification of biomarkers of smoking exposure and susceptibility. The term molecular cancer epidemiology was first reported in 1982 by Perera and Weinstein who proposed combining the concepts of classic epidemiology with the use of advanced laboratory technology to identify specific environmental hazards and host factors that modify susceptibility, and to elucidate mechanisms of carcinogenesis. For this purpose, four categories of molecular markers (i.e., biomarkers) were described: internal dose, biologically effective dose, response, and susceptibility. The ultimate goal of implementing biomarkers in epidemiology is to improve the characterization of exposure providing an objective, quantifiable measure for a more precise evaluation of human cancer risk. This is particularly useful when evaluating complex exposures and weak effects which are difficult to assess through traditional epidemiology approaches (Perera and Weinstein, 1982; Perera, 1987, 2000). Molecular epidemiology aims to gather evidence that specific compounds pose carcinogenic hazard to human health through the definition of exposure-cancer associations and the identification of populations at greater risk. Conversely, factors that protect against carcinogenesis could be defined. Therefore, based on the notion that up to some extent an important proportion of cancer could be prevented, modifiable risk factors represent a great opportunity for primary cancer prevention through avoidance or minimization of key exposures or through the promotion of protective factors. Moreover, molecular epidemiology represents a unique opportunity of increasing our understanding of the mechanisms underlying carcinogenesis hence shaping our knowledge towards the biological plausibility behind carcinogenic exposure. The fast development of technology, particularly high-throughput technologies, has resulted in the rapid growth and evolution of molecular epidemiology in the past decades. More sophisticated molecular platforms have allowed to refine exposure assessment and to improve the definition of disease phenotype. Therefore, progress in the field has provided us with the valuable opportunity

q

Change History: February 2018. Paulina Gomez-Rubio and Evangelina López de Maturana are involved in preparing the update. No change history of section/ figures/tables is provided since, as previously discussed with the Editor, this article was written from stratch due to the tremendous evolution of the field in the past decade. This article is an update of Frederica P. Perera, Molecular Epidemiology and Cancer Risk, in Encyclopedia of Cancer (Second Edition), edited by Joseph R. Bertino, Academic Press, 2002, Pages 213–220.

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.96132-4

487

488

Molecular Epidemiology and Cancer Risk

of increasing our knowledge and understanding of cancer risk. However, we have also encountered several difficulties that hamper their practical application in risk assessment. In the present article we discuss key contributions of molecular epidemiology in cancer risk assessment, prediction and prevention as well as some of its most relevant challenges in the current scientific landscape. In addition, while the role of genetic factors in disease is formally studied under the umbrella of genetic epidemiology, we will briefly comment on genetic susceptibility to cancer due to important interactions with other cancer risk factors, as well as their relevance in the omics and data integration fields.

Epidemiologic Study Designs for the Evaluation of Biomarkers Molecular epidemiology builds upon classic epidemiology designs. Among these, the case-control studies are the most commonly used, entailing the retrospective recruitment of disease and disease-free subjects. While ascertainment of the diagnosis of cases before recruitment is one of the main advantages of case-control studies, these suffer from different drawbacks, one of them specifically pertaining to the difficulty of establishing reverse causation. For this purpose, prospective studies are ideal because information about exposure and biological samples are collected at baseline from a group of “healthy” subjects who are followed-up during a period of time in order to identify those who develop the disease of interest. Since exposure is assessed before disease development, prospective studies are less likely to introduce selection and information bias such as recall bias. Moreover, due to the prospective nature of this design reverse causation is precluded. Prospective studies also offer the possibility of collecting information and samples at different time points during the follow-up time. However, these studies are costly and time consuming particularly for uncommon diseases for which larger study populations and longer time periods are necessary to identify sufficient number of subjects with the disease of interest. For different reasons, many large cohort studies are unable to measure the biomarker(s) of interest in the whole study population. Under this circumstance, nested case-control studies are an efficient sampling scheme. In this design, all cases identified up to a particular point in time are included in the study while controls are composed from a random sample of subjects free of the disease of interest by the time of the case diagnosis or by the end of the follow-up time. Similarly useful is the case-cohort design, in which disease-free subjects are sampled as controls at baseline from the pool of all subjects enrolled having the same disease risk as anyone at time zero of the study. This type of design requires specific statistical considerations since some controls could also be included as cases in the study (García-Closas et al., 2006, 2011).

Contributions of Molecular Epidemiology Assessment of Exposure The use of biomarkers in molecular epidemiology The use of molecular markers to assess the precise contribution of individual factors, their interactions and those with the host result in a more objective measure of exposure, one that is exempt from the individual’s perception or recall. Moreover, biological markers measured with laboratory methods more closely reflect the level of exposure that affects the target cells or organs (i.e., biological effective dose). Risk assessment has particularly benefited from the advancement of refined laboratory techniques such as those used to measure metabolites of carcinogens and their interaction with DNA or proteins (adducts). DNA adducts are quantitative indicators of exposure used to evaluate carcinogenicity of specific compounds like those contained in tobacco. Tobacco was initially identified as a potent carcinogen in the mid-twentieth century and it is known to contain at least 70 human carcinogens including polycyclic aromatic hidrocarbons (PAHs). PAHs in tobacco smoke are metabolized and activated in the human body forming DNA adducts that have the potential to induce mutations and ultimately initiate a carcinogenic process (IARC, 2004). Consequently, DNA adducts are suitable biomarkers of carcinogenesis since they indicate a genetic damage that is relevant to such process not only reflecting exposure but also the host’s ability to metabolize these compounds (Perera and Weinstein, 1982). Levels of PAH DNA adducts have been reported to be higher in smokers than nonsmokers in different human tissues including those not directly exposed to tobacco smoke such as bladder, liver or pancreas (Phillips and Venitt, 2012). Accordingly, higher DNA adduct levels have been suggestive of increased risk of lung, bladder and breast cancer (Agudo et al., 2017; Jin et al., 2017; Munnia et al., 2017). Moreover, the role of PAHs in cancer is highlighted by reports that show that single nucleotide polymorphisms (SNPs) in key metabolism and conjugation genes are associated with PAH-DNA adducts levels (Etemadi et al., 2013). As a result of the extensive research performed in the field during the past decades, different countries worldwide have applied smoking bans that along with smoking cessation strategies have had an impact in lowering the incidence of smoking related cancers (McAfee et al., 2015).

The emergence of omics data in molecular epidemiology Cancer is a complex condition that entails many biological changes within the affected cells and the surrounding microenvironment. These include mutations and modifications of gene and protein expression through different mechanisms. Advances in high-throughput technology during the past two decades coupled with sound biostatistics and bioinformatics methodology

Molecular Epidemiology and Cancer Risk

489

have allowed exploring the collective of different small biological molecules such as genes (genomics), gene expression (transcriptomics), epigenetic modifications (epigenomics), and metabolites (metabolomics), among others. These technologies represent an invaluable tool for a more comprehensive insight of human exposure, which is particularly useful in the inquiry of molecular mechanism of carcinogenesis. During decades, researchers have been focusing in the study of single factors. Based on an a priori knowledge of the molecule’s biological function, candidate studies were pioneer in the identification of association with disease (Zhu and Zhao, 2007). However, given that the available functional information is limited elaborating a priori hypothesis may be impaired. As a result, after its emergence, high-throughput technology has been widely used as an agnostic approach in an attempt to identify risk factors for diseases such as cancer. In genomics, through the analysis of thousands of genetic variants in a single run, genome-wide association studies (GWAS) have identified common low-penetrance alleles providing evidence of polygenic susceptibility. Among the different discoveries reached through GWAS, the region 8q24.21 is perhaps one of the most intriguing, harboring multiple independent risk loci for several tumor types including breast, bladder, colorectal, ovarian, pancreatic and prostate cancer (Al Olama et al., 2014; Michailidou et al., 2015; Sud et al., 2017). DNA methylation, a form of epigenetic mechanism that does not alter DNA sequence but may affect gene expression, has also been identified as an important mechanism of carcinogenesis. As such, aberrant DNA methylation has been reported in different cancers. Methylation arrays allow the rapid assessment of global methylation patterns through the analysis of thousands of CpGs throughout the genome. Studies have reported that such epigenome-wide DNA methylation analysis could have a potential use in the identification of risk markers in cancers such as breast cancer (Severi et al., 2014). Other omics platforms have been similarly applied in an attempt to better characterize cancer leading to advances in risk assessment through a greater understanding of this disease.

The exposome and its potential use in risk assessment The exhaustive consideration of exposure in risk assessment was initially addressed by Wild in the early 2000s, who proposed the concept of exposome as a means to account for all exposures an individual encounters during their lifetime. The concept encompasses internal exposure, to which molecular and omics data contribute to; specific external exposure that includes pollutants, infectious agents, lifestyle factors or diet; and generalized external exposure such as social, economic and psychological exposures (Wild, 2005, 2012). Therefore, while molecular information can greatly contribute to a better definition of exposure it is unable to cover the totality of a subject’s exposure. The integration of epigenomics, metabolomics and microbiome (microbial load and diversity data), with exposure information collected via epidemiological questionnaires and biological markers such as vitamin D or cholesterol, is likely to provide a more precise characterization of the exposome (López de Maturana et al., 2016). The accurate assessment of the exposome would require prospective studies in which multiple repetitive measures are collected throughout the participants’ lifetime. Therefore, to date, it is still debatable to what extent the assessment of the exposome and its common use in scientific research and risk assessment is pragmatic.

Definition of Phenotype The multi-factorial etiology of cancer is highlighted by its great heterogeneity of both clinico-pathological and molecular features. Accordingly, there have been numerous reports on the wide diversity of the natural course of tumors and the way these respond to treatments. Consequently, the use of a single phenotype to describe a mixture of closely related subtypes of diseases (i.e. subphenotypes) is gradually becoming recognized as an unsuitable approach for the proper assessment of disease risk. Improving the phenotyping accuracy has been an important focus of molecular epidemiology in recent years. Innovative technologies have given place to the definition of novel molecular tumor subtypes in different cancers. In this section, we focus in inter-patient tumor heterogeneity. For information about intra-tumor heterogeneity, the reader can refer to further reading. Cancer subphenotyping was first explored in breast cancer, one of the most commonly diagnosed cancers in women worldwide and for which different treatment responses and clinical outcomes have been widely reported (Zardavas et al., 2015). Four main breast tumor subtypes were established through genome-wide expression: luminal A, luminal B, HER2 enriched and basal like, which reflect relevant biological entities regarding variation of growth rate, activation of specific signaling pathways and cellular compositions (Perou et al., 2000). Following its initial characterization, the evolution of technology allowed exploring these subtypes through a broad array of molecular platforms as mRNA expression microarrays, DNA methylation chips, SNP arrays or miRNA sequencing providing a deeper insight into each subtype like specific mutational processes or signaling pathways (TCGA, 2012). Accurate molecular subphenotyping of breast cancer has an important clinical impact, particularly in therapeutic decision making (Zardavas et al., 2015). Moreover, tumor subtypes might potentially reflect etiological differences with important implications in risk prediction. Up to date however, there has been a lack of consensus on the characterization of molecular breast cancer subtypes in epidemiological studies that, along with the limited statistical power in the less common subtypes, have hampered their use in etiological research (Ellingjord-Dale et al., 2017). Regarding risk, at present, the most consistent findings show a stronger protective effect of breastfeeding on the basal-like subtype; while, other factors such as parity have been consistently associated with decreased risk of luminal subtypes (Holm et al., 2017). The effort of utilizing molecular information and its translation to specific tumor subtypes has expanded beyond breast cancer. With support of The Cancer Genome Atlas and The International Cancer Genome Consortiums different cancers have

490

Molecular Epidemiology and Cancer Risk

been sub-phenotyped in the past years including colorectal, cervical and urothelial (Guinney et al., 2015; TCGA, 2017; Robertson et al., 2017). Along with the clinical impact of the subclassification of tumor types, the vision of a more precise definition of cancer etiologies and risk gradually becomes more tangible. However, there is still a long way ahead before establishing the practical use of this wealth of information in the public health arena.

High-Risk Populations and Primary Prevention Strategies Measure of inter-individual variability The concept of human variability, that is, diverse responses to similar exposures to carcinogens, is not new. Classic epidemiology has laid the groundwork for discerning between vulnerable groups. Illustrative is that initial work in the field was paramount to establish that intrauterine and preconceptional exposure to ionizing radiation was associated with the development of childhood cancer (Wakeford, 1995). Molecular epidemiology has also helped in the identification of interactions that explain a certain fraction of variability in risk. In addition to familial aggregation and high penetrance genetic variants, low penetrance variants can affect the way a person processes and reacts to an exposure thus affecting their vulnerability to a carcinogen. For example, arsenic, an environmentally ubiquitous element and whose metabolites are detected through mass spectrometry in urine, is classified as carcinogenic to humans by the International Agency for Research on Cancer (IARC). Since the discovery of the enzyme involved in arsenic metabolism (arsenic 3 methyl transferase, AS3MT), polymorphisms in the coding gene have been associated with arsenic methylation efficiency and consequently with cancer risk, helping in the identification of subsets of the population with different degrees of vulnerability to arsenic-associated carcinogenesis (Engström et al., 2015; Gomez-Rubio et al., 2010). Additionally, other nongenetic factors have been associated with susceptibility to arsenic toxicity such as sex and body mass index (Vahter et al., 2007; GomezRubio et al., 2011). The definition of key factors modifying cancer risk is essential for a better understanding of interindividual variability. Failure to account for this variation can result in overlooking important sensitive subgroups leading to inadequate health policies and measures.

Definition of high risk populations One of the main goals of molecular epidemiology is disease prevention through the better understanding of exposure-susceptibility interactions. The appropriate definition of high-risk populations has been pivotal for risk assessment guiding the implementation of numerous health prevention measures. For instance, smoking cessation strategies have been implemented by different healthcare systems worldwide with the final aim of reducing smoking-related health effects such as lung and oropharyngeal cancers (McAfee et al., 2015). Moreover, molecular epidemiology studies associating occupational exposures with carcinogenicity have helped in the implementation of more healthy work policies and practices (Espina et al., 2013). For instance, in Sweden, incidence of mesothelioma linked to occupational exposure to asbestos, leveled off after being one of the first countries restricting asbestos use in the 70s (Hemminki and Shehnaz, 2008). Other characteristics defining cancer high-risk populations include genetic predisposition, age, sex or ethnicity. Appropriate screening and informed decisions of lifestyle modifications or even participation in chemoprevention programs could be expected as a result of proper risk stratification. Though the definition of high-risk populations is not a substitute for general prevention efforts applicable to the whole populations (e.g., promotion of a healthy weight, or reduction of alcohol and tobacco consumption), certain interventions to high-risk individuals could be more effective, particularly when those interventions result in hazardous side effects (e.g., chemoprevention agents) and/or are too expensive to be applied in the whole population.

Improving the definition of risk The underlying interplay of multiple etiological factors in human carcinogens is a reality scientists recognize and attempt to address. Up to date, the most common type of analysis performed has independently considered each biological process and their relationship with cancer status. While we have gained important insight with this approach, a great deal of disease complexity might be hidden behind networks of interactions between several disease processes. For example, higher risk of cancer has been reported when sets of morbidities are jointly analyzed in comparison with their individual analysis (Gomez-Rubio et al., 2017). Moreover, innovative statistical approaches have been implemented to integrate more than two types of omics data (e.g., genetic variants, DNA methylation and gene expression) with the aim of pinpointing potential carcinogenic mechanisms in different malignancies such as bladder and breast cancer (Sun et al., 2011; Pineda et al., 2015). Hence, the combination of multiple sources of data including molecular markers as well as lifestyle factors could provide with a more thorough and comprehensive definition of risk by accounting for the interplay among different risk factors. In breast cancer, the integration of genetic risk factors with the Gail’s risk model based on demographic and clinical data conferred a modest improvement in classification of breast cancer risk compared to either the Gail’s risk or the genetic score alone in a population of white non-Hispanic postmenopausal women (Mealiffe et al., 2010). The clinical utility of risk models for disease prediction depends in their ability to accurately define population strata with different risk levels. Despite significant efforts devoted to developing integrative approaches for risk assessment, its transition into clinics is not currently regular practice. Therefore, risk prediction tools properly validated in prospective cohorts are urgently needed in order to adequately incorporate risk stratification in primary prevention strategies.

Molecular Epidemiology and Cancer Risk

491

Challenges of Molecular Epidemiology in the New Era In complex diseases, omics data can help to achieve a better and more accurate characterization of the exposure, genetic factors and phenotype providing a holistic view of the human health states by including both the combined exposure of all sources that reach the internal chemical environment and the response of our body to environmental influences, including metabolic processes (Rappaport and Smith, 2010; Miller and Jones, 2014). Technological progress in high-throughput biological data generation have provided with massive molecular omics data from different layers of the biological processes of the host (e.g., genomics, epigenomics, transcriptomics, proteomics or metabolomics, among others) and the nonhost, such as microbiome. Unfortunately, this substantial data production has not been accompanied with the development of next generation statistical methods to integrate in a computationally timely manner all the omics data needed to better characterize the phenotype, the exposome and the genome (Lopez de Maturana et al., 2014). The integration of this type of data is crucial and can be as costly as omics data generation (Mardis, 2010). Although some progress has been achieved in the field, there is still a need of developing sophisticated analytical strategies that can handle the most common characteristic to omics-based studies: the vast number of variables versus a limited number of individuals, the so-called large p small n problem. Apart from volume, computational approaches should consider the complexity of each omics data type jointly with potential interactions among omics variants of the same or different type, correlated variants and possible (non)linear relationships (Green and Guyer, 2011; Tini et al., 2017). Moreover, these analytical strategies should also consider the integration of omics data with epidemiological data from questionnaires. Integration of both types of data represents a new challenge requiring of both sufficient understanding of the underlying biological concepts and of analytical algorithms to build hierarchical models and theories of disease causation. Two main analytical strategies have been applied for data integration: the multistage and the meta-dimensional analyses (Ritchie et al., 2015). While meta-dimensional analysis combines multiple types of data for simultaneous analysis producing complex models, multistage (or multistep) approach first looks for the relationship between the data types and then, between the data types and the trait of interest. At present, multistage approach is adopted in most of the integrative studies; however, its main drawback is that due to its step-wise nature, this approach is not able to capture interactions. Therefore, despite the current efforts devoted to keeping up with the high biostatistics and bioinformatics demands of the existing data, there are still many issues that need to be addressed in order to take full advantage of this information.

Summary and Prospective Vision Molecular epidemiology has experienced an important evolution during the past 15 years. Improved molecular technologies have given place to the opportunity to refine our accuracy of cancer risk estimation contributing in the implementation of proper prevention strategies. However, despite great efforts devoted to cancer risk assessment, and some important advances, the full potential of cancer prevention is yet to be attained. Practical application of cancer risk assessment is further hampered by the complexity of newly available high-dimensional data. As technology evolves and massive amounts of biological information are being generated, new methodology for data integration and analysis is demanded. Along with better methods, cancer consortia will prove to be fundamental for testing and validation of more complex risk assessment models.

Acknowledgments We would like to thank Núria Malats and Esther Molina-Montes for their kind suggestions to this manuscript.

See also: Aspirin and Cancer. Enhancers in Cancer: Genetic and Epigenetic Deregulation. Environmental Exposures and Epigenetic Perturbations. Epigenetic Therapy.

References Agudo, A., Peluso, M., Munnia, A., et al., 2017. Aromatic DNA adducts and breast cancer risk: A case-cohort study within the EPIC-Spain. Carcinogeneisis 38 (7), 691–698. Al Olama, A.A., Kote-Jarai, Z., Berndt, S.I., et al., 2014. A meta-analysis of 87,040 individuals identifies 23 new susceptibility loci for prostate cancer. Nature Genetics 46 (19), 1103–1109. Ellingjord-Dale, M., Vos, L., Tretli, S., et al., 2017. Parity, hormones and breast cancer subtypes- results from a large nested case-control study in a national screening program. Breast Cancer Research 19 (1), 10. Engström, K.S., Vahter, M., Fletcher, T., et al., 2015. Genetic variation in arsenic (þ 3 oxidation state) methyltransferase (AS3MT), arsenic metabolism and risk of basal cell carcinoma in a European population. Environmental and Molecular Mutagenesis 56 (1), 60–69. Espina, C., Porta, M., Schüz, J., et al., 2013. Environmental and occupational interventions for primary prevention of cancer: A cross-sectorial policy framework. Environmental Health Perspectives 121 (4), 420–426. Etemadi, A., Islami, F., Phillips, D.H., et al., 2013. Variation in PAH-related DNA adduct levels among non-smokers: The role of multiple genetic polymorphisms and nucleotide excision repair phenotype. International Journal of Cancer 132 (12), 2738–2747.

492

Molecular Epidemiology and Cancer Risk

Fitzmaurice, C., Allen, C., Barber, R.M., et al., 2017. Global, regional, and national cancer incidence, mortality, years of life lost, years lived with disability, and disability-adjusted lifeyears for 32 cancer groups, 1990 to 2015: A systematic analysis for the global burden of disease study. JAMA Oncology 3 (4), 524–548. García-Closas, M., Varmeulen, R., Sherman, M.E., et al., 2006. Application of biomarkers in cancer epidemiology. In: Schottenfeld, D., Fraumeni Jr., J.F. (Eds.), Cancer epidemiology and prevention, 3rd edn. Oxford University Press, New York (NY), pp. 70–80. García-Closas, M., Vermeulen, R., Cox, D., et al., 2011. Population-based study designs in molecular epidemiology. IARC Scientific Publications 163, 241–259. Gomez-Rubio, P., Meza-Montenegro, M.M., Cantu-Soto, E., Klimecki, W.T., 2010. Genetic association between intronic variants in AS3MT and arsenic methylation efficiency is focused on a large linkage disequilibrium cluster in chromosome 10. Journal of Applied Toxicology 30 (3), 260–270. Gomez-Rubio, P., Roberge, J., Arendell, L., et al., 2011. Association between body mass index and arsenic methylation efficiency in adult women from soythwest U.S. and Northwest Mexico. Toxicology and Applied Pharmacology 252 (2), 176–182. Gomez-Rubio, P., Rosato, V., Márquez, M., et al., 2017. A systems approach identifies time-dependent associations of multimorbidities with pancreatic cancer risk. Annals of Oncology 28 (7), 1618–1624. Green, E.D., Guyer, M.S., 2011. Charting a course for genomic medicine from base pairs to bedside. Nature 470, 204–213. Guinney, J., Dienstmann, R., Wang, X., et al., 2015. The consensus molecular subtypes of colorectal cancer. Nature Medicine 21 (11), 1350–1356. Hemminki, K., Shehnaz, H., 2008. Mesothelioma incidence has leveled off in Sweden. International Journal of Cancer 122, 1200–1201. Holm, J., Eriksson, L., Ploner, A., et al., 2017. Assessment of breast cancer risk factors reveals subtype heterogeneity. Cancer Research 77 (13), 3708–3717. IARC working group on the evaluation of carcinogenic risk to humans, 2004. Tobacco smoke and involuntary smoking. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans 83, 1–1438. Jin, F., Thaiparambil, J., Donepudi, S.R., et al., 2017. Tobacco-specific carcinogens induce hypermethylation, DNA adducts, and DNA damage in bladder cancer. Cancer Prevention Research 10 (10), 588–597. Lopez de Maturana, E., Ibanez-Escriche, N., Gonzalez-Recio, O., et al., 2014. Next generation modeling in GWAS: Comparing different genetic architectures. Human Genetics 133 (10), 1235–1253. López de Maturana, E., Pineda, S., Brand, A., Van Steen, K., Malats, N., 2016. Toward the integration of Omics data in epidemiological studies: Still a “long and winding road”. Genetic Epidemiology 40 (7), 558–569. Mardis, E.E., 2010. The $1,000 genome, the $100,000 analysis? Genome Medicine 2 (11), 84. McAfee, T., Babb, S., MnNabb, S., Fiore, M.C., 2015. Helping smokers quit: Opportunities created by the affordable care act. The New England Journal of Medicine 372 (1), 5–7. Mealiffe, M.E., Stokowski, R.P., Rhees, B.K., et al., 2010. Assessment of clinical validity of a breast cancer risk model combining genetic and clinical information. Journal of the National Cancer Institute 102 (21), 1618–1627. Michailidou, K., Beesley, J., Lindstrom, S., et al., 2015. Genome-wide association analysis of more than 120,000 individuals identifies 15 new susceptibility loci for breast cancer. Nature Genetics 47 (4), 373–380. Miller, G.W., Jones, D.P., 2014. The nature of nurture: Refining the definition of the exposome. Toxicological Sciences 137 (1), 1–2. Munnia, A., Giese, R.W., Polvani, S., et al., 2017. Bulky DNA adducts, tobacco smoking, genetic susceptibility, and lung cancer risk. Advances in Clinical Chemestry 81, 231–277. Perera, F.P., Weinstein, I.B., 1982. Molecular epidemiology and carcinogen-DNA adduct detection: New approaches to studies of human cancer causation. Journal of Chronic Diseases 35 (7), 581–600. Perera, F.P., 1987. Molecular cancer epidemiology: A new tool in cancer prevention. Journal of the National Cancer Institute 78, 887–898. Perera, F.P., 2000. Molecular epidemiology: On the path to prevention? Journal of the National Cancer Institute 92, 602–612. Perou, C.M., Sørlie, T., Eisen, M.B., et al., 2000. Molecular portraits of human breast tumours. Nature 406 (6797), 747–752. Phillips, D.H., Venitt, S., 2012. DNA and protein adducts in human tissues resulting from exposure to tobacco smoke. International Journal of Cancer 131 (12), 2733–2753. Pineda, S., Real, F.X., Kogevinas, M., et al., 2015. Integration analysis of three omics data using penalized regression methods: An application to bladder cancer. PLoS Genetics 11 (12), e1005689. Rappaport, S.M., Smith, M.T., 2010. Epidemiology. Environment and disease risk. Science 330 (6003), 460–461. Ritchie, M.D., Holzinger, E.R., Ruowang, L., Pendergrass, S.A., Kim, D., 2015. Methods of integrating data to uncover genotype-phenotype interactions. Nature Reviews Genetics 16, 85–97. Robertson, A.G., Kim, J., Al-Ahmadie, H., et al., 2017. Comprehensive molecular characterization of muscle-invasive bladder cancer. Cell 171 (3), 540–556.e25. Severi, G., Southey, M.C., English, D.R., et al., 2014. Epigenome-wide methylation in DNA from peripheral blood as a marker of risk for breast cancer. Breast Cancer Research and Treatment 148 (3), 665–673. Samet, J.M., 2016. Epidemiology and the tobacco epidemic: How research on tobacco and health shaped epidemiology. American Journal of Epidemiology 183 (5), 394–402. Sud, A., Kinnersley, B., Houlston, R.S., 2017. Genome-wide association studies of cancer: Current insights and future perspectives. Nature Reviews Cancer 17 (11), 692–704. Sun, Z., Asman, Y.W., Kalari, K.R., et al., 2011. Integrated analysis of gene expression, CpG island methylation, and gene copy number in breast cancer cells by deep sequencing. PLoS One 6 (2), e17490. Tini, G., Marchetti, L., Priami, C., Scott-Boyer, M.P., 2017. Multi-omics integration-a comparison of unsupervised clustering methodologies. Briefing in Bioinformatics. https:// doi.org/10.1093/bib/bbx167. The Cancer Genome Atlas Network, 2012. Comprehensive molecular portraits of human breast tumours. Nature 490 (7418), 61–70. The Cancer Genome Atlas Network, 2017. Integrated genomic and molecular characterization of cervical cancer. Nature 543 (7645), 378–384. Vahter, M., Akesson, A., Lidén, C., Ceccatelli, S., Berglund, M., 2007. Gender differences in the disposition and toxicity of metals. Environmental Research 104 (1), 85–95. Wakeford, R., 1995. The risk of childhood cancer from intrauterine and preconceptional exposure to ionizing radiation. Environmental Health Perspectives 103 (11), 1018–1025. Wild, C.P., 2005. Complementing the genome with an “exposome”: The outstanding challenge of environmental exposure measurement in molecular epidemiology. Cancer Epidemiology, Biomarkers & Prevention 14 (8), 1847–1850. Wild, C.P., 2012. The exposome: From concept to utility. International Journal of Epidemiology 41 (1), 24–32. Zardavas, D., Irrthum, A., Swanton, C., Piccart, M., 2015. Clinical management of breast cancer heterogeneity. Nature Reviews Clinical Oncology 12 (7), 381–394. Zhu, M., Zhao, S., 2007. Candidate gene identification approach: Progress and challenges. International Journal of Biological Sciences 3 (7), 420–427.

Further Reading Chatterjee, N., Shi, J., García-Closas, M., 2016. Developing and evaluating polygenic risk prediction models for stratified disease prevention. Nature Reviews Genetics 17 (7), 392–406. McGranahan, N., Swanton, C., 2017. Clonal heterogeneity and tumor evolution: Past, present, future. Cell 168 (4), 613–628. Rothman, K.J., Sander, G., Lash, T.L. (Eds.), 2008. Modern epidemiology, 3rd edn. Wolters Kluwer Health/Lippincott Williams & Wilkins, Philadelphia (PA).

Molecular Epidemiology and Cancer Risk

Relevant Websites http://cancergenome.nih.gov/dThe cancer genome atlas. http://www.exposomicsproject.eu/dThe EXPOsOMICS project. http://icgc.orgdThe International Cancer Genome Consortium project.

493

Multiple Myeloma: Pathology and Genetics Marie Yammine, University of Lille, INSERM UMR-S 1172, Lille, France Salomon Manier, University of Lille, INSERM UMR-S 1172, Lille, France; University of Lille, CHU Lille, Lille, France; and Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, United States © 2019 Elsevier Inc. All rights reserved.

Introduction Multiple myeloma (MM) is a hematologic malignancy that originates from plasma cells in the bone marrow (BM). According to the National Cancer Institute, the prevalence of MM was estimated at 89,650 people in the United States in 2012, with an annual incidence of 6.3 new cases per 100,000 individuals. All MM are preceded by a monoclonal gammopathy of undetermined significance (MGUS) and smoldering myeloma (SMM). The diagnosis of MM is often suspected because of the presence of clinical symptoms such as bone pain with lytic lesions, an increased total serum protein concentration, the presence of a monoclonal protein in the urine or serum, anemia or hypercalcemia. MM is incurable, due to the lack of treatment able to target specific oncogenes involved in the pathogenesis of the disease. Genomic events such as chromosomal translocations, copy number variation and somatic mutations as well as epigenetic modifications are responsible of deregulation of specific oncogenes or tumor suppressor genes and their corresponding pathways, leading to MM proliferation.

Plasma Cell Differentiation and Myeloma Initiation As MM is characterized by the production of a monoclonal immunoglobulin (Ig), its initiation is linked to B cell development. Bcells undergo rearrangement of their Ig genes into the bone marrow, generating a functional B cell receptor precursor, and migrate to lymph nodes where they encounter antigens. B cells are then activated and reach the germinal center where they undergo affinity maturation in response to antigen presented on antigen-presenting cells. This process is called somatic hypermutation (SHM) and induces mutations of the hypervariable regions of the immunoglobulin heavy chain locus (IGH), thus producing highly specific and avid antibodies. The functionality of these antibodies is increased by class switch recombination (CSR), which produces antibodies of different immunoglobulin isotypes. SHM and CSR require the expression of activation-induced deaminase (AID), which induces double strand DNA breaks (DSBs) in the Ig loci. AID-induced DSBs are mostly repaired locally, however they can be joined to DSBs occurring elsewhere in the genome, resulting in aberrant chromosomal translocations. These translocations occur preferentially into actively transcribed genes; however, the range of CSR-driven translocations in MM is limited. This may explain why only a subset of these translocations causes a proliferation advantage. Once SHM and CSR completed, selected B-cells leave the germinal center and migrate to the bone marrow where they become mature plasma cells, secreting antibodies. This process requires cell cycle arrest, compaction of chromatin, silencing of cellular functions that are unnecessary for antibody production and activation of key programs that are required to generate and secrete antibodies. Deregulation of these pathways could potentially lead to malignant transformation. Normal plasma cell differentiation is regulated by different transcription factors. IRF4 downregulates BCL-6, resulting in the upregulation of BLIMP1, which in turn leads to downregulation of PAX5 and upregulation of XBP1. Expression of IRF4, BLIMP1 and XBP1 is necessary for survival of plasma cells. After exiting the germinal center, plasma cells migrate to the bone marrow where either they rapidly undergo apoptosis, or reside in specialized niches where they can survive for many years as memory plasma cells.

Inherited Variants Although lifestyle or environmental exposures have not been consistently linked to the incidence of MM, there seems to be a two- to fourfold elevated risk of MM in relatives of individuals with the disease. This has been postulated to be a consequence of the coinheritance of multiple low-risk variants. Investigating these families further and performing genome-wide association studies (GWAS) on large patient populations, three genetic loci were associated with a modest but increased risk of developing MM. These include 3p22.1 (rs1052501, in ULK4), 7p15.3 (rs4487645, surrounding by DNAH11 and CDCA7L) and 2p23.3 (rs6746082, surrounding by DNMT3A and DTNB). A follow-up study by the same group, including 4692 individuals with MM and 10,990 controls, revealed four new loci: 3q26.2 (rs10936599, surrounding by MYNN and TERC), 6p21.33 (rs2285803 in PSORS1C2), 17p11.2 (rs4273077 in TNFRSF13B) and 22q13.1 (rs877529 in CBX7). These seven identified loci provide further evidence for an inherited genetic susceptibility to MM and reportedly account for
13% of the familial risk of MM. The complete functional role of each of these candidate genes remains to be elucidated. The authors found no association between genotypes and the expression level of their genes. Interestingly, in another GWAS study, the same team identified a strong association between the variant rs603965, responsible for c807G > A polymorphism in CCND1 and the translocation t(11;14)(q13;q32), in which CCND1 is placed under the control of the immunoglobulin heavy (IGH) chain enhancer. In this model, a constitutive genetic factor is associated with risk of a specific chromosomal translocation. Based on these initial studies, it is likely that more susceptibility loci will be identified in the future and possibly correlated with specific MM subtypes. For example, African-Americans have a higher risk of

494

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65260-1

Multiple Myeloma: Pathology and Genetics

495

developing MM than Caucasians; however, no potential genetic variants have been identified to date. Moreover, uncovering the functional role of these seven SNPs significantly associated with MM might help to advance our understanding of MM oncogenesis.

Chromosomal Translocations In MM, most chromosomal translocations involve chromosome 14, and specifically the IGH locus on 14q32.33, placing a partner gene under the control of the IGH enhancer. These translocations are generated by abnormal class switch recombination (CSR) events and are usually present in all clonal cells. They are also detectable in monoclonal gammopathy of unknown significance (MGUS), consistent with their early development in MM oncogenesis. Five major chromosomal partnersdt(4;14), t(6;14), t(11;14), t(14;16), and t(14;20)dseem to impart a selective advantage to the clone by up regulating expression of specific oncogenesdMMSET and FGFR3, CCND3, CCND1, MAF and MAFB, respectively. It is likely that all these translocations lead to deregulation in the cell cycle G1/S transition, which has been described as a key early molecular abnormality in MM. This can be direct through t(11;14) and t(6;14) deregulating CCND1 and CCND3 respectively. In t(14;16), this is modulated through MAF which up regulates CCND2 by directly binding to its promoter while in t(4;14), the exact mechanism is still uncertain but the translocation of FGFR3 and MMSET to the IGH enhancer is known to also up regulate CCND2. Recently mutations involving the MYC locus have been identified in MM. Translocation (4;14) is observed in about 15% of MM cases and has been associated with adverse prognosis in a variety of clinical settings. The juxtaposition results in deregulation of the expression of FGFR3 and MMSET/WHSC1. The breakpoints all reside between FGFR3 and MMSET, resulting in overexpression of FGFR3 in 70% of cases and MMSET in all cases. MMSET is a methyl-transferase protein, whose up regulation leads to methylation of histone H3K36, which regulates expression of several genes. MMSET has been shown also to regulate histone H4K20 methylation and recruit 53BP1 at DNA damage sites. FGFR3 is a tyrosine kinase receptor oncogene activated by mutations in several solid tumor types. Notably, FGFR3 is up regulated in only 70% of patients with the translocation because of an unbalanced translocation with loss of the telomeric part of chromosome 4, bearing FGFR3. This suggests that MMSET is the main molecular target of the translocation. Interestingly, despite the poor prognosis associated with t(4;14), a survival advantage in these patients has been demonstrated through early treatment with the proteasome inhibitor Bortezomib. Translocation (6;14) is a rare translocation present in only about 2% of MM patients and results in the direct up regulation of CCND3 via juxtaposition to the IGH enhancers. The breakpoints are all located 50 of the gene. The overall prognostic impact of this translocation is neutral. Translocation (11;14) is the most frequent translocation cited as being present in about 15%–20% of patients with MM. Normally B cells express cyclin D2 and D3 but not D1. However, due to the translocation juxtaposing CCND1 to the IGH enhancer, its expression is deregulated. The breakpoints seem to be located 50 of CCND1. In terms of prognosis, this translocation is considered as neutral, however it has been shown recently that in 10% of t(11;14) a CCND1 mutation co-occurs and the combination is associated with a poor prognosis when compared with non-mutated t(11;14) patients. Translocation (14;16) is estimated to be present in about 5%–10% of patients with MM and results in overexpression of the MAF oncogene splice variant c-MAF, a transcription factor which up regulates a number of genes, including CCND2, by binding directly to its promoter. Breakpoints are located 30 of MAF within the last exon of WWOX, a known tumor suppressor. Though t(14;16) was associated with poor prognosis in several studies. A more recent retrospective multivariable analysis on 1003 newly diagnosed MM patients showed that t(14;16) is not associated with poor prognosis. Translocation (14;20) is present in about 1% of patients and is the rarest translocation of the major five. It results in upregulation of the MAF gene paralog MAFB. According to microarray studies, MAFB overexpression results in a similar gene expression profile (GEP) as that seen with c-MAF, implying common downstream targets including CCND2. The translocation is associated with poor prognosis when present in MM, but interestingly correlates to long-term stable disease when found in precursor conditions like MGUS and smoldering MM (SMM). This suggests that the translocation itself is not responsible for the poor prognosis, but additional genetic events are likely required to accumulate imparting this negative prognosis. MYC translocations have been recently identified in a cohort of 463 whole exome sequencing including extra baits on the MYC locus. MYC translocations were found in 85 patients (18.4%). Partner genes include IGH, IGL and IGK loci, as well as FAM46C, FOXO3, BMP6 and rarely XBP1, TXNDC5, CCND1 and CCND3. These translocations lead to significant overexpression of MYC, probably resulting from juxtaposition of super-enhancers surrounding the partner gene to MYC locus. MYC translocations are associated with a poor outcome.

Hyperdiploidy Hyperdiploidy (HRD) is defined as a chromosome number between 48 and 74. HRD MM are characterized by multiple chromosomal gains, preferentially trisomy of chromosomes 3, 5, 7, 9, 11, 15, 19, and 21. The mechanism underlying this is not known but one hypothesis suggests that the gain of multiple whole chromosomes occurs during a single catastrophic mitosis rather than through the serial gain of chromosomes over time. Nearly half of MGUS and MM tumors are hyperdiploid. Only a few HRD tumors have a co-existing primary IgH translocationdabout 10% of the casesdwhereas non-HRD tumors usually have an IgH

496

Multiple Myeloma: Pathology and Genetics

translocation. Interestingly, when HRD and IGH translocations coexist, HRD may precede IGH translocations in a proportion of patients, as revealed by single cell sequencing analysis. In terms of signaling pathways, HRD tumors display biological heterogeneity. Some harbor high expression of proliferation-associated genes while others are characterized by genes involved in tumor necrosis factor/nuclear factor-kB (TNF/NFkB) signaling pathway. HRD is associated with a more favorable outcome in general, however coexistent adverse cytogenetic lesions (del 17p, t(4;14) and gain of 1q) shorten survival in MM patients with HDR tumors.

Copy Number Variations Copy number variations (CNVs) represent a common feature of MM and are thought to be secondary events, involved in tumor progression. CNVs result from gain and loss of DNA, which can be focal or involve an entire chromosome arm. Like single nucleotide mutations, CNVs are probably both driver and passenger events. Highly frequent and recurrent CNVs are likely to be driver, suggesting that the minimal amplified or deleted regions contain important genes involved in the development and progression of MM. 1q gain: Duplication of the long arm of chromosome 1 is present in 35%–40% of patients. This is known to have an adverse effect on overall survival. Gain of 1q21, detected with a specific probe for CKS1B, is an independent prognostic factor and remains when other adverse cytogenetic lesions that frequently coexist are removed. Though the relevant genes on 1q have not yet been fully explored, a minimally amplified region was identified between 1q21.1 and 1q23.3 containing 679 genes. Among these candidate oncogenes are CKS1B, ANP32E, BCL9, and PDZK1. Of these genes, ANP32E, a protein phosphatase 2A inhibitor involved in chromatin remodeling and transcriptional regulation, is particularly interesting and has been shown to be independently associated with shortened survival. These findings reinforce the role of gain of 1q in MM pathogenesis and suggest that patients with this type of CNV may benefit from specific inhibitors of these candidate genes and pathways that have been identified. 1p deletion: Deletions of 1p are observed in approximately 30% of MM patients and are associated with poor prognosis. Two regions of the 1p arm are of interest in MM pathogenesis when deleted: 1p12 and 1p32.3. 1p12 contains the candidate tumor suppressor gene FAM46C, of which expression has been correlated to that of ribosomal proteins and eukaryotic initiation/elongation factors involved in protein translation. This gene is frequently mutated in MM and this has been independently correlated with poor prognosis. Region 1p32.3 may be hemi- and homozygously deleted and contains the two target genes CDKN2C and FAF1. CDKN2C is a cyclin-dependent kinase 4 inhibitor involved in negative regulation of the cell cycle, whereas FAF1 encodes a protein involved in initiation and enhancement of apoptosis through the Fas pathway. Deletion 1p is associated with adverse overall survival. 13q deletion: Monosomy of the long arm of chromosome 13 is present in about 45%–50% of patients and is commonly associated with non-hyperdiploid tumors. In approximately 85% of cases, deletion of chromosome 13 constitutes monosomy or loss of the q arm, whereas in the remaining 15% various interstitial deletions occur. Chromosome 13 has been extensively investigated as prognostic factor and as a location of tumor suppressor genes. The minimally deleted region lies between 13q14.11-13q14.3 and contains 68 genes including RB1, EBPL, RNASEH2B, RCBTB2, and the microRNA miR-16-1 and miR-15a. Molecular studies have shown that the tumor suppressor gene RB1 is significantly under-expressed in these deletions, which may result in inferior negative cell cycle regulation. Establishing a prognostic significance of deletion 13 is challenging because it is frequently associated with other high risk cytogenetic lesions such as t(4;14). As such, the historic link between deletion 13 and poor prognosis is a surrogate of its association with high-risk lesions. 17p deletion: Most chromosome 17 deletions are hemizygous and involve the whole p arm, a genetic event observed in around 10% of newly diagnosed MM cases, with an increasing frequency in later stages of the disease. The minimally deleted region includes the tumor suppressor gene TP53. While cases without del(17p) have a rate of TP53 mutation that is < 1%, cases with the deletion show a higher rate of mutation at 25%–37%dsuggesting that mono-allelic 17p deletion contributes to the disruption of the remaining allele. The TP53 gene, which has been mapped to 17p13, is known to function as a transcriptional regulator influencing cell cycle arrest, DNA repair, and apoptosis in response to DNA damage. Loss of 17p is associated with an adverse overall survival. The deletion is also linked to an aggressive disease phenotype, more extra-medullary disease, and shortened survival.

Somatic Mutations The generalization of next generation sequencing a few years ago has enabled high throughput whole-exome sequencing in several cancers, including MM. The frequency of somatic mutations in MM is at the median across cancer types, with an average of 1.6 mutations per Mb, as compared to < 0.5 per Mb in pediatric cancer, such as rhabdoid tumor or Ewing sarcoma, and about 10/Mb in melanoma and lung cancer. Studies by whole genome sequencing (WGS) and whole exome sequencing (WES) of patients with MM, comparing sequences from each tumor to its corresponding normal germline sample, allowed identification of tumorspecific mutations. Significantly mutated genes included three that were previously reported as being implicated in MM: KRAS, NRAS, and TP53 as well as two newly described genes FAM46C and DIS3. Several new oncogenic mechanisms were suggested by the pattern of somatic mutations across this data set. Nearly half the patients showed mutations of genes involved in protein translation. One of these is the DIS3 gene, also known as RRP44, which encodes a highly conserved RNA exonuclease and serves as the catalytic component of the exosome complex involved in regulating

Multiple Myeloma: Pathology and Genetics

497

the processing and abundance of all RNA species. DIS3 mutations, postulated to be loss of function, cluster in the catalytic pocket of the enzyme and deregulate protein translation as an oncogenic mechanism. Another significantly mutated gene, FAM46C, is less well characterized but thought to be functionally related to translation regulation. The same team (Lohr et al., 2014) next reported massively parallel sequencing of a large cohort of patients with MMdincluding those previously studied. Beyond the five significantly mutated genes previously described, they identified another six significantly mutated genes (BRAF, TRAF3, PRDM1, CYLD, RB1, and ACTG1). Overall in this study, 65% of the patients had mutations in one or more of the 11 recurrently mutated genes. Like KRAS and NRAS, BRAF is a known oncogene playing a role in regulating the MAP kinase pathway. Strikingly, mutations in KRAS, NRAS, and BRAF can be both clonal and sub-clonal. However, if mutations in these genes sometimes co-exist in the same tumor, they are almost never simultaneously clonal, indicating that they probably rarely occur in the same clone but rather in different sub-clones. In contrast, KRAS and DIS3 mutations are reported to be often simultaneously clonal and therefore probably co-occurring in the same clone. TRAF3 and CYLD are part of the NFkB pathway, which is also the case for nine other mutated genes of significance in this cohort (BTRC, CARD11, IKBKB, MAP3K1, MAP3K14, RIPK4, TLR4, and TNFRSF1A), which reaffirms the central role of the NFkB pathway in MM. Another significantly mutated gene is PRDM1 (also called BLIMP1), a transcription factor involved in plasma cell differentiation. Loss of function mutations of BLIMP1 occur in diffuse large B cell lymphoma. The oncogene IRF4, a transcriptional regulator of PRDM1 was also frequently mutated, in addition to mutations of PRDM1. In parallel, a WES and copy number analysis of a series of MM samples led to the identification of two new recurrently mutated genes, SP140 and LTB. SP140 is a lymphoid restricted homologue of SP100 that encodes a nuclear body protein implicated in antigen response of mature B cells, and is truncated in several cases. LTB, a type II membrane protein of the TNF family involved in lymphoid development, also harbor truncated mutations. Finally, WES of a large number of patients enrolled in a phase III clinical trial (Walker et al., 2015) brought the list to 15 significantly mutated genes, comprising KRAS, NRAS, TP53, FAM46C, DIS3, BRAF, HIST1H1E, RB1, EGR1, TRAF3, LTB, CYLD, IFR4, MAX, and FGFR3. Interestingly, mutations in RAS (43% of the cases) and NFkB (17% of the cases) were prognostically neutral. In contrast, mutations in CCND1 and the DNA repair pathway (TP53, ATM, ATR, and ZFHX4) were associated with a negative impact on survival in contrast to those in IRF4 and EGR1 that are associated with a favorable overall survival. Identification of driver mutations in MM holds great promise for personalized medicine, whereby patients with particular mutations would be treated with the appropriate targeted therapy. However, if the mutation is present in only a fraction of the cells, one might doubt whether such targeted therapy would be clinically efficacious.

Clonal Heterogeneity and Clonal Evolution in Multiple Myeloma In addition to the genetic complexity of MM, intra-clonal heterogeneity has emerged as a further level of complexity. Analysis of clonal heterogeneity by WES (Lohr et al., 2014) showed that most patients harbor at least three detectable sub-clones, some having as many as seven, which reaffirms that MM tumors are highly heterogeneous. The finding that tumors contain on average at least five sub-clones is even an underestimation of the clonal diversity in MM as the applied method only allows for the detection of subclones representing at least 10% of the entire tumor sample. It has become clear that following disease initiation, the steps necessary for MM development do not occur through a linear fashion but rather via branching, non-linear pathways, as proposed by Darwin in explaining the evolution of species. This idea is based on the notion that mutations occur randomly and are selected and propagated based on the clonal survival advantage that they confer. A phenomenon of parallel evolution, whereby independent but not distantly related clones might acquire similar mutations conferring important growth or survival advantages, is revealed in single-cell level studies showing more than one alteration of same genetic pathway (RAS/MAPK) within the same tumor, but in separately evolving divergent clones. In a series of t(11;14) MM, evidence for the persistence of the earliest MM progenitor cell clone was found in two cases, characterized by the presence of a sub-clone carrying t(11;14) as the sole abnormality, validating that this translocation is an early event in myeloma pathogenesis. The clonal diversity is present at all stages of the disease. Although less genetically complex than MM, the pre-malignant stages MGUS and SMM harbor clonal heterogeneity. By studying sequential samples of SMM and overt MM, it was shown that the predominant clone of MM is already present at the SMM stage.

Treatment Implications One of the main areas of development in genomics of MM is the critical need to comprehensively implement precision medicine. Perhaps more importantly, precision medicine aims to tailor the appropriate therapy to a patient in a personalized fashion, based on the disease genomic information. To this end, identification of driver mutations in MM holds great promise for precision medicine, whereby patients with particular mutations would be treated with the appropriate targeted therapy. A recent report describes a durable response to the mutation-specific BRAF inhibitor Vermurafenib in a patient harboring an activating mutation of BRAF. Several of the recurrently mutated genes in MM are potentially “actionable,” meaning they can be targeted by therapies specifically

498

Multiple Myeloma: Pathology and Genetics

inhibiting the mutated or activated oncogene. However, this targeted therapy approach must be considered in light of the clonal heterogeneity and clonal competitions co-occurring in the cancer cell population. In fact, BRAF inhibitors are known to paradoxically activate the MAPK pathway in the event of co-existent KRAS-, NRAS-mutated or BRAF-WT sub-clones. This could be abrogated by the combination of BRAF and MEK inhibitors. Clonal heterogeneity poses a significant challenge; in fact, in most cases relapsed disease shows marked differences in genetic makeup compared with pre-treatment disease. This brings up the challenge of determining the most effective combinations of therapy. Thus, the comprehensive understanding of clonal heterogeneity is crucial to further develop targeted therapies in MM.

Conclusion MM is a genetically complex and heterogeneous disease, combining primary events, secondary events and clonal diversity, leading to tumor development and progression from MGUS to late stages of MM. It is likely that many driver events need to co-occur for MM development and progression. This genomic complexity presents a challenge towards the cure of MM. In recent years, a tremendous amount of information has been obtained by next generation sequencing of MM tumors. Now that we have a good comprehension of the genomic landscape in MM at presentation, the near future should provide some insights regarding pre-malignant stages and resistance to treatment.

Further Reading Lohr, J.G., et al., 2014. Widespread genetic heterogeneity in multiple myeloma: Implications for targeted therapy. Cancer Cell 25, 91–101. Manier, S., et al., 2017. Genomic complexity of multiple myeloma and its clinical implications. Nature Reviews. Clinical Oncology 14, 100–113. Morgan, G.J., et al., 2012. The genetic architecture of multiple myeloma. Nature Reviews. Cancer 12, 335–348. Palumbo, A., Anderson, K., 2011. Multiple myeloma. New England Journal of Medicine 364, 1046–1060. Walker, B.A., et al., 2015. Mutational spectrum, copy number changes, and outcome: Results of a sequencing study of patients with newly diagnosed myeloma. Journal of Clinical Oncology 33, 3911–3920.

Mutational Signatures and the Etiology of Human Cancers Ludmil B Alexandrov, University of California San Diego, San Diego, CA, United States Maria Zhivagui, International Agency for Research on Cancer (WHO), Lyon, France © 2019 Elsevier Inc. All rights reserved.

Glossary Mutational process A mixture of DNA damaging and repair mechanisms that act collectively and have the ability to cause mutations in cells. Mutational catalogue The conglomeration of all detected somatic mutations in the genome of a cell. Mutational signature A characteristic pattern of somatic mutations imprinted by a mutational process on the genome of a cell. Nonnegative matrix factorization A mathematical approach that allows unscrambling mixed signals and identifying meaningful components, such as, mutational signatures. Somatic mutation Any change in the nucleotide sequence of DNA that is present in the genome of a somatic cell and has occurred after conception.

Nomenclature AA Aristolochic acid AFB1 Aflatoxin B1 APOBEC Apolipoprotein B mRNA editing enzyme, catalytic polypeptide-like BaP Benzo[a]pyrene BSS Blind source separation DNA Deoxyribonucleotide acid EBV Epstein Barr virus HBV Hepatitis B virus HCC Hepatocellular carcinoma HMEC Human mammary epithelial cell HPV Human papilloma virus Hupki Human TP53 knock-in ICGC International Cancer Genome Consortium Indels Small insertions and deletions MEF Mouse embryonic fibroblast MMR DNA mismatch repair mRNA Messenger RNA NGS Next-generation sequencing NMF Nonnegative matrix factorization PCR Polymerase chain reaction ROS Reactive oxygen species SBS Single-base substitutions TSG Tumor suppressor gene UTUC Upper tract urothelial carcinoma UV Ultraviolet light

Introduction The term “cancer” encompasses a broad group of over two hundred different diseases characterized by an abnormal cellular growth. The transformation from a normally functioning cell to a neoplastic cell is due to the accumulation of changes in the genetic material of the cell. These changes are known as somatic DNA mutations. Analogous to Darwinian evolution, some mutations can have no effect on the cell, some can be deleterious thus killing the cell in which they arouse, and others can provide a selective growth advantage. Mutations providing growth advantage ultimately allow cancer cells to survive, to proliferate, to invade, and to metastasize to other tissues in the body.

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65046-8

499

500

Mutational Signatures and the Etiology of Human Cancers

Somatic mutations can be caused by many different mutational processes. These mutational processes can be triggered by the internal molecular mechanisms of cells (e.g., mutations occurring due to normal cellular division), tissues-specific processes (e.g., inflammation), external environmental sources (e.g., exposure to radiation), and lifestyle choices (e.g., tobacco smoking). Each mutational process imprints a characteristic pattern of somatic mutations on the genome of cancer cells, termed, mutational signature. Mutational signatures can be deciphered from genomes of cancer cells by using next-generation sequencing data and advanced computational and mathematical approaches. This article reviews our current understanding of mutagenesis, mutational signatures in human cancer, and mathematical approaches for deciphering mutational signatures from nextgeneration sequencing data.

Somatic Mutations By definition, a somatic mutation is a change in the DNA sequence of a somatic cell. This change can be a single base substitution, multi-base tandem substitutions, small insertion, small deletion, complex insertion and/or deletion, chromosomal rearrangement, copy number variation, or introduction of a foreign DNA (Fig. 1). A single base substitution (SBS) modifies one DNA base pair to another. For example, a C:G base pair is substituted with an A:T base pair. Multi-base tandem substitutions simultaneously change multiple adjacent DNA base pairs. For example, a CC:GG sequence is substituted with a TT:AA sequence. Small insertions and/or deletions, commonly termed indels, result from the loss or the gain of a small number of adjacent nucleotide base pairs (Fig. 1). Most indels in a cancer genome have a length of less than 10 base pairs but some can be as long as a few hundred base pairs. Depending on the base changes and positions, single base substitutions and indels can have distinct effects on cellular functionality. For example, silent or synonymous mutations do not alter the translated protein sequence of the transcripted RNA. However, they may affect the translation fidelity of the transcripted RNA in the ribosomal subunits. Conversely, non-synonymous mutations change the protein sequence by introducing amino-acid changes, frameshifts, or stop codons. Further, many indels and SBSs occur in the non-coding region of the genome where they can disrupt gene expression by, for example, affecting the binding affinity of transcription factors. Single Base Substitution (SBS) 5’ A C C T T G G A

Multi-base tandem substitutions 3’

C G

5’

C T T T G G A A A C

3’

C G

5’ A C C T T G G A

5’

C T T T G G A A A C

3’

5’

3’ C G

C T T T G G A A A C

3’

C G

3’

A C C T T G G A

3’ A T

A C C T T T G G A A

3’

Inter-chromosomal rearrangements

T G A C 5’

Intra-chromosomal rearrangements

C T T T G G A A A C 5’

deletion of CC:GG insertion of AAAA:TTTT

insertion of AA:TT 5’

5’

3’

A C C T T G G A 5’

C T T T G G A A A C

3’

5’

5’

3’ A C C T G G G A

Complex insertion and deletion

5’

5’

5’

T T T T G A A A A C

3’

Small insertion A C C T T G G A

T A

C T T T G G A A A C

deletion of C:G

3’

A C C T T G G A

C G

3’

5’

CC:GG > TT:AA 3’

T A

A C C T T G G A

C T T T G G A A A C

3’

C:G > T:A

3’

5’

3’

A C C T T G G A 5’

Small deletion

5’

3’

A C C T T G G A

A T

A A A T T T T T A A

3’

T G A C 5’

Copy number variations

(i)

Insertion of a foreign DNA sequence

(ii)

Fig. 1 Illustrative examples of the different types of genomic alterations present in cancer genomes. Small changes of the genome consist of single base substitutions, multi-base tandem substitutions, small insertions, small deletions, and complex insertions and deletions. Chromosomal rearrangements comprise intra-chromosomal translocations and inter-chromosomal translocations. Copy number variation of a specific DNA sequence leading to an increased (i) or a depleted (ii) copy of genome region. Insertion of new DNA sequence (e.g., viral DNA). Each alteration is represented relative to its position on chromosomes.

Mutational Signatures and the Etiology of Human Cancers

501

In addition to indels and SBSs, chromosomal rearrangements and copy number variations are commonly found in many types of human cancer. Chromosomal rearrangements occur when DNA segments break off and re-attach at different genomic locations. These locations can be on the same chromosome or on a different chromosome, resulting in intra- or inter-chromosomal rearrangements, respectively (Fig. 1). Chromosomal rearrangements have many functional implications and can lead, for example, to gene disruption, to fusion of two distinct genes, or to abnormal gene expression. Copy number changes of a DNA sequence can increase its two copies in a normal diploid genome to several hundred copies in a cancer genome. This increase is referred to as genomic amplification, which is a common mechanism for the activation of many proto-oncogenes. Copy number changes can also reduce the copies of a DNA sequence due to large chromosomal deletions. These deletions may induce the complete absence of a DNA segment resulting in the loss of an associated gene; a commonly observed mutational mechanism for tumor suppressor genes. Insertion of a foreign DNA sequence, originating from an exogenous source, is another important mechanism unambiguously implicated in the development of different types of cancer. Notable examples include incorporation of the sequence of the human papilloma virus (HPV), Epstein Barr virus (EBV), or hepatitis B virus (HBV).

Driver and Passenger Somatic Mutations Mutations accumulate progressively during the lifespan of an individual in every single cell of the body. In general, it is accepted that mutations occur somewhat randomly across the genome and that they can be broadly separated into two categoriesd(i) mutations that provide selective advantage for clonal expansion and (ii) mutations that do not result in any growth advantage. The latter have been termed passenger mutations, while the former are referred to as driver mutations. It is widely believed that the number of driver mutations in a cancer genome is limited to a handful, usually between two and ten mutations. In contrast, the genome of a cancer can harbor more than a million somatic mutations most of which are considered passengers. Passenger mutations are not per se involved in cancer development but are rather the residual molecular fingerprints of the operative mutational processes.

Mutational Catalogues of Cancer Genomes Even before the official start of the Human Genome Project, it was hypothesized that systematically analyzing the genetic information of cancer cells at a single base resolution will provide significant insights into the mechanisms of cancer development. While previous approaches allowed identification of large genomic events (e.g., copy number changes, chromosomal translocations, etc.) examining cancer genes by interrogating their sequence at a single-base resolution held the promise for observing previously unseen mutational events. At first, such sequencing examinations were performed using polymerase chain reaction (PCR)-based capillary sequencing for a targeted set of genes; however, the development of next-generation sequencing methods allowed rapidly sequencing of the complete set of exons in a cancer genome (i.e., an approach known as whole-exome sequencing) and, currently, even whole-genome sequences of cancer samples can be generate at a low cost. Regardless of the experimental approach, the idea behind sequencing cancer genomes (or parts of these genomes) is simple. Genomic DNA is extracted from both the cancer and the normal tissue (which is usually but not always blood) and then these genomic DNAs are sequenced separately. The identified normal and cancer nucleotide sequences are aligned to the reference human genome, they are compared to it, and then they are compared to each other. The nucleotides that differ from the reference genome and that are found in both the normal and the cancer tissues are attributed to germline polymorphisms. Conversely, DNA sequence changes identified only in the cancer tissue, but not in the normal tissue, are attributed to somatic mutations specific for the cancer. All DNA changes identified only in the cancer tissue constitute the mutational catalogue of the cancer genome. An illustrative example identifying a somatic base substitution and a single nucleotide polymorphism from next generation sequencing (NGS) reads is provided in Fig. 2. The majority of somatic mutations identified in the mutational catalogues of cancer genomes are passenger mutations. The ability to examine hundreds and even thousands of mutational catalogues of cancer genomes has resulted in the development of advanced statistical methods that allow pinpointing a handful of driver mutations from an ocean of passenger mutations. In simple terms, these algorithms evaluate which genes are mutated more often than purely expected by chance while correcting for a multitude of different factors. Using a targeted capillary sequencing approach, an early cancer genomics study demonstrated that mutations in the BRAF gene are found in
70% of melanomas. This was followed by subsequent studies identifying PIK3CA and EGFR as genes commonly mutated in human cancer. These early successes and their clinical significance made the identification of cancer genes through the systematic sequencing of cancer genomes, one of the main topics in cancer research. The emergence of NGS technologies allowed rapid and cheap examination of the genetic material of cancer cells. This led to the formation of the International Cancer Genome Consortium (ICGC). The goal of the ICGC is the identification of novel cancer genes through the molecular characterization of 50 types of human cancer (and their matched-normal tissues) from more than 25,000 patients. Nowadays, large-scale initiatives, such as the ICGC, continue to identify genes causally implicated in tumorigenesis and the census of genes driving human cancer gets updated on nearly a monthly basis.

502

Mutational Signatures and the Etiology of Human Cancers

Fig. 2 Somatic mutations in cancer and nucleotide polymorphisms in the germline. Illustrated example demonstrating the identification of germline polymorphisms and somatic mutations from sequencing data.

Endogenous and Exogenous Mutational Processes The genome of a cancer cell can be examined as a historical record that bears the imprints of the different mutagens to which the lineage of this cell has been exposed. Mutations can be caused by a plethora of endogenous factors, such as defects in the DNA repair machinery, or a variety of exogenous factors, such as dietary compounds, smoking, viruses, as well as some occupational and environmental carcinogens. Differences in incidence rates across cancer types have stimulated a discussion on whether stochastic DNA changes (i.e., somatic mutations due to the variations in stem cell divisions in different organs) could explain these variations. Correlative analyses of the lifetime risk for developing a certain cancer with the number of stem cells divisions within the same organ implied that replicationrelated mutations are associated with neoplastic development. Mathematical analyses leveraging DNA sequencing and epidemiological data from 69 countries, representing 4.8 billion people, estimated that at least 29% of mutated driver genes are linked to environmental factors, whereas 66% can be attributed to replicative errors. However, it is impossible to completely separate replication-related mutations from exogenous factors. For example, consumption of very hot beverages, which has been linked to oesophageal carcinogenesis, induces severe damage to the cellular lining of the esophagus, which in turn will trigger stem cells located in the deep layers to divide in order to replace the damaged cells. Therefore, divisions of stem cells, caused by an exogenous factor can introduce replication-related mutations. Moreover, the endogenous production of reactive oxygen species (ROS), which can trigger replicative mutations, has been associated with several environmental compounds that act indirectly on DNA. Examples of such compounds are pesticides, mycotoxins, and heavy metals. Numerous epidemiological studies emphasize the contribution of the environment to the cancer burden observed in particular populations. For instance, hepatocellular carcinoma (HCC) shows a high incidence rate in regions with a high risk of exposure to aflatoxin B1 (AFB1) compared to other regions where AFB1 exposure is minimal. In addition, comparing cancer incidence of Japanese subjects residing in Hawaii versus those living in Japan, especially ones living in Okinawa, revealed a dramatic decrease in cancers of the mouth, pharynx, and esophagus in all Japanese migrants. This result suggests that Japanese migrants have escaped exposure to mysterious environmental cancer risk factor(s) particular to the Okinawa region. Multiple epidemiological studies employing different approaches have estimated that extrinsic factors contribute 70%–90% of the risk for cancer development, with the rest being due to intrinsic factors. Thus, current epidemiological estimates indicate that the majority of global cancer risk is preventable. Taking the ongoing discussion regarding the contribution of replication errors into account, a smaller proportion of cancers would be amenable to a reduction of environmental exposures, while the majority would require cancer prevention measures based on early detection and early intervention. Regardless of the cancer prevention approach, uncovering the causes that lead to cancer will allow developing measures that seek to reduce or completely eradicate the exposure factors.

Cancer Etiology and Approaches to Identify the Sources of Somatic Mutations All cancer genomes harbor mutational patterns that reflect tumor heterogeneity and that stem from exposures to multiple mutagenic carcinogens. Some mutagenic carcinogens leave a specific SBS imprint on DNA, exemplified by tobacco smoke carcinogens in lung (viz., C:G > A:T), ultraviolet light in most cancers of the skin (viz., C:G > T:A), and AFB1 in liver cancers from certain regions around the world (viz., C:G > A:T). A further refined classification of these SBS mutations can be applied by taking into consideration the immediate 50 and 30 nucleotide bases flanking the somatic mutation. This elaborated classification makes it possible to discriminate between the mutational patterns of C:G > A:T transversions observed in lung and liver cancers. Currently, the analysis of human cancer mutational spectra offers the possibility to study cancer etiology with much greater resolution. In the next few

Mutational Signatures and the Etiology of Human Cancers

503

sections, we discuss leveraging somatic mutations in single genes as well as somatic mutations across larger parts of the genome for elucidating cancer etiology.

Single gene approaches Previously, mutagenicity and genotoxicity evaluation of compounds relied on simple assays employing prokaryotic systems, such as the Ames test, as well as assays that are laborious, such as the comet and micronucleus assays. However, these assays do not provide insights regarding the specific base changes and their sequence context. Leveraging single-gene sequencing in experimental models and primary human tumors provides an alternative to study the mutagenic processes associated with specific carcinogenic exposures. The experimental systems used for this purpose depend either on a phenotypic selection method (e.g., bacterial reporter genes) or on genes that are frequently mutated in human cancers. Commonly utilized in vitro reporter genes rely on endogenous genes (e.g., HPRT, DHFR, or TK) to convert certain media supplements to toxic metabolites implying the occurrence of genetic changes in the encoded genes. In contrast, animal in vivo model systems include the genomic integration of a transgene consisting of a reporter gene (e.g., lacI, lacZ, gpt, gpa, hprt, aprt, supF, or cII) and a viral shuttle vector. After exposure, the transgene is packaged into phage particles ensuring the efficient delivery of the target gene into a bacterial host. Mutation detection is examined using chromogenic or viability selection. Reporter gene assay allowed the assessment of the mutagenicity of a number of carcinogens, for instance, the heterocyclic amine 2-amino-1-methyl6-phenylimidazo[4,5-b]pyridine, a common dietary carcinogen in cooked meat, which was associated with an increased rate of C:G > A:T transversions. Other examples include the dietary carcinogen acrylamide (viz., T:A > A:T and C:G > G:C), AFB1 (viz., C:G > A:T), and the chemotherapeutic agent 8-methoxypsoralen (viz., T:A > A:T). Sequencing of cancer genes facilitates the identification of driver mutations in a cancer type. This approach also provided the first evidence of the molecular mechanisms by which environmental carcinogens leave characteristic imprints on DNA. Examples of the most frequently mutated genes in human tumors include TP53, KRAS, and BRAF genes. Single cancer gene sequencing in skin and lung tumors identified mutational patterns characteristic of exposures to ultraviolet light (UV-light) and benzo[a]pyrene (BaP), respectively. UV-light induces C:G > T:A transitions at dipyrimidines in skin cancers, and tobacco smoke prompts C:G > A:T transversions in lung tumors (Fig. 3). These tumor-associated mutational patterns are in agreement with results from in vitro controlled experimental exposure studies of UV-light and BaP. Indeed, sequencing of the TP53 gene in cancer patients with different exposures history (i.e., exposed vs. non-exposed) strengthens the link between environmental factors and cancer development. Lung tumors from smokers display a predominant C:G > A:T mutational pattern which is not evident in lung tumors from non-smokers, and the amount of the C:G > A:T imprint correlates with the level of tobacco consumption. Human exposure to the plant carcinogen aristolochic acid (AA) has been linked to the endemic Balkan nephropathy as well as upper tract urothelial carcinoma (UTUC) and cancers of the liver, bladder, and kidneys. Indeed, TP53 mutation screening of these

UV radiation

Smoking A:T>C:G (3.44%) A:T>G:C (11%) A:T>T:A (5.93%) CC tandem (0.09%) complex (0.26%) del (9.02%) G:C>A:T (14.52%) G:C>A:T at CpG (10.57%) G:C>C:G (10.65%) G:C>T:A (32.39%) ins (1.8%) NA (0.09%) tandem (0.26%)

A:T>G:C (3.23%) A:T>T:A (6.45%) CC tandem (29.03%) del (3.23%) G:C>A:T (35.48%) G:C>A:T at CpG (6.45%) G:C>T:A (9.68%) tandem (6.45%)

Aristolochic acid

Aflatoxin B1

A:T>C:G (1.72%) A:T>G:C (8.05%) A:T>T:A (58.05%) del (1.15%) G:C>A:T (10.34%) G:C>A:T at CpG (8.62%) G:C>C:G (3.45%) G:C>T:A (8.62%)

A:T>C:G (3.77%) A:T>G:C (11.32%) A:T>T:A (3.77%) G:C>A:T (7.55%) G:C>C:G (1.89%) G:C>T:A (71.7%)

Fig. 3 Data extracted from the International Agency for Research on Cancer’s TP53 database. Pie chart representations of the proportion of TP53 single base substitutions observed in human cancers of the skin, lung, kidney, and liver. Each of the pie charts corresponds to a specific exposure linked with one or more of these cancer types. The database can be found at: http://p53.iarc.fr/TP53SomaticMutations.aspx.

504

Mutational Signatures and the Etiology of Human Cancers

cancers identified a pronounced T:A > A:T mutational fingerprint associated with the unique mutation pattern of AA observed in laboratory experiments. Finally, the TP53 gene exhibits a unique mutational profile characterized by predominant C:G > A:T transversions in HCC cases from regions where AFB1 exposure is prevalent. Liver cancers from other populations, where AFB1 exposure is minimal and other risk factors prevail, exhibit other distinct TP53 mutational fingerprints. Different in vivo and in vitro experimental systems contributed to the identification of TP53 mutational patterns, representing “rudimentary signatures” of mutagens, and their putative associations to specific cancer types. These models include a genetically engineered mouse system, harboring the human TP53 gene, from which embryonic fibroblasts were derived. These Hupki (Human TP53 knock-in) mice were exposed to UV-light inducing characteristic TP53 gene mutations similar to those predominantly observed in human skin cancer (viz., C:G > T:A). Moreover, Hupki mouse embryonic fibroblasts (Hupki MEFs) were exposed to a number of carcinogens revealing an analogy between the Sanger sequenced TP53 genes from the in vitro assays and human tumors associated with the same exposures. Yeast systems were also exploited for TP53 mutagenesis using a strain transfected with an expression vector harboring human TP53 cDNA that had been in vitro irradiated with UV-light. The results revealed CC:GG> TT:AA mutations which is consistent with the observations from data derived from sequencing UV-light associated skin cancers. Normal human fibroblasts were also treated with other known carcinogens, such as BaP, AFB1, and acetaldehyde. The observed mutational patterns of TP53 were evaluated and they were consistent to the mutational patterns observed in some human cancers. Overall, the experimental identification of carcinogen-specific mutational patterns has demonstrated a convergence between mutational data and epidemiological studies. These mutational data can be effectively utilized for establishing of causal associations between environmental exposures and human cancers. Despite their significant contribution to understanding the sources of somatic mutations in cancer, single gene sequencing studies harbor major limitations. First, many samples from a specific cancer type are needed to accumulate enough alterations to extract a specific mutational profile. This profile is specific for a cancer type and, due to the small numbers of mutations in a gene, one cannot examine individual samples. Second, the patterns of mutations in single genes are usually affected by selection as, in many cases, these genes provide a selective growth advantage. The selection imprint can severely bias the observed mutational spectrum. Third, single genes have specific nucleotide sequences that are not representative of the human genome. The nucleotide structure of genes affects the opportunity for observing a somatic mutation and it further biases the observed mutational spectrum. Advances in NGS and bioinformatics analyses have addressed these challenges and allowed efficient testing of hypotheses regarding putative cancer-risk factors.

Genome-wide mutational patterns and deciphering mutational signatures from human cancers Massively parallel sequencing has revolutionized many aspects of biology, including mutation research, due to high speed sequencing capacities and the reduction in the overall sequencing cost. NGS enables examining the complete genomes of cancer cells at a single-base level, providing an unprecedented resolution of the mutational processes that have molded these genomes through the lineage of the cancer cells. The generation of large mutational datasets required the development of novel mathematical models and computational frameworks that allow meaningful interpretation of these data. Below, we provide a description of a mathematical model that was previously used to identify the signatures of the mutational processes operative in human cancer by examining thousands of cancer genomes. A mutational catalogue of a cancer genome can encompass a diverse set of mutation classes including single base substitutions, insertions/deletions, structural rearrangements, copy number changes, and others (Fig. 1). Each class of mutation can then be further sub-classified. For example, SBSs can be sub-classified according to the six unique types of single base substitutions. For simplicity, the pyrimidine of the Watson–Crick base pair is usually used as a reference for a SBS mutation, thus, a C:G> A:T mutation will be written as a C> A mutations. Using this abbreviate nomenclature the six unique SBS types are: C > A, C > T, C> G, T> A, T> C, and T > G. This classification can be further elaborated to include a variety of mutational features such as the sequence context of the mutated base and the transcriptional strand on which the substitution has arisen. For the purpose of mathematical modeling, a limited number of features of a mutational catalogue need to be selected. The choice of features may be influenced by prior biological knowledge. The choice is also often constrained by statistical considerations and the available data. Mathematically, a set of mutational features can be expressed as a finite alphabet X with K letters, where each letter corresponds to a mutational feature. The simplest alphabet for SBSs, X6 contains K ¼ 6 letters. X6 is based on the six types of single base substitutions and it has six letters: C > A, C> T, C> G, T> A, T > C, and T> G. It should be noted that this alphabet of mutation types could be easily extended by, for example, including other mutation types such as doublet base substitutions. Here, we focus on mathematical alphabets representing SBSs as these alphabets have been the ones most commonly used to examine mutational signatures. The X6 alphabet is perhaps the simplest possible alphabet as it considers only the six types of somatic substitutions. This alphabet is generally not used in any analysis but, rather, its simplicity is leveraged to provide examples and visual representations that clarify mathematical concepts. The X96 alphabet provides a greater resolution for examining the six types of single nucleotide variants (i.e., the X6 alphabet) by including the immediate sequence context of each mutated base. In this alphabet, a mutation type contains a somatic substitution and both the 50 and 30 base immediately next to the somatic mutation. For example, a C > T mutation can be characterized as .TpCpG. > .TpTpG. (mutated base underlined and presented as the pyrimidine partner of the mutated base pair in the

Mutational Signatures and the Etiology of Human Cancers

505

trinucleotide sequence) generating 96 possible mutation typesd(six types of substitutions) * (four possible types of 50 base pairs) * (four possible types of 30 base pairs). The X1536 further extends X96 by including two base pairs 50 and 30 to the mutated base resulting in 1536 possible mutated pentanucleotidesd(six types of substitutions) * (16 possible types of the two immediate 50 base pairs) * (16 possible types of the two immediate 30 base pairs). For example, using the X1536 alphabet, one of the 256 subclasses of a C > T mutation is .ApTpCpGpC. > .ApTpTpGpC. (mutated base underlined and presented as the pyrimidine partner of the mutated base pair in the pentanucleotide sequence). Lastly, X192 elaborates X96 by considering the transcriptional strand on which a substitution resides. In contrast to all previously discussed alphabets, X192 is defined only in the regions of the genome where transcription occurs, which in most analyses has been limited to the genomic footprints of protein coding genes. Thus, the previously defined 96 substitution types are extended to 192 mutation types. For example, C > T mutations at TpCpA are split into two distinct categories: C > T mutations at TpCpA occurring on the untranscribed strand of genes and C > T mutations at TpCpA occurring on the transcribed strand of genes. In general, one would expect that these two numbers are approximately the same unless the mutational processes are influenced by the activity of the transcriptional machinery. This could happen, for example, due to recruitment of the transcription-coupled component of nucleotide excision repair. For example, if a mutational process has a higher number of C > T substitutions on the transcribed strand compared to C> T substitutions on the untranscribed strand (note that a C > T mutation on the untranscribed strand is the same as a G > A mutation on the transcribed strand), this could indicate that the mutations caused by this process are being repaired by transcription-coupled nucleotide excision repair, although other explanations are also possible. A known example of such strand-bias due to an interplay between a DNA damaging process and a repair mechanism is the formation of photodimers due to ultraviolet light exposure that are repaired by transcription-coupled nucleotide excision repair, resulting in a higher number of C> T mutations on the untranscribed strand of UV-light associated cancers. Quantitatively, a mutational catalogue of a cancer genome is a vector, m, containing the number of somatic mutations of a genome, g, defined over a finite alphabet of mutational types X. Mathematically, a mutational catalogue is a morphism between the pre-defined finite alphabet, X, and a set of K nonnegative integers, N0 K, that is, m : X / N0 K. Thus, for a given genome, its mutational catalogue can be expressed as a K-tuple of natural numbers, m ¼ [m1, m2, ., mK]T. Comparing the mutational catalogues of two cancer genomes, m1 ¼ [m11, m12, ., m1K]T and m2 ¼ [m21, m22, ., m2K]T, requires that both mutational catalogues are defined over the same mutational alphabet X. A comparison can be performed either by using a cosine distance to compare the difference between the mutational patterns in two catalogues or by using a Euclidean distance to compare the absolute mutational difference between two catalogues. Different cancer genomes can be exposed to a particular mutational process at different intensities. For example, a mutational process may cause 350 mutations in one cancer genome while causing 350,000 in another. We refer to this number of mutations as a mutational exposure (or simply exposure) of a signature of a mutational process in a cancer genome. Hence, one may say that a mutational process with a signature P has an exposure e,corresponding to the number of mutations caused by this process in a mutational catalogue m of a cancer genome. Multiple mutational processes can be operative in a single cancer genome and each of these processes can have a distinct mutational exposure. Thus, the mutational catalogue of a cancer genome m ¼ [m1, m2, ., mK]T,defined over the mutation alphabet X with K letters, is a superposition of the signatures of the N operative mutational processes Pi ¼ [pi 1, pi 2, ., pi K]T, i ¼ 1.N,each defined over the mutation alphabet X, with their respective exposures ei, i ¼ 1.N. This superposition is never perfect as DNA sequencing, bioinformatics analyses, and other pre- and post-processing of the data causes non-systematic noise, n. Taking into account this noise, the number of the j-the mutational type in an m mutational catalogue can be examined as: mj ¼

N X

j

P i ei þ n j

(1)

i¼1

Note that in this definition, m, e, and n are vectors, while P is expressed as a matrix. Indeed, the signature P1 of a mutational process, defined over an alphabet X with K letters, can also be expressed as a nonnegative K-tuple, P1 ¼ [p11, p12, ., p1K]T, where PK i i i¼1 p1 ¼ 1 and p1 is the probability of the mutational processes P1 to cause the mutation type corresponding to the i-th letter of the alphabet X. As previously described, a set of N mutational signatures can be expressed as a nonnegative mutational signature 3 2 1 1 p1 p2 / p1N1 p1N matrix P ¼ 4 « « / « « 5 with size K  N,where K is the number of mutation types and N is the number of signatures. pK1 pK2 / pKN1 pKN The subscript index indicates the signature, while the superscript index corresponds to the mutation type. The mutational catalogue of a cancer genome, defined over the alphabet of mutation types X, is represented by a morphism m, where m : X / N0 K. For a given genome,i, its mutational catalogue can be expressed as a nonnegative K-tuple, mi ¼ [mi 1, mi 2, ., mi K]T. Hence, the mutational catalogues of G cancer genomes can be expressed as a nonnegative matrix of mutational catalogues 2 1 3 m1 m12 / m1G1 m1G M¼4 « « / « « 5 of size K  G. In this case, the mutational catalogues are the columns of the matrix, where K is the K K K m1 m2 / mG1 mKG

506

Mutational Signatures and the Etiology of Human Cancers

number of mutation types and G is the number of genomes. The subscript index indicates the mutational catalogue while the superscript index corresponds to the mutation type. The exposure to a mutational process with a signature P1 ¼ [p11, p12, ., p1K]T is a scalar describing the number of mutations, e1 ˛ N0, attributed to that signature in a given mutational catalogue. In this notation, the product p12  eg1 is the number of mutations of type corresponding to the second letter of alphabet X caused by the mutational process P1 in a cancer genome with number g. 3 2 1 1 e1 e2 / e1G1 e1G Hence, one can define a set of exposures of G genomes to a set of N processes as a nonnegative matrix E ¼ 4 « « / « « 5 N N eN / eN 1 e2 G1 eG with size N  G. Here, the subscript index indicates the genome while the superscript index corresponds to the signature. In addition to the signatures of the operative mutational processes, the mutational catalogue of a cancer genome also reflects the effect of random error processes, which may occur due to the used experimental approach (e.g., DNA sequencing) and/or bioinformatics methods (e.g., algorithms for identifying somatic mutations from next-generation sequencing data). This noise term n reflects an additive white Gaussian noise that occurs due to non-systematic errors. The noise term is specific to each mutational catalogue and it is defined over the alphabet X of the mutational catalogue, where n : X / RK. Hence, for a set of mutational catalogues 3 2 1 1 n1 n2 / n1G1 n1G of G cancer genomes, the noise term can be expressed as a matrix N ¼ 4 « « / « « 5 of real numbers with size

nK1 nK2 / nKG1 nKG K  G. The subscript index indicates the noise term for the mutational catalogue while the superscript index corresponds to the mutation type. The signatures of N different mutational processes and their respective exposures need to be extracted from the mutational catalogues of M cancer genomes. Using the introduced notations, this could be expressed as: 3 2 1 1 3 2 1 1 3 2 1 1 3 2 1 p1 p2 / p1N1 p1N e1 e2 / e1G1 e1G n1 n2 / n1G1 n1G m1 m12 / m1G1 m1G 4 « (2) « / « « 5¼4 « « / « « 54 « « / « « 5þ4 « « / « « 5 mK1 mK2

/

mKG1 mKG

pK1 pK2

/

pKN1 pKN

N eN 1 e2

N / eN G1 eG

nK1 nK2

/ nKG1 nKG

or one can simplify Eq. (2) using matrix notations as: M¼PEþN

(3)

In practice, one knows only the mutational catalogues in the matrix M as M corresponds to the mutations identified by NGS. To identify the signatures of the operative mutational processes one needs to decompose M and identify P and E such that these matrices best describe the original matrix without over-fitting the data. Finding an optimal solution for Eq. (3) can be achieved by leveraging a blind source separation (BSS) technique. BSS problems involve unscrambling latent (i.e., not observed) signals from a set of mixtures of these signals, without knowing or assuming anything about the mixing process. In the cases of mutational signatures, each cancer genome can be examined as a mixture of distinct mutational processes that have imprinted their mutational signatures on this genome. In principle, BSS algorithms are capable of revealing hidden features and dependencies in large sets of observed data. Then, these features can be used to build a representation of the data that can contribute to understanding of the underlying mechanisms behind these data. The unmixing and reconstruction of the original signals is usually based on some constrained and/or regularized optimization procedure minimizing an objective (i.e., cost) function together with a few imposed constraints, such as: maximum variability, statistical independence, nonnegativity, smoothness, sparsity, simplicity, etc. The choice of the optimization constraints is usually based on a priori knowledge about the examined data, and, hence, constraints are usually specifically tailored for a particular problem and/or a specific mathematical model. Unscrambling mutational signatures from the genomes of cancer cells can be effectively done by leveraging nonnegative matrix factorization (NMF). NMF is a BSS approach that assumes simple nonnegativity of the different signals. In turn, the nonnegativity makes each of the identified latent variable (i.e., mutational signature) a component of the original data, thus, resulting in easily identifiable and interpretable latent variables. Moreover, NMF contains a natural sparsity constraint that allows separating distinct latent variables only when there are sufficient data for such separation. In 2013, a computational framework, based on NMF, was first developed and later used by Alexandrov and colleagues to identify signatures of mutational processes from 7042 cancer patients with 30 different types of human cancer. More than 20 distinct mutational signatures were identified, many of which were attributed to either exogenous or endogenous mutational processes. In the past few years, the map of mutational signatures in human cancer has been further refined and more than 30 distinct mutational signatures have been discovered. The newest set of mutational signatures in human cancer can be found on the COSMIC website: http://cancer.sanger.ac.uk/cosmic/signatures.

Signatures of Mutational Processes With Proposed Etiology The analysis of thousands of cancer genomes has revealed more than 30 signatures of mutational processes in human cancer. From some of these signatures, previous analyses have been able to identify or, at least, to propose a putative etiology. Tobacco-associated

Mutational Signatures and the Etiology of Human Cancers

507

lung cancers are overwhelmed by signature 4 (Fig. 4), which displays a predominant C > A pattern with transcription strand bias indicative of damage to guanine. This signature matches well the in vitro pattern of mutations induced by tobacco smoke carcinogens and, most specifically, the pattern induced by in vitro exposure to BaP. Bladder, kidney, and liver cancers from certain regions around the world exhibit a unique mutational signature known as signature 22 (Fig. 4). Signature 22 is characterized by T> A mutations and it has been ascribed to exposures to AA. Skin-cancers associated with ultraviolet-light harbor large-numbers of mutations that are attributed to signature 7 (Fig. 4). Signature 7 is predominantly characterized by C > T mutations at dipyrimidine nucleotides with a transcriptional strand bias that is consistent with the formation of photodimers. Importantly, the hitherto described mutational signatures (and others) confirmed prior observations derived from single-gene sequencing of TP53 and provided a strong mutational link between a specific cancer type and its main aetiological factor. The mutational catalogue of a cancer is a composite of superimposed mutational signatures left by various mutagenic insults. In HCC, many individual cancer samples exhibit more than five distinct mutational signatures, reflecting the variety of carcinogens that can cause mutations in hepatocytes. Some of the known risk factors attributed to liver cancer are HBV infection, HCV infection, alcohol consumption, and exposure to AFB1. Amongst the identified signatures, signature 16, characterized by T> C transitions at ApTpN sites, is observed exclusively in the majority of liver tumors. Nevertheless, its etiology is still unknown. Signature 24 was uncovered in AFB1-exposed liver cancers and it is characterized by a transcription asymmetry of C > A transversions (Fig. 4). Integrated experimental analysis across human cell lines, animal models, and primary HCC tumors further confirmed the association between signature 24 and exposure to AFB1. Interestingly, endogenous mutational processes constitute almost half of the identified mutational signatures. Signature 1, attributed to the spontaneous deamination of 50 -methylcytosine, is seen in almost all tumors and it contributes up to 80% of mutations in some cancer types, for example, in myeloid leukemia. Together, signatures 1 and 5 has been termed clock-like mutational signatures as the number of mutations attributed to each of these signatures in cancer (and normal somatic) cells are proportionate to the chronological age of the person from which the cancer (or normal cells) were taken. A number of mutational signatures have been attributed to the disruption of processes regulating DNA homeostasis. Signature 9 reflects infidelity of the DNA polymerase eta, while signature 10 is due to a malfunction of the DNA polymerase epsilon. Defects in DNA mismatch repair (MMR) were found to generate at least five distinct mutational signatures depending on the nature of MMR failure: signatures 6, 15, 20, 21, and 26. Failure of DNA double strand repair by homologous recombination also causes a specific mutational signature, termed, signature 3. Signature 30 was attributed to failure of base-excision repair due to defects in the cancer predisposition gene NTHL1. Notably, in more than half of the cancer types analyzed to date, mutational signatures attributed to the APOBEC (apolipoprotein B mRNA editing enzyme, catalytic polypeptide-like) family of deaminase, known as signatures 2 and 13, have been identified. The patterns of the APOBEC signatures have been further supported by extensive experimental work utilizing various model systems. The APOBEC deaminases have been implicated in virus restriction and suppression of retrotransposition. Signatures 2 and 13 have been overwhelmingly found in cervical cancer as well as in cancers of the head and neck, both of which are related to HPV infection. Moreover, signatures 2 and 13 are highly elevated in samples with prior evidence for HPV infection, implying the recurrent activation of APOBEC upon viral infection. In addition to mutational signatures with known etiology, the mechanisms underlying many of the identified mutational signatures still remain mysterious. A comprehensive information of the current set of mutational signatures in human cancer can be found on the COSMIC website.

Orphan Mutational Signatures and Approaches for Understanding Them Genome-wide mutational analysis of thousands of human cancers highlighted a number of mutational signatures attributed to endogenous and exogenous mutagens. In order to reveal the causal factors underlying a mutational signature, the convergence of multiple lines of evidence from different areas of research is required including experimental studies, epidemiology data, and individual patient history including exposomics data. Amongst the 30 distinct mutational signatures thus far identified, 40% remain with unknown etiology reflecting the need for controlled experimental studies. In vivo exposure bioassays as well as in vitro exposure assays are two roads that can lead to a controlled assessment of the genotoxicity, mutagenicity, and carcinogenicity of a compound. Ideally, such exposure studies would use model systems that enable the testing of a large number of compounds within a reasonable timeframe. Normal primary cells, both human and rodent, have a limited lifespan when transferred from their in vivo environment to culture. These cells undergo stress-associated senescence, with a permanent exit from the cell cycle and eventual death of the cell population. Occasionally, one cell may bypass this fate due to genetic changes and resume the cell cycle producing a clone of immortalized cells that replicates indefinitely in vitro. Clonal expansion is a prerequisite property of the system allowing the investigation of the acquired somatic mutations in more or less homogeneous cell populations by deep sequencing analysis. Cellular models suitable for mutational signatures analyses should include a bottleneck step followed by a clonal expansion, thus, mimicking key steps of carcinogenesis: (i) initiation via exposures, (ii) promotion, (iii) and progression. There are two approaches that should be considered for an in vitro system: (i) bypassing a biological barrier, such as crisis or senescence, and emergence of an immortalized clonal population; or (ii) single cell subcloning after exposure for cells for which a selective biological bypass step is not applicable; also known as clonal expansion. Moreover, these experimental models should recapitulate key aspects of human biology (e.g., metabolic activity as well as activity of DNA repair pathways).

508

Mutational Signatures and the Etiology of Human Cancers

Fig. 4 Examples of mutational signatures identified from examining next-generation sequencing data from human cancers. Signature 4 has been overwhelmingly found in lung cancer from smokers and it has been attributed to smoking tobacco cigarettes. Signature 7 generates the majority of mutations in cancers of the skin and it has been attributed to exposure to ultraviolet light. Signature 22 and 24 has been found in cancers of the liver from specific regions around the world and they have been attributed to exposure to aristolochic acid and aflatoxin B1, respectively.

Mutational Signatures and the Etiology of Human Cancers

509

Through extensive laboratory work, multiple mouse-derived and human-derived cells have been employed to experimentally investigate the genome-wide mutagenic effects of candidate cancer-risk factors. First, Hupki MEFs have proven suitable for cell immortalization due to their sufficiently long telomeres. Interestingly, the scope of overlap between the genome-wide mutational patterns observed in human datasets and immortalized MEF cell lines, exposed to a number of strong carcinogens, includes: (i) the SBS mutational pattern based on the six possible types of SBS mutations, (ii) the transcription strand bias of specific SBSs, and (iii) the immediate sequence context of the dominant mutation type. Second, a number of human cell line models have been explored for the analysis of mutational patterns, namely a proximal tubule human kidney cell line, viz., HK-2, as well as HepaRG and HepG2 cells derived from HCC, and the human mammary epithelial cells (HMEC). As for most human model systems, their use requires long-term exposure and the number of compounds that can be tested is a limiting factor. Nevertheless, the use of human cancer target cell models is an elegant strategy to identify the mutational pattern of a carcinogen triggering cancer development in a specific site. Other emerging models such as induced pluripotent stem cells and organoids are versatile systems capable of clonal expansion that can be generated from different cell and tissue types. However, these models lack the biological barrier step of HMEC and Hupki MEFs and it is not clear whether this will influence the imprinted mutational signatures. Further, taking into consideration in vivo carcinogen-animal models may provide new avenues for a better understanding of the molecular alterations observed in human diseases, and thus improve our knowledge on cancer initiation, progression, diagnosis, and treatment. In addition to experimental examination, future epidemiological studies can be used to identify novel mutational signatures as well as to reveal the identity of known mutational signatures with mysterious etiology. Coupling of epidemiology and cancer genomics approaches can provide synergistic results by allowing association of epidemiological features with mutational signatures or other molecular cancer data. Further, cancer risk assessment through the establishment of new epidemiological studies with thorough follow-up of human subjects can enable the exploration of carcinogenic risk associated to specific carcinogens and eventually the categorization of mutational signatures to cancer risk factors.

Prospective Vision Characterization of novel mutational signatures specific to cancer-risk agents may ultimately contribute to the overall interdisciplinary mission of cancer research and provide novel avenues for cancer prevention. Multiple national and international initiatives have started with the goal for better understanding mutational signatures in human cancer. Notably, the Mutographs of Cancer project is investigating whether different mutational signatures can explain geographical differences in cancer incidence across the world. This international project will collect cancer samples and information about cancer-causing exposures from 5000 patients with colorectal, oesophageal, pancreatic, and kidney cancer living in regions of high or low cancer incidence across five continents. The complete DNA of these cancers (and their matched normal tissues) will be sequenced and the mutational signatures will be comparted and linked to cancer risk factors. Further, Mutographs will also identify specific causes of mutational signatures by sequencing the DNA of rodent cancers and cultured human cells experimentally exposed to more than 150 cancer-causing agents, thus assembling a compendium of mutational signatures associated with known causes of cancer. The overall outcome of the Mutographs project will provide a comprehensive understanding of mutational signatures in cancer and may lead to new approaches to prevent cancer as well as provide opportunities for more effective application of existing cancer therapies. Interestingly, in recent years, evidence has emerged linking epigenetics modifications to environmental factors and human malignancies. Cancer genomes frequently undergo epigenetic changes that follow the Darwinian natural selection process and favor the growth of cells with characteristically altered chromatin structure and deregulated gene expression. These changes are brought through epigenetic modifier genes that are frequently mutated in human cancer. Some examples include DNA methyltransferases, DNA demethylases, histone modifiers, ATP-dependent chromatin remodelers, and other. The results of epigenetic deregulation can range from dysregulated expression of individual oncogenes or tumor suppressor genes to large-scale chromatin structure alterations and genomic instability. Well-established cancer-risk agents and lifestyle factors have been studied in terms of epigenome deregulation, improving the understanding of their long-lasting effects on cancer outcome. Tobacco smoking, diet, infections, inflammation and age are known to affect epigenetic states and can play a role in the early onset of cancers through different mechanisms. Smoking, which is the strongest exposure factor causing lung cancer, harbors an epigenetic signature characterized by consistent methylation changes in the Aryl-hydrocarbon receptor repressor gene. Age is the strongest demographic risk factor for cancer and, interestingly, DNA methylation profiles of chronological age established an “epigenetic clock” that can be affected by different external and endogenous factors. Progress in epigenetic research can open the door to a new era where epigenetic signatures can serve as surrogate for diagnostics and risk stratification of cancer in tissues and can provide evidence on the interactive role of epigenetic deregulation in the roadmap between environmental exposures and cancer.

See also: Aflatoxins. Carcinogen-DNA Adducts. Dietary Factors and Cancer. Role of DNA Repair in Carcinogenesis and Cancer Therapeutics.

510

Mutational Signatures and the Etiology of Human Cancers

Further Reading Alexandrov, L.B., Nik-Zainal, S., Wedge, D.C., Campbell, P.J., Stratton, M.R., 2013. Deciphering signatures of mutational processes operative in human Cancer. Cell Reports 3, 246–259. Alexandrov, L.B., Nik-Zainal, S., Wedge, D.C., Aparicio, S.A.J.R., Behjati, S., Biankin, A.V., Bignell, G.R., Bolli, N., Borg, A., Børresen-Dale, A.-L., et al., 2013. Signatures of mutational processes in human cancer. Nature 500, 415–421. COSMIC database of mutational signatures, http://cancer.sanger.ac.uk/cosmic/signatures. Hollstein, M., Alexandrov, L.B., Wild, C.P., Ardin, M., Zavadil, J., 2017. Base changes in tumor DNA have the power to reveal the causes and evolution of cancer. Oncogene 36, 158–167. Zhivagui, M., Korenjak, M., and Zavadil, J. (2017). Modeling mutation spectra of human carcinogens using experimental systems. Basic and Clinical Pharmacology and Toxicology, 121 Suppl 3: 16-22: https://doi.org/10.1111/bcpt.12690, Epub 2017 Feb 3.

Mutations in Chromatin Remodeling Factors Katherine A Giles, UNSW Sydney, Sydney, NSW, Australia Phillippa C Taberlay, University of Tasmania, Hobart, TAS, Australia © 2019 Elsevier Inc. All rights reserved.

Glossary ATPase A motor enzyme that performs a reaction by ATP hydrolysis. Chromatin The multilayered protein-DNA structure that packages DNA inside a nucleus. Chromatin remodeling An ATP-dependent process that mobilizes nucleosomes either by sliding them through DNA or the removal of histone proteins from chromatin. Facilitated by chromatin remodeler complexes. DNA accessibility Refers to how open or relaxed the chromatin structure is. More “open” chromatin has longer linker DNA between nucleosomes or has a higher nucleosome turnover and is permissible for DNA regulatory factors to bind. DNA repair The process by which an error in DNA is corrected. There are several pathways for DNA repair that is dependent on the type of damage. Histone modification The addition of chemical moieties to histone proteins, such as methylation or acetylation. Nucleosome The basic repeating unit of chromatin that consists of an octamer of histone proteins that accommodates 147 bp of DNA in approximately two helical turns. Transcription RNA polymerase transcribes or “copies” the coding strand of the DNA into an RNA molecule.

Introduction DNA does not exist as a linear molecule within a cell; instead it is wrapped around proteins into a dynamic multilayer structure called chromatin. The basic repeating unit of chromatin is the nucleosome, the core of which is comprised of an octamer of histone proteins (two of each histone H2A, H2B, H3 and H4) that accommodates 147 base pairs (bp) of DNA in approximately two helical turns. The H1 linker histone sits outside of the core and influences the orientation of DNA at the entry and exit points of the nucleosome, locking them into position. The length of the linker (inter-nucleosomal) DNA between two nucleosomes varies from
20– 90 bp depending on the degree of chromatin compaction and physiological conditions. A single piece of linker DNA coupled with the nucleosome forms the chromatasome and an array of these structures forms the characteristic “beads on a string” chromatin fiber in the presence of low salt. Chromatin protects DNA from damage and naturally represses gene transcription, unless this state is actively circumvented. Therefore, nucleosomes must be mobilized so that the DNA strand becomes physically accessible to transcription factors, the transcriptional machinery and other chromatin binding factors in order for gene expression to be activated. To do so, two major subclasses of protein complexes are actively recruited to specific genomic regions to control transcription. These are: “chromatin posttranslational modifiers” and “ATP-dependent chromatin remodelers.” The first broad group of chromatin modifiers refers to those that alter histone protein posttranslational modifications (PTMs). These complexes are the epigenomic “writers” and “erasers” that add or remove chemical groups for other complexes to “read” and interpret. PTMs include, but are not limited to: acetylation, methylation, phosphorylation, ubiquitination, sumoylation and polyADP-ribosylation. Most of the currently known modifications occur on the histone tails that project out from the nucleosome core, although the globular domains can also be modified, and contribute to the epigenetic landscape of a genomic locus. These epigenetic histone modifications play a critical role in many DNA-based processes and in chromatin stability. There are four broad classes of proteins that are able to “read” histone modifications including; chromatin architectural proteins, ATP-dependent chromatin remodeling enzymes, chromatin modifiers (as many of the enzymes that “write” and “erase” these modifications also contain domains to “read” them) and adapter proteins that recruit molecular machinery for processes such as transcription. For example, the ATP-dependent chromatin remodeler, CHD1, utilizes its tandem chromodomains to recognize and remodel nucleosomes carrying histone 3 lysine 4 is tri- methylation (H3K4me3), a modification found at active gene promoters. When recruited, CHD1 localizes with transcription elongation factors to positively promote transcription at highly transcribed genes. While histone posttranslational modifications are important in chromatin biology, this chapter will instead focus on the role of ATP-dependent chromatin remodelers. ATP-dependent chromatin remodelers mobilize nucleosomes and alter their physical positions. First, by loosening the DNA around the nucleosome, then by either translocating the nucleosome through the DNA or completely removing the nucleosome from the DNA strand (Fig. 1). Conversely, these complexes can re-position nucleosomes during the process of gene repression and chromatin compaction (Fig. 1). Nucleosome movement is not only essential for transcriptional control, but also for normal cellular processes including cell cycle progression, DNA replication and DNA repair. For example, progression through the cell cycle requires nucleosome compaction for DNA condensation and sister chromatid segregation. During DNA replication, histone cores are loaded onto newly synthesized DNA while the DNA repair process is impaired by the presence of

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65225-X

511

512

Mutations in Chromatin Remodeling Factors

Assembly

Surveillance & Recruitment

x

ATPase

ADP ATP

ADP

DNA translocation

ATP NDR

Sliding

x

x

Insertion

NDR

Ejection

Fig. 1 Mechanisms of ATP-dependent chromatin remodeling for the alteration of DNA accessibility. Utilizing ATP hydrolysis, chromatin remodeler protein complexes act on chromatin to modulate the physical positioning of nucleosomes to allow or inhibit the binding of regulatory factors to DNA. Chromatin remodelers do not recognize a particular consensus sequence and continually “survey” chromatin for suitable substrates. On recruitment, chromatin remodelers may slide, insert of eject nucleosomes from chromatin. Mobilization by sliding nucleosomes alters the position of the nucleosome and simultaneously changes the length of the internucleosomal DNA fragment. This may involve lengthening (depicted by the green arrow), when a nucleosome depleted region (NDR) is formed, or shortening (depicted by the red arrow), causing loss of the detectable NDR. Alternatively, nucleosome positions may be altered by deposition or eviction. In some cases, such as with the INO80-like family, remodelers are able to exchange canonical histones for histone variants, such as dimer exchange of H2A-H2B for H2A.Z-H2B. A generic chromatin remodeling complex (pink) with an ATPase subunit (green) is depicted. The complex may be recruited to active chromatin (unmethylated at CpG dinucleotides (small white circles) and enriched with active histone posttranslational modifications (green circles)), poised chromatin (unmethylated and enriched with permissive histone posttranslational modifications (gray circles)) or silenced chromatin (methylated and enriched with repressive histone posttranslational modifications (red circles) to coordinate sliding, insertion or ejection of nucleosomes (large gray circles). Transcriptional start sites are indicated by arrows, and may be active (green line) or inactive (red cross).

nucleosomes, which must be removed to make DNA accessible to the repair machinery. Chromatin remodelers are also responsible for exchange of canonical histones for variant histones. These histone variants, such as H3.3 and H2A.Z, provide additional resolution at which chromatin structure is controlled by altering the affinity between DNA and the nucleosome core on a finer scale. Hence, chromatin remodelers may be described as the “gate-keepers” of chromatin structure that facilitate or impede access to DNA. There are four structural families of chromatin remodelers: Switch/Sucrose Non-Fermenting (SWI/SNF), Imitation Switch (ISWI), Inositol 80 (INO80) and chromodomain-helicase-DNA binding (CHD) (Fig. 2). Common features of these complexes include: that they have a high affinity for the nucleosome, have an essential catalytic subunit that utilizes ATP hydrolysis to mobilize nucleosomes, a high degree of similarity between their helicase ATPase domain structures (DExx and HELICc), and that they interact with other regulatory factors (both protein and RNA) to facilitate chromatin control. Despite their common properties, each remodeling family has evolved their own unique mechanistic properties and functions (Fig. 3). SWI/SNF ATPases have a bromo-domain

Mutations in Chromatin Remodeling Factors

SWI/SNF BAF

CHD

BCL11A/B

BAF155

OTHER

BAF45A/D

BAF53A BAF170

BAF47

B-Actin

PBAF NBAF EBAF

ARID1A/B

BRD9 SS18 BAF57

BAF60A/B/C

NuRD-Mi-2

OTHER

MTA1/2/3

CHD3/4

x

CHD3/4

BRG1 or BRM

BPTF RBAP48

CHRAC15

CHRAC

INO80-like

BPAP48

SNF2L

OTHER NoRC RSF WICH b-WICH CERF

SNF2L

CHRAC1

RBAP48

p66a/b RBAP46

BCL7A/B/C

x

ISWI

CHD1 CHD5 (NuRD-like) CHD6-9

HDAC1/2

MBD2/3

BRG1 or BRM

NURF

513

INO80 ARP4/5/8

TFPT INO80D INO80B

INO80E RUVBL1/2 MCRS1

NFRKB

YY1

INO80

OTHER Tip60/p400 SRCAP

UCHL5 INO80C

ACF1

SNF2H H2A.Z-H2B (variant dimer)

H2A-H2B (canonical dimer)

SNF2H ACF1

ACF

SNF2H

SNF2H

Fig. 2 The classical organization of ATP-dependent chromatin remodeling families based on the SNF2 ATPase catalytic subunit structure. Typical subunit composition is shown for common SWI/SNF, ISWI, CHD and INO80-like complexes. BAF is a SWI/SNF complex containing either BRG1 or BRM as the ATPase, and is present in most cell types. Other SWI/SNF complexes are present only in certain cell types, such as embryonic BAF (EBAF) in embryonic stem cells. Most SWI/SNF complexes can remodel chromatin ready for transcription activation or repression. NURF and CHRAC are both ISWI complexes with known roles in transcriptional activation. NURF contains the SNF2L ATPase and aberrantly activates oncogenes in cancer cells while CHRAC contains the SNF2H ATPase and is known to activate transcription in both normal and cancer cells. ACF contains the SNF2H ATPase and remodels nucleosomes after DNA damage (orange lightning bolt). The NuRD/Mi-2 complex containing the CHD3/4 ATPase is representative of the CHD family that has roles in remodeling chromatin for transcriptional activation or repression. INO80 from the INO80-like complexes contains INO80 as the ATPase and is activates transcription by nucleosome mobilization and facilitating exchange of the canonical H2A-H2B dimer for the variant H2A.Z-H2B. Nucleosomes, large gray circles; CpG sites, small white circles; active histone posttranslational modifications, green circles; repressive posttranslational histone modifications, red circles; transcriptional start site, arrow.

recognizing acetylated histones; ISWI ATPases have a HAND/SLANT/SLIDE domain that recognizes nucleosomes and internucleosomal DNA; INO80 complexes have a longer peptide chain between their DExx and HELICc domains that has been proposed to accommodate replication forks and Holliday junction DNA structures, and CHD ATPases have tandem chromo-domains recognizing methylated histones (Fig. 3). The inclusion of divergent additional domains within the ATPase subunit, as well as the variety of accessory subunits recruited, suggests that each remodeler complex can uniquely decode chromatin, accounting for their differing roles within the cell. Chromatin remodelers are known to utilize two primary mechanisms to identify a target nucleosome. Firstly, they can continually scan chromatin for an appropriate nucleosome substrate, relying on their recognition of histone PTMs and inter-nucleosomal DNA. Alternatively, a remodeler can be recruited to and interact with a sequence-specific protein or noncoding RNA (ncRNA) species at a particular genomic locus when required. Since ATP-dependent chromatin remodelers act generically in the absence of consensus sequences, these cofactors provide an additional layer of chromatin regulation that ensures remodeler complexes are recruited to the right place at the right time.

514

Mutations in Chromatin Remodeling Factors

SnAC

SWI/SNF

QLQ

HSA

DEXDc

Bromodomain

HELICc AT Hooks HAND

ISWI

DEXDc AutoN

CHD

NegC DEXDc

PHD Chromodomain

INO80-like

HSA

HELICc

DEXDc

SANT

SLIDE

HELICc SANT

SLIDE

HELICc

Fig. 3 Domain structures of the ATPase catalytic subunit of the remodeller families containing their signature motifs. Each chromatin remodeler contains a core SNF2 ATPase domain consisting of an N-terminal DEXDc and C-terminal HELICc domains (yellow/green). The SWI/SNF remodelers also contain a N-terminal helicase-SANT (HSA) domain (blue) for actin binding and a C-terminal bromodomain (gray) for the recognition of acetylated histones. The ISWI family also contain a C-terminal HAND/SANT/SLIDE domain (gray/pink/green) for the recognition of nucleosomes and regulatory domains N-terminal AutoN (gray) and NegC (gray) C-terminal to the ATPase domain. The CHD family each contain tandem chromodomains (gray) and a PHD domain (gray, Class II only) for recognizing methylated histones at the N-terminal, and various other domains depending on whether they fall into subclass I, II and III (not all shown). The ISWI remodelers have a split ATPase with a spacer between the DEXDc and HELICc domains, and a N-terminal HSA domain (blue), similar to the SWI/SNF family.

The outcomes of recent genomic sequencing studies have revealed that chromatin remodelers are frequently mutated in multiple cancer types. Indeed, the frequency of mutation of the SWI/SNF complex alone rivals that of TP53, which is otherwise known as the “guardian of the genome.” There are conflicting reports of chromatin remodelers functioning as both tumor suppressors and oncogenes, highlighting the diverse roles that they are likely to play in tumor biology. Modulating chromatin is critical for cell proliferation and it is unsurprising that the factors involved play an important role in cancer initiation and disease progression. Sequencing of cancer genomes has demonstrated that chromatin organization is directly associated with genome-wide mutation rates, where open chromatin as a result of remodeling is more likely to contain indels and substitutions than closed chromatin. As informative links between chromatin remodelers and their role in tumor development have been established, a desire for further insight into both the normal function of these complexes and the mechanisms by which cancer develops as a consequence of their mutation is rising.

SWI/SNF Of all the remodeling complexes, SWI/SNF has been the most comprehensively studied. It is a large multisubunit complex initially identified in Saccharomyces cerevisiae and highly conserved from yeast to humans. Core subunits of this complex include a catalytic ATPase, either the mutually exclusive Brahma (BRM, encoded by SMARCA2) or Brahma-related gene 1 (BRG1, encoded by SMARCA4), SNF5 (also known as BAF47 and INI1, encoded by SMARCB1), BAF170 (encoded by SMARCC2) and BAF155 (encoded by SMARCC1). Accessory subunits such as the mutually exclusive ARID1A and ARID1B that bind to AT-rich, nonsequence specific DNA are responsible for the targeting, assembly and regulation of SWI/SNF. SWI/SNF mutant cancers demonstrate a role for the complex in DNA repair, proliferation, motility, invasion and metastasis, and genes involved in these processes are generally upregulated in SWI/SNF deficient tumors. Therefore, it is not unexpected that the expression of SWI/SNF subunits and activity of the complex is altered in several cancer types. The SWI/SNF complex has a well-characterized role in the normal DNA repair process; defective DNA repair directly results in genome instability and cancer development. There are several reports demonstrating that loss of functional SWI/SNF leads to an increased susceptibility to DNA damage, most likely mediated through interactions between SWI/SNF and the histone variant, gH2A.X. BRG1 normally promotes double stranded break (DSB) repair by binding to acetylated H3 histones at the site of damage in the presence of gH2A.X; thus, loss of BRG1 slows the repair process leading to inefficient repair. Contrary, loss of other subunits within the SWI/SNF complex does not necessarily affect DSB repair pathways. SNF5 and ARID1A deficient tumors exhibit little genome instability, remain diploid and are characteristically normal on single nucleotide polymorphism (SNP) arrays, again highlighting the multifaceted functions that each subunit can have within this complex. SWI/SNF interacts with and regulates several known cancer associated genes, including genes involved in cell cycle progression (RB, E2Fs, CCND1, MYC, GLI1), nuclear receptor signaling (glucocorticoid (GR), estrogen (ER), androgen (AR), vitamin D (VDR) and retinoic acid (RARA)), and cell-to-cell interaction/adhesion proteins (integrin’s and CD44). Interestingly, SWI/SNF can both upand down- regulate the same gene; the outcome being dependent on cellular context. For example, in response to estrogen, SWI/SNF is bound by ER and recruits p300 to activate ER-responsive genes, while estrogen antagonists, such as the tumor suppressor protein Prohibitin, recruit SWI/SNF and HDAC1 to repress these same genes. Similarly, SWI/SNF has a complex role in regulating MYC oncogene expression. MYC over-expression in a variety of cancers disrupts the expression of genes involved in cell cycle progression, apoptosis and differentiation. SWI/SNF can normally directly repress MYC, but in certain cancers such as leukemia, over expression

Mutations in Chromatin Remodeling Factors

515

of the SWI/SNF complex causes chromatin de-condensation at cell-specific enhancer regulatory elements to increase MYC expression. The following section outlines the roles played by significant subunits of the SWI/SNF complex in cancer.

SNF5 Some of the strongest evidence linking chromatin remodeling and cancer comes from studies of rhabdoid tumors (RT). RTs are an aggressive, soft-tissue childhood cancer with a poor response to therapy and bi-allelic loss of SNF5 is a defining featuredoccurring in 95% of these cancers (Table 1). RTs can derive from both somatic and familial mutations in SMARCB1 (encoding SNF5), which causes gene inactivation by deletion, nonsense, missense and frameshift mutations. Schwannomatosis and occasionally other sarcomas also report a bi-allelic loss of SNF5, and other cancers demonstrating high mutation frequency in SNF5 include colorectal, gastric and bladder (Table 1). The use of knockout mouse models further confirm the functional role of SNF5 in RT initiation; SNF5 knockout mice develop RT with a rapid onset, indicating that SNF5 is likely a driver mutation in these cancers. Loss of SNF5 in RT occurs concomitant with an increase in expression of the Polycomb group (PcG) protein, EZH2. EZH2 is the catalytic subunit of Polycomb repressive complex 2 (PRC2), which contributes to the repression of genes through the trimethylation of histone 3 lysine 27 (H3K27me3). Depletion of EZH2 in SNF5 deficient tumors is associated with a decrease in cellular proliferation, suggesting that EZH2 is required for the growth of these tumors. This is in part due to the role that SNF5 and the PcG proteins play in the regulation of key tumor suppressor genes. For example, SNF5 is required for activation of the INK4b-ARF-INK4a locus and in the absence of SNF5, epigenetic repression of this region is initiated by EZH2 and the PcG proteins. This study also demonstrated that with re-expression of SNF5 into a SNF5-deficient RT human cell line, PcG proteins were evicted from the INK4b-ARF-INK4a locus and that the activity of the SWI/SNF complex was restored. Moreover, the authors noted that the mixed lineage leukemia 1 (MLL1) histone posttranslational modifier is recruited following PcG eviction, in order to restore an active epigenetic signature on histone 3 lysine 4 (H3K4). Precisely how atypical SNF5 expression disrupts the function of SWI/SNF is unclear. It is possible that aberrant SNF5 expression alters the composition of the SWI/SNF complex, driving atypical interactions with other proteins in a context dependent manner to promote tumorigenesis. In SNF5-deficient cell lines, continued growth is reliant on increased expression of BRG1, which could explain why loss of SNF5 in cancers activates gene expression pathways to promote growth whilst in primary cells it triggers cell cycle arrest. Indeed, several cell cycle dependent proteins are consequently disrupted following loss of SNF5. One example is of p16INK4A, a cyclin-dependent kinase (CDK) inhibiting cyclinD1-CDK4 phosphorylation, which the downregulated concomitant with SNF5 inactivation. Conversely, expression of cyclinD1 itself, MYC and E2F target genes are elevated. SNF5 loss has been linked to the disruption of cancer promoting signaling pathways. Hedgehog signaling is essential for the growth of RTs and is up-regulated in SNF5-deficient tumors. RHOA stimulates cell migration, actin stress fiber formation and contractility. When SNF5 is lost, RHOA expression increases which is correlated with poor prognosis in cancer. These data suggest that SNF5 has broad roles in establishing the cancer phenotype, potentially by driving aberrations in histone posttranslational modification marking at gene regulatory regions.

BRG1 and BRM A SWI/SNF complex always contains either BRG1 or BRM as two possible alternative ATPases. While they share approximately 75% sequence homology, BRG1 has a unique N-terminal domain not existent in BRM, through which it can bind to zinc-finger proteins. BRG1 and BRM have partially overlapping functions and therefore may offer some compensation when one of them is lost or disrupted in cancer. However, the ATPases also have clearly defined roles since BRG1, but not BRM, is embryonic lethal. Both BRG1 and BRM suppress cancer development through control of the cell cycle and cell-to-cell adhesion. In cancer cell lines deficient of both functional BRG1 and BRM, re-expression of either ATPase can induce Rb-mediated cell cycle arrest. Similarly, loss of BRG1 and/or BRM in lung cancer drives cancer development and alters cell-to-cell contacts through an increased VIM expression and down-regulation of CDH1, CD44 and CSF-1. Furthermore, BRG1 has a synthetic lethal relationship with BRM and research has shown that cancers depend on BRM for growth and survival. Loss of BRM in these cancers results in cell cycle arrest, senescence and increased tri-methylation of H3K9 (H3K9me3). The cancer-specific differences in SWI/SNF activity may be partially explained by the variance in expression of BRG1 and BRM in tumors, which likely alters feedback loops that balance tumor suppressive and oncogenic activity of SWI/SNF. SMARCA4 (encoding BRG1) is down-regulated in bladder, colon, nontriple negative breast cancers, head and neck, esophageal, melanoma, pancreatic and ovarian cancers; where mutation rates have been reported between 4% and 13% in these cancers (Table 1). Conversely, SMARCA4 is over-expressed in cancers of the prostate, triple negative breast cancers and some leukemias (Table 1). Interestingly, no SMARCA4 mutations have been reported in cancers over-expressing BRG1, suggesting that atypical SMARCA4 expression may have an epigenetic basis, although this has not been experimentally confirmed. The contextdependent function of BRG1 makes it difficult to fully define its role in cancer; however, there is sufficient evidence that the balance between tumor suppressive and oncogenic activity of BRG1 is important. Functional studies have assessed the outcomes of both depletion and restoration of the BRG1 protein. Aberrant BRG1 expression disrupts proliferation, cell motility and epithelial to mesechymal transition (EMT). For example, in combination with lung-specific carcinogens, haploinsufficiency of SMARCA4 increases the likelihood of lung cancer development while conversely, re-expression of BRG1 protein in lung cancer cells reduces proliferation, metastasis and EMT via up-regulation of miR-148b. Mechanistic studies in cell lines demonstrate changes in cell motility occur concomitant with BRG1 depletion and similarly, the loss of BRG1 in a pancreatic cancer cell line stimulates actin stress fibers and motility. Conversely, in a cervical cancer cell line deficient in BRG1, restoration

516

Summary of SWI/SNF subunits with an identified role in cancer as either a tumor suppressor and/or an oncogene

Gene

Protein

Function

Tumor suppressor

SMARCB1

SNF5/INI1/BAF47

Binds to lateral surface of the nucleosome, scaffold

SMARCA4

BRG1

ATPase, mutually exclusive with BRM, interacts with acetylated histones

SMARCA2

BRM

ARID1A

ARID1A/BAF250a

ATPase, mutually exclusive with BRG1, interacts with acetylated histones Nonsequence specific DNA binding, mutually exclusive with ARID1B

ARID1B

ARID1B/BAF250b

SMARCE1 BRD7 ARID2

BAF57 BRD7 ARID2

Nonsequence specific DNA binding, mutually exclusive with ARID1A Interacts with nuclear receptors and DNA, HMG domain Interacts with acetylated histones Role in NER damage response pathway

Bladder, colorectal, gastric, rhabdoid tumors, sarcoma, Schwannomatosis Bladder, breast (except triple negative), cholangiocarcinoma, colorectal, esophagus, gastric, head and neck, lung, medulloblastoma, melanoma, ovarian, pancreatic, uterine Adenoid, bladder, colorectal, gastric, lung, prostate, uterine Bladder, breast, Burkett’s lymphoma, cervical, cholangiocarcinomas, colorectal, endometrial, esophagus, gastric, heptoblastoma, liver, lung, medulloblastoma, melanoma, neuroblastoma, ovarian clear cell, pancreatic, prostate, renal, sarcoma Bladder, breast, cervical, colorectal, esophagus, gastric, melanoma, lung, sarcoma

PBRM1

BAF180

Interacts with acetylated histones

SMARCC2 SMARCC1 SMARCD1/2/3

BAF170 BAF155 BAF60A/B/C

ATRX

ATRX

Role in neural differentiation Interacts with SNF5 through SWIRM domain Role in cellular senescence, lipid, nutrient and energy, metabolism ATRX/DAXX complex, ATPase, DNA damage response, heterochromatin formation, deposits H3.3 into chromatin

Breast, colorectal, gastric, lung, ovarian, prostate Bladder, colorectal, esophagus, gastric, hepatocellular, lung, melanoma, NSCLC, uterine Bladder, breast, cholangiocarcinoma, clear cell renal cell carcinoma, gastric, head and neck Bladder, gastric Colorectal, gastric, lung Colorectal Brain, osteosarcomas

Oncogene

Colorectal, leukemia, prostate, triple negative breast cancer

Breast, endometrial, ovarian, prostate

Mutations in Chromatin Remodeling Factors

Table 1

Mutations in Chromatin Remodeling Factors

517

of BRG1 resulted in reduced actin stress fibers and increased expression of ROCK1, a major driver of cell motility. By contrast, overexpression of BRG1 promotes proliferation and cancer development. Several studies have reported that SMARCA4 over-expression in prostate cancer is a feature correlated with disease progression and invasiveness. In leukemia, BRG1 plays a role in regulating the actin cytoskeleton and proliferation, where loss of BRG1 down-regulates hedgehog signaling, decreases proliferation via downregulation of MYC. Unlike the divergent BRG1 expression profiles observed between cancers and cancer subtypes, the BRM ATPase is typically depleted. This is consistent with the observations that BRG1 has both tumor suppressive and oncogenic roles, while all studies of BRM in tumors suggest that it acts solely as a tumor suppressor. This is supported by the observation that BRM loss correlates with advanced disease in prostate and lung cancers (Table 1). Indeed, there is an anticorrelation between the proliferative markers, ki-67 and E2F1, and BRM in these cancers. In prostate cancer, BRM loss occurs concomitant with an increased rate of EMT and perturbed cell cycle progression. Moreover, decreased BRM expression leads to increased androgen-dependent growth, and homozygous deletion of BRM in prostate epithelial cells causes acquisition of lobe-specific, castration-resistant proliferation. BRM is silenced in 15% of nonsmall cell lung cancer (NSCLC), and lung tumors with loss of BRM have a lepidic growth pattern pathology and EGFR mutations. Two studies conducted in cancer cell lines found no evidence of deleterious BRM mutations, and in both of these studies, BRM expression was restored following treatment with histone deacetylase (HDAC) inhibitors. These data suggest that BRM is epigenetically silenced and is supported further by work showing that certain mutations, BRM-741 (rs34480940) and BRM1321 (rs3832613), in the BRM promoter facilitate the binding of HDACs. Interestingly, tumors with these genetic variants are associated with worse overall survival. BRG1 and BRM expression changes can at least be partially attributed to epigenetic mechanisms, indicating that epigenetic therapies could be effective for cancers deficient in either BRG1 or BRM. Epigenetic therapies to alter BRG1 expression are of particular interest in cancers traditionally treated with the chemotherapeutic agent, cisplatin; cancers with reduced BRG1 are more sensitive to this line of treatment. Work in head and neck cancer cell lines with knockdown of either BRG1 or BRM found the enhanced sensitivity of cancer cells to cisplatin was through the reduced repair of intrastrand adducts and interstrand crosslink formation and cisplatin-induced DSB repair was impaired from ineffective recruitment of ERCC1 for later stage repair. Furthermore, this study demonstrated that altered cell cycle checkpoint activation and enhanced apoptosis in these cancers. In BRM depleted cancers, there is potential use for HDAC inhibitors. Treatment of lung cancer cells with HDAC inhibitors restored BRM expression, but inhibited BRM function. Removal of HDAC inhibitors kept BRM expression elevated for a few days and the complex was functional before BRM levels again were depleted. Future targeting of this pathway must refine the HDAC inhibitors to reduce the impact on BRM function.

ARID1A and ARID1B The AT-rich interactive domain (ARID) subunits, ARID1A and ARID1B, are mutually exclusive within the SWI/SNF complex. They facilitate interaction with DNA that is not sequence specific. While these subunits share approximately 60% sequence homology, they are nonredundant. SWI/SNF complexes containing ARID1A repress MYC, but complexes with ARID1B increase MYC expression. Additionally, as cells progress through the cell cycle the expression of ARID1A decreases and ARID1B increases. ARID1A is one of the most frequently mutated genes in a variety of primary tumor types. Its down-regulation positively correlates with tumor progression, indicating that its main role is that of a tumor suppressor. Loss of function mutations in ARID1A have been identified in bladder, colorectal, breast, Burkitts Lymphoma, cervical, endometrial, gastric, melanoma, medulloblastoma, uterine corpus endometrioid carcinomas, neuroblastoma, cholangiocarcinomas, carcinosarcomas, ovarian clear cell, lung, pancreatic, prostate, renal and heptoblastoma cancers (Table 1). Many of these cancers have heterozygous mutations in ARID1A and express a lower level of the protein. Indeed, it has been observed that loss of ARID1A, BRG1 or SNF5 can occur without complete loss of function in some leukemias. Conversely, in ovarian clear cell carcinomas, one of the most lethal subtypes of ovarian cancer, approximately 50% of cases present with a mutation in ARID1A and the majority of these express a truncated form of the ARID1A protein. Interestingly, ARID1A mutations are more frequent in solid tumors, particularly in the gastrointestinal tract, than hematopoietic cancers. ARID1B mutations have been identified in bladder, breast, colorectal, carcinosarcomas, lung, melanoma, cervical and gastric cancer (Table 1). Loss of ARID1B is detrimental to cells and mutations are known to cause a host of other diseases and syndromes relating to intellectual disability and developmental delays. A recent study has found that ARID1B expression is required in order for ARID1A-deficent tumors to survive. Cancers with a homozygous deletion of ARID1A can tolerate a heterozygous mutation in ARID1B, where only one functional copy is sufficient for tumor survival. The loss of both ARID1A and ARID1B destabilizes SWI/SNF and reduces proliferation of affected cells. This suggests that ARID1B could be a therapeutic target in ARID1A deficient cancers, providing another therapeutic angle through which to target SWI/SNF mutant cancers, and development is underway for small molecule inhibitors of ARID1A and ARID1B.

Other SWI/SNF subunit mutations BAF57 is an accessory subunit of SWI/SNF that is only present in higher eukaryotes. BAF57 plays a direct role in determining how the SWI/SNF complex interacts with chromatin, as it contains both a high mobility group domain and DNA binding properties. There is no evidence supporting a direct role for BAF57 in the initiation of cancer but rather, in cancer progression where increased BAF57 expression correlates with tumor grade, invasion and metastasis. In breast, ovarian, prostate and endometrial carcinoma BAF57 tends to have elevated expression (Table 1); indeed, 38% of endometrial carcinomas are reported to exhibit increased

518

Mutations in Chromatin Remodeling Factors

BAF57 expression. Mutations have only been reported in breast cancer cell lines. As part of its normal and cancer related functions, BAF57 interacts and promotes the activities of GR, AR and ER nuclear hormone receptors. In prostate cancer, BAF57 interacts with AR to increase the expression of a2-integrin, giving those cells a metastatic advantage. In breast cancer, BAF57 expression correlates with expression of PTK2, a key player in migration, motility and adhesion to promote metastasis. Loss of BAF57 decreases the rate of breast cancer metastasis to the lung and furthermore, BAF57 depletion causes a subsequent down-regulation of ER target genes. In ovarian cancer, BAF57 expression strongly correlates with sensitivities to cisplatin, doxorubicin and 5-fluorouracil, and depletion of BAF57 in ovarian cells increases the number of cells arresting in the G1 phase of the cell cycle. BRD7 is a tumor suppressor that is down-regulated in breast, lung, gastric, colorectal, ovarian and prostate cancer (Table 1). While low levels of BRD7 have determined it to function as a tumor suppressor, no significant mutations have been identified. BRD7 has a bromodomain, indicating that it contributes to the localisation of SWI/SNF at loci enriched with acetylated histones. In breast cancer, BRD7 binds to the ESR1 promoter (an ER protein) and while BRD7 depletion does not affect the recruitment of SWI/SNF and RNA polymerase II (RNA PolII) to ESR1, it does reduce ER expression. BRD7 is also a binding partner of BRCA1 and p53, coregulating many of the same genes. In lung cancer, when BRD7 is over-expressed it inhibits proliferation through the downregulation of CCND1 and MYC. It can also down regulate AKT to promote apoptosis. ARID2 is only found in the PBAF form of the SWI/SNF complex, and is mutated in bladder, colorectal, esophageal, gastric, hepatocellular, lung, melanoma, NSCLC and uterine at a rate of approximately 5%–10% (Table 1). It is one of the most frequently homozygous deleted genes in NSCLC and has been identified as a driver mutation in melanoma. Additionally, the four major subtypes of hepatocellular carcinoma all harbor increased ARID2 mutation rates. Little is known about the mechanistic effects of ARID2 mutations or atypical expression of ARID2 in cancer, but it is proposed that ARID2 alterations disrupt the nucleotide excision repair (NER) pathway by inhibiting the recruitment of xeroderma pigmentosum complementation group G (XPG). BAF180 has essential functions with the cell, is important for normal development and is embryonic lethal in mice. It has been suggested that BAF180 provides functional specificity via its six bromodomains that each have different targeting affinities, as well as a high-mobility group domain recognizing nucleosomes. Loss of function mutations in the BAF180 subunit have been identified in clear cell renal cell carcinoma (ccRCC) and breast cancer. It is the second most commonly mutated gene in ccRCC after PTEN, where it is inactivated in 40% of these cancers (Table 1). Loss of BAF180 promotes proliferation while re-expression of BAF180 causes G1 arrest, likely through its ability to inhibit p21 expression. BAF180 also plays a role in regulating p53 activity, suggesting that aberrations in this gene would have a general and widespread role in cancer development.

ISWI ISWI is a large and diverse family of remodeler complexes. Each has specialized functions including DNA repair, replication and transcription. At the core of each ISWI complex is an ATPase: either SNF2H or SNF2L. The SNF2 ATPase forms the ISWI complexes; WSTF-ISWI chromatin remodeling complex (WICH) with WSTF, B-WICH complex containing WSTF and nuclear myosin I, Nucleolar remodeling complex (NoRC) with TIP5, Remodeling and spacing factor (RSF) with RSF, ATP-utilizing chromatin remodeling and assembly factor (ACF) with ACF1 and chromatin assembly complex with ACF1, CHRAC15 and CHARC17. SNF2L forms the ISWI complex Nucleosome Remodeling factor (NURF) with BPTF, RBAP46, and RBAP48 and both SNF2L and SNF2H can form CERC2-containing remodeling factor (CERF) with CECR2. Homologues of ISWI are found in a range of species, and have a highly conserved role in chromatin biology. ISWI proteins are generally tumor suppressors and display loss of function mutations in cancer. However, there are cases of ISWI-associated subunits acting as oncoproteins. ISWI has a well-studied role in DNA repair, forming part of the homologous repair (HR), nonhomologous end joining (NHEJ) and NER pathways.

SNF2H and SNF2L SNF2H mutations have been identified in approximately 5% of colorectal and uterine cancers (Table 2). Alternatively, downregulation of SNF2H in cancer can be mediated by miR-99a and miR-100 species. Loss of functional SNF2H affects the fidelity of DNA repair in cancer. Ordinarily, the ACF and WICH SNF2H complexes are both recruited to the site of DNA damage by transcription factors and histone modifications for the repair process. During DSB repair SIRT6 deacetylates histone 3 lysine 56 (H3K56), and both SIRT6 and SNF2H coordinate to incorporate gH2A.X into chromatin, then the ACF SNF2H remodeling complex relaxes chromatin at the break site. In response to UV damage, ACF and WICH SNF2H complexes respond and require a functional SLIDE domain of SNF2H to proceed with repair. After deposition at the damage site, SNF2H uses its HAND domain to reposition itself away from the center of the damage ready for the next stage of repair. SNF2H also has an important role in chromatin architecture. SNF2H, but not SNF2L is involved in the positioning and phasing of nucleosomes around CTCF binding sites. Defects and delays in DNA repair, along with loss of well-positioned nucleosomes at CTCF sites can lead to genomic instability and cancer progression in cases where SNF2H function is lost. SNF2L is mutated at a higher rate than SNF2H, at around 4%–12% in cholangiocarcinomas, uterine, colorectal and gastric cancers (Table 2). However, there have been conflicting reports regarding differences in expression and function of SNF2L. In melanoma, it was found that SNF2L protein levels were undetectable in malignant cells compared to normal melanocytes. In the HeLa cervical cancer cell line, a functional study has indicated that atypical SNF2L expression causes an increase in cell proliferation and migration, through the activation of Wnt signaling pathways. Conversely, the results of another study that compared multiple cell lines indicated that SNF2L expression was highly similar between normal and cancer cells, but that cancer cells are sensitive to SNF2L knockdown. This same study found that SNF2L loss inhibits growth, down regulates p53 and p21, up regulates cell cycle

Mutations in Chromatin Remodeling Factors Table 2

Summary of ISWI subunits with an identified role in cancer as either a tumor suppressor and/or an oncogene

Gene

Protein

Function

Tumor suppressor

SMARCA5

SNF2H

Colorectal, uterine

SMARCA1

SNF2L

BAZ1A

ACF1

BAZ1B

WSTF

BAZ2A

TIP5

CECR2

CECR2

BPTF

BPTF

RSF1

RSF1

ATPase, nucleosome phasing at CTCF sites, DSB damage response ATPase, regulates transcription, DNA damage response ACF ISWI complex, found at replicating pericentromeric heterochromatin, relaxes chromatin at DSB WICH ISWI complex, regulates genes in EMT and proliferation pathways, involved in UV DNA damage response NoRC ISWI complex, shifts nucleosomes to induce DNA methylation, H4 hypoacetylation and H3K9 methylation for heterochromatin formation CECR2 ISWI complex, inhibits gH2A.X stability, recognizes acetylated histones, histone chaperone NURF ISWI complex, recognizes residues 1–15 of H3K4, regulates HOX gene expression, up regulates EMT and pro survival pathways RSF ISWI complex, histone chaperone, DSB repair

Cholangiocarcinomas, colorectal, gastric, uterine Colorectal, esophageal, gastric, melanoma, renal cell carcinomas, squamous cell carcinoma, uterine

519

Oncogene

Colorectal, lung

Colorectal, lung, melanoma, uterine

Prostate

Melanoma, Oligodendroglioma

Adrenal gland, breast, colorectal, esophagus, liver, lung, ovarian, uterine Heptacellular, lung, NSCLC, melanoma

Bladder, cholangiocarcinoma, colorectal, gastric, lacrimal gland adenoid cystic carcinomas, lung, melanoma, prostate Colorectal, gastric, melanoma

Breast, lung

checkpoint genes, increases the phosphorylation of DNA damage response proteins as well as the level of DNA damage, and increases apoptosis. As SNF2H and SNF2L are generally not part of the same ISWI complex, the sensitivity of cancer cells to SNF2L loss could be attributed to lack of compensation from other chromatin remodeling complexes.

ACF1 ACF1 (ATP-utilizing chromatin assembly and remodeling factor 1, encoded by BAZ1A) is mutated in approximately 5% of all cancers (Table 2). Renal cell carcinomas have diverse copy number variants, many involving ACF1. Additionally, chromosome 14 q12-q13, where ACF1 is located is frequently amplified and over-expressed in esophageal squamous cell carcinoma. Enrichment of ACF1 is found at replicating pericentromeric heterochromatin, which is a barrier to DNA repair and correlates with somatic mutation rates in cancer. ACF1-containing ISWI complexes have a role in DNA repair, by relaxing chromatin at the DSB site. Indeed, the accumulation of DNA damage as a result of ACF1 repression further delays progression through S-phase of the cell cycle.

WSTF WSTF (Williams syndrome transcription factor, encoded by BAZ1B) is partnered with SNF2H, in the WICH ISWI complex and is mutated in 5% of cancers (Table 2). A study examining any protein containing a kinase domain across colorectal cancer development, reported WSTF was significantly increased in these cancers. Moreover, over expression of WSTF in lung cancer promotes several cellular processes essential for cancer development including proliferation, colony formation, migration, invasion and EMT. Specifically, the EMT gene fibronectin and EMT-inducing genes FOS, CEACAM6 and N-cadherin are up-regulated concomitant with increased WSTF. There is also up regulation of the PI3K/AKT pathway. In T-cells, it has been noted that WSTF senses L-arginine levels to promote cellular survival. Elevated L-arginine levels induce a metabolic change, which directly impacts T-cell metabolic fitness and is crucial for their antitumor response. As is observed for many chromatin remodelers, WSTF also has a role in DNA repair. It forms a complex with RUNX2, INTS3 and gH2A.X in response to UV damage and interacts with phosphorylated gH2A.X (pTyr142) to maintain gH2A.X at sites of DNA damage. Following depletion of the oncoprotein, RUNX2, the interaction between WSTF and gH2A.X is lost, while WTSF deletion causes abnormal gH2A.X signals in early pachytene of meiosis.

TIP5 TIP5 (encoded by BAZ2A) is part of the ISWI NoRC complex. This complex shifts nucleosomes to induce DNA methylation, H3K9 methylation, and hypoacetylation of H4. TIP5 has a TAM (TIP5/ARBP/MBD) domain that contains a unique C-terminal extension for binding to RNA. The TAM domain binds to ncRNA, which targets the NoRC complex to the genome by recognition of histone 4 lysine 16 acetylation (H4K16ac). Binding to this mark by NoRC mediates heterochromatin formation at rRNA genes, centromeres and telomere repetitive elements to preserve the structural integrity of chromatin. Loss of a functional NoRC complex causes

520

Mutations in Chromatin Remodeling Factors

centromeric and telomeric heterochromatin relaxation, abnormalities in mitotic spindle assembly, impaired chromosome segregation during mitosis and increased genome instability. Conversely, when critical sites of the TAM domain are mutated, ncRNAs can no longer bind and expression of rRNA genes increase. Silencing of rRNA genes drives cells into senescence, therefore an increase in TIP5 expression may induce senescence in cancer cells where TIP5 has been lost. TIP5 is over-expressed in prostate cancer and a potential marker for disease reoccurrence and metastasis (Table 2). TIP5 directly interacts with the PRC2 protein EZH2 and together they regulate genes involved in metastasis progression. In addition, over-expression is associated with the CpG Island methylator phenotype (CIMP) in prostate cancer.

CECR2 Cat eye syndrome chromosome region candidate 2 (CECR2) forms ISWI complexes with both SNF2H and SNF2L. In the testis, CECR2 predominately associates with SNF2H and in the neuronal cells of an embryo, with SNF2L. It contains a bromodomain, recognizing acetylated histones. CRCR2 has diverse roles in cancer and is mutated in approximately 5% of cases including in oligodendroglioma and melanomas (Table 2). It is up-regulated in cancers of the adrenal gland, breast, colorectal, liver, lung, ovarian, esophagus and the urinary tract. In leukiemas, it is a fusion partner of ASXL proteins. CECR2 is a novel DNA damage repair protein and has strong inhibitory activity on gH2A.X stability, though little else is known about the function of CECR2 in remodeling and cancer development.

BPTF The bromodomain and PHD domain transcription factor (BPTF) is the largest subunit in the NURF ISWI complex. It recognizes residues 1–15 of H3K4, specifically binding when it is tri-methylated at active gene promoters. BPTF has a key role in development, regulating HOX gene expression and development of the mesoderm, endoderm and ectoderm. It is also essential for proper differentiation of mammary stem cells, T-cells and melanocytes. As with CECR2, BPTF has diverse affects in different cancers. BPTF is mutated in almost 30% of lacrimal gland adenoid cystic carcinomas, involved in translocations of both prostate and lung cancer and frame shift mutations in gastric and colorectal cancer (Table 2). BPTF is over-expressed in melanoma, NSCLC, lung adenocarcinomas and HCC, where it is associated with poor prognosis and metastasis. The tumor promoting activity of BPTF has been linked to the up-regulation of EMT and pro-survival pathways, through partnership with c-MYC. It has been demonstrated that depletion of BPTF from T-cells improves antigen processing and tumor antigenicity, while in mammary cells, it results in loss of open chromatin, particularly at enhancers.

INO80-Like The INO80-like family includes the ATPases INO80, SRCAP and p400. These complexes are involved in transcription, but have a more prominent role in DNA repair, DNA replication and exchange of histone variants. This includes the insertion of H3.3 and exchange of the canonical H2AeH2B dimer for H2A.Z-H2B at active promoters, active enhancers and at sites of DNA damage. Interestingly, while all INO80-like complexes can exchange histone variants, only INO80 and not SRCAP and p400, can translocate nucleosomes and only the p400 complex has the ability to acetylate H2A.Z through its subunit Tip60.

INO80 complexes A significant rate of mutations in the INO80 ATPase has not been found in cancer. However, over-expression of INO80 has been reported in melanoma, NSCLC, cervical tumors and cervical cancer cell lines. In melanoma (Table 3), INO80 is required for expression of known oncogenes and tumor growth. In these cancers, INO80 occupies and increases chromatin accessibility at enhancers of tumor development genes. It is also known that recruitment of INO80 to Sox9 and MITF target genes activates genes for melanoma progression and when INO80 is silenced in melanoma cell lines there is an impairment of BRAF- and NRAS-directed tumorigenesis, as well as tumor maintenance. INO80 has a similar role in NSCLC, where it is required to facilitate chromatin de-compaction at enhancers of lung cancer associated genes and therefore, INO80 expression negatively correlates with lung cancer prognosis. In cervical cancer cells, inhibition of INO80 hinders cancer growth by initiating G1/G0 arrest, but does not affect migration and metastasis. INO80 complexes can also contribute to cancer progression through their involvement in DNA repair, telomere stability, replication and transcription. INO80 interacts with YY1 for HR directed repair and is important for the recruitment of 53BP1, but not Rad51. This suggests that INO80 complexes have a role in the early stages of the DNA repair pathway. Deletion of either the INO80 ATPase or accessory subunit, ARP5, hampers the repair of UV-induced DNA lesions, which occurs due to the inability of DNA repair factors to assemble at the site of damage. Loss of the INO80 complex inhibits proliferation in a p21 dependent manner and creates defects in telomere structure. INO80 has a known role in normal cells promoting progression of the replication fork through the chromatin template after stalling. This is mediated by an interaction between INO80 with CDC48, and together they evict RNA pol II from transcribed genes after a collision with the replication fork and mediate its degradation. Aberrant INO80 function preventing replication fork progression has the potential to jeopardize the completion of genomic DNA replication, and a subsequent loss of genome integrity that is likely to contribute to cancer progression.

Mutations in Chromatin Remodeling Factors Table 3

521

Summary of INO80 like complex subunits with an identified role in cancer as either a tumor suppressor and/or an oncogene

Gene

Protein

Function

INO80

INO80

SRCAP

SRCAP

ATPase, loads H2A.Z-H2B dimers onto chromatin, translocates nucleosomes ATPase, loads H2A.Z-H2B dimers onto chromatin

EP400

p400

ERCC2 ACTR5

ERCC2/XPD ARP5

KAT5

TIP60

TRRAP

TRRAP

RUVBL1/2

RUVBL1/2

ATPase, loads H2A.Z-H2B dimers onto chromatin Role in transcription and UV damage and NER Stimulates ATP hydrolysis and nucleosome sliding p400 complex, acetyltransferase of H2A, H2A.Z, H2A.X, H4 p400 complex, required for MYC transformation, reduces proliferation gene, recruits p400 to sites of damage AAAþ ATPase, All INO80 like complexes, regulates TERT and proliferation genes

Tumor suppressor

Bladder, colorectal, gastric, head and neck, lung, melanoma, nasopharyngeal, prostate, uterine Breast, colorectal, gastric, head and neck, lymphoma, prostate Gastric, melanoma, uterine Colorectal, gastric, melanoma Breast, head and neck squamous cell carcinoma, lymphoma, prostate Brain, melanoma

Oncogene Cervical, melanoma, NSCLC

Brain Colorectal, gastric, pancreatic

SRCAP complexes Like INO80, Snf-2-related CREB-binding protein activator protein (SRCAP) complexes have a role in histone variant exchange. SRCAP is a known coactivator of transcription factors that interact with CBP and knockdown of SRCAP reduces the level of H2A.Z deposition within chromosomes. However, as H2A.Z is generally associated with active regions of the genome and decondensed chromatin, it is interesting that SRCAP can be found at both active and inactive promoters. This suggests the SRCAP complex is poised at its target genes for activation upon the correct cellular signaling cascade. Mutations in SRCAP have been identified in bladder, colorectal, gastric, head and neck, lung, melanoma, nasopharyngeal, prostate and uterine cancers ranging from 5% to 10% (Table 3) and colorectal cancers with mutated SRCAP demonstrate a high level of chromosome instability. In prostate cancer, SRCAP is a known coactivator of AR and directly targets PSA (KLK3). When SRCAP is lost in prostate cancer cells there is a subsequent decrease in the amount of H2A.Z enrichment at the KLK3 promoter, a decrease in PSA gene expression and inhibited cell growth. As with INO80 complexes, SRCAP can also contribute to cancer through the DNA damage pathway. SRCAP facilitates repair by exchanging H2A.Z-H2B dimers and is also important for the recruitment of RPA and Rad51 to break sites for DSB repair. In colorectal cancer, mutated SRCAP confers resistance to DNA damage and increases genomic instability.

p400 Complexes p400 is a large complex with tumor suppressor activity in both primary human tumors and mouse cancer models. Mutations have been reported in head and neck, breast, prostate, colon and gastric cancers as well as lymphoma. Expression of the p400 complex is significantly down-regulated in breast cancer and in 61% of gastric cancers (Table 3), displaying a high correlation with invasion and metastasis in these cancers. p400 has essential roles in transcription and DNA damage. It plays an important role in ER signaling in breast cancer, where p400 loads histone variant H2A.Z at ER responsive promoters. Inhibition of p400 and H2A.Z slows growth of breast cancer cells, and conversely over-expression of p400 increases proliferation. p400 is required for efficient DNA repair where its recruitment to sites of DNA damage is independent of ATM signaling. At DSBs p400 is required for RNF8-directed ubiquinitation of chromatin and the recruitment of repair factors BRCA1, 53BP1 and Rad51. TIP60 is a lysine acetyl transferase (KAT), also known as KAT5, of the MYST family of acetyltransferases. TIP60 is only present in the p400 complex and not in other INO80-like complexes; its role is to acetylate H2A.Z, H2A.X, H2A and H4. TIP60 is frequently mutated in breast cancer, head and neck squamous cell carcinoma, lymphoma and prostate cancer (Table 3). Free H2A.Z is known to stably associate with TIP60 in the nucleus; H2A.ZK5ac strengthens this interaction and assists p400-driven deposition of H2A.ZH2B dimers into chromatin. TIP60 is an essential part of the DNA damage response and is required to acetylate histone proteins at the site of damage, which facilitates the chromatin opening. Depletion of TIP60 reduces the rate and efficiency of DNA repair causing an accumulation of DSBs and widespread genome instability. After damage is repaired, TIP60 is required to acetylate H2A.X, which signals for H2A.X’s removal from chromatin. TIP60 also has roles in transcription activation. It is a coactivator of AR and prostate cancer biopsies show TIP60 accumulates in the nucleus compared to normal prostate cells. In both xenografts and LNCaP (Lymph Node Cancer of the Prostate) cells withdrawal of androgens increases the levels of TIP60. Accessory subunit of p400, Transformation/Transcription domain associated protein (TRRAP), has been reported as both a tumor suppressor and oncogene in brain cancer (Table 3). Loss of TRRAP reduces proliferation and increases sensitivity to apoptotic stimuli in brain tumors; however, TRRAP also interacts with MYC and E2F oncoproteins and is essential for MYC induced transformation. TRRAP helps recruit the p400 complex to sites of DNA damage, where TIP60 then normally acetylates histones, and

522

Mutations in Chromatin Remodeling Factors

allows p400 to remodel chromatin into an open configuration allowing entry of repair machinery. TRRAP is also mutated in 4% of melanomas (Table 3).

RUVBL1 and RUVBL2 RUVBL1 and RUVBL2, also known as TIP48 and TIP49, are present in each of the INO80-like complexes. RUVBL1/2 are AAA þ ATPases with a hexamer ring structure and together they form a heterodimer. Depleting both these proteins in human cells and also in mouse models has shown they are essential for cellular proliferation. Studies have shown that cells in which RUVBL2 is knocked down are moderately more sensitive to anticancer therapies including cisplatin and 5-Aza-20 -deoxycytidine, and there is reduced cell growth after treatment. Driver mutation events have to date not been identified in RUVBL1/2, but they are overexpressed in colorectal, gastric and pancreatic cancers contributing to cancer growth through a number of mechanisms (Table 3). Pancreatic ductal adenocarcinoma (PDAC) is known as an extremely invasive cancer that will metastasise early. Loss of RUVBL1 in PDAC reduces actin polymerization and the concentration of globular actin in the cell protrusions of migrating cells; reducing their invasiveness and motility. In NSCLC cells loss of RUVLB1/2 reduces growth and causes G2/M arrest, but there is no increase in apoptosis. There is also a role for RUVB1/2 in chromosome stability, where loss of either protein results in chromosome alignment and segregation defects, and disruption to the mitotic spindle assembly. In colon cancer, RUVBL2 regulates human telomerase reverse transcriptase (hTERT) and the increased expression of both of these is associated with advanced nodel disease. The RUVBL1/2 proteins also associate with several oncogenic factors such as c-Myc and E2F1, and several proteins in the DNA damage repair pathway to drive cancer progression.

CHD The ATPases of this remodeling family are highly diverse, with three subfamilies found in humans. Class I contains CHD1 and CHD2, Class II is CHD3 and CHD4, which form part of the NuRD complex and CHD5, and Class III comprises of CHD6CHD9. All CHD ATPase remodelers contain a SNF2-like ATPase domain and tandem chromodomains, but vary in their other protein domains. CHD1 and CHD2 have a DNA binding domain located towards the c-terminal; CHD3, CHD4 and CHD5 have PHD Zinc-finger-like domains, and CHD5 also contains a DEAH-box domain; CHD6 has a Pdxk domain with pyridoxal kinase function and CHD7-CHD9 have BRK domains with unknown function. As more is known about the precise role of the NuRD complex in cancer, this section will first discuss the NuRD complex and key subunits within it, followed by the remaining CHD ATPases.

NuRD complex The Nucleosome Remodeling Deacetylase (NuRD, also known as Mi-2) complex is highly conserved from plants to animals and was first described for its role in gene repression. It is now known that it’s roles in chromatin biology extend to transcriptional activation, DNA repair, replication and genome stabilitydall processes that must be tightly regulated to ensure proper cell function and to inhibit cancer progression. NuRD has more specific roles in development including hematopoietic stem cell maintenance and differentiation, controlling key genes in B- and T-cell development. The NuRD complex contains two catalytic subunits; an ATPase that is either one of the mutually exclusive CHD3 or CHD4, as well as either one of histone deacetylase 1 (HDAC1) or 2 (HDAC2). Other key subunits include the methyl-CpG-binding domain 2 (MBD2) and 3 (MBD3), metastasis-associated gene 1 (MTA1), 2 (MTA2) and 3 (MTA3) and retinoblastoma-binding protein 4 (RBBP4) and 7 (RBBP7). Like SWI/SNF, the NuRD complex both promotes and represses tumor development, which is dependent on subunit composition and cell type. In addition to mutations, many subunits of this complex can contain PTMs and perturbed remodeling function may arise from disruption of this. Several lines of evidence have demonstrated that the NuRD complex associates with oncogenic transcription factors to promote the repression of tumor suppressors, as well as interacting with known fusion genes. Promyelocytic leukemia (PML)dretinoic acid receptor-a (RARa) is a fusion oncoprotein transcription factor that recruits the NuRD complex to its target loci through protein-protein interactions in order to repress tumor suppressors such as retinoic acid receptor b2. As stated above, in other cases the NuRD complex itself directly functions as a tumor suppressor. The NuRD complex has an essential role in proliferation and is implicated in the regulation of both the G1/S and G2/M checkpoints. Knockdown of either CHD4 or MTA1 blocks G1/S transition and G2/M progression is inhibited by unrepaired DNA damage following CHD4 loss. NuRD can also prevent deacetylation of p53, which results in increased p21 levels and cell cycle arrest. In lymphocytes, NuRD foci have been observed at the pericentromeric heterochromatin of chromosomes 1, 9 and 16 during Sphase. The NuRD foci correlate with proteins associated with active replication forks such as PCNA and CAF1, suggesting a role for NuRD in assembly, stability and chromatin condensation at these regions. Interestingly, PcG complexes are usually observed at these pericentromeric heterochromatin regions, but are absent in lymphocytes with NuRD foci, suggesting a unique cell specific role for the NuRD complex.

CHD3 and CHD4 CHD3 and CHD4 are highly similar in structure and are the alternative ATPase subunits of the NuRD complex. They are both known to interact with transcription factors and cofactors in order to direct the NuRD complex to its target sites. The PHD domains of CHD3/4 recognize the tails of H3, enabling transcriptional regulation; interestingly, while H3K9me3 enhances the binding of these

Mutations in Chromatin Remodeling Factors

523

remodelers to chromatin, H3K4me3 abolishes this binding. CHD4 complexes that contain HDAC1 are shown to mediate transcriptional repression via colocalisation with DNA methyltransferase 1 (DNMT1) at tumor suppressor gene promoters. CHD3 and CHD4 are known to interact with several regulatory factors to coordinate transcriptional control in cancer. ZIP, a transcriptional corepressor for cell proliferation, survival and migration pathways, preferentially interacts with CHD3/4 from the NuRD complex in breast cancer and when ZIP is lost there is aggressive tumor growth in mouse xenografts. A similar role has been reported for CHD3/4 binding to ZGPAT in breast cancer, which also represses genes for cellular proliferation, survival and migration and when interacting with NGFI-A Binding Protein 2 (NAB2), CHD3/4 repress early growth response 1 (EGR1) activities. Although they share highly similar functions for transcriptional control, CHD3 and CHD4 have differing roles in DNA damage. At sites of DSBs, KAP1 phosphorylation disrupts the interaction with and releases CHD3 from chromatin, while CHD4 is rapidly recruitment to site of damage. There are limited studies on the effects of CHD3 mutations in cancer. However, there are reports of mutations in colorectal, gastric and prostate cancers (Table 4). In ovarian cancer, CHD3 is silenced due to DNA hypermethylation of its promoter, which is correlated with tumor resistance to platinum-based therapies, more invasive tumors and an EMT phenotype. It has been proposed that CHD3 would be a useful prognostic marker for ovarian cancers, but it remains unclear whether this is likely to extend to other types of cancer.

Table 4

Summary of CHD and NuRD complex subunits with an identified role in cancer as either a tumor suppressor and/or an oncogene

Gene

Protein

Function

Tumor suppressor

CHD1 CHD2

CHD1 CHD2

ATPase, promotes HR over NHEJ ATPase, deposits H3.3 onto chromatin

CHD3

CHD3

CHD4

CHD4

CHD5

CHD5

NuRD complex, ATPase, H3 tail recognition NuRD complex, ATPase, H3 tail recognition, DNA damage response ATPase, binds unmodified H3 tails, forms NuRD like complex

CHD6

CHD6

ATPase, transcription activation

CHD7

CHD7

ATPase, transcription activation

CHD8

CHD8

CHD9

CHD9

ATPase, establishes heterochromatin to euchromatin boundaries ATPase

MBD2

MBD2

Colorectal, gastric, prostate Bladder, chronic lymphoblastic leukemia’s, colorectal, gastric, monoclonal B lymphocytosis, uterine Bladder, colorectal, gastric, lung, melanoma, ovarian, prostate, uterine Bladder, colorectal, endometrial, gastric, lung, uterine Breast, colorectal, gallbladder, gastric, glioblastoma, head and neck, heptacellular, laryngeal squamous cell carcinoma, leukemia, lung, melanoma, neuroblastoma, ovarian, pancreatic, renal cell carcinoma Colorectal, gastric, lung, melanoma, prostate, transitional cell bladder, prostate, uterine Colorectal, gastric, lung, medulloblastoma, melanoma Breast, colorectal, esophagus, gastric, lung, prostate, uterine Colorectal, gastric, lung, melanoma, neuroblastoma, uterine Gastric, colorectal, prostate

MBD3

MBD3

MTA1

MTA1

MTA2

MTA2

MTA3

MTA3

HDAC1/2

HDAC1/2

RBBP4

RBBP4

RBBP4

RBBP4

NuRD complex, binds methylated DNA, mutually exclusive with MBD3 NuRD complex, binds unmethylated DNA, mutually exclusive with MBD2 NuRD complex, mutually exclusive with MTA2 and MTA3, binds transcription factors NuRD complex, mutually exclusive with MTA1 and MTA3, binds transcription factors NuRD complex, mutually exclusive with MTA1 and MTA2, binds transcription factors NuRD complex, Histone deacetylase NuRD complex, provides structural support NuRD complex, provides structural support

Colorectal, endometrial, gastric, pancreatic

Oncogene

Breast, pancreatic Endometrial, ovarian, prostate

Brain, breast, colorectal Brain Breast, endometrial, esophageal, hepatocellular, lung, lymphoma, ovarian, pancreatic, prostate Breast

B-cell lymphomas, breast, ovarian, prostate

Lung, prostate Cholangiocarcinomas

Breast, cervical, colorectal, endometrial, gastric, hepatocellular, lung, pancreatic, prostate, thyroid

524

Mutations in Chromatin Remodeling Factors

CHD4 mutations have been found in a variety of cancers. It is mutated in 17% of endometrial cancer and 23.7% uterine carcinomas with these mutations predicted to impair function (Table 4). Loss of CHD4 also occurs in gastric and colorectal cancers that have microsatellite instability. The contribution of CHD4 to cancer progression is intriguingly diverse. Following DSB breaks, CHD4 is rapidly recruited to sites of damage, and its activity is dependent on binding to the PARP complex. Phosphorylation of CHD4 is then catalyzed by ATM enabling CHD4 to interact with RNF8 to mediated chromatin unfolding. Mutations in CHD4 at S1349dthe site of phosphorylationdreduces its association with chromatin at break sites. Similarly, depletion of CHD4 results in hypersensitivity to DNA damage and accumulation of unrepaired breaks. CHD4 is also important for the cell cycle, where its loss results in cell cycle delay through the G2/M checkpoint.

MBD2 and MBD3 MBD proteins recognize methylated or unmethylated DNA and provide a mechanism for crosstalk to occur between the DNA sequence and chromatin. DNA methylation primarily follows a CpG context and can occur either in response to chromatin compaction or provide the signal that chromatin compaction is required. CpG sites are not distributed evenly throughout the genome; instead, they tend to be clustered into regions of high CpG density, called CpG Islands. CpG Islands are generally found at the promoters of ubiquitously expressed genes and are unmethylated in normal somatic cells. DNA methylation is frequently and ubiquitously disrupted in cancers and is characterized by genome-wide DNA hypomethylation, and site-specific DNA hypermethylation. Promoter DNA hypermethylation, particularly at CpG Islands, is associated with tumor suppressor transcriptional silencing. MBD2 and MBD3 are mutually exclusive subunits of the NuRD complex. They share 71% sequence homology; however, MBD2 is capable of binding to methylated DNA, while MBD3 is not. This difference is due to two amino acid changes in the methylbinding domain; K30 to H30, and Y34 to F34. Interestingly, MBD2 knockout causes only mild defects in cells, while MBD3 loss is embryonically lethal. Mutations in these proteins have been found in neurological disorders and tend to be truncating missense mutations. MBD2 has a prominent role in suppressing transcription and over-expression induces gene silencing across the genome. Overlapping with the methyl-binding domain (MBD) is a transcriptional repressor domain (TRD)dthat is absent from MBD3dsuggesting that the methyl- binding and repressive functions of MBD2 are tightly connected. MBD2 is up-regulated in colorectal cancer cell lines, breast cancer and in brain cancer, but down-regulated in prostate and gastrointestinal cancers (Table 4). MBD2 contributes to tumor development by silencing the tumor suppressor genes: GSTP1 in prostate cancer, CDKN2A in colorectal cancer, and hTERT in the HeLa cell line by promoter DNA hypermethylation. It has been demonstrated in cell lines that the knockdown of MBD2 allows for the re-expression of tumor suppressors p16INK4A and p14ARF. Furthermore, MBD2 depletion within the NuRD complex increases chromatin accessibility. MBD3 is normally enriched at unmethylated promoters and enhancers, containing the active marks H3K4me3 and histone 3 lysine 27 acetylation (H3K27ac) respectively. It is also reported to bind to 5-hydroxymethylated regions in brain tissue and in ES cells. MBD3 has inactivating mutations in cancer and is down-regulated in gastrointestinal, endometrial and pancreatic cancer. However, there are reports that it is up-regulated in brain tumors. MBD3 can function as a tumor suppressor, demonstrated by it binding to the un-phosphorylated JUN oncoprotein to repress JUN’s transcriptional activity. When MBD3 is inactivated in mice, the expression of JUN target genes increases, leading to hyper-proliferation and susceptibility to tumor development. Functional studies to understand the oncogenic roles of MBD3 found it recruits HDACs to silence tumor suppressor promoters including CDKN1A, and interactions with the proto-oncogene, FBI-1 in transformed cell lines.

MTA1, MTA2 and MTA3 The functions of the MTA subunits are very well studied in cancer development. Each subunit is associated with transcription factor binding but mutually exclusive within the NuRD complex. Within the MTA family there are three alternate subunits: MTA1, MTA2 and MTA3. MTA1 and MTA2 promote cancer development, while MTA3 is a tumor suppressor (Table 4). Increased MTA1 levels have been detected in cancers of breast, esophageal, endometrial, pancreatic, ovarian, lunge, prostate, liver and in lymphoma and is positively correlated with higher tumor grade, invasion and poor prognosis. This is in part due to activation of MTA1 by the MYC oncoprotein, which requires MTA1 for transformation of cells. In breast cancer, up-regulation of the HER2 pathway increases MTA1 expression, which then interacts with ER to suppress transcription, including the BRCA1 gene. MTA1 has also been linked to breast cancer through tumor hypoxia, since hypoxia induces MTA1 expression to turn on hypoxiainducible factor 1a (HIF1a) promoting angiogenesis. MTA1 can also inactivate p53 inhibiting growth arrest and apoptosis. Like MTA1, MTA2 is known to promote breast cancer development (Table 4). In these cancers, MTA2-containing NuRD complexes associate with TWIST, a key player in EMT and metastasis, and it is this association that causes subsequent repression of CDH1, promoting EMT. MTA2 also interacts with the ER receptor and MYC to promote breast cancer development. MTA3 has opposing roles to both MTA1 and MTA2, acting as a tumor suppressor. Mutations have been reported in breast, ovarian and prostate cancers. MTA3 inhibits EMT through the suppression of SNAIL, a factor known to decrease cell adhesion and increase motility. MTA3 is also required for BCL-6 repression of genes involved in plasma cell differentiation in B-cell lymphomas.

HDAC1 and HDAC2 HDAC1 and HDAC2 are the only two histone deacetylases to form a part of the NuRD complex and share 80% sequence homology. As they can also form part of the CoREST and SIN3 complexes that are both associated with transcriptional repression, it can make

Mutations in Chromatin Remodeling Factors

525

their role within the NuRD complex difficult to determine. It is unknown whether NuRD complexes containing HDAC1 or HDAC2 work synergistically or on different regions of the genome. However, it is clear that they have differing roles in development where HDAC1 is essential for stem cell survival and is embryonically lethal, while HDAC2 is not. HDAC1/2 have a low frequency of mutations, but have a tendency to be over-expressed in cancers of the colon, breast, prostate, thyroid, cervical, gastric, hepatocellular carcinoma, pancreatic, endometrial and lung cancers (Table 4). A negative correlation with overall survival has been reported in colorectal cancer, hepatocellular carcinoma, pancreatic, endometrial, lung cancer. Additionally, an association with advanced disease, metastasis and aggressive disease has been found in hepatocellular carcinoma, colorectal cancer, prostate, NSCLC, breast cancer. Furthermore, HDAC2 truncating mutations have been reported in cancers with microsatellite instability, with these cells more resistant to the antiproliferative and pro-apoptotic effects of HDAC inhibitors. As with most chromatin remodeling complexes, HDAC1/2 has roles in transcription and promotion of cancer development pathways. HDAC1/2 over-expression results in increased cell proliferation, migration, angiogenesis and invasion, and decreased apoptosis. HDAC1 is known to promote proliferation of breast cancers by repressing ER transcription. Loss of HDAC1/2 causes arrest in G1 or in G2/M and reduces proliferation in breast cancer.

RBBP4 and RBBP7 The RBBP proteins directly associate with histone tails. Like HDACs, they can also be subunits within other complexes making their role complicated to understand. It is assumed that these subunits provide structural support and promote protein-protein interactions within the NuRD complex. RBBP4 is mutated in approximately 5% of lung and prostate cancers, and RBBP7 is mutated in 12.5% of cholangiocarcinomas (Table 4). Functional studies in LNCaP prostate cancer cells have shown that over-expression of either RBBP4 or RBBP7 results in reduced proliferation.

CHD5 More recently, CHD5dalso a Class II CHD ATPasedhas been demonstrated to form NuRD-like complexes containing each of the main subunits detailed above. CHD5 preferentially binds to unmodified H3K4 tails via it’s dual PHD domains, but can also bind H3K27me3 through its tandem chromodomains. CHD5 is predominately expressed in neurons and the testis; however, low expression has been associated with poorer survival and advanced tumors in a variety of cancer tissue types. Mutations in key residues of CHD5 prevent it binding to H3, and in turn, prevent it from inhibiting tumor growth. Several tumors including breast, lung, colorectal cancer, ovarian, gastric, laryngeal squamous cell carcinoma, gall bladder, hepatocellular carcinoma, pancreatic, leukemia and renal cell carcinoma (Table 4), have demonstrated significant hypermethylation at the CHD5 promoter, concomitant with a decrease in expression. Several of these studies demonstrate that CHD5 expression can be restored with 5-Aza-20 -deoxycytidine treatment, which reduced cellular proliferation. Deletion of the 1p36 chromosome locus containing CHD5 has also been found in glioblastoma, neuroblastoma, renal cell carcinoma and ovarian cancer. CHD5 suppresses genes involved in motility, proliferation, apoptosis, EMT markers and epigenetic regulators. CHD5 also has a role in the DNA damage response with an increase in gH2A.X associated with CHD5 down-regulation in pancreatic cancer. CHD5 controls proliferation through activation of CDKN2A to mediate Rb and p53 pathways.

Class I and class III CHD ATPases CHD1 is a tumor suppressor and plays a significant role in the development of prostate cancer. CHD1 is located at 5q21, and deletions within this region are one of the most frequent to occur in prostate cancer, second only to PTEN. Approximately 17% of prostate cancers contain a CHD1 mutation or deletion (Table 4). CHD1 has a synthetic-lethal relationship with PTEN, where in PTEN-deficient prostate cancer, CHD1 is stabilized. Loss of CHD1 in these cancers reduces proliferation and survival of cells supporting the hypothesis that CHD1 is important in the early stages of prostate cancer development. Deletion of CHD1 in mouse prostate epithelial cells results in morphological changes and increased invasiveness, but not transformation. CHD1 loss has also been linked to PSA failure and ERG-fusion absence and is required for androgen receptor (AR) transcription and regulates tumor suppressor genes including NKX3–1, FOXO1 and PPARg. Prostate cancer patients with CHD1 depletion are more sensitive to PARP inhibitors that target tumors with HR DSB defects. It has been suggested that CHD1 is a prognostic marker for stratification of prostate cancer patients for PARP inhibitor therapeutics. CHD1 is also mutated in 5%–10% of cancers of the digestive tract. (Table 4). CHD1 deletion affects HR DSB repair but not NHEJ. CHD1 is required to open chromatin around the break site, in order to facilitate HR repair protein entry. When CHD1 is depleted in cancer, the tumors switch from utilizing HR to NHEJ mechanisms, leading to increased error prone repair. CHD1 depleted cancers tend to have increased genome instability, concomitant with increased chromosome deletions, particularly on chromosomes 2q, 5q and 6q. Like CHD1, CHD2 is also a tumor suppressor, but less information about its role in cancer exists. Mutations in CHD2 have been reported in cancers of the digestive tract ranging from 5% to 10%, 5.3% in chronic lymphoblastic leukemia’s (CLL) and 7% of monoclonal B lymphocytosis that can evolve to CLL (Table 4). The majority of known mutations are truncating, or occurring in a functional domain. Functional studies following CHD2 depletion in mice have shown heterozygous mutations increase their susceptibility to lymphoma due to the abnormal differentiation of stem cells. Loss of CHD2 also decreases the level of H3.3 deposition in chromatin at promoters and at sites of DNA damage. Cells containing CHD2 mutations also have a higher level of gH2A.X remaining after damage and an aberrant DNA damage response after x-ray radiation. Furthermore, loss of functional CHD2 results in growth and viability defects.

526

Mutations in Chromatin Remodeling Factors

CHD6 was discovered in 2002 and studies of the role of CHD6 in cancer are also limited. Mutations have been reported in 19.5% of colorectal cancers and 7% in transitional cell bladder cancer (Table 4). CHD6 has a high affinity for short linker DNA, down to 20 bp and disrupts nucleosomes in a nonsliding mechanism. CHD6 colocalizes with hypo- and hyper- phosphorylated forms of RNA PolII and is present at the sites of RNA synthesis, suggesting that it positively regulates transcription. CHD7 has been reported as both a tumor suppressor and an oncogene. CHD7 is enriched at sites marked by H3K4me1/2, located near the TSS and near DNase hypersensitive sites, suggesting a pro-transcriptional role. Mutations have been reported in 11.1% of gastric cancers and 42% of colorectal cancer cases (Table 4). CHD7 mutations have been linked with lung cancer in heavy smokers and a number of lung cancer cell lines display PVT1-CHD7 fusion genes. CHD7 has been described as an oncogene and associated with more aggressive subtypes in breast cancer. Deletion of CHD7 in breast cancer cell lines inhibits proliferation and is associated with down-regulation of known oncogenes, such as NRAS. In pancreatic cancer, patients with low CHD7 expression are more sensitized to gemcitabine, the primary treatment for these cancers, and have increased overall survival. CHD7 could therefore be used to stratify patients predicted to have a positive response to gemcitabine in pancreatic cancer. CHD8 plays a role in both transcription and at the boundaries between heterochromatin and euchromatin. CHD8 binds to active promoters in the G1/S transition and is required for the loading of E2F1 and E2F3 onto genes expressed exclusively during S-phase; MLL recruitment to these genes is impaired in the absence of CHD8. CHD8 is a marker of aggressive gastric cancer, with 15.9% of cases containing mutations and 35.75% exhibiting with a loss of expression. In these cancers, CHD8 regulates b-catenin/ Wnt pathways and low expression of CHD6 is considered to be a prognostic marker for poor outcome. CHD8 has a bimodal role in prostate cancer; the CHD8 promoter is hypermethylated in 45% of cases, but high CHD8 expression results in poor clinical outcomes and metastasis. Conversely, CHD8 is a coregulator of androgen dependent transcription and loss of CHD8 is sufficient to stimulate proliferation of prostate cancer. CHD8 mutations are associated with increased risk of cancer development in those carrying familial BRCA1 mutations. In ovarian cancer, amplification of the NSD3-CHD8-BRD4 pathway is associated with worse survival rates. This amplification also occurs in 9% of serous endometrial cancer and again, is associated with worse overall survival. In AML, these proteins colocalize across the genome and are released from chromatin by BET inhibitor treatment. Little is known about the role of CHD9 in cancer, though a study of neuroblastoma found mutations in approximately 4% of patients. Down-regulation of CHD9 in these cancers was correlated with increased bone metastasis and decreased patient survival (Table 4). The mutation sites identified were close to a phosphorylation site for CHD9, potentially inhibiting the ability of this remodeler to be phosphorylated and therefore, for CHD9 to be activated. CHD9 is also mutated in cancers of the digestive tract, including 11.1% of gastric cancers. To date, growing evidence suggests that CHD9 is a tumor suppressor.

ATRX ATRX an ATP-dependent chromatin remodeler with high homology to SWI/SNF and with roles in genome stability, DNA damage and heterochromatin formation. ATRX forms a complex with DAXX, a chaperone of histone H3.3. Somatic mutations in ATRX have been identified in brain tumors and osteosarcomas, and are positively correlated with increased rates of single nucleotide variation in cancer. ATRX binds to the 30 exons of zinc finger genes to maintain genomic stability and also has a major role in the formation of heterochromatin. ATRX deposits the histone variant H3.3 at pericentric heterochromatin and at telomeres to maintain H3K9me3 enrichment. Reduced ATRX is associated with loss of H3K9me3 and telomere lengtheningda feature of many cancers. ATRX also maintains the silencing of imprinted genes and retrotransposons through the same mechanism.

Concluding Remarks Chromatin modifying proteins have an integral role within a cell to maintain proper chromatin integrity and genome stability (Fig. 4). Chromatin remodelers and modifiers control access to the genes needed for development and progression of cancer. In normal cells, there is tight regulation around these genes and disruption to chromatin remodelers in cancer either by way of mutation and/or changes in expression driven by epigenetic mechanisms can lead to the altered chromatin structure. DNA is therefore extremely vulnerable to further mutations from damaging agents, either intra- (e.g., DNA methylation) or extra- cellular (e.g., UV damage). Particularly, the unrestricted access to genes due to global chromatin instability and “loosening” is all that cancer cells need to gain a simple proliferative and survival advantage. Aberrant function of chromatin remodelers is known to directly result in tumor development, and many cancer types contain mutations in at least one chromatin remodeler.

Prospective Vision Moving forward, it will be particularly important to characterize the role of epigenetic mechanisms in driving atypical changes in chromatin remodeler expression. Moreover, it will be essential to determine whether epigenetic reprogramming of the cancer genomes causes additional mutations affecting chromatin remodeler activity. Future study into the role of ncRNA targeting chromatin remodellers to the genome is an exciting area of development. ncRNA is ideal for this purpose due to the sequence specificity and the dynamic secondary structure. DNA methylation and activity of the APOBEC family of cytidine deaminases involved in C to U mRNA repair may be of particular interest as the field seeks to characterize the evolution of chromatin remodeller mutations over

527

G1

Mutations in Chromatin Remodeling Factors

A DN AIR P RE

CHR S SEG OMOSO REG ME ATIO N

A M

G2

X

AL ION RIPT ON I NSC TRA GULAT RE

HIGHER ORDER STRUCTURE

EP DN LI A C A TI ON

R

Fig. 4 The mutation or disruption of chromatin remodellers has the potential to reduce genome integrity in cancer. Atypical expression or activity of chromatin remodelers by mutation or epigenetic reprogramming has the potential to alter higher order chromatin structures, including enhancerpromoter interactions by DNA looping; transcriptional regulation (abnormal gene expression patterns); DNA repair (misrepair or incomplete repair), DNA replication (unfaithful) and chromosome segregation (impaired) from the mitotic spindle during the cell cycle. Nucleosomes, large gray circles; CpG sites, small white circles; active histone posttranslational modifications, green circles; repressive posttranslational histone modifications, red circles; transcriptional start site, arrow.

the course of cancer progression. The interplay between chromatin remodellers and the epigenome is likely to be complex, and bidirectional. The influence of somatic chromatin remodeller mutations on the epigenome are likely to hold key clues to cancer etiology and identification of the most promising epigenetic therapies.

See also: Mutations in DNA Methyltransferases and Demethylases. Mutations in Histone Lysine Methyltransferases and Demethylases.

Further Reading Brachmann, R.K., Davis, P.K., 2003. Chromatin remodeling and cancer. Cancer Biology & Therapy 2 (1), 23–30. Chen, J., Herlong, F.H., Stroehlein, J.R., Mishra, L., 2016. Mutations of chromatin structure regulating genes in human malignancies. Current Protein and Peptide Science 17, 411–437. Clapier, C.R., Iwasa, J., Cairns, B.R., Peterson, C.L., 2017. Mechanisms of action and regulation of ATP-dependent chromatin-remodeling complexes. Nature Reviews Molecular Cell Biology 18, 407–422. Deindl, S., Hwang, W.L., Hota, S.K., et al., 2013. ISWI remodelers slide nucleosomes with coordinated multi-base-pair entry steps and single-base-pair exit steps. Cell 152 (3), 442–452. Lai, A.Y., Wade, P.A., 2011. Cancer biology and NuRD: A multifaceted chromatin remodeling complex. Nature Reviews Cancer 11, 588–596. Nair, S.S., Kumar, R., 2012. Chromatin remodeling in cancer: A gateway to regulate gene transcription. Molecular Oncology 6 (6), 611–619. Mayers, K., Qiu, Z., Alhazmi, A., Landry, J.W., 2014. ATP-dependent chromatin remodeling complexes as novel targets for cancer therapy. Advanced Cancer Research 121, 183–233. Marfella, C.G.A., Imbalzano, A.N., 2007. The CHD family of chromatin remodelers. Mutation Research 618 (1–2), 30–40. Skulte, K.A., Phan, L., Clark, S.J., Taberlay, P.C., 2014. Chromatin remodeler mutations in human cancers: Epigenetic implications. Epigenomics 6 (4), 397–414. Watanabe, R., Kanno, S., Roushandeh, A.M., Ui, A., Yasui, A., 2017. Nucleosome remodeling, DNA repair and transcriptional regulation build negative feedback loops in cancer and cellular aging. Philosophical Transactions of the Royal Society B 371, 1731. Wilson, B.G., Roberts, C.W.M., 2011. SWI/SNF nucleosome remodelers and cancer. Nature Reviews Cancer 11, 481–492.

Mutations in DNA Methyltransferases and Demethylases Xiao-Jian Sun, Zhu Chen, and Sai-Juan Chen, State Key Laboratory of Medical Genomics, Shanghai Institute of Hematology, Rui Jin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China © 2019 Elsevier Inc. All rights reserved.

Glossary CpG islands Genomic regions with a high frequency of CpG sites. In mammalian genomes, CpG islands are typically 200– 3000 base pairs in length, less methylated than other CpG sites cross the genome, and many of them are associated with gene promoters and play a role in gene transcriptional regulation. Epigenetics Stable heritable traits that cannot be explained by changes in DNA sequence. The epigenetic inheritance is maintained through cell divisions and passed from one cell generation to the next even though they do not involve changes in the DNA sequence of the organism. The known mechanisms of epigenetic inheritance involve regulation of gene activity and expression by covalent modifications of chromatin (DNA and histones), nucleosome positioning, RNA transcripts, and prions, etc. Hematopoiesis The process of generating all types of blood cellular components from hematopoietic stem cells (HSCs). This process involves the maintenance of the HSCs by a self-renewal mechanism and step-wise differentiation from HSCs into different lineages of mature blood cells. The cell fate determination during hematopoiesis is controlled by cellular autonomous mechanisms, which include epigenetic regulation, transcription factors and signaling pathways, as well as extracellular signals induced by cytokines and cell–cell interactions. Leukemia A group of hematopoietic cancers that usually begin in the bone marrow and result in high numbers of abnormal white blood cells which are not fully differentiated. The major types of leukemia include acute myeloid leukemia (AML), acute lymphoblastic leukemia (ALL), chronic myeloid leukemia (CML) and chronic lymphocytic leukemia (CLL). Myelodysplastic syndromes (MDS) A heterogeneous group of hematopoietic malignancies characterized by impaired peripheral blood cell production and most commonly a hypercellular, dysplastic-appearing bone marrow. Tricarboxylic acid (TCA) cycle A looped metabolic pathway that consists of series of enzyme-catalyzed chemical reactions occurred in living cells to release energy through oxidation of acetate, in the form of acetyl-CoA, into carbon dioxide and water. It is a key metabolic pathway that connects carbohydrate, fat, and protein metabolism, as it provides precursors of certain amino acids, as well as the reducing agent NADH that are used in numerous other biochemical reactions. It is also known as citric acid cycle (CAC) and the Krebs cycle.

Introduction Recent cancer genomic studies have led to the discovery of many previously unrecognized cancer-causing genes. Functional classification of these genes facilitates us to pinpoint essential cellular regulatory processes that contribute to tumorigenesis. DNA methylation is one of such processes, since several newly identified mutated genes in cancers encode enzymes responsible for “writing” and “erasing” an important epigenetic mark of the genomed5-methylcytosine (5mC; i.e., methylation of the DNA base cytosine at the 5th atom in the pyrimidine ring). DNA methylation plays important roles in many biological processes in mammalians, including transcriptional repression, genomic imprinting and X chromosome inactivation, as well as cell fate decision. As illustrated in Fig. 1, multiple members of three families of enzymes that control the key steps of DNA methylation and demethylation have been found being recurrently mutated in cancer: (1) DNMT3A belongs to a family of DNA methyltransferases (DNMTs) that transfer the methyl group from S-adenosyl methionine (SAM) to cytosine. (2) TET2 (ten-eleven translocation 2) belongs to a family of TET dioxygenases that convert the 5mC to 5-hydroxymethylcytosine (5hmC) through a hydroxylation reaction, followed by further processing to complete the demethylation. (3) IDH1 and IDH2 belong to a family of isocitrate dehydrogenases (IDHs) that convert isocitrate to a-ketoglutarate, which is an essential co-factor for the TETs. Thus, DNMTs and TETs act as “writers” and “erasers” of DNA methylation, respectively, whereas IDHs are important co-regulators; mutations in these enzymes cause aberrant DNA methylation patterns in cancer. Accumulating studies seek to explore the mechanism of how the mutations in these enzymes contribute to tumorigenesis and to identify potential new therapeutics to treat cancer. In this article, we will introduce the basic concepts and newly acquired knowledge regarding the functions of the DNA methyltransferases and demethylases in development and cancer. An adequate understanding of the underlying mechanisms should lead to identification of new therapeutic strategies for cancer treatment.

Location and Biological Functions of DNA Methylation Location of DNA Methylation in Human Genome In mammals, the majority of 5mCs occurs on the cytosine nucleotide followed by a guanine nucleotide (CpG site), although few non-CpG 5mCs have also been observed in specific cell types such as embryonic stem cells, some neural cells, and hematopoietic

528

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65054-7

Mutations in DNA Methyltransferases and Demethylases

529

Fig. 1 Mutated enzymes in the pathway of DNA methylation and demethylation. Asterisks denote the enzymes known to be affected by somatic mutations in cancer. The DNA methyltransferases (DNMTs) catalyze DNA methylation by transferring a methyl group to cytosine (C) to form 5methylcytosine (5mC). The TET methylcytosine dioxygenases oxidize 5mC to 5-hydroxymethylcytosine (5hmC), and further oxidation reactions results in the successive conversion of 5hmC to 5-formylcytosine (5fC) and 5-carboxylcytosine (5caC). The 5hmC can also be deaminated to 5hydroxymethyluracil (5hmU) by AID/APOBEC (activation-induced cytidine deaminase/apolipoprotein B mRNA-editing enzyme complex). These oxidation products can be removed by the TDG (thymine DNA glycosylase) and BRE (repaired by the base excision repair) enzymes, returning to an unmodified cytosine. Activities of the TETs are dependent on a-ketoglutarate (a-KG), the natural product of isocitrate dehydrogenases (IDHs), but are inhibited by 2-hydroxyglutarate (2-HG), the product of cancer-associated IDH mutants.

progenitors. Mammalian genomes are depleted of CpG sites, probably because 5mC can be spontaneously deaminated to thymine and therefore is vulnerable to mutagenic potentials. Indeed, the total number of CpG sites in humans is approximately 28 million, which is much lower than what is expected according to the average GC content of human genome. The CpG sites in the genome are overall heavily methylated, except for some specific genomic regions such as CpG islands. In different genomic regions, the localization pattern, regulatory mechanism and biological function of the methylated CpG sites could be different (Fig. 2). In the specific genomic regions named CpG islands, which have a higher CpG density than the rest of the genome, the CpG sites are much less methylated. As many mammalian genes have CpG islands associated with their promoters, the 5mCs in CpG islands play an important role in gene transcriptional regulation. It has been shown that the presence of multiple methylated CpG sites in the CpG islands results in silencing of gene expression. Given the relative stability of 5mC, the gene silencing mediated by methylation of CpG islands is usually stable. This mechanism is particularly important for genomic imprinting, an epigenetic phenomenon in which certain genes are expressed from only one of the two parental chromosomes and their expression is determined in a parent-of-origin specific manner. Besides the imprinted genes, 5mCs also regulate gene expression in development and tumorigenesis. Notably, however, the CpG islands associated with gene promoters rarely show tissue-specific methylation pattern. Instead, the differences of CpG methylation between tissues, or between normal and cancer samples, are observed in the regions called CpG island shores, located a short distance from the CpG islands. The CpG methylation also locates in the gene body, and its enrichment in these regions is positively correlated with gene expression levels. In mouse embryonic stem cells, this distribution pattern has been shown to be established and maintained through a direct recognition of histone lysine 36 trimethylation (H3K36me3) by the Pro-Trp-Trp-Pro (PWWP) domain of the DNMT3B. Given the high similarity between DNMT3A and DNMT3B, DNMT3A should be able to recognize H3K36me3 as well, though probably in different cell types and/or genome sites. The intragenic DNA methylation has been suggested to play a role in repression of cryptic transcriptional initiation and regulation of splicing, although the detailed mechanisms remain to be further clarified. DNA methylation plays an important role in permanent silencing of transposable and viral elements, which comprise approximately 45% of the human genome. If expressed, these elements are potentially harmful to the cell. However, it is interesting to note that these elements may also provide a resource for stimulating immune responses. Indeed, recent studies have shown that some

530

Mutations in DNA Methyltransferases and Demethylases

Fig. 2 Localization and basic functions of DNA methylation in human genome. The majority of DNA methylation (5mC) occurs on the CpG sites, and most CpG sites tend to be methylated, except for the CpG islands, which are less methylated. The CpG islands associated with gene promoters play an important role in transcriptional regulation, whereas those located in the gene body may regulate cryptic intragenic transcriptional initiation and splicing. The methylated CpG sites located in the intergenic regions are important for silencing the transposable elements. Modified from Wikipedia (https://en.wikipedia.org/wiki/File:DNAme_landscape.png).

demethylating agents can induce a cell autonomous immune activation response by stimulating expression of the endogenous retroviruses. This response probably explains some of the anti-tumor activities of these drugs.

DNA Methylation Dynamics in Development and Cancer Despite its stability, the patterns of DNA methylation can be substantially altered in development and cancer. In mammals, there are two waves of genome-wide DNA demethylation and re-methylation during the development. The first wave occurs in early embryogenesis. DNA methylations on both paternal and maternal genomes are largely erased shortly after fertilization, and a de novo methylation then takes place during the implantation stage of the embryo. The second wave of genome-wide DNA demethylation occurs during the specification of primordial germ cells (PGCs), the first germline cells established during embryogenesis and the precursors for both the oocytes and sperms. At the beginning of their specification and migration, PGCs are shown to have the same DNA methylation patterns as the epiblast cells from which the PGCs are generated. However, by the time they arrive at the genital ridge, their DNA methylations are almost completely erased. This DNA methylation reprogramming is possibly required for totipotency of the germ cells in the newly formed embryos. Even before the discovery of the mutations of the DNA methyltransferase and demethylases in cancer, altered DNA methylation patterns have been heavily implicated in various cancers. As a stable gene silencing mechanism, DNA hypermethylation has been commonly observed in cancers to silence tumor-suppressor genes. Particularly, in some familial cancers, when one allele of the key tumor-suppressor gene has mutated in germ line, there is a fair chance that the second allele is silenced by DNA methylation. Besides the individual cancer-related genes that are directly targeted by DNA methylation, alterations of DNA methylation patterns in broad genomic regions, such as repetitive elements, low-density CpG regions, and lamin-associated domains, have also been observed in cancer cells, although their functional roles in cancer remain to be defined. It has been widely accepted that a genome-wide profile of DNA methylation (methylome) of cancer is a useful method not only for understanding the molecular mechanism of epigenetic regulation of cancer, but also for precise diagnosis and treatment of cancer. Differentially methylated regions (DMRs) are commonly used to describe potentially functionally relevant genomic regions that are differentially methylated between different cells/tissues. In another way, the concept of CpG island methylator phenotypes (CIMPs) has been invented to identify concordant CpG methylations of a group of genes in a certain type of cancer and thereby to classify a subtype of cancer from others based on their specific methylation patterns.

Mutations in DNMT3A DNMT3A is one of the most frequently mutated genes in adult hematopoietic malignancies. It belongs to an evolutionarily conserved family of methyltransferases. Except for DNMT3A, none of other members of the family have been found being recurrently mutated in cancer, though DNMT3B, the closest homologue of DNMT3A, is mutated in patients with a rare autosomal recessive immune disorder termed immunodeficiency, centromeric instability, and facial anomalies (ICF) syndrome. It is also unanswered why DNMT3A mutation is only heavily implicated in hematopoietic malignancies, but not other types of cancer. Understanding the basic function of DNMT3A in DNA methylation regulation and normal hematopoiesis would be helpful to clarify the functions and mechanisms of DNMT3A in leukemogenesis.

Mutations in DNA Methyltransferases and Demethylases

531

The DNMT Family The DNMT family in humans contains five members: DNMT1, DNMT2 (also known as TRDMT1), DNMT3A, DNMT3B and DNMT3L (Fig. 3). They are evolutionarily conserved as each of them has homologue(s) found in vertebrates and they likely share a common ancestor in invertebrates such as fruit fly (Fig. 3A). They share a homologous catalytic domain in their C-terminuses, and some of them have acquired additional domains that may mediate substrate recognition, protein interaction and other regulatory activities (Fig. 3B). Indeed, the biological functions of the DNMT family members are highly specified. Although DNMT2 is the most conserved member in the family, it has been shown not being able to methylate DNA, instead it can methylate the transfer RNA (tRNA) Asp-GTC on cytosine 38 in the anticodon loop. The other DNMTs possess DNA methylation activity but they are functionally classified as two types: de novo and maintenance methyltransferases. DNMT3A, DNMT3B and DNMT3L are de novo DNMTs, as they create hemimethylated CpG sites in double-strand DNA, and are responsible for setting up the pattern of methylation. The high similarity between DNMT3A and DNMT3B implies that they may exert redundant functions, but they may also play different roles in gene- and cell type-specific manners. While DNMT3L is catalytically inactive, it can facilitate the activities of DNMT3A and DNMT3B. DNMT1 is the maintenance DNMT that adds methyl group to the cytosine when one strand has already been methylated, and is responsible for maintaining the methylation pattern that had been established by the de novo DNMTs.

Structure and Function of DNMT3A DNMT3A contains three conserved domains (Fig. 3B). The PWWP domain can bind both DNA and histone marks such as H3K36me3, through which DNMT3A is recruited to the H3K36me3 enriched genomic regions, including gene body region and

Fig. 3 The DNMT family. (A) Phylogenic tree of human (Homo sapiens, HS) and mouse (Mus musculus, Mm) DNMT family members. Construction of the tree is based on an alignment of the amino acid sequences of the catalytic domains. The methyltransferase 2 (Mt2) of fruit fly (Drosophila melanogaster, Dm) represents the common ancestor of the mammalian DNMTs. (B) Domain architectures of the human DNMT family members. The DNMT3A mutational hotspot R882 is localized within the catalytic domain.

532

Mutations in DNA Methyltransferases and Demethylases

pericentromic heterochromatic foci. The ATRX-DNMT3-DNMT3L (ADD) domain binds H3 tail peptides with or without lysine 4 (H3K4) methylation, and, in the presence of H3K4me3, acts as an autoinhibitor for the enzymatic activity of DNMT3A. The C-terminal catalytic domain is a C5-type S-adenosyl methionine (SAM)-dependent methyltransferase domain. It has been shown that the catalytic domain preferentially binds unmethylated DNA. DNMT3A can form oligomers, including dimer and tetramer, through two distinct binding interfaces in the catalytic domain. The oligomeric states of DNMT3A, as well as its heterooligomerization with DNMT3B and DNMT3L, are important for their catalytic activities. Regulation of the oligomerization thus greatly contribute to the function of DNMT3A in both hematopoiesis and tumorigenesis.

DNMT3A in Hematopoiesis Deletion of Dnmt3a in mouse models leads to enhanced self-renewal of hematopoietic stem cells (HSCs). Although the Dnmt3anull HSCs are phenotypically indistinguishable from wild-type HSCs, with identical cell surface markers, intact abilities to differentiate to all hematopoietic lineages, and similar levels of proliferation and apoptosis, serial bone marrow transplantation assays indicate a gradual expansion of the Dnmt3a-null long-term hematopoietic stem cells (LT-HSCs). Transplantation of the Dnmt3a-null HSCs lead to development of myeloid and lymphoid malignancies, which recapitulate many clinical features of human malignancies caused by DNMT3A mutations. Dnmt3a-null HSCs display significant genome-wide hypomethylation with focal areas of hypermethylation. The DMRs identified in Dnmt3a-null HSCs are significantly enriched at the edges of large hypomethylated regions known as methylation canyons. Many methylation canyon-associated genes involved in HSC self-renewal are up-regulated in Dnmt3a-null HSCs and remain inappropriately activated in differentiated cells, suggesting that the Dnmt3a deficiency results in an inability of the cells to turn off the genes that regulate differentiation.

DNMT3A Mutations in Hematopoietic Malignancy DNMT3A mutations are presented in approximately 20% of patients with de novo and secondary acute myeloid leukemia (AML), especially in the French–American–British (FAB) classified M4 and M5 subtypes. DNMT3A mutated AMLs frequently harbor NPM1 (nucleophosmin 1) and FLT3 (fms related tyrosine kinase 3) mutations. Although DNMT3A mutations have been defined as an earliest driver event in tumorigenesis, the impact of DNMT3A mutations in AML prognosis is under debate. Besides AML, DNMT3A mutations have also been reported in other myeloid malignancies, including myelodysplastic syndrome (MDS), myeloproliferative neoplasm (MPN) and myelodysplastic/myeloproliferative overlapping disease (MDS/MPN), although their prevalence is much lower than in AML. Furthermore, DNMT3A mutations have also been recurrently detected in lymphoid malignancies. Notably, in T-cell acute lymphoblastic leukemia (T-ALL), the prevalence of DNMT3A biallelic mutation is higher than in myeloid malignancies, implying potentially different mechanism for DNMT3A-driven lymphoid and myeloid neoplastic transformation. The majority of DNMT3A mutations detected in hematopoietic malignancies occur within the C-terminal catalytic domain, with a significant enrichment for mutations at amino acid arginine 882 (R882), although non-R882 missense and truncating mutations are found with much lower frequency in other domains. In myeloid malignancies, R882 mutations are usually heterozygous, whereas in T-ALL, there is a high frequency of non-R882 biallelic mutations. The DNMT3A R882H mutant has been shown to exert a dominant-negative effect against the wild-type protein. Mechanistically, the R882H mutant can dimerize with wild-type DNMT3A, but prevent the formation of a tetramer, which represents a stronger enzymatic activity. Thus, the existence of the R882H mutant can reduce the DNA methyltransferase activity in the cells, explaining the genome-wide hypomethylation observed in patients with DNMT3A R882 mutations. In addition, DNMT3A R882 mutants may also exert its function through affecting other proteins/pathways. For example, DNMT3A R882 mutants can interact with the polycomb repressive complex 1 (PRC1) to downregulate specific target genes involved in hematopoietic differentiation. It can also activate the mechanistic target of rapamycin (mTOR) pathway and increase the protein level of cyclin-dependent kinase 1 (CDK1), which are associated with a dysregulation of cell cycle of HSCs. Indeed, the mTOR inhibitor rapamycin elicited a significant therapeutic response in the knock-in and bone marrow transplantation mouse models of the DNMT3A R882H mutation.

Mutations in TET2 TET2 mutations have been found to be associated with multiple hematopoietic malignancies, as well as melanoma and few cases of glioma. TET2 somatic mutations and loss of heterozygosity (LOH) occur in patients with MDS, MPN and AML. TET2 mutations have also been reported in T-cell lymphomas. Unlike DNMT3A, IDH1 and IDH2 (see below), TET2 does not display a dominant hotspot mutation site, instead nonsense and missense mutations, as well as deletions, are found across multiple exons.

The TET Family The founding member of the TET protein family was first identified as a fusion partner of MLL (mixed-lineage leukemia; also known as lysine methyltransferase 2A, KMT2A) in patients with t(10;11)(q22;q23) AML, and therefore this gene was named as ten-eleven translocation 1 (TET1). The TET family contains three members, TET1, TET2 and TET3 (Fig. 4A), all of which have been identified as

Mutations in DNA Methyltransferases and Demethylases

533

Fig. 4 The TET family. (A) Phylogenic tree of human (HS) and mouse (Mm) TET family members. Construction of the tree is based on an alignment of the amino acid sequences of the catalytic domains. The fruit fly (Dm) Tet protein represents the common ancestor of the mammalian TETs. (B) Domain architectures of the human TET family members.

methylcytosine dioxygenases that catalyze the conversion of 5mC to 5hmC. The three TET family members share a catalytic domain in their C-terminuses; TET1 and TET2 also contain a N-terminal CXXC domain, whereas TET2 does not (Fig. 4B).

Structure and Function of TET2 The dioxygenase activity of the TET family members is mediated by the double-stranded beta helix (DSBH) fold domain termed base J-binding protein (JBP) domain. This JBP domain virtually defines a much larger superfamily of 2-oxoglutarate (2-OG)and ferrous iron (Fe(II))-dependent dioxygenases (2OGFeDO), which are found in various eukaryotes, bacteria and bacteriophages. Notably, the co-factor of TETs, 2-OG (also known as a-ketoglutarate (a-KG)), is an important metabolite in the tricarboxylic acid (TCA) cycle (also known as citric acid cycle or Krebs cycle), and it is converted from isocitrate by the IDHs. Thus, the discovery of TETs as key enzymes for DNA demethylation makes a direct connection between epigenetic regulation and cellular metabolism. Alterations of both aspects had already been considered as hallmarks of cancer but the mechanistic link between them has otherwise been lacking. Besides the catalytic JBP domain, TET1 and TET3, but not TET2, contain a conserved CXXC domain that is shown to bind unmethylated CpG sites. Interestingly, although TET2 does not contain the CXXC domain, a CXXC domain-containing gene, IDAX (also known as CXXC4), locates upstream of the TET2 genomic locus, suggesting that an ancient CXXC-containing TET2 gene was split into two separate genes through evolution. Nevertheless, biochemical studies revealed that the IDAX protein is still able to interact with TET2, and that this interaction is associated with caspase-mediated TET2 degradation, providing a flexible regulatory mechanism for TET2 stabilization.

TET2 in Hematopoiesis Deletion of TET2 in hematopoietic cells in mice leads to an expansion of hematopoietic stem/progenitor cells (HSPCs) in vivo. These cells show increased re-plating potential and colony formation ability in vitro. In competitive bone marrow transplantation assays, TET2-null HSPCs are capable of outcompeting wild-type cells. These observations suggest that TET2 deficiency results in increased self-renewal of HSCs, which may lead to neoplastic transformation. Indeed, aged TET2 knockout mice have been reported to develop myeloid malignancies, which are mostly characterized as a chronic myelomonocytic leukemia (CMML)-like disease. In addition, other hematopoietic malignant phenotypes, such as a MDS-like disease and increased immature thymic T cells and mature splenic B cells, were observed in other Tet2 knockout mouse models. These diverse phenotypes of the Tet2-deleted mice may attribute to different approaches to target Tet2 and/or different genetic background. It is also conceivable that, because there was always a considerable latency before the mice developed full-blown hematopoietic malignancies, secondary genetic events acquired through time could contribute to tumorigenesis and specify disease phenotypes.

534

Mutations in DNA Methyltransferases and Demethylases

TET2 Mutations in Hematopoietic Malignancy Somatic deletions and mutations in TET2 were first found in patients with MDS, MPN and AML based on mapping of loss-ofheterozygosity and microdeletions within a minimal region of chromosome 4q24. It turned out that TET2 mutations are among the most common genetic abnormalities observed in patients with MDS (6%–26%) and CMML (20%–58%). TET2 mutations are also frequently observed in primary and secondary AMLs (12%–32%). In addition, TET2 mutations are detected in B-cell and T-cell lymphomas. This broad spectrum of hematopoietic malignancies associated with TET2 mutations in humans is consistent with the above described diverse phenotypes of Tet2-knockout mice. TET2 mutations can be observed at both early and late stages of the development of hematopoietic malignancies, and they may exert different function at different stages. TET2 mutations are mostly heterozygous, in which the wild-type allele of TET2 in the patients remains expressing, whereas biallelic TET2 inactivation occurs in less than 10% of patients with leukemia. Furthermore, there is not a prominent hotspot mutation site in TET2. These observations suggest that TET2 can function as a haploinsufficient tumor suppressor in most patients. By a curious paradox, mutations of the two enzymes, DNMT3A and TET2, which apparently take opposite biochemical actions on DNA methylation (i.e., writing versus erasing), result in hematopoietic malignancies similarly. Furthermore, DNMT3A and TET2 have been found to be co-mutated in some patients with hematopoietic malignancies, and, as described above, loss of Dnmt3a or Tet2 in mice similarly results in expansion of HSCs. These observations suggest that DNMT3A and TET2 may work in parallel and/or cooperatively in hematopoiesis and malignancies. Consistent with this notion, it has been shown that Dnmt3a and Tet2 can cooperatively repress the erythroid transcription factor Klf1 (Kruppel like factor 1), and that the double knockout of Dnmt3a and Tet2 synergistically up-regulate Klf1 and related pathway, which may contribute to malignant transformation.

Mutations in IDH1 and IDH2 Interference with cellular energy metabolism has recently been defined as one of the emerging hallmarks of cancer. IDH1 and IDH2 are key enzymes involved in the TCA cycle; frequent mutations of both enzymes in cancer provide a prominent example to support this notion. IDH1 mutations were first identified in glioblastoma through whole-exome sequencing, and IDH2 mutations were identified in AML through a candidate gene approach. Mutations of IDH1 and IDH2 have now been detected in a broad spectrum of cancer types including astrocytoma, glioblastoma and AML.

The IDH Family The IDHs catalyze the oxidative decarboxylation of isocitrate to a-ketoglutarate, which is an essential step in the TCA cycle. There are five IDHs in human genome and they belong to two distinct subclasses: IDH1 and IDH2 utilize NADP(þ) as the electron acceptor, whereas IDH3 a, b and g use NAD(þ). They also show different subcellular localization: while IDH2, IDH3 a, b and g localize to mitochondria, IDH1 predominantly functions in the cytoplasm. The IDH family is extremely conserved through evolution, as both of the NADP(þ)- and NAD(þ)-dependent IDHs have homologues in all eukaryotes ranging from yeasts to humans. Unlike the DNMT and TET families in which the different members in mammals share a common ancestor in invertebrates, the homologues of both IDH1 and IDH2 are found in invertebrates such as the nematode Caenorhabditis elegans (Fig. 5A), suggesting that their different roles were specified much earlier through evolution than that of the DNMT and TET family members.

Gain-of-Function Mutations in IDH1 and IDH2 Mutations of IDH1 and IHD2 identified in cancer are heterogenous, and, as same as DNMT3A, they clearly show mutational hotspotsdmost of the mutations occur at one of three highly conserved arginine residues: R132 of IDH1 and R140 and R172 of IDH2 (Fig. 5B). Based on the evolution analysis and sequence alignment, the R132 in IDH1 is exactly the homologous residue of R172 in IDH2 (Fig. 5C). As there has not been evidence for haploinsufficiency of IHD1 and IDH2 in cancer, their mutational pattern suggests a possibility that these mutants might be gain-of-function. Supportive of this possibility, the mutant IDH1 and IDH2 have been found to acquire a neomorphic enzymatic activity that converts a-KG to 2-hydroxyglutarate (2-HG). It has also been shown that patients with AML harboring IDH1 or IDH2 mutations have significant higher level of 2-HG in their serum, suggesting that 2HG could be considered as a biomarker of IDH-mutated AMLs. Furthermore, 2-HG has been shown to be sufficient to promote malignant transformation of hematopoietic cells, and this effect is reversible upon removal of 2-HG. Thus, 2-HG represent an “oncometabolite,” which is produced specifically in cancer cells and exert a cancer-inducing activity.

Mutual Exclusivity Between IDH1/2 and TET2 Mutations The structural similarity between a-KG and 2-HG suggests that the 2-HG could act as an inhibitor of the enzymes that naturally require a-KG as a co-factor. Indeed, members of several subfamilies of a-KG-dependent dioxygenases have been shown to be inhibited by 2-HG. These dioxygenases include the TETs, which, as described above, are responsible for erasing the DNA methylation and are also frequently mutated in hematopoietic malignancies. Interestingly, genomic and epigenomic studies of a large cohort of patients with AML have demonstrated that IDH mutations and TET2 mutations are mutually exclusive, and that patients

Mutations in DNA Methyltransferases and Demethylases

535

Fig. 5 The IDH1/2 family and the mutational hotspots. (A) Phylogenic tree of the IDH1/2 family members in human (HS), mouse (Mm), fruit fly (Dm), nematode (Caenorhabditis elegans, Ce), fission yeast (Schizosaccharomyces pombe, Sp) and baking yeast (Saccharomyces cerevisiae, Sc). Note that the Caenorhabditis elegans contains two IDHs (idh-1 and idh-2) related to the mammalian IDH1 and IDH2, respectively. (B) Domain architectures of human IDH1 and IDH2. A mitochondrial localization signal (MLS) is on the N-terminus of IDH2. The three mutational hotspots are localized in the catalytic domains of IDH1 and IDH2; the R132 of IDH1 is corresponding to the R172 of IDH2. (C) Alignment of the amino acid sequence context of the mutational hotspots in the IDH1/2 family members, showing the high conservation of these two R residues as well as their sequence context.

harboring mutations in TET2 or IDH1/2 share similar genome-wide methylation profiles, thus suggesting that the TET2 and IDH1/2 mutations affect the same pathway.

2-HG Is Involved in Many Cellular Processes Through Various a-KG-Dependent Dioxygenases Besides the TETs, many other a-KG-dependent dioxygenases can also be inhibited by 2-HG. Given that these enzymes play important roles in different signaling pathways, the cancer-associated IDH1 and IDH2 mutations and their product 2-HG are involved in many cellular processes and thereby can regulate different aspects of development and cancer. First, the Jumonji C domain-containing histone lysine demethylases are a-KG dependent dioxygenases sensitive to 2-HG. Ectopic expression of IDH1/2 mutants leads to dramatic accumulation of lysine-methylated histones. Compared to the wild-type cells, IDH1 mutated oligodendroglioma shows elevated H3K9me3 and H3K27me3.

536

Mutations in DNA Methyltransferases and Demethylases

Second, a pair of a-KG-dependent dioxygenases, the fat mass and obesity-associated protein (FTO) and the alkB homologue 5 (ALKBH5) RNA demethylases, are responsible for removing the RNA N6-adenosine methylation (m6A); the m6A modulates several aspects of RNA functions including splicing, nuclear export, stability and translation. There has been evidence suggesting that the IDH mutants affects mRNA m6A levels in leukemia cells likely through 2-HG inhibition of FTO activity. Third, in the hypoxia signaling pathway, the stability of the hypoxia-inducible factors (HIFs) is mainly regulated through HIF hydroxylation mediate by two families of enzymes, namely the three prolyl hydroxylase domain proteins (PHD1, PHD2 and PHD3) and an asparaginyl hydroxylase factor inhibiting HIF (FIH). These hydroxylases require a-KG as a co-factor and potentially can be inhibited by 2-HG. The correlations between IDH mutation, DNA methylation and hypoxia signaling turn out to be cellular context dependent; the detailed underlying mechanisms require further clarification. Lastly, the type IV collagen can be hydroxylated at their proline and lysine residues by the a-KG-dependent prolyl 4-hydroxylases 1, 2 and 3 (P4HA1, 2 and 3) and procollagen-lysine 2-oxoglutarate 5-dioxygenases 1, 2 and 3 (PLOD1, 2 and 3), respectively. Through 2-HG inhibition of type IV collagen hydroxylation, the IDH mutations may trigger collagen maturation defects which possibly contribute to tumor behaviors such as angiogenesis.

Targeted Therapy The discovery of frequent mutations in epigenetic modifiers provide an enormous potential for targeted therapy of cancer. In contrast to the transcription factors, which also play an important role in gene regulation and heavily implicated in cancer but difficult to be targeted, inhibition of the enzymes responsible for DNA methylation and other epigenetic modifications is a much easier strategy. As many of the mutations in DNA methyltransferases and demethylases have been shown to be driver mutations occurred in cancer stem cells, inhibition of these mutated enzymes provides an opportunity to eliminate the cancer stem cells which otherwise are barely targeted by chemotherapy. Furthermore, the cross-talks between DNA methylation and cellular metabolism, as well as other signaling pathways, suggest a potential for combination therapy with DNA methylation agents and other available therapeutics, such as receptor tyrosine kinase inhibitors, DNA damaging agents, mTOR inhibitors, and immune therapy. Two DNA methyltransferase inhibitors, 5-azacytidine (azacytidine, AZA) and decitabine (5-aza-20 -deoxycytidine, DAC) have already obtained approval for treatment of MDS and leukemia by the US Food and Drug Administration. Both drugs are cytidine analogues that can incorporate into DNA, block the catalytic activities of DNMTs and trigger their degradation, leading to a global DNA demethylation in the cells. Many studies have suggested that the promoter demethylation and derepression of the aberrantly silenced tumor suppressor genes underlies the anti-tumor effects of these DNMT inhibitors. However, recent studies have provided another mechanism that the DNMT inhibitors can induce the expression of double strand RNAs derived at least in part from endogenous retroviral elements, leading to an activation of interferon signaling pathway and an immune response. These studies open a new perspective on application of DNMT inhibitors in cancer immunomodulation. Many pharmacological inhibitors of mutated IDH1 and IDH2, which are expected to diminish 2-HG production in IDH1 and IDH2 mutated cells, have been developed and are currently under clinical investigation. Among these inhibitors, for example, AGI6780 is capable of binding in an allosteric manner at the dimer interface of the mutated IDH2 (IDH2-R140Q) but is less potent against wild-type IDH2. Interestingly, treatment with AGI-6780 induced differentiation of human acute myelogenous leukemia cells, providing a great potential for a differentiation therapy of cancer. In summary, since the initial identification of recurrent IDH mutations in glioblastoma and TET2 and DNMT3A mutations in myeloid malignancies, encouraging and rapid progress has been made to frame new concepts, clarify underlying mechanisms and develop targeted therapeutic strategies. While the detailed molecular mechanisms of epigenetic regulation of development and cancer need to be more thoroughly investigated, the development of drugs by mechanism-based design and functional screening approaches, as well as the translational and clinical studies, are of great importance for grasping the favorable opportunity to target these enzymes and metabolites, thereby completing the bench-to-bedside transition efficiently.

Acknowledgments This work is supported by the National Key Basic Research Project of China (2013CB966801); the Chinese Ministry of Health (201202003); the National Natural Science Foundation of China (81470316, 81670094, 81123005); the 1000 Talents Program for Young Scholars; and the Shanghai Municipal Education Commission-Gaofeng Clinical Medicine Grant Support (20152506).

See also: Acute Myelogeneous Leukemia: Diagnosis and Treatment. Mutations in Chromatin Remodeling Factors. Mutations in Histone Lysine Methyltransferases and Demethylases.

Further Reading Bowman, R.L., Levine, R.L., 2017. TET2 in normal and malignant hematopoiesis. Cold Spring Harbor Perspectives in Medicine 7, a026518. https://doi.org/10.1101/ cshperspect.a026518.

Mutations in DNA Methyltransferases and Demethylases

537

Brunetti, L., Gundry, M.C., Goodell, M.A., 2017. DNMT3A in Leukemia. Cold Spring Harbor Perspectives in Medicine 7, a030320. Dai, Y.J., Wang, Y.Y., Huang, J.Y., Xia, L., Shi, X.D., Xu, J., et al., 2017. Conditional knockin of Dnmt3a R878H initiates acute myeloid leukemia with mTOR pathway involvement. Proceedings of the National Academy of Sciences of the United States of America 114, 5237–5242. Dang, L., Su, S.M., 2017. Isocitrate dehydrogenase mutation and (R)-2-hydroxyglutarate: from basic discovery to therapeutics development. Annual Review of Biochemistry 86, 305–331. Dang, L., White, D.W., Gross, S., Bennett, B.D., Bittinger, M.A., et al., 2009. Cancer-associated IDH1 mutations produce 2-hydroxyglutarate. Nature 462, 739–744. Delhommeau, F., Dupont, S., Della Valle, V., James, C., Trannoy, S., et al., 2009. Mutation in TET2 in myeloid cancers. New England Journal of Medicine 360, 2289–2301. Jones, P.A., Baylin, S.B., 2002. The fundamental role of epigenetic events in cancer. Nature Reviews Genetics 3, 415–428. Ko, M., An, J., Pastor, W.A., Koralov, S.B., Rajewsky, K., Rao, A., 2015. TET proteins and 5-methylcytosine oxidation in hematological cancers. Immunological Reviews 263, 6–21. Ley, T.J., Ding, L., Walter, M.J., McLellan, M.D., Lamprecht, T., et al., 2010. DNMT3A mutations in acute myeloid leukemia. New England Journal of Medicine 363, 2424–2433. M Gagné L, Boulay, K., Topisirovic, I., Huot, M.É., Mallette, F.A., 2017. Oncogenic activities of IDH1/2 mutations: from epigenetics to cellular signaling. Trends in Cell Biology 27, 738–752. https://doi.org/10.1016/j.tcb.2017.06.002. Moore, L.D., Le, T., Fan, G., 2013. DNA methylation and its basic function. Neuropsychopharmacology 38, 23–38. Parsons, D.W., Jones, S., Zhang, X., Lin, J.C., Leary, R.J., et al., 2008. An integrated genomic analysis of human glioblastoma multiforme. Science 321, 1807–1812. Rasmussen, K.D., Helin, K., 2016. Role of TET enzymes in DNA methylation, development, and cancer. Genes & Development 30, 733–750. Shih, A.H., Abdel-Wahab, O., Patel, J.P., Levine, R.L., 2012. The role of mutations in epigenetic regulators in myeloid malignancies. Nature Reviews Cancer 12, 599–612. Wang, J.H., Chen, W.L., Li, J.M., Wu, S.F., Chen, T.L., et al., 2013. Prognostic significance of 2-hydroxyglutarate levels in acute myeloid leukemia in China. Proceedings of the National Academy of Sciences of the United States of America 110, 17017–17022. Weisenberger, D.J., 2014. Characterizing DNA methylation alterations from The Cancer Genome Atlas. Journal of Clinical Investigation 124, 17–23. Wu, S.C., Zhang, Y., 2010. Active DNA demethylation: many roads lead to Rome. Nature Reviews Molecular Cell Biology 11, 607–620. Yan, X.J., Xu, J., Gu, Z.H., Pan, C.M., Lu, G., et al., 2011. Exome sequencing identifies somatic mutations of DNA methyltransferase gene DNMT3A in acute monocytic leukemia. Nature Genetics 43, 309–315.

Mutations in Histone Lysine Methyltransferases and Demethylases Sara Weirich and Albert Jeltsch, Stuttgart University, Stuttgart, Germany © 2019 Elsevier Inc. All rights reserved.

Glossary Compound heterozygous mutations Compound heterozygous mutations consist of two different mutations in the paternal and maternal alleles of a gene. Dominant negative A dominant negative variant acts to reduce the activity of the wild-type gene. Gain-of-function mutation A mutation that confers a new molecular function on the mutated gene. Genetic dominance One allele in a heterozygous genotype masks the effect of the other. Haploinsufficiency A single functional copy of a gene cannot provide sufficient gene product to preserve the wild-type phenotype leading to an altered or diseased state. Heterozygous mutation A mutation of only one allele of a gene. Homozygous mutation An identical mutation of both, the paternal and maternal alleles, of a gene. Loss-of-function mutation A mutation that results in the gene product having less or no function (being partially or completely inactivated). Epigenetic modification Covalent modification of the DNA or chromatin that encodes epigenetic information. Epigenetics Stable heritable traits (or “phenotypes”) that are not caused by changes in DNA sequence but often are not fully stable. SET domain A 130 to 140 amino acid protein domain initially characterized in the Drosophila proteins Su(var)3–9, Enhancerof-zeste and Trithorax that contains the active center of one family of PKMTs. JmjC domain A protein domain that adopts the cupin fold and contains the active center of one family of KDMs. PKMT Enzymes that methylate lysine residues in proteins. KDM Enzymes that demethylate methyllysine residues in proteins.

Introduction Setting and Removing of Histone Lysine Methylation All human cells contain the same genetic information, yet they follow different developmental pathways and differentiate into the numerous different cells types found in the human body. The cellular fate and phenotype are determined by epigenetic regulatory mechanisms, which modulate the chromatin structure and control gene expression. The basic functional unit of chromatin is the nucleosome, which contains 147 base pairs of DNA wrapped around a core histone octamer comprising of two subunits of each of the histones H2A, H2B, H3, and H4. Epigenetic signals include DNA methylation, noncoding RNAs, and modifications of the histone proteins mainly within the N-terminal tails of H3 and H4, which are subject to a large variety of posttranslational modifications like acetylation, methylation, phosphorylation, or ubiquitylation. Methylation of histone tails can occur on Lys and Arg residues. Lysine methylation is a dynamic process and methylation levels change during biological processes such as cellular differentiation or carcinogenesis. Histone lysine methylation is introduced by protein lysine methyltransferases (PKMTs, also abbreviated as KMTs) which use S-adenosyl-L-methionine (AdoMet) as methyl group donor and belong to two distinct families of proteins (see Table 1 for a compilation of all human PKMTs and KDMs). Most histone methylating PKMTs contain a SET domain of approximately 130–140 amino acids, while currently known examples of seven-ß-strand PKMTs with histone substrates are limited to DOT1L. For the removal of lysine methylation, two families of lysine demethylase enzymes (KDMs) are responsible, the LSD family and enzymes containing Jumonji C (JmjC) domains. LSD1 and LSD2 are members of the monoamine oxidase superfamily, which are using flavin adenine dinucleotide and molecular oxygen as cofactors for the oxidation of CeN single bonds to an unstable imine intermediate generating hydrogen peroxide as byproduct. Subsequently, the imine is hydrolyzed and the methyl group is released as formaldehyde. This catalytic mechanism permits demethylation of secondary and tertiary but not of quaternary amines, limiting the substrate to mono- and dimethylated lysines. The second and much larger family of JmjC domain containing KDMs comprises about 20 human enzymes subgrouped into five subfamilies. These enzymes utilize oxygen, a-ketoglutarate, and Fe(II) ions as cofactors in the hydroxylation of lysine-bound methyl groups which is followed by the elimination of the hydroxymethyl group as formaldehyde. Enzymes of this family catalyze the removal of methyl groups from mono-, di-, and trimethylated lysine residues at various sites of the histone tails. PKMTs and KDMs are involved in several diseases including cancer.

538

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65056-0

539

Mutations in Histone Lysine Methyltransferases and Demethylases Table 1

List of human PKMTs and KDMs with their different abbreviations, main histone substrates, and the corresponding UniProtKB entry

Enzyme name

Full name

DOT1L EZH1 EZH2 G9a

DOT1-like protein Enhancer of zeste homolog 1 Enhancer of zeste homolog 2 Euchromatic histone-lysine Nmethyltransferase 2 GLP (EuHMT1) G9a-like protein 1 MLL1 (KMT2A) Myeloid/lymphoid or mixed-lineage leukemia protein 1 MLL2 (KMT2B) Myeloid/lymphoid or mixed-lineage leukemia protein 4 MLL3 (KMT2C) Myeloid/lymphoid or mixed-lineage leukemia protein 3 MLL4 (KMT2D) Myeloid/lymphoid or mixed-lineage leukemia protein 2 SETD1A SET domain-containing protein 1A SETD1B SET domain-containing protein 1B NSD1 Nuclear receptor-binding SET domaincontaining protein 1 NSD2 Histone-lysine N-methyltransferase NSD2 NSD3 Nuclear SET domain-containing protein 3 SET7/9 (SETD7) SET domain containing protein 7 SETD2 SET domain-containing protein 2 SETD8 (PRHistone-lysine N-methyltransferase KMT5A SET7) SETDB1 SET domain bifurcated 1 SETDB2 Histone H3-K9 methyltransferase SMYD1 SET and MYND domain-containing protein 1 SMYD2 SET and MYND domain-containing protein 2 SMYD3 SET and MYND domain-containing protein 3 SMYD4 SET and MYND domain-containing protein 4 SMYD5 SET and MYND domain-containing protein 5 SUV39H1 Suppressor of variegation 3–9 homolog 1 SUV39H2 Suppressor of variegation 3–9 homolog 2 SUV420H1 Suppressor of variegation 4–20 homolog 1 SUV420H2 Suppressor of variegation 4–20 homolog 2 KDM1A Lysine-specific histone demethylase 1A KDM1B KDM2A KDM2B

Lysine-specific histone demethylase 1B Lysine-specific demethylase 2A Lysine-specific demethylase 2B

KDM3A KDM3B KDM4A

Lysine-specific demethylase 3A Lysine-specific demethylase 3B Lysine-specific demethylase 4A

KDM4B KDM4C

Lysine-specific demethylase 4B Lysine-specific demethylase 4C

KDM4D KDM4E KDM5A KDM5B KDM5C KDM5D KDM6A

Lysine-specific demethylase 4D Lysine-specific demethylase 4E Lysine-specific demethylase 5A Lysine-specific demethylase 5B Lysine-specific demethylase 5C Lysine-specific demethylase 5D Lysine-specific demethylase 6A

KDM6B

Lysine-specific demethylase 6B

Abbreviation and abbreviation of selected synonyms

Main histone substrate(s) KB#

DOT1L, KMT4 ENX-2 EZH2, KMT6 BAT8, C6orf30, G9A, KMT1C, NG36

H3K79 H3K27 H3K27 H3K9

Q8TEK3 Q92800 Q15910 Q96KQ7

EUHMTASE1, GLP, KIAA1876, KMT1D MLL1, KMT2A

H3K9 H3K4

Q9H9B1 Q03164

HRX2, KIAA0304, MLL2, MLL4, TRX2, WBP7 H3K4

Q9UMN6

MLL3, KMT2C

H3K4

Q8NEZ4

ALR, MLL2, MLL4

H3K4

O14686

KMT2F, SET1, SET1A KIAA1076, KMT2G, SET1B NSD1, KMT3B

H3K4 H3K4 H3K36

O15047 Q9UPS6 Q96L73

NSD2, MMSET, WHSC1 WHSC1L1 KIAA1717, KMT7, SET7, SET9 SETD2, KMT3A PRSET7, SET07, SET8, SETD8

H3K36

O96028 Q9BZ95 Q8WTS6 Q9BYW2 Q9NQR1

KIAA0067, KMT1E C13orf4, CLLD8, KMT1F / KMT3C ZMYND1, ZNFN3A1 KIAA1936 RAI15 KMT1A, SUV39H KMT1B KMT5B KMT5C AOF2, KDM1, KIAA0601, LSD1

H3K4 H3K36 H4K20 H3K9 H3K9 H3K4 H3K4 and H3K36 H3K4 and H3K5

H3K9 H3K9 H4K20 H4K20 Methylated H3K4 and H3K9 AOF1, C6orf193, LSD2 H3K4 CXXC8, FBL7, FBXL11, JHDM1A, KIAA1004 Methylated H3K36 CXXC2, FBL10, FBXL10, JHDM1B, PCCX2 Methylated H3K4 and H3K36 JHDM2A, JMJD1, JMJD1A, KIAA0742, TSGA Methylated H3K9 C5orf7, JHDM2B, JMJD1B, KIAA1082 Methylated H3K9 JHDM3A, JMJD2, JMJD2A, KIAA0677 Methylated H3K9 and H3K36 JHDM3B, JMJD2B, KIAA0876 Methylated H3K9 GASC1, JHDM3C, JMJD2C, KIAA0780 Methylated H3K9 and H3K36 JHDM3D, JMJD2D Methylated H3K9 KDM4DL Methylated H3K9 JARID1A, RBBP2, RBP2 Methylated H3K4 JARID1B, PLU1, RBBP2H1 Methylated H3K4 DXS1272E, JARID1C1, SMCX1, XE169 Methylated H3K4 HY, HYA, JARID1D, KIAA0234, SMCY Methylated H3K4 UTX (Ubiquitously transcribed X-chromosome Methylated H3K27 tetratricopeptide repeat protein) JMJD3, KIAA0346 Methylated H3K27

Q15047 Q96T68 Q8NB12 Q9NRG4 Q9H7B4 Q8IYR2 Q6GMV2 O43463 Q9H5I1 Q4FZB7 Q86Y97 O60341 Q8NB78 Q9Y2K7 Q8NHM5 Q9Y4C1 Q7LBC6 O75164 O94953 Q9H3R0 Q6B0I6 B2RXH2 P29375 Q9UGL1 P41229 Q9BY66 O15550 O15054

540

Mutations in Histone Lysine Methyltransferases and Demethylases

Critical Enzymatic Properties of PKMTs and KDMs PKMTs were initially identified as histone lysine methyltransferases and shown to have very important roles in the regulation of gene expression and chromatin biology. Later, they were also shown to methylate several nonhistone substrates. Similarly, KDMs show activity on histone and nonhistone proteins, which both have roles in cancer. Locus-specific histone lysine methylation can signal either gene activation or repression, depending on the site of modification. For example, H3K4me3 has an activating function, while H3K9me3 and H3K27me3 are repressive marks. For the biological outcome of the lysine methylation it is also important to consider that a lysine side chain can be mono-, di-, or trimethylated and different methylated forms in general cause different effects. For example, H3K4me1 is associated with enhancers, while H3K4me3 is associated with promoters of active genes. The biological effects of lysine methylation are mediated by reading domains, which specifically interact with methylated lysine residues and trigger downstream signaling. Lysine methylation of nonhistone proteins can influence their stability and interaction with other proteins. However, the biological effects of nonhistone lysine methylation are not well understood in most cases. The substrate spectrum of PKMTs and KDMs is determined by the recognition of the target peptide in the active site, which includes readout of the peptide sequence. Moreover, the access of PKMTs and KDMs to target lysine residues is controlled by the interaction of the enzyme with substrate proteins. In the case of histone methylation and demethylation, the locus-specific targeting of the PKMTs and KDMs by either endogenous binding domains or (more often) other proteins and RNAs is essential for the biological activity. In the cell, PKMTs and KDMs form large complexes with a number of additional proteins, which have roles in their regulation and targeting. These complexes often contain other epigenetic writers or erasers and they regularly contain several DNA binding and epigenetic reading domains for locus-specific targeting. For example, while isolated EZH2 is weakly active, the PRC2 complex consisting of EZH2 and six protein partners, which has a molecular weight of approximately 500 kDa, is a highly active H3K27 trimethyltransferase. Regulation of PKMTs and KDMs often includes conformational changes between active and auto-inhibited states that prevent the aberrant activity at nontarget sites. Moreover, the ability of PKMTs and KDMs to generate different methylation states is an important property of these enzymes. More to that, PKMTs and KDMs differ in their ability to use differentially methylated lysine residues as substrates. For example, the SUV420H1 enzyme introduces H4K20 trimethylation. However, it is only weakly active on unmethylated H4, but needs monomethylated H4K20 as substrate, which is generated by the SET8 PKMT. Hence SUV420H1 is dependent on lysine methylation introduced before by SET8, which restricts its biological effects. Similarly, LSD family KDMs are not able to remove methylation from trimethylated lysine residues, which limits their potential effects. In summary, the critical enzymatic properties of PKMTs and KDMs are their amino acid sequence recognition, the preference for particular methylation states of the target lysine, their ability to generate mono-, di-, or trimethyllysine, protein–protein interaction, and their regulation and targeting. All these properties can be potentially altered by mutations in cancer cells, which in turn can cause massive changes in the signaling potential of the mutated enzymes.

Somatic Cancer Mutations in Epigenetic Enzymes Carcinogenesis is driven by the acquisition of a series of mutations and epigenetic changes, many of which ultimately confer a growth advantage upon the cells in which they have occurred. Epigenetic enzymes that modify histone proteins and DNA are involved in several diseases including cancer. Various genome-wide exome sequencing studies revealed several genes that are recurrently mutated in tumor tissues. Among them are mutations in genes that are involved in epigenetic functions either directly or indirectly, including many PKMTs and some KDMs, which will be described in this article in detail. The prevalence of somatic mutations in chromatin-modifying genes highlights the central role of transcriptional dysregulation in tumorigenesis. Somatic mutations in these enzymes might lead to either loss- or gain-of-function similar to mutations in other genes. Loss-of-function refers to mutations that reduce or disrupt the enzymatic activity, while gain-of-function refers to mutations, which confer novel activities on the mutated enzymes. Loss-of-function mutations include not only nonsense mutations, frameshift mutations, and deletion, which all directly disrupt protein synthesis, but also some missense mutations, which affect amino acid residues essential for catalysis or protein folding, for example. Somatic mutations can occur on one allele only (heterozygous) or on both alleles. If both alleles are affected, usually they carry different mutations (compound heterozygous). Gain-of-function effects are caused by missense mutations leading to amino acid exchanges, which alter critical enzymatic properties. They are almost always dominant yielding biological effects in heterozygous state. Loss-of-function mutations have biological effects if they occur as compound heterozygous mutation such that both alleles are affected. Heterozygous loss-of-function mutations (still combined with one wild-type allele) can have biological effects if the mutations act in a dominant negative or haploinsufficient manner meaning that the mutated gene reduces the activity of the remaining wild-type gene or the reduced amount of the wild-type gene product is not sufficient to fulfill its biological task. The physiological effects of loss-of-function mutations are often predictable because they directly relate to the function of the corresponding gene product. In contrast, understanding the pathogenic mechanism of gain-of-function mutations is complex and it requires an in-depth experimental investigation of each specific mutation of the corresponding protein. On the other hand, understanding the mechanisms of gain-of-function mutations provides unique information about carcinogenic pathways and it may aid in the development of individualized cancer therapies.

Mutations in Histone Lysine Methyltransferases and Demethylases

541

Somatic Cancer Mutations in PKMTs and KDMs This review will focus on the description of somatic cancer mutations in PKMTs and KDMs (Fig. 1). The strongest enrichment of somatic cancer mutations among PKMTs is observed in EZH2, but MLL3, MLL4, SETD2, SYMD1, SMYD3, and SUV420H1 also show relatively high frequencies of somatic cancer mutations when compared to the size of the proteins. Among KDMs, the highest frequency of mutations is observed in KDM6A followed by KDM4D. Of note, this distribution may be influenced by sample bias, because the initial observation of mutations in one enzyme will always stimulate targeted follow-up work that may lead to the discovery of additional mutations. In addition, the mutational spectrum depends on the tumor type, but currently databases are dominated by blood cancers (due to the relative ease of preparing pure tumor cells for analysis). Therefore, the picture may change in future when more data for various other cancers will become available. As described above, somatic cancer mutations in PKMTs and KDMs could have loss-of-function or gain-of-function effects. While frame-shifts and deletion mutations will mainly have loss-of-function effect, the situation is less clear for missense mutations, which cause an amino acid exchange in the protein. Such point mutations could inactivate the enzyme, if they disrupt cofactor binding, folding, or the binding of the substrate peptide. However, point mutations can also have gain-of-function effects, like the alteration of the peptide interaction potentially leading to the methylation or demethylation of novel substrates, the change of the product methylation state or preferred substrate methylation state, or an alteration of the interaction with other proteins leading to changes in the regulation or targeting of the PKMT (Fig. 2). Depending on the original role of the mutated PKMT or KDM in the regulation of cancer-related genes by histone methylation, somatic mutations in PKMTs or KDMs can lead to the overexpression of oncogenes or loss of expression of tumor suppressor genes via different pathways. If effects are based on the altered methylation of nonhistone proteins, oncogenic proteins can be stabilized or activated or proteins with tumor suppressor activity can be destabilized or inactivated. To allow an initial prediction of the main effect of somatic cancer mutations in enzymes, one can calculate the ratio of the missense mutations divided by the sum of frame-shift and nonsense mutations (Sense Score) (Fig. 3). This number corresponds to the probability of a widespread gain-of-function mechanism, because a missense mutation has the potential to cause gain-of-function effects, while frame-shift or nonsense mutations will lead to loss-of-function. Thus, a high Sense Score as seen in DOT1L, GLP, SETD1A, SMYD1, SMYD4, SMYD5, SUV39H1, KDM1B, KDM4A, and KDM4B proposes a gain-offunction effect of many mutations in these enzymes. These effects are often dominant and the mutated enzymes act as oncogenes. In contrast, low Sense Scores as in MLL4, SETD2, or KDM6A predict loss-of-function effects as main mechanism suggesting that the corresponding wild-type proteins have a tumor suppressor activity. However, the Sense Score can only provide a rough approximation, since each mutation obviously has its own effects on the enzyme that may differ from the “average” effects of all other mutations. Moreover, epigenetic enzymes can have different roles in different tumors acting as oncogenes in some cases and tumor suppressors in others. In the following sections, we will introduce several important families of PKMTs and KDMs, briefly summarize their role in cancer, and describe the effects of somatic cancer mutations in the respective enzymes.

Fig. 1 Graphical representation of the Mutational Scores of PKMTs and KDMs. Data were retrieved in September 2017 from the Cosmic database. The Mutational Score is defined by the ratio of the total number of somatic tumor mutations in the corresponding enzyme divided by the length of the protein in amino acids.

542

Mutations in Histone Lysine Methyltransferases and Demethylases

Fig. 2 Potential effects of somatic cancer mutations in PKMTs and KDMs. Mutations may interfere with AdoMet binding or protein folding, which will lead to loss of activity. Mutations may alter the methylation level preference of the PKMTs and KDMs with respect to either the product or the substrates representing a gain-of-function effect. In addition, mutations may change the interaction with other proteins and alter the regulation or targeting of the PKMT. Finally, mutations may change the recognition of the target peptide or binding of the substrate protein leading to either loss of activity on the original substrate or eventual activity on novel substrates. From Kudithipudi, S. and Jeltsch, A. (2014). Biochimica et biophysica acta (BBA)dreview on cancer 1846, 366–379 with permission.

Fig. 3 Graphical representation of the Sense Scores of somatic cancer mutations in PKMTs and KDMs. Data were retrieved in September 2017 from the Cosmic database. The Sense Score is defined by the ratio of the missense mutations (which depending on the exchange can cause loss-offunction or gain-of-function effects) divided by the number of frame-shift and nonsense mutations (which will usually cause loss-of-function effects).

Specific Roles of PKMTs and Their Known Somatic Mutations in Cancer The EZH2 H3K27 PKMT The Polycomb repressive complex 2 (PRC2) introduces H3K27me3, which is an important repressive chromatin mark. In mammals, the PRC2 core complex consists of EED, SUZ12, NURF55, Rbap46/48, and the catalytic subunits EZH1 or EZH2. It also has some variable additional subunits including PHF1, JARID2, and AEBP2, which further regulate the PRC2 activity. The SET domain containing EZH1 and EZH2 subunits represent the catalytically active components that introduce up to three methyl groups on H3K27. The interaction between EZH1/2 and the embryonic ectoderm development (EED) protein in the PRC2 complex is particularly essential for activity because EED stimulates the catalytic activity of EZH2 as a scaffolding protein. Moreover, EED contains a seven-bladed b-propeller WD-40 domain forming an aromatic cage that binds trimethylated H3K27 and mediates the association of PRC2 with H3K27me3 containing chromatin thereby supporting the propagation of H3K27me3. Genes repressed by PRC2 include HOXC8, HOXA9, MYT1, CDKN2A, and retinoic acid target genes, which all have important roles in development

Mutations in Histone Lysine Methyltransferases and Demethylases

543

and differentiation. In addition, Polycomb proteins play an important role in epigenetic gene silencing in X-chromosome inactivation and imprinting. Furthermore, Polycomb target genes include transcription factors (TFs), signaling genes that play a main role in cell fate determination, and tumor suppressor genes, which function to prevent uncontrolled cell proliferation. The PRC2 complex may also serve as a recruiting platform for DNA methyltransferases, thereby linking these two epigenetic repression systems.

The Role of EZH2 and Its Somatic Mutations in Cancer Mutations in Polycomb proteins cause developmental defects due to the deregulation of gene expression of key TFs, such as the Homeobox proteins. Polycomb-mediated silencing of those genes prevents the cell from undergoing senescence, which might ultimately lead to the occurrence of cancer. EZH2 overexpression is observed in many tumors and it correlates with a poor prognosis in several tumor types. EZH2 overexpression can lead to the hyper-silencing of differentiating genes or tumor suppressor genes, which can be reverted pharmacologically with EZH2 inhibitors. However, other studies reported that EZH2 has both oncogenic and tumor suppressive roles. The oncogenic histone H3K27M missense mutation identified in pediatric glioblastoma inhibits the PRC2 complex by binding to the catalytic active site of EZH2 and blocking its activity. Moreover, EZH2 methylates nonhistone proteins including the TF GATA4 and retinoic acid receptor alpha, which also could have an effect in carcinogenesis. Next-generation sequencing of follicular lymphoma and diffuse-large B-cell lymphoma has revealed recurrent somatic mutations within the catalytic SET domain of EZH2 very frequently at Y641 (mainly to F, N, S, and H) and less frequently at A677 (mainly to G). Heterozygosity and the presence of equal quantities of both mutant and normal mRNA as well as expressed protein suggest a dominant gain-of-function mode of action. The Y641 residue is part of the so-called aromatic cage in the active center of EZH2 that surrounds the target lysine ε-amino group. For several PKMTs it was reported that mutations of aromatic cage residues affect the substrate preference and product pattern. The structure of the wild-type and mutant SET domain of EZH2 showed that the hydroxyl group of Y641 (supported by the carbonyl of A677) forms a hydrogen bond to one proton of the lysine ε-amino group, which prevents trimethylation of the target lysine. As a consequence, wild-type EZH2 introduces mono- and dimethylation of H3K27 preferentially using the unmethylated peptide as substrate, while the Y641 mutants prefer mono- and dimethylated peptide substrates to introduce trimethylation. Therefore, these mutations change the methylation level of the product of the EZH2 reaction. EZH2 overexpression or activating mutations likely have a common carcinogenic mechanism based on the hyper-silencing of differentiating genes or tumor suppressor genes. The dominant mode of action of these hyperactive variants suggests EZH2 inhibitors as therapeutic strategy for cancers containing this mutation or overexpressed EZH2. Indeed, several efficient EZH2 inhibitors were developed showing promising effects in animal models and first clinical studies. This development illustrates that understanding of the mechanism of cancer mutations in PKMTs is an important step for targeted and individualized cancer treatment.

The MLL1 and MLL3 H3K4 PKMTs The mixed lineage leukemia (MLL) family of PKMTs are composed of six members: MLL1–4, SET1A, and SET1B. Each enzyme has H3K4 methyltransferase activity, but they are not redundant and have specific cellular functions. H3K4 can be mono-, di-, and trimethylated and specific MLL proteins introduce defined methylation states of H3K4 at distinct genomic regions. MLL1/2 (KMT2A/B) are trimethyltransferases at promotor regions of genes involved in early development and hematopoiesis, MLL3/4 (KMT2C/D) introduce monomethylated H3K4 at enhancer elements, and the SET1A and SET1B proteins contribute to the bulk of H3K4 trimethylation in mammals. To achieve full methyltransferase activity, MLL proteins require the formation of a core complex containing tryptophan-aspartate repeat protein-5 (WDR5), retinoblastoma-binding protein-5 (RBBP5), and absent small homeotic-2-like (ASH2L) (WRA complex). Interaction with complex partners reorients the SET-I helix in the SET domain switching the enzyme from an inactive autoinhibited state into an active conformation. This effect is mainly caused by the interaction with the RBBP5/ASH2L heterodimer, leading to efficient activation of MLL3 in the presence of these two factors. The interaction of MLL1 with the RBBP5/ASH2L heterodimer is weaker compared to the other MLL enzymes. Therefore, MLL1 requires WDR5 for efficient binding of the RBBP5/ASH2L heterodimer such that MLL1 only exhibits full methyltransferase activity in the presence of all three complex partners. The isolated SET domains of MLL1 and MLL2 also show activity and monomethylate H3K4. In complex with WDR5, ASH2L, and RBBP5 they trimethylate the substrate lysine residue. Although the MLL1 protein contributes to only about 5% of the global H3K4 trimethylation, it is majorly responsible for the regulation of HOX genes and loss of WDR5 expression in cells reduced the H3K4 methylation of HOX gene promoters and decreased Hox gene expression. In addition to the core complex partners, different MLL complexes contain variable additional subunits involved in more specific targeting and regulation.

The Role of MLL1 and Its Somatic Mutations in Cancer MLL1 undergoes numerous chromosomal translocations with more than 70 different partner genes that are found in several human acute leukemia. AF4, AF9, AF10, and ENL are reported to form fusion partners in more than 75% of MLL associated leukemia; these fusion genes are sufficient to transform normal cells to leukemic cells. In these chromosome translocations, the N-terminus of MLL1 is fused in frame with the partner proteins such that the fusion proteins retain the DNA binding and targeting domains of MLL1, but they lose the C-terminal catalytic SET domain. This leads to dysregulation of developmental genes, like HoxA9 and Meis1.

544

Mutations in Histone Lysine Methyltransferases and Demethylases

Aggressive leukemia in children and adults with MLL1 gene translocations show high expression of HOX genes. MLL1 knockdown affects cell cycle-regulating genes especially cyclin A, cyclin B, and p57, which results in a cell cycle arrest in the G2/M phase and apoptosis in cultured cells. Malignant cells are more sensitive to MLL1 knockdown than the normal cells, which might be due to the essential requirement of the MLL1 gene for the high proliferation of malignant cells. Consequently, knockdown of MLL1 in a xenograft mouse tumor model was shown to suppress tumor growth. In summary, MLL1 is an oncogenic protein; it is overexpressed in tumors or it forms fusion partners with other genes finally enhancing the expression of HOX genes. Similarly, MLL2, the closest paralog of MLL1, functions as oncogene and its deletion led to a reduction in the survival of MLL–AF9 leukemia cells. In addition, the combined loss of MLL1 and MLL2 further reduced AML growth. These discoveries indicate that addressing MLL1 and/or MLL2 could be valuable therapeutic strategies. Several somatic cancer mutations were found in MLL1 leading to nonsense, missense, and frame-shift mutations (Fig. 1), but only few of them have been studied mechanistically. The catalytic activity of four MLL1 cancer mutants found in different tumors was determined in absence and presence of the complex partners WDR5, RBBP5, and ASH2L. Two mutants, R3903H and S3865F, showed partial or complete loss of activity. This result was in agreement with the relatively low Sense Score of MLL1 mutants. However, two mutants, R3841W and R3864C, were more active than wild-type MLL1 and both were not stimulated by the WRA complex. Because of their hyperactivity and the lost control of activity by the WRA complex, these mutants are likely to cause gain-of-function effects. Mechanistically, all these effects were triggered by conformational changes of the enzyme into more stable active or inactive states. These data show that different somatic cancer mutations of MLL1 found in different tumors can lead to inhibition or stimulation of MLL1 combined with a loss of the regulation of its methyltransferase activity by complex partner proteins. This finding highlights the multifaceted roles of MLL1 in cell fate determination and gene regulation. Furthermore, the inhibition of the cancer mutants by the MM-102 inhibitor was tested, which disrupts the interaction between MLL1 and WDR5 leading to growth inhibition of MLL-AF9 leukemia cells. In line with the enzymatic results described above, the activity of R3841W and R3864C, which did not depend on WDR5 binding for activity, was not reduced by MM-102, indicating that this compound would not be a therapeutic option for patients with these mutations in MLL1. This observation highlights the importance of genetic analysis of tumor samples combined with specific functional investigation of the effect of somatic tumor mutations in the design of individualized therapies.

The Role of MLL3 and Its Somatic Mutations in Cancer MLL3 is often deleted in myeloid leukemia and inactivation of MLL3 in mice leads to epithelial tumor formation, which suggests that the methyltransferase activity of MLL3 suppresses tumorigenesis. MLL3 expression was also decreased by at least twofold in almost half of the breast cancer patients when compared to the matched normal tissue, but the expression changes did not correlate with the tumor progression and the pathological effects of changes in the MLL3 expression in breast cancer are still ambiguous. In addition, in colorectal cancer cell lines the MLL3 promoter is hypermethylated, which results in the repression of MLL3 protein expression. Taken together, these findings indicate that MLL3 has a tumor suppressor role in several cancers. MLL3 and MLL4 are frequently mutated in cancers (Fig. 1) but their pathological effects are not known in most cases. MLL3 shows a high frequency of loss-of-function mutations (Fig. 3), which are spread over the entire protein including the PHD fingers and SET domain. Three mutations located in the catalytic SET domain of MLL3 (S4757C, N4848S, and Y4884C) were investigated regarding their effects on critical enzymatic properties. One mutation, N4848S, led to a loss of catalytic activity, which is in agreement with the location of N4848 in the AdoMet binding site of MLL3. This result is in line with the evidence for a tumor suppressive function of MLL3 described above. However, another mutation, Y4884C, located in the aromatic pocket of the SET domain, showed a striking gain-of-function effect. Kinetic data demonstrate that MLL3 wild-type preferentially transfers one single methyl group to unmethylated H3K4 generating H3K4me1. In contrast, in vitro methylation studies showed that the Y4884C mutation converts MLL3 into a trimethyltransferase with H3K4me1 as preferred substrate. This result was further confirmed by global histone H3K4me3 methylation analysis from human cell lines. Hence, the mutagenic phenotype and pathogenic mechanism of Y4884C mirrors that of Y641 mutations in EZH2. By generation of abnormal H3K4me3 instead of H3K4me1, enhancer regions could be reprogrammed to function as promoters leading to abnormal gene expression. Interestingly, like in MLL1, these results suggest that each MLL3 mutant has individual roles during tumorigenesis, which may be related to the corresponding tumor types and tumorigenic pathways in different cell types. MLL3 inhibitors might be beneficial in patients carrying the Y4884C mutation, while they are unlikely to be effective in cases with inactive MLL3. MLL4 is a H3K4 monomethyltransferase belonging to the same subfamily as MLL3. It is often found mutated in cancers and a large fraction of the MLL4 cancer mutations are nonsense or frame-shift mutations which inactivate the enzyme (Fig. 3), indicating a tumor suppressor role of MLL4 at least in some cell types. Somatic loss-of-function mutations of MLL4 were observed on a single copy of the gene suggesting that MLL4 acts as a haploinsufficient tumor suppressor.

The DOT1L H3K79 PKMT DOT1L was initially identified in a genetic screen aiming to identify proteins involved in telomeric silencing in yeast. The yeast and human homolog was later demonstrated to methylate K79 of histone H3, which is located in the core structure of histone H3. DOT1L is a unique histone lysine methyltransferase because it does not contain a SET domain. Instead, its active site is located in a seven-ß-strand Rossmann-fold domain, also found in other class I methyltransferases, like protein arginine and DNA

Mutations in Histone Lysine Methyltransferases and Demethylases

545

methyltransferases and other PKMTs, which methylate nonhistone substrates. H3K79 methylation is associated with active gene transcription and it is also involved in cell cycle regulation. H3K79me3 and me2 are majorly localized in the coding regions of active genes, whereas H3K79me1 is broadly distributed across the coding region. DOT1L-mediated H3K79 methylation is important for embryonic development and a germline knockout of the mouse DOT1L homolog causes embryonic lethality. DOT1L is the sole enzyme that methylates H3K79 indicating that it represents the main target protein in diseases related to the aberrant H3K79 methylation.

The Role of DOT1L and Its Somatic Mutations in Cancer During the last years, several reports have demonstrated that DOT1L is linked to MLL-mediated leukemia. As described above, 70% of the infant leukemia patients were found to have MLL translocations, in which the N-terminal part of MLL containing its DNA recognition motifs is expressed as a fusion protein with several partner proteins like AF10, AF9, AF4, or ENL. These fusion proteins retain the MLL DNA binding and chromatin targeting domains, which results in an altered MLL target gene regulation depending on the fusion partner. Interestingly, it was found that all of these fusion proteins physically interact with DOT1L either directly or indirectly. This leads to a recruitment of DOT1L to MLL target loci. The resulting increased H3K79 methylation is a positive transcription mark that, bypassing the normal transcription regulation, causes the expression of proleukemogenic genes (like HOXA9 and MEIS1) and by this stimulates the development of leukemia. H3K79 trimethylation plays a crucial role in the regulation of HOX genes such as Hoxa9, Hoxa7, and Meis1, which are overexpressed after aberrant H3K79 methylation in leukemic cells. Thus, DOT1L is an example in which the mistargeting of a PKMT causes methylation at aberrant genomic loci. By rational drug design and genetic approaches, it was shown that either inhibition of the activity of DOT1L or loss of the DOT1L protein has dramatic effects on the expression of HOX genes. A DOT1L-specific competitive inhibitor of AdoMet binding selectively inhibited H3K79 methylation and blocked the expression of leukemogenic genes. This compound was efficient in killing leukemic cells with MLL translocations with relatively small effect on non-MLL leukemic cells. This result shows that the methyltransferase activity of DOT1L plays a central role in the MLL fusion-mediated cellular transformation and maintenance of the transformed state and it is a very important drug target. In addition to its pathological function in MLL-mediated leukemia, H3K79 hypermethylation was observed in lung cancer cell lines and tissues. Downregulation of DOT1L in lung cancer cell lines blocked the proliferation of cells, which further supports the oncogenic role of DOT1L. Several somatic cancer mutations in DOT1L have been identified (Fig. 1), but their mechanism has not yet been described. Based on the known role of DOT1L in cancer, gain-of-function or hyperactivation effects of the mutations are likely, which is also in agreement with the high Sense Score of DOT1L somatic tumor mutations (Fig. 3).

The SETD2 H3K36 PKMT SETD2 was first described as Huntingtin-interacting protein B (HYPB). Later, studies have uncovered that it can trimethylate H3K36 via its SET domain, leading to its renaming to SETD2. The SETD2 protein interacts with the Ser2/Ser5 hyperphosphorylated RNA polymerase 2 during transcriptional elongation via its SRI (Set2 Rpb1 interacting) domain, which explains why H3K36 trimethylation is enriched in the body of actively transcribed genes. In humans, H3K36 mono- and dimethylation is introduced by several PKMTs (including the NSD enzymes), while H3K36me3 is only generated by SETD2.

The Role of SETD2 and Its Somatic Mutations in Cancer The SETD2 protein interacts with p53 via its C-terminal SET and WW domains and stimulates p53 activity. Knockdown of SETD2 lowered the expression of different p53 target genes such as Puma, Noxa, Huntingtin, and p21. In addition, SETD2 also increases the stability of the p53 protein by lowering the expression of the HDMT2 Ring finger-type E3 ubiquitin ligase, which triggers degradation of p53 by ubiquitination. SETD2 mRNA expression was lost in human breast malignant tissue and the SETD2 expression levels were negatively correlated with the tumor stage. These data suggest that SETD2 can be used as a biomarker for the breast cancer and SETD2 acts as a tumor suppressor gene. SETD2 is among the PKMTs with the highest frequency of somatic cancer mutations (Fig. 1). Somatic mutations of SETD2 are often observed in high grade gliomas, clear cell renal cell carcinoma, and leukemias and the majority of them are nonsense and frame-shift mutations, which lead to loss-of-function (Fig. 3). Gliomas with disruption of SETD2 showed a considerable decrease in H3K36 trimethylation, indicating that the SETD2 mutants negatively affect the activity of the enzyme. SETD2 truncating mutations are majorly found in the C-terminus, which either leads to the loss of the SET or SRI domains; the former leads to loss of PKMT activity and the latter disrupts the interaction with the phosphorylated RNA polymerase II. The SETD2 D1616N mutation that is located in the SET domain of the protein has been identified in the cRCC cell lines, which show a considerable loss of H3K36 trimethylation, similarly as other cRCC tumor cell lines containing mutations in the SETD2 protein. Finally, K36M mutations in histone H3, found in chondroblastomas at frequencies of up to 96%, inhibit SETD2 and lead to genome wide loss of K36me3 accompanied by gains in K27me3 and changes in the expression of cancer-associated genes. All these results document that the loss of SETD2 function plays a pivotal role in the initiation and progression of cancer, and somatic cancer mutations of SETD2 are examples of loss-of-function.

546

Mutations in Histone Lysine Methyltransferases and Demethylases

The NSD Family of H3K36 PKMTs The nuclear receptor-binding SET domain containing protein (NSD) family consists of three members: NSD1, NSD2, and NSD3. They differ in protein size and domain organization and have specific biological functions in normal development and pathophysiology. All NSD enzymes contain a catalytic SET domain, in addition to several PHD domains and two PWWP domains. The SET domain introduces H3K36 mono- and dimethylation using unmethylated H3K36 as substrate in vivo and in vitro. The other domains mediate the interaction with chromatin and other proteins. NSD1 and NSD2 were reported to methylate several other substrates including H1.5 K168, which is efficiently methylated by NSD1, and H4K44, which is methylated by NSD1 and NSD2. Defects in the NSD family proteins are implicated in several diseases. Mutations and deletions in NSD1 lead to SOTOS syndrome, a developmental disease associated with fetal overgrowth. Loss of NSD2 activity causes Wolf-Hirschhorn syndrome (WHS) which is characterized by several developmental defects, growth retardation, and brain anomalies leading to mental retardation.

The Role of NSD Enzymes and Their Somatic Mutations in Cancer An increasing number of reports document defects in NSD family members in several cancers: NSD1 is associated with acute myeloid leukemia and multiple leukemia, NSD2 is linked to prostate cancer and myeloid leukemia, and NSD3 is associated with breast cancer and lung cancer. Protein expression studies with 3000 different tumor samples revealed higher NSD2 expression when compared to normal tissue. Moreover, NSD2 expression positively correlated with the progression and advanced tumor stage. The cells with high levels of NSD2 contained enriched H3K36me2 and lower levels of H3K27 trimethylation and they showed a change in the distribution of H3K36me2, which is enriched in gene bodies in normal cells, but dispersed in the NSD2 overexpressing cells. Depletion of NSD2 in a xenograft mouse tumor model showed a decrease of H3K36me2 and also a reduction in tumor growth. Moreover, NSD proteins are involved in chromosomal translocations forming different fusion proteins in several cancers including fusions to the nucleoporin gene (NUP98) in NUP98-NSD1 or NUP98-NSD3, which are found in acute myeloid leukemia. The fusions enhance HOXA5-A10 gene expression via increasing H3K36 dimethylation. In summary, the deregulation of NSD proteins plays a causative role in cancers and the histone methyltransferase activity of these proteins is critical for promoting malignant growth. Recent genomic sequence data of cancer cell lines revealed somatic mutations in NSD enzymes (Fig. 1), which especially in the case of NSD2 suggest a gain-of-function mechanism (Fig. 3). The E1099K mutation was found at very high frequency in NSD2 mostly in pediatric lymphoid malignancies, such as hypodiploid acute lymphoid leukemia (ALL), chronic lymphocytic leukemia (CLL), multiple myeloma, and lung or stomach adenocarcinoma. E1099 is located in the SET domain in a loop next to the histone binding groove and in vitro methylation assays have shown higher activity of NSD2 E1099K on nucleosomes. In addition, increased H3K36 dimethylation and decreased H3K27me3 were observed in cell lines harboring the E1099K mutant, also suggesting an increased H3K36 methyltransferase activity. However, different from the EZH2 mutations described above, the E1099K mutation enhances the methyltransferase activity of NSD2 without apparent changes of the product specificity. In addition, ALL tumors frequently contain a D1125N mutation in NSD2; this variant is also associated with higher H3K36me2 methyltransferase activity. Hence, the somatic cancer mutations in NSD2 are examples of a gain-of-function similarly as the overexpression of NSD2. These observations indicate that NSD2 has a vital role as oncoprotein and is a very promising drug target in particular for tumors with overexpressed or hyperactive NSD2.

The SMYD PKMT Family The SMYD PKMT protein family consists of five members named SMYD1–5. SMYD proteins are not well characterized; among them SMYD2 and SMYD3 are best studied. They were grouped based on their similar domain architecture containing a split SET domain, which carries an inserted MYND (Myeloid, Nervy, and DEAF-1) domain. The MYND domain is responsible for protein–protein interactions and the split SET domain contains the catalytic elements as in other SET domain enzymes. SMYD2 has been reported to monomethylate several lysine residues on histone and nonhistone proteins. It was initially shown to methylate H3K36 but later it was reported that the interaction with HSP90a changes its specificity toward H3K4. Additional studies showed that SMYD2 also methylates K266 of estrogen receptor alpha, K370 of p53, and K810 and K860 of the retinoblastoma (RB) protein. SMYD3 was initially identified as H3K4 di- and trimethyltransferase. Later reports showed that SMYD3 can also methylate H4K20 with a preference for the dimethylated lysine as a substrate, and that it trimethylates histone H4 at K5. Apart from that, SMYD3 also methylates nonhistone proteins, like vascular endothelial growth factor receptor 1 (VEGFRI) at K831, which enhances its kinase activity. Overall, the methylation site(s) of SMYD3 are still ambiguous and the full spectrum of substrate proteins is not well known.

The Role of SMYD Family Enzymes and Their Somatic Mutations in Cancer Methylation of nonhistone targets by SMYD2 has clear connections to cancer and overexpression of SMYD2 has a tumor-promoting effect. For example, SMYD2 methylation of p53 at K370 impairs the tumor suppressor activity of p53. Moreover, methylation of the Rb protein at K810 enhances its phosphorylation at S807 and S811 leading to the dissociation of Rb from E2F, which enhances its transcriptional activity and promotes cell cycle progression. SMYD2 is significantly overexpressed in several cancers. For example,

Mutations in Histone Lysine Methyltransferases and Demethylases

547

the SMYD2 mRNA levels were reported to be almost 8-fold increased in leukemic cells, when compared with normal bone marrow controls and patients with high SMYD2 expression showed lower survival rate. Similarly, SMYD2 protein overexpression was identified in esophageal squamous cell carcinoma (ESCC), bladder, and gastric cancer. Knockdown of SMYD2 in ESCC cell lines with overexpression of SMYD2 led to the suppression of proliferation due to a G0–G1 arrest, suggesting that SMYD2 inhibitors might be good cancer drugs in these cases. In summary overexpression of SMYD2 is frequently observed in several cancers and it is correlated with lower survival rate indicating that SMYD2 acts as an oncoprotein. Similarly as SMYD2, SMYD3 acts as an oncogenic protein. Elevated levels of SMYD3 expression were observed in colorectal carcinoma (CRC), hepatocellular carcinoma (HRC), breast, pancreatic, gastric, and lung cancer cells, whereas weak or no expression was observed in noncancerous cells of other tissues. In agreement with this, suppression of SMYD3 significantly reduced the growth of CRC and HRC cells. In breast cancer cell lines, a truncated SMYD3 protein lacking the first 34 amino acids was observed as well, which was shown to have a higher activity than the full-length SMYD3. Moreover, it has been shown that SMYD3 upregulates the expression of the oncogene matrix metalloproteinase MMP-9, which is involved in tumor progression and metastasis by stimulating cell migration, tumor invasion, and angiogenesis. Suppression of SMYD3 reduces MMP9 gene expression supporting its oncogenic property. Furthermore, SMYD3 methylation of additional nonhistone substrates is a key event in Ras-driven carcinomas. SMYD3 methylates the MAP3K2 kinase at K260, which increases MAP kinase signaling and leads to the formation of Ras-driven carcinomas. In addition, MAP3K2-K260 methylation inhibits binding of PPP2 phosphatase complex, which is a cellular phosphatase and a key negative regulator of MAP kinase signaling pathway. Low SMYD4 gene expression has been reported in breast cancer cells where it was correlated to higher expression of the plateletderived growth factor receptor a (Pdgfr-a). In several mammalian cancers, Pdgfr-a is an oncogene, which is involved in the proliferation and survival of tumors. Re-expression of SMYD4 in breast cancer cell lines repressed the Pdgfr-a gene, suggesting that SMYD4 acts as potential tumor suppressor at least in this cellular model system. Altogether these data suggest that the SMYD proteins play vital roles in regulating the expression of oncogenes in several cancers with SMYD4 acting as tumor suppressor while SMYD3 and SMYD2 function as oncogenes. Recent genomic sequencing in cancer patients revealed several somatic mutations in SMYD proteins with SMYD1 and SMYD3 showing the highest proportion of mutations (Fig. 1). The somatic mutations in SMYD PKMTs include missense and nonsense mutations as well as insertions and deletions. Interestingly, the mutational patterns are indicative of a gain-of-function effect in particular for SMYD1, but also SMYD4 and SMYD5 (Fig. 3). This finding is not in agreement with the observation that SMYD4 acts as tumor suppressor in breast cancer cells as described above, suggesting that the role of SMYD4 is complex and tumor type specific. Since SMYD proteins methylate several histone and nonhistone substrates they may play roles in gene expression regulation and tumor formation. It will be important to determine the full substrate spectrum of all SMYD proteins and understand the pathogenic mechanism of the somatic missense mutations on the methylation of different substrate proteins. Hyperactive somatic mutations could have the same effect as the overexpression of SMYD2 or SMYD3 in cancer cells and they may, for example, aberrantly increase the lysine methylation of tumor suppressor p53 and decrease its tumor suppressor function. Additionally, it will be interesting to see whether SMYD protein mutations may change the product specificity of the PKMT, although such an effect still awaits experimental validation.

Roles of KDMs and Their Known Somatic Mutations in Cancer The UTX H3K27 KDM Family The Ubiquitously Transcribed Tetratricopeptide repeat on Chromosome X (UTX) demethylase, also known as KDM6A, is a member of the Jumonji-containing lysine demethylase family. UTX catalyzes the demethylation of H3K27me2/3 in an Fe(II)-dependent dioxygenase reaction using a-ketoglutarate and oxygen as cosubstrates with the production of formaldehyde and succinate. UTX was found as a specific complex partner of the MLL3/4, which catalyze the H3K4 monomethylation at enhancers of active genes. H3K4 and H3K27 have antagonistic roles, as H3K27me2/3 leads to the repression of the same target genes, which are activated by H3K4 methylation. By demethylation of H3K27, UTX supports the maintenance of the H3K4 activation mark set by MLL3/4. UTX knockdown studies have documented its essential role in embryonic development, tissue-specific differentiation, and its requirement for correct reprogramming.

The Role of UTX and Its Somatic Mutations in Cancer Deletion of UTX and point mutations in UTX cause the Kabuki Syndrome (KS) characterized by distinctive facial features, developmental delay, mild to moderate intellectual disability, postnatal growth retardation, and additional features such as skeletal anomalies. Previously, mutations in MLL4 were found in about half of all KS patients indicating that the switching from H3K27 methylated silenced to H3K4 methylated active chromatin states is disturbed in this disease. Moreover, UTX was the first demethylase that was found somatically mutated and associated with human cancer. Cancer mutations in the UTX gene include homozygous (in females) or hemizygous (in males) large deletions, nonsense mutations, small frame-shifting insertion/deletions, and consensus splice site mutations that lead to aberrant splicing and premature termination codons. Inactivating UTX mutations and deletion were found in several cancer types, including multiple myeloma, leukemia, esophageal, and renal cancer (Figs. 1 and 3). These data document that UTX acts as a tumor suppressor. Still, it was found that GSKJ4, an inhibitor of UTX and

548

Mutations in Histone Lysine Methyltransferases and Demethylases

JMJD3, is a promising suppressor of breast cancer stem cells (BCSCs) due to its efficient repression of the expansion and self-renewal capacity of BCSCs. In addition, xenograft models confirmed a significant inhibition of tumor progression in vivo. This study suggested that targeting UTX is a therapeutic option for cancer treatment, despite of its tumor suppressive role in some cancers, indicating that PKMTs and KDMs (and their somatic mutations) have tumor-specific effects.

The KDM1 Family The human KDM1 family of histone demethylases, LSD1 (KDM1A) and LSD2 (KDM1B), catalyze the demethylation of mono- and dimethyl marks of H3K4 and H3K9. The enzymes contain a SWIRM and Amine oxidase domain, LSD2 also an N-terminal Zinc finger domain. In LSD1 the Amine oxidase domain is split by an inserted Tower domain. The substrate specificity of LSD1 is influenced by its association with different partners. LSD1 generally demethylates H3K4me1/2, but when LSD1 interacts with the androgen receptor (AR) its enzymatic specificity switches to H3K9me1/2. This dual activity enables LSD1 to regulate both the repression and activation of genes. LSD1 is a component of large transcription regulatory complexes including the nucleosome remodeling and deacetylating (NuRD) complex, RCOR1 complex, and the CtBP corepressor complex. The interaction with RCOR1 is especially relevant, since it stabilizes the protein and strongly promotes its H3K4me1/2 demethylation activity. In embryonic stem cells, KDM1A has been identified at enhancers, in association with the NuRD complex. When associated with these complexes, LSD1 promotes gene silencing by removing activating methyl marks from H3K4. KDM1A-containing complexes are recruited to their respective target genes in the genome by a series of TFs that act in specific cell types during normal development or, potentially aberrantly, in human disease. In contrast, when LSD1 interacts with the androgen or estrogen receptor, it promotes transcriptional activation by demethylating the repressive H3K9me2 (dimethylation of histone H3 at K9) histone modification. In addition, the enzyme can work on nonhistone substrates including p53, DNMT1, E2F1, MYPT1, STAT3, HIV Tat, HSP90, and MTA1. LSD2 has been described to demethylate H3K4me1/2 but the biological roles of LSD2 (KDM1B) are less known.

The Role of KDM1 Enzymes and Their Somatic Mutations in Cancer LSD1 and its downstream targets are involved in a wide range of normal physiological processes, including embryonic development, stem cell maintenance, and differentiation, but it is also a key player in oncogenic processes, including tumor-cell growth and metastasis, compromised differentiation, enhanced cell motility, and metabolic reprogramming. LSD1 has been reported to be overexpressed in a variety of tumors and inactivating LSD1 or downregulating its expression inhibits cancer-cell development. The function of somatic tumor mutations in LSD1 and LSD2 has not yet been described, but the mutational spectrum of LSD2 (Fig. 3) suggests the occurrence of gain-of-function mutations.

The Role of KDM4 Enzymes and Their Somatic Mutations in Cancer The KDM4 family comprises five members (KDM4A-E). The KDM4A, B, and C proteins are very similar to each other and contain a JmjN and JmjC domain, two PHD-, and two Tudor domains. KDM4D and KDM4E lack the C-terminal region including the PHD and Tudor domains. All KDM4 enzyme demethylate H3K9, KDM4A-C also H3K36, and some of them also H3K27. KDM4 enzymes have roles in steroid receptor regulation and they can activate the expression of oncogenes. KDM4 genes were found amplified and overexpressed in various tumor types. The effects of somatic cancer mutations have not yet been described, but a very high Sense Score of KDM4B and KDM4A mutations (Fig. 3) is in agreement with an oncogenic role of these enzymes.

Somatic Mutations in Related Proteins Examples of other frequently mutated epigenetic enzymes not described in detail in this article include the DNA methyltransferase 3A (DNMT3A), the Ten-eleven-translocation 2 methylcytosine hydroxymethyltransferase (TET2), and SWI/SNF chromatin remodeling complexes. DNMT3A is an essential DNA methyltransferase frequently containing an R882H mutation in blood cancer, which has been reported to reduce its activity and changes its oligomerization in a dominant negative way and by this it may alter the DNA methylation pattern. Somatic TET2 mutations are frequently observed in various cancers. These include nonsense, deletion, and missense mutations at very crucial residues for the enzymatic activity thus predicted to inactivate the enzyme. Finally, mutations that inactivate SWI/SNF chromatin remodeling complexes subunits are found in nearly 20% of human cancers. These complexes catalyze the opening or condensation of chromatin and by this help to execute the epigenetic program. In addition, recurrent somatic mutations were found in genes indirectly related to PKMTs and KDMs. Different isocitrate dehydrogenase (IDH) isoforms catalyze the oxidative decarboxylation of isocitrate to a-ketoglutarate (a-KG) in the TCA cycle, while reducing NADþ or NADPþ depending on the isoform catalyzing the reaction. Unexpectedly, frequent mutations were found in IDH enzymes and shown to be involved in the progression of tumors. All discovered IDH mutations are missense mutations in the active site of the enzymes, often affecting R132 in IDH1, and either the analogous arginine residue (R172) or R140 in IDH2. The mutations occur heterozygously, suggesting that the alterations lead to a dominant gain-of-function effect. Unexpectedly, the mutant IDH enzymes were found to catalyze the NADPH-dependent reduction of a-KG to 2-hydroxyglutarate (2-HG). This oncometabolite has a similar structure as a-KG and acts as competitive inhibitor for enzymes that employ a-KG as cofactor, like

Mutations in Histone Lysine Methyltransferases and Demethylases

549

the TET enzymes and the Jumonji family of histone lysine demethylases. In agreement with this observation, hypermethylation of many histone methylation marks was observed following introduction of IDH1 mutations into cells and DNA hypermethylation by IDH mutations has been observed as well. Another example of somatic cancer mutations with indirect epigenetic effects are mutations in the histone proteins themselves. As already mentioned, different cancers frequently contain missense mutations in genes encoding histone H3.3 (H3F3A) and H3.1 (HIST3H1B), which alter methylated lysine residues to methionine, like K27M or K36M. When incorporated into chromatin, the histone mutants were shown to inhibit the corresponding PKMTs leading to massive disturbances of chromatin modification patterns.

Prospective Vision Proteomic and genomic approaches have allowed addressing several mechanisms of carcinogenesis, but cancer is continuously confronting us with new challenges. It is well established that cancers arise as a result of gene mutations and clonal proliferation of cells containing growth-promoting alterations. The specific mutations identified in defined cancer subtypes have important diagnostic significance. For instance, Janus kinase 2 mutation (JAK V2617F) is present in many myoproliferative neoplasma and polycythemia vera and now this mutation is used as diagnostic marker for this condition. In addition, epigenetic alterations were recognized to play important roles in tumor formation. The recent discovery and characterization of somatic tumor mutations in epigenetic enzymes connect these two processes, because in these cases the genetic changes drive the appearance of epigenetic alterations. Two large classes of epigenetic enzymes are PKMTs and KDMs and the investigation of the role of somatic mutations in these enzymes in carcinogenesis is an emerging approach to the mechanistic understanding of tumor formation and progression. While the mechanism of loss-of-function mutations in PKMTs and KDMs can be deduced from the function of the corresponding gene, this is not necessarily true for gain-of-function mutations, which might have diverse pathogenic effects: they might enhance the enzymatic activity, increase the stability of the protein, or alter the cellular targeting. Somatic mutations in PKMTs and KDMs might also alter their association with interacting partners leading to changes in the substrate profile, or recruit the enzymes to specific chromatin loci resulting in the altered expression of particular oncogenes or tumor suppressor genes. The higher activity of a mutated PKMT or KDM might lead to abnormal methylation of histones at particular loci, which either enhance or inhibit the transcription of oncogenes or tumor suppressor genes. Mutations in the catalytic SET domain of PKMTs and JmjC domain of KDMs might change their peptide sequence specificity leading to the methylation or demethylation of novel targets. Mutations may also change the product specificity leading to a change in the final methylation state of the target lysine residues. In most cases, the mechanistic consequences of somatic mutations in PKMTs and KDMs are not yet known and careful biochemical studies will be needed for their elucidation. The study of the pathomechanism of somatic cancer mutations in PKMTs and KDMs holds great promises for several reasons: (1) Different studies have demonstrated that depending on the tumor type, inhibition of a PKMT or KDM or its hyperactivity and loss of regulation can promote tumor formation illustrating the complex and multifaceted role of these epigenetic enzymes in cell fate determination and gene regulation. Therefore, dedicated biochemical investigations are needed for each somatic tumor mutation of important PKMTs and KDMs in order to decipher its pathological role. (2) It will be important in future to investigate the effects of somatic mutations in order to develop individualized cancer treatments. For example, inhibitors developed against specific PKMTs or KDMs might assist treatment of cancers with hyperactive enzyme mutants, but these inhibitors cannot be used to treat the cancers caused by the loss of enzyme activity. This makes the investigation of the pathomechanism of somatic cancer mutations very relevant for the development of novel more causative and individualized tumor therapies. (3) The investigation of somatic cancer mutations in PKMTs and KDMs is also imperative for the general understanding of the role of these enzymes in carcinogenesis. Different processes such as expression changes, abnormal posttranslational modifications (PTM), or somatic mutation can alter the activity of PKMTs or KDMs in cells. However, expression levels and PTM patterns of proteins are very difficult to measure in tumor tissues and these changes may occur only transiently during a critical phase of cellular transformation. In contrast, somatic cancer mutations are stable, and they can be studied in tumor samples, cell lines, or in vitro much more easily. Therefore, studying and understanding of the mechanism of somatic cancer mutations is a powerful tool to unravel the role of the corresponding PKMT and KDM even in cases not involving mutations. (4) The screening for somatic cancer mutations holds great promises in clinical practice, since the detection of mutations occurs at DNA level, which can be done from small samples of cancer tissue. In contrast, the detection of expression changes or aberrant PTMs needs RNA and proteomic methods, which are much more demanding with respect to sample size and sample treatment. Furthermore, simple patient blood samples can be used as normal controls in mutational screenings, while the acquisition of matching “normal” tissue for comparison is sometimes very difficult in expression studies and proteomic studies. In summary, the investigation of the pathomechanism of somatic cancer mutations in PKMTs and KDMs is an important and urgent challenge in biomedical research. For this, relevant assay systems need to be established to investigate the properties of the wild-type enzymes and their cancer mutants in vitro and in cellular settings. These studies will also pave the way toward the generation of novel and more potent and specific PKMT and KDM inhibitors and guide their rational application in clinical studies.

550

Mutations in Histone Lysine Methyltransferases and Demethylases

See also: Mutations in DNA Methyltransferases and Demethylases. Mutations in Chromatin Remodeling Factors.

Further Reading Bannister, A.J., Kouzarides, T., 2011. Regulation of chromatin by histone modifications. Cell Research 21, 381–395. Chi, P., Allis, C.D., Wang, G.G., 2010. Covalent histone modifications–miswritten, misinterpreted and mis-erased in human cancers. Nature Reviews Cancer 10, 457–469. Dawson, M.A., Kouzarides, T., 2012. Cancer epigenetics: From mechanism to therapy. Cell 150, 12–27. Forbes, S.A., Beare, D., Boutselakis, H., Bamford, S., Bindal, N., Tate, J., Cole, C.G., Ward, S., Dawson, E., Ponting, L., Stefancsik, R., Harsha, B., Kok, C.Y., Jia, M., Jubb, H., Sondka, Z., Thompson, S., De, T., Campbell, P.J., 2017. COSMIC: Somatic cancer genetics at high-resolution. Nucleic Acids Research 45, D777–D783. Greer, E.L., Shi, Y., 2012. Histone methylation: A dynamic mark in health, disease and inheritance. Nature Reviews Genetics 13, 343–357. Hamamoto, R., Saloura, V., Nakamura, Y., 2015. Critical roles of non-histone protein lysine methylation in human tumorigenesis. Nature Reviews Cancer 15, 110–124. Huether, R., Dong, L., Chen, X., Wu, G., Parker, M., Wei, L., Ma, J., Edmonson, M.N., Hedlund, E.K., Rusch, M.C., Shurtleff, S.A., Mulder, H.L., Boggs, K., Vadordaria, B., Cheng, J., Yergeau, D., Song, G., Becksfort, J., Lemmon, G., Weber, C., Cai, Z., Dang, J., Walsh, M., Gedman, A.L., Faber, Z., Easton, J., Gruber, T., Kriwacki, R.W., Partridge, J.F., Ding, L., Wilson, R.K., Mardis, E.R., Mullighan, C.G., Gilbertson, R.J., Baker, S.J., Zambetti, G., Ellison, D.W., Zhang, J., Downing, J.R., 2014. The landscape of somatic mutations in epigenetic regulators across 1,000 pediatric cancer genomes. Nature Communications 5, 3630. Kudithipudi, S., Jeltsch, A., 2014. Role of somatic cancer mutations in human protein lysine methyltransferases. Biochimica et biophysica acta (BBA)dreview on cancer 1846, 366–379. Morera, L., Lübbert, M., Jung, M., 2016. Targeting histone methyltransferases and demethylases in clinical trials for cancer therapy. Clinical Epigenetics 8, 57. Suva, M.L., Riggi, N., Bernstein, B.E., 2013. Epigenetic reprogramming in cancer. Science 339, 1567–1570. Tian, X., Zhang, S., Liu, H.M., Zhang, Y.B., Blair, C.A., Mercola, D., Sassone-Corsi, P., Zi, X., 2013. Histone lysine-specific methyltransferases and demethylases in carcinogenesis: New targets for cancer therapy and prevention. Current Cancer Drug Targets 13, 558–579. Watson, I.R., Takahashi, K., Futreal, P.A., Chin, L., 2013. Emerging patterns of somatic mutations in cancer. Nature Reviews Genetics 14, 703–718. Weirich, S., Kudithipudi, S., Jeltsch, A., 2017. Somatic cancer mutations in the MLL1 histone methyltransferase modulate its enzymatic activity and dependence on the WDR5/ RBBP5/ASH2L complex. Molecular Oncology 11, 373–387. Weirich, S., Kudithipudi, S., Kycia, I., Jeltsch, A., 2015. Somatic cancer mutations in the MLL3-SET domain alter the catalytic properties of the enzyme. Clinical Epigenetics 7, 36. You, J.S., Jones, P.A., 2012. Cancer genetics and epigenetics: Two sides of the same coin? Cancer Cell 22, 9–20.

Relevant Websites UniProtKB: http://www.uniprot.org/uniprot/dDescription: The UniProt Knowledgebase (UniProtKB) is the central hub for the collection of annotated, functional information on proteins. Cosmic: http://cancer.sanger.ac.uk/cosmicdDescription: COSMIC is designed to store and display somatic mutation information and contains information relating to human cancers. HIstome: The Histone Infobase: http://www.actrec.gov.in/histome/dDescription: A database of human histones, their posttranslational modifications and modifying enzymes. dbEM A Database of Epigenetic Modifiers: http://crdd.osdd.net/raghava/dbem/dDescription: dbEM is a database of epigenetic modifiers curated from cancerous and normal genomes. It is created to study role of epigenetic proteins in oncogenesis and cancer drug resistance. EpiFactors http://epifactors.autosome.ru/dDescription: EpiFactors is a database for epigenetic factors, corresponding genes and products.

Mutations: Driver Versus Passenger Scott W Piraino, Valentina Thomas, Peter O’Donovan, and Simon J Furney, Royal College of Surgeons in Ireland, Dublin, Ireland © 2019 Elsevier Inc. All rights reserved.

Glossary Driver mutation A driver mutation is causally implicated in tumor development. It has been positively selected for and has contributed to the survival and proliferation of a tumor. Genomic mutation rate This refers to the number of mutations in a genome. In cancer genomic mutations rates are heterogeneous can vary considerably both between individuals and within tumor genomes. Mutational signature A mutational signature is a distinctive profile that has been inferred from mutations in tumor genomes. These signatures are the consequence of exposure to multiple mutational processes. Passenger mutation A passenger mutation is a mutation that has not contributed to tumor development. Proto-oncogene Proto-oncogenes normally require a gain-of-function mutation or chromosomal gain to drive tumorigenesis, and when mutated in cancer they promote proliferation. Tumor heterogeneity This refers to genomic or epigenomic differences between different patients, different tumors from the same patient, or different regions in an individual tumor. Tumor suppressor gene Tumor suppressor genes are generally antiproliferative and in general require inactivation of both alleles to contribute to tumor development.

Introduction In the 1950s Armitage and Doll provided evidence, based on cancer incidence, that tumorigenesis is the end-result of several successive events (e.g., mutations) in a cell. Since then molecular studies have described the transformation of a normal cell into a cancer cell as a multistep process, with each intermediate stage conferring a selective advantage on the cell. For example, in colorectal cancer, the canonical pathway of tumorigenesis involving the gradual accumulation of mutations in APC, KRAS, and then genes such as PIK3CA, SMAD4, and TP53 has been well described. These alterations to specific genes or regions of the genome are a result of aberrations in the DNA sequence or structure (e.g., translocations, mutations, and copy-number alterations). One of the consequences of the inherent genomic instability underlying the tumorigenic process is the collateral accumulation of passenger mutations in cancer genomes that are unlikely to contribute to tumor formation. Only a subset of the mutations present in a tumor cell actually contribute to the survival and proliferation of that tumordin fact, a significant majority of the mutations observed in a tumor cell are “passenger mutations,” which have a neutral or even negative effect on the cell’s survival. In contrast, a small number of mutations in cancer cells are “driver mutations” which actively benefit the survival of the tumor. A driver mutation is not necessarily required for the survival of the tumor at every stage in its life, but by definition must have been actively selected for at some point in a tumor’s evolutionary history. Genes with the potential to carry driver mutations are called cancer genes. The identification of cancer genes and the specific mutations in these genes that act as drivers is a key focus of cancer research. Driver genes may be classified according to the cellular pathway that a gene affectsdgenes that operate in the same pathway tend to be mutated in a mutually exclusive manner in an individual tumor, as a mutation that has the same effect as an already existing mutation will not be selected for. Some cancer genes are mutated more frequently than othersdfor most cancers, the genomic “landscape” of mutations features a small number of “mountains” (genes that are very frequently mutated in that cancer) and a far greater number of “hills” (genes that are only rarely mutated in that cancer). The process of tumorigenesis is driven by alterations to specific genes. At a rudimentary level, these cancer-driving genes can be classified as either proto-oncogenes (e.g., BRAF, KRAS, and MYC) or tumor suppressor genes, for example TP53, PTEN, and CDKN2A. Proto-oncogenes normally require a gain-of-function mutation or chromosomal gain to drive oncogenesis, and when mutated or dysregulated in cancer, they promote proliferation. On the other hand, tumor suppressor genes are generally antiproliferative and in general require inactivation of both alleles to contribute to tumor development. This can occur in many ways, such as loss of function mutations, deletions, or epigenetic silencing. In addition to proto-oncogenes and tumor suppressor genes, genes directly involved in DNA damage response pathways (e.g., base excision repair and mismatch repair genes), which repair various types of damage to DNA, have been proposed as an additional type of cancer gene. More recently, genes with novel roles in tumorigenesis, such as splicing and chromatin remodeling genes, have been implicated in cancer. Driver mutations may activate oncogenes or inactivate tumor suppressor genes. Because there are far more ways to inactivate a protein than to activate it, oncogene driver mutations tend to be focal amplifications or missense mutations at specific codons, whereas tumor suppressor mutations tend to be deletions, frameshifts, splice-site mutations, and nonsense mutations found in

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65045-6

551

552

Mutations: Driver Versus Passenger

a wider range of sites across the gene. This has led to a “20/20” rule of thumb for predicting whether a known cancer gene is more likely to be an oncogene or a tumor suppressordif over 20% of the mutations in a cancer gene are missense and in consistent locations, that gene is likely to be an oncogene, whereas if over 20% of the mutations in a cancer gene are nonsense or otherwise inactivate the protein then the gene is likely to be a tumor suppressor. It is important to note the distinction between a driver gene and a driver mutation. A driver cancer gene is a gene for which there is very strong evidence of a role in tumorigenesis, and which may be affected by driver mutations. A list of cancer driver genes is curated and updated in the Sanger Institute Cancer Gene Census (http:// cancer.sanger.ac.uk/census). However a driver gene may also contain passenger mutations, which do not confer a gain or loss of function at the protein level. Since driver mutations are far rarer than passenger mutations, it would not be practical to functionally test all mutations identified in cancer genomes. Instead, researchers examining cancer genome data predict which mutations are most likely to be drivers. The main methods for doing this are examining the frequency of the mutation across different tumors and predicting the likely functional impact of the observed mutation. Frequency based approaches are based on the fact that driver mutations are by definition actively selected for. This means that a driver mutation should appear more frequently across many samples of a given cancer type than would be expected given the background mutation rate. Function based approaches examine the likely biological impact of a mutation and try to identify putative drivers by looking more closely at mutations likely to inactivate a tumor suppressor or activate an oncogene. Analyzing the likely impact of a mutation may involve predicting the impact of that mutation on the structure of the protein, the evolutionary conservation of the region mutated, the biochemical similarity of the original amino acid and its replacement, the protein domain affected or the protein family of the identified cancer gene. The recent advances in sequencing technologies have revolutionized cancer genome analysis and have created an unprecedented amount of genomic data from tumors. Interpretation of these data has required the development of new analytical and statistical approaches for the identification of cancer driver genes. Furthermore the progress in the field in the last decade has unearthed previously unknown mechanisms of tumor development and has implicated a host of novel driver genes. Strides have been made in understanding factors affecting genomic mutation rates, the mutational processes impacting tumor genomes, and intratumor heterogeneity. Whole genome sequencing data have facilitated the search for noncoding driver mutations in cancer. In the following sections we discuss coding and noncoding cancer driver identification and highlight the progress that has been made, the issues and considerations in driver and passenger mutation identification, discuss how passenger mutations can provide clinically relevant information, and summarize the main cancer genome analysis initiatives worldwide.

Cancer Gene Identification Chromosomal Alterations Aneuploidy is observed frequently in tumor cells. Cytogenetic techniques such as spectral karyotyping (SKY), fluorescence in situ hybridization (FISH), comparative genomic hybridization (CGH), and SNP arrays have been employed in the past to identify small and large-scale chromosomal alterations (e.g., amplifications, gains, losses, and deletions) in cancer. In addition, chromosomal translocations and gene fusion events have been shown to play an important role in tumor development in some cancer types, particularly in leukemias and lymphomas. Perhaps the most well-known chromosomal translocation in cancer is the Philadelphia chromosome, which results in the chimeric, constitutively active tyrosine kinase BCR–ABL fusion protein in chronic myeloid leukemia (CML). This driver translocation event has been the focus of much attention due to the effectiveness of the tyrosine kinase inhibitor Gleevec (imatinib) in treating CML. Numerous other chromosomal translocations in cancer have been described and are catalogued in the Mitelman Database of Chromosome Aberrations in Cancer (https://cgap.nci.nih.gov/Chromosomes/Mitelman). This database is manually curated from published literature and relates the aberrations to tumor characteristics.

Coding Sequence Mutations and Mutational Screens Somatic mutations in the coding sequence of a gene, for example single nucleotide variants (SNVs) and small insertions or deletions (indels) can have varying consequences (Fig. 1). SNVs can be classified as (1) synonymous (no amino acid change), missense (amino acid change), or nonsense (change of the amino acid to a stop codon); and (2) loss of function or gain of function. Indels can result in frameshift mutations (change of the reading frame) or in-frame mutations (no change to the reading frame). Different types of mutations affect genes altered in cancers. For example, oncogenes usually sustain gain-of-function mutations, such as the BRAF V600E mutation in melanoma and other cancers. This mutation leads to the constitutive activation of the protein even in the absence of an activating signal resulting in sustained signaling of the MEK/ERK pathway. Conversely, tumor suppressor genes can be inactivated by loss-of-function mutations. For example, a host of loss of function missense and nonsense mutations has been found in the cell cycle gene CDKN2A in multiple cancer types. To identify novel cancer genes mutational screens in the early 21st century used capillary sequencing to focus on previously implicated signaling pathways and gene families such as the tyrosine kinases, lipid kinases, tyrosine phosphatases, and tyrosine kinase receptors. These studies highlighted the significance of members of these gene families in cancer by identifying recurrent mutations in genes such as BRAF, PI3KCA, and EGFR in multiple cancer types. In 2006, the first genomic mutational screen of the somatic landscape of cancers was conducted in human breast and colorectal tumors. In an initial screen, the authors sequenced the coding sequences of > 13,000 genes in 11 breast and 11 colorectal tumors, removing any silent changes, and known germline

Mutations: Driver Versus Passenger

(A)

(C)

de novo MYB binding site

A C GG T T A

enhancer codon change

GGC

2bp insertion

553

promoter

GA C UTR

KRAS G12D mutation

exon intron

(B)

(D)

de novo ETS binding site

C C CT T C C C228T

NOTCH1 3’ UTR mutation

CAGAA splice-effecting mutation

Fig. 1 Driver mutations can operate by a variety of functional mechanisms. Mutations in enhancers (A) or promoters (B) can alter the regulation of driver genes, effecting their expression or response to transcriptional signals. Coding mutations (C) change the amino acid sequence of a gene and can alter the function of the encoded protein. UTR mutations (D) can have a variety of effects, including altering the splicing of the encoded driver gene. From Piraino, S. W., Furney, S. J. (2016). Beyond the exome: The role of non-coding somatic mutations in cancer. Annals of Oncology 27(2), 240–248.

events or SNPs. Candidate mutations were then sequenced in 24 additional breast and colorectal tumors. They developed a statistical score (cancer mutation prevalence scoredCaMP) to discriminate driver mutations from passenger events. The CaMP score adjusts for the background mutation rate and takes into account the type of the base mutated, the resulting base change, the 50 and 30 neighbors, and the codon usage. Using this approach the study identified 122 and 69 candidate genes for the breast and colorectal tumors, respectively. Among the lessons learnt from this landmark study, and an expanded follow-up in 2007, were that the majority of the genes identified had not been previously implicated in cancer, and that genes showed different biases in the type of nucleotide changes. Furthermore, it highlighted that different genes were mutated in the two cancer types and that samples of the same cancer type showed heterogeneous mutational landscapes. This first large-scale sequencing effort identified important principles and issues in the identification of candidate driver genes and mutations and laid the foundation for subsequent cancer genome profiling initiatives, such as The Cancer Genome Atlas (TCGA), which were to take advantage of the development of new sequencing technologies.

Advances in Sequencing Technologies Rapid technological advances in sequencing technologies over the last decade have paved the way for massively parallel sequencing (MPS) or next-generation sequencing (NGS) enabled cancer genomics. The first cancer genome sequence was reported by the Genome Center at Washington University in 2008. Using MPS, the study sequenced DNA from tumor and normal skin cells from a patient with acute myeloid leukemia (AML) at a coverage of 33 times for the tumor genome and 14 times for the normal genome, finding only eight heterozygous, nonsynonymous somatic single nucleotide variants in the tumor genome. In 2010, the Cancer Genome Project at the Sanger Institute published the whole genome sequences of a malignant melanoma cell line and a small-cell lung cancer cell line. They compared each cell line to the respective normal genomes to identify somatic variants, alterations and distinctive mutational signatures corresponding to DNA damage due to ultraviolet light in the melanoma genome and tobacco smoke in the lung cancer genome. Building on these foundations, the enormous NGS-driven activity in the field of cancer genomics in the last decade has unearthed previously unknown mechanisms of tumor development and has implicated a host of novel driver genes. One of the most prominent examples is the discovery of recurrent somatic mutations in genes that encode spliceosome components in multiple cancer types, including chronic lymphocytic leukemia, uveal melanoma, and breast cancer. The majority of spliceosome-related mutations are found in splicing factor 3B subunit 1 (SF3B1), serine/arginine-rich splicing factor 2 (SRSF2), U2 small nuclear RNA auxiliary factor 1 (U2AF1), and zinc finger RNA-binding motif and serine/arginine-rich 2 (ZRSR2). In addition, mutations in genes implicated in epigenetic regulation and chromatin remodeling processes have been identified in a large number of studies. The mutated genes are involved in a range of processes, including DNA methylation (e.g., DNMT3A, DNMT1), DNA demethylation (e.g., TET1, TET2), histone methylation (e.g., the lysine methyltransferases EZH2, SETD2, and MLL family genes), histone acetylation (e.g., the histone acetyl transferases CREBBP and EP300, and histone deacetylase (HDAC) family genes), the SWI/SNF complex chromatin remodeling complex (e.g., ARID1A, ARID1B, ARID2, SWI/SNF-related

554

Mutations: Driver Versus Passenger

matrix-associated actin-dependent regulator of chromatin (SMARC) genes, and PBRM1), and alternative lengthening of telomeres (e.g., ATRX and DAXX).

Cancer Gene Census In 2004, the Cancer Genome Project published a census of cancer genes, curated from the literature. This database (http://cancer. sanger.ac.uk/census), which is updated regularly, is a list of genes in which mutations have been causally implicated in cancer and contains relevant information including the tumor types in which mutations are found and the type of mutation/alteration affecting each gene. Currently, there are > 560 cancer genes in this database, of which  100 are mutated in the germline setting and > 500 are mutated somatically in cancer. Translocations are the commonest mechanism of altering a gene (55%), followed by missense mutations (43%), nonsense mutations (27%), frameshift mutations (27%), splicing mutations (13%), large deletions (7%), and amplifications (4%). In addition to the discovery of new driver mutations and genes, whole genome sequencing and pan-cancer genome analyses have identified a variety of factors and mutational processes that affect mutation rates in cancer genomes, and confound the differentiation of driver and passenger mutations. For instance, there is considerable variation in mutation rates both within and between cancer types (Fig. 2). In the next two sections, we explore the factors that affect mutation rates and discuss the mutational processes that have been identified in cancer genomes. In this light, it is important to acknowledge the importance of analyzing both driver and passenger mutations to elucidate the mechanisms and variables influencing the acquisition of somatic mutations in cancer genomes. Accounting for these factors can then assist in the refinement of approaches for driver mutation identification.

Factors Impacting Genomic Mutation Rates Somatic mutation rates vary considerably across regions of the genome. Understanding the factors that contribute to this variation is essential for methods for the detection of driver genes, and is also independently of interest in terms of understanding cancer biology (Fig. 3). One of the first genomic features to be associated with genomic mutation rate is gene expression level. Highly transcribed genes generally show lower mutation rates compared to weakly expressed genes. It has also been noted that the transcribed strand of genes shows lower mutations rates compared to the nontranscribed strand, which has been interpreted as suggesting that the impact of gene expression on mutation rate may be due to transcription-coupled nucleotide excision repair. DNA repair has emerged as a central factor in mutation rate variation in cancer genomes. Early efforts to characterize the variability of mutation rate across the genome identified chromatin markers associated with chromatin state as being particularly associated with cancer genome mutation rate. Markers of closed chromatin have generally been associated with higher mutation rate, while open chromatin states have been associated with lower mutation rate. This observed association was given a mechanistic interpretation when chromatin state was linked to access of DNA repair machinery to the underlying DNA in the context of mutation rates. It has been observed that while mismatch repair proficient tumors demonstrate substantial mutation rate variability across the genome, this variability is considerably reduced in mismatch repair deficient tumors. Combined with the observed association of mutation rate with chromatin state, these results suggest a model in which open chromatin state makes DNA more available for repair, resulting in lower mutation rates in these regions when DNA repair is proficient. Other studies have suggested a plausible role for the interaction of DNA repair and chromatin state in more specific chromatin interactions. Reduced mutation rates have been observed in DNase I hypersensitive sites, while this hypomutation is absent in tumors with deficient nucleotide excision repair, again suggesting easier access of repair machinery in open chromatin as a factor influencing mutation rates. While DNase I hypersensitivity sites show reduced mutation rates, increased mutation rates have been

Fig. 2 Mutation rates per Mb across different cancer types. The black bar is the median mutation rate of each cancer type. There is considerable variation in mutation rates both within and between cancer types. Some of the top cancer in terms of median mutation rate (LUSC and LUAD) have a known environmental exposure that can causes large numbers of mutation (tobacco smoke). UCEC and COLO have moderate median mutation rates and long tail of samples with very high mutation rates, possibly as a result of hypermutator tumors within these cancer types. BRCA, breast adenocarcinoma; LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma; UCEC, uterine corpus endometrial carcinoma; GBM, glioblastoma multiforme; HNSC, head and neck squamous cell carcinoma; COLO, colon and rectal carcinoma; BLCA, bladder urothelial carcinoma; KIRC, kidney renal clear cell carcinoma; OV, ovarian serous carcinoma; LAML, acute myeloid leukemia. Data was obtained from the supplemental materials of https:// www.nature.com/nature/journal/v502/n7471/full/nature12634.html.

Mutations: Driver Versus Passenger

555

Genomic factors

Replication and repair processes stochastic errors

transcription chromatin state

DNA polymerase proofreading DNA damage response

replication time APOBEC exposure Immunoglobulin gene hypermutation

tobacco smoke UV light aristolochic acid chemotherapy agents

Environmental exposures Fig. 3 Many factors influence mutations rates both within individual cancer genomes and across individuals. Here we give several examples of factors that influence mutation rates from three categories: environmental exposures, factors that impact DNA replication and repair, and features that vary across the genome. From Piraino, S. W., Furney, S. J. (2016). Beyond the exome: The role of non-coding somatic mutations in cancer. Annals of Oncology 27(2), 240–248.

noted at the actual site of protein binding. Transcription factor binding sites coincide with decreased excision repair activity as measured by excision repair sequencing, and the excess mutation rate at these sites relative to flanking regions is disrupted in XPC-deficient skin cancers, suggesting that this mutation rate variability is again caused by access to DNA repair factors. In colorectal cancer, CTCF binding sites have an excess of mutations, while colorectal tumors harboring mutations that disrupt the proofreading activity of POLE actually have fewer mutations at these sites compared to background mutations, again implicating an interaction between protein binding and DNA repair. Understanding mutation rate variability both within and between cancer genomes has important implications for the detection of driver genes. The number of mutations available for driver gene discover influences the power to detect driver genes. Mutation rate variability can also lead to “false positives” in analyses, genes that have high mutation rates but are not driver genes. Accounting for factors that influence mutation rate can help improve the performance of statistical methods and decrease the likelihood of false positives. Recently, ratiometric methods which use the ratio of different types of mutations as opposed to raw rates have been suggest as a method that can improve performance in the presence of un-modeled variation in mutation rates. In addition to varying across the genome, mutation rates also vary substantially across individuals. Environmental exposures such as tobacco smoke, UV light, and aristolochic acid can result in increased mutation rates in cancer genomes. Mutation rates across individuals are also impacted by variability in the activity of certain cellular processes. In the next section we discuss some of the environmental and endogenous-driven mutational processes that contribute to the heterogeneity of mutation rates.

Mutational Signatures DNA is exposed to constant attacks from a variety of exogenous and endogenous sources. These exposures often trigger mutational processes and result in DNA damage, which results in the accumulation of somatic mutations. Each mutational process affects specific nucleotides and generates patterns of base substitutions, indels or structural variations, thus leaving recognizable imprints on the cancer genome, known as mutational signatures. The recent deluge of data produced by global sequencing initiatives, in tandem with mathematical modeling approaches, has allowed the identification of a growing number of mutational signatures in human cancer and the completion of mutational catalogues. At least 30 distinct mutational signatures have been identified so far, and classified into signatures associated with endogenous mutational processes, signatures associated with exogenous mutational processes and signatures of mutational processes with unknown origins. Mutational signatures offer evidence of the biological history of a cancer, and pan-cancer analyses of next generation sequencing data have revealed that tumor types are characterized by discrete genome-wide mutation profiles, which are a consequence of different exposure to mutational processes. Such mutational records therefore provide the opportunity to identify and examine new mechanisms of mutagenesis.

556

Mutations: Driver Versus Passenger

Current efforts have identified distinct mutational signatures from the analysis of mutations from thousands of cancer genomes (http://cancer.sanger.ac.uk/cosmic/signatures). About half of these have been attributed to known exogenous carcinogenic exposure or endogenous mutational processes (Fig. 4). Experimental exposure to exogenous mutagens such as tobacco smoke and ultraviolet (UV) radiation has been shown to produce specific classes of mutations (mainly G ∙ C/T ∙ A transversions in the former, and C ∙ G/ T ∙ A transitions in the latter) that matched those emerging from analyses of mutated genes in lung cancers and melanomas, respectively. Other mutations, however, occur as a result of disruptions in endogenous processes, such as mismatch repair (MMR) deficiency and DNA replication errors that result in discrete mutational signatures. For example, Signature 1 is the result of an endogenous mutational process characterized by spontaneous deamination of 5-methylcytosine and is correlated with age of cancer diagnosis. The etiologies of several of the mutational signatures remain unelucidated.

Mutational Signatures Linked to Specific Driver Genes A number of the mutational signatures have been linked to defects in or the actions of specific genes. The APOBEC gene family of deaminases has a major role in viral restriction and suppression of retrotransposition, and aberrant activity of APOBEC in human cancer is reported to produce mutational signature 2 and 13. Genes involved in DNA damage response pathways are known to contribute to tumorigenesis in many cancer types. Defects in gene involved in homologous recombination, in particular BRCA1 and BRCA2, result in a range of mutations, which correspond to particular mutational signatures. Defects in genes involved in mismatch repair, which results in microsatellite instability and an increased mutation rate, are associated with four specific mutational signatures. Altered activity of the POLE gene, which encodes the catalytic subunit of an enzyme involved in DNA proofreading, is proposed as the mutational process underlying another mutational signature and has been associated especially with colorectal cancers. Therefore the combination of driver and passenger mutations can provide a powerful insight into the processes underlying tumor development. In addition, recent studies have highlighted how signatures and mutational burdens can be relevant clinically. For example, whole genome sequencing of pancreatic cancers revealed that the BRCA-associated mutation signature

Fig. 4 Some of the mutational signatures identified to date have significant clinical or epidemiological implications. Signatures such as those associated with tobacco smoke (A), ultraviolet light (B) and alkylating agents (E), can reveal previous mutagenic exposure. Signatures associated with altered DNA damage response including deficiency in POLE (D), BRCA (F) and mismatch repair genes (G) may serve as markers for prognosis and response to certain types of therapy. Further signatures have been attributed to the activity of the AID/APOBEC family of cytidine deaminases (C). From Piraino, S. W., Furney, S. J. (2016). Beyond the exome: The role of non-coding somatic mutations in cancer. Annals of Oncology 27(2), 240–248.

Mutations: Driver Versus Passenger

557

was correlated with genomic instability as well as sensitivity to platinum-based therapies, suggesting that some of the mutational signatures may be potential biomarkers for therapeutic response. Furthermore, a number of studies have suggested that sensitivity to immune checkpoint inhibition is associated with the mutational landscape of tumors, with factors such as mutation burden, neoantigen burden, and DNA repair deficiency playing a role in response to therapy.

Methods of Driver Identification Driver mutations confer a selective advantage to the cancer cells in which they occur, and as a result are under positive selection within the tumor. In cancer sequencing data, this positive selection is expected to result in properties that distinguish driver mutations from passenger mutations. Several classes of methods have been developed which exploit these properties in order to use sequencing data to identify putative driver mutations (Fig. 5). The previous two sections have outlined some of the mutational processes and factors that can confound the systematic identification of driver genes. In this section, we discuss the properties that driver mutations display and methods that have been developed to detect driver mutations based on these properties. It is fundamental to note that the foundation of all the approaches we discuss is the identification of somatic variants in cancer genomes, a task in itself that is not trivial. A number of prominent somatic mutation calling algorithms have been developed including

Sequencing data mutations high functional impact

GGAAATATATTCCTGGA (A)

(B)

Impact-based

Cluster-based

GGAAATATATTCCTGGA

GGAAATATATTCCTGGA

annotate function impact

(C)

Rate-based

GGAAATATATTCCTGGA background rate

Identify hotspots

GGAAATATATTCCTGGA GGAAATATATTCCTGGA

test mutation rate vs background

excess of high FI mutations

GGAAATATATTCCTGGA

excess mutations

GGAAATATATTCCTGGA

GGAAATATATTCCTGGA

Prioritised Mutations Fig. 5 Examples of DNA sequences that might be identified by different driver detection methods. In (A), color indicates the functional impact of a mutation, and the sequence shows a large number of highly functionally impactful mutations, and therefore could be identified by functional impact-based methods. In (B) the sequence has highly localized hotspot regions, and could be identified by clustering-based methods. In (C) the sequence shows a large, evenly distributed mutation rate, and could be identified by rate-based methods. From Piraino, S. W., Furney, S. J. (2016). Beyond the exome: The role of non-coding somatic mutations in cancer. Annals of Oncology 27(2), 240–248.

558

Mutations: Driver Versus Passenger

Varscan (http://dkoboldt.github.io/varscan/), SomaticSniper (http://gmt.genome.wustl.edu/packages/somatic-sniper/), Mutect (https://software.broadinstitute.org/gatk/documentation/tooldocs/current/org_broadinstitute_gatk_tools_walkers_cancer_m2_ MuTect2.php), CaVEMan (https://github.com/cancerit/CaVEMan), and Strelka (https://github.com/Illumina/strelka). Each of these approaches for somatic single nucleotide variant identification has its own approach, biases, and limitations. As such, users must remain cognizant that any list of somatic variants output by an algorithm is a combination of true and false positives, and may omit false negative mutations (those that are present in the tumor but were not identified by the algorithm). Furthermore, the computational identification of short insertions and deletions has proved to be a far more difficult task than point mutations to date. Although driver mutations mechanistically impact cancer cells at the level of the individual mutation, mutations within the same functional elements may have similar biological effects. As a result, mutations are often considered in terms of their impact on functional elements. In particular, much cancer genomic research has focused on the identification of driver genes, protein-coding genes that contain driver mutations. We focus on the identification of driver genes, although the concepts discussed here are applicable to other functional elements as well. One of the most prominent properties expected of driver mutations in their increased occurrence in tumor genomes. Driver mutations are not simply generated by mutational processes in cancer genomes and remain unselected for like passenger mutations, but are instead causally related to the development of the tumor. As a result, driver mutations are expected to appear more frequently across multiple independent tumors in cancer sequencing data relative to the base rate that these mutations would occur at due to mutational processes alone. This concept has given rise to a class of methods that seek to identify regions of the genome that display an excess of mutations relative to an expected base rate that is due to mutational processes. Under this model, an excess of mutations is a signature of positive selection, and genes that are mutated more frequently than expected are possible driver genes. The baseline or background model of mutation rates is a critical component of these methods. Seminal work by multiple groups has demonstrated that the mutation rate within cancer cells is not constant across the genome, but instead varies along with genomic features such as gene expression, chromatic state, epigenetic markers, and base composition. As a result, rate-based methods for the identification of driver genes typically use a model for the neutral mutation rate of a gene and statistically compare the observed mutation counts to their expectation under the background model. Examples of rate-based methods include MutSigCV (http://software.broadinstitute.org/cancer/software/genepattern/modules/docs/MutSigCV), and Genome MuSiC (http://gmt. genome.wustl.edu/packages/genome-music/). In additional to methods based on mutation rate, another class of methods focuses on the functional impact of mutations that occur within a gene. Genes under positive selection are expected to display a “functional impact bias,” an excess of mutations that have functional effects compared to mutations that are not as impactful in terms of biological function. Functional impact based methods start with a method of identifying the degree of functional impact that individual mutations have. For example, a simple method could be based on the idea that nonsynonymous mutations are more likely to have large functional consequences compared to synonymous mutations. The functional bias method then looks for genes that tend to have large numbers of high functional impact mutations compared to lower impact mutations. Driver mutations have functional effects that contribute to tumorigenesis, and as a result, functional impact methods identify driver genes as genes that display an excess of functionally impactful mutations. These methods can be extended in novels ways by incorporating different measures of functional impact and by including a model of the expected functional impact score for a gene or genomic element that accounts for potential confounding factors, such as mutational processes. Examples of functional impact based methods include OncodriveFM and its extension OncodriveFML. Appropriately chosen measures of functional impact may also assist in the identification of interesting categories of driver genes. For example, looking for genes with an excess of truncating or other deleterious mutations can help in the identification of tumor suppressor genes, while identification of genes with an excess of mutations targeting phosphorylation sites can be used to identify genes involved in aberrant phosphorylation signaling. A critical component of functional impact methods is variant annotation. Many software tools, databases, and other resources aim to facilitate this task, and some resources are available that bring together multiple existing resources to address this challenge. Within the wider field of genomics and analysis of genetic variants, a host of tools exist to annotate genetic variations, including the Variant Effect Predictor (VEP, http://www.ensembl.org/info/docs/tools/vep/index.html), and ANNOVAR (http://annovar. openbioinformatics.org/en/latest/), which provide general annotation of genetic variants. These general annotations also give access to more specific functional annotations, including tools such as SIFT (http://sift.jcvi.org), PolyPhen (http://genetics.bwh.harvard. edu/pph2/), MutationAssessor (http://mutationassessor.org/r3/), and many others aimed at predicting the functional impact of mutations. Cancer-specific mutation annotation such as CHASM (http://wiki.chasmsoftware.org/index.php/Main_Page), and Oncotator (https://portals.broadinstitute.org/oncotator/) also exist. A third method for the identification of driver genes is based on the observation that driver mutations which target a specific functional element within a protein (e.g., a phosphorylation site) will often be clustered near to each other in the linear sequence of the gene. Genes can contain domains and elements that make up a single functional protein component or serve a common function. If a functional element can be impacted by multiple different mutations along its length, then driver mutations targeting that element may cluster within the region of the gene containing the functional element. As a result, genes that display significant spatial clustering of mutations as opposed to mutations evenly distributed throughout the gene can be classified as putative driver genes.

Mutations: Driver Versus Passenger

559

Noncoding Driver Mutations In addition to coding driver mutations, recent efforts directed at the noncoding regions of the genome have revealed the presence of noncoding driver mutations. Major efforts to characterize the functional properties of noncoding DNA, particularly regulatory regions, have created the opportunity to discover important insights about the role of noncoding variation in disease, including cancer. We review important examples of noncoding driver mutations and the role they play in cancer biology. Among the most important and well-known noncoding driver mutations are TERT promoter mutations (Fig. 1). TERT mutations occur in a mutually exclusive manner and result in the creation of an ETS transcription factor binding site, resulting in increased expression of the TERT transcript. TERT promoter mutations are extremely frequent across multiple cancer types and have been associated with clinical features in multiple cancer types. TERT promoter mutations are an example of noncoding driver variants caused by point mutations. Another noncoding driver mutation that has been identified is generated by a small indel in T-cell acute lymphoblastic leukemia. It has been observed that insertions create a MYB binding site upstream of the TAL1 oncogene, resulting in increased TAL1 expression. There have also been several attempts to systemically identify putative noncoding driver mutations in cancer. These efforts have employed a variety of methods for the identification of putative noncoding driver regions, including rate and clustering-based methods similar to those used to examine coding driver genes, methods based on annotations of noncoding mutations such as the impact of the mutation on regulatory motifs, and methods based on correlating the presence of mutations with gene expression across tumors. In addition to identifying prominent noncoding drivers such as TERT, these efforts have also revealed other putative noncoding driver regions such as regions upstream of PLEKHS1, WDR74, SDHD, and MED16. In addition to pan-cancer studies of potential noncoding driver mutations, several large genomic studies in individual cancer types have identified recurrent noncoding mutations. Analysis of breast tumors identified recurrent promoter mutations upstream of one protein-coding gene (FOXA1), and two noncoding RNA genes (RMRP and NEAT1). An analysis in chronic lymphocytic leukemia revealed mutations in the 30 -untranslated region of the gene NOTCH1, resulting in aberrant splicing of the NOTCH1 transcript, as well as mutations in an enhancer putatively targeting the gene PAX5. Notwithstanding the discoveries made by these studies to date, the interrogation of the somatic noncoding genome is complex and encumbered by our lack of knowledge about the function of noncoding DNA relative to coding sequences. The generation of a greater number of whole genome sequences of tumors, through initiatives such as the pan-cancer analysis of whole genomes, in tandem with efforts to understand functional elements in the genome, for example the Encyclopedia of DNA Elements (ENCODE) Project, should elucidate the importance of non-coding elements in tumor development further.

Tumor Heterogeneity Tumor heterogeneity is thought to play a significant role in treatment resistance and failure. Tumor heterogeneity can be classified as: (i) Intertumor heterogeneity, which is differences between tumors in different patients (Fig. 2). This is the focus of many cancer studies, contributes to differential patient responses to therapy and is the basis for precision medicine approaches. (ii) Intersite heterogeneity, which describes differences between distinct tumors within an individual patient (e.g., between primary and metastatic tumors, or between multiple metastatic sites). (iii) Intratumor heterogeneity, which refers to differences between cellular populations in a distinct tumor, the extent of which has been demonstrated through recent multiregion next generation sequencing analyses. In this section, we discuss intersite and intratumor genetic heterogeneity, which are based on the view of cancer as an evolutionary process that can give rise to genetically divergent subclonal populations within tumors. Morphological differences in tumors have been described in numerous pathology studies and cytogenetic studies have revealed intratumor heterogeneity in copy number alterations. Intratumor heterogeneity refers to the fact that cells within a tumor do not share identical genomesddifferent cells acquire different mutations as the tumor grows. In addition to the clonal mutations that are shared by all cells in the tumor and that were presumably present in the tumor’s ancestral cell, tumor cells may also harbor subclonal mutations, which are unique to a cell or subset of cells in the tumor. These genomic differences can take the form of point mutations, copy number variations or differences in chromosomal structure or number. With the use of next generation sequencing, the extent of intratumor heterogeneity is beginning to be understood more clearly, as is the fact that some tumor types show a greater degree of heterogeneity than others. Furthermore, some driver mutations are more likely to be subclonal than othersdfor example, across several cancer types it has been suggested that mutations in genes involved in the PIK3-AKT-mTOR pathway are more likely than genes involved in the RAS-MAPK pathway to be subclonal. The likelihood of a mutation being clonal or subclonal can depend on the cancer typedfor example, mutations in TP53 are almost always clonal in ovarian cancer and nonsmall cell lung cancer, but are often subclonal in chronic lymphocytic leukemia and clear cell renal cell carcinoma. Since intratumor heterogeneity is only detectable in sequenced samples, detecting subclonal mutations and working out whether apparently clonal mutations are truly clonal is dependent on how many tumor regions are sequenced. Sampling bias in methods used to select regions to sequence may lead to subclonal mutations appearing to be clonal, or to subclonal clinically

560

Mutations: Driver Versus Passenger

relevant driver mutations being missed in sequencing. For example, the TRACERx lung cancer study has shown that significantly more driver events are identified with the use of multiregion sequencing than by analysis of a single sample. In fact, > 75% of the tumors in this study harbored a subclonal driver alteration, with genes such as TP53, PIK3CA, KRAS, and NF1 being affected. As well as directly impacting the biology and clinical course of a cancer, intratumor heterogeneity allows researchers to analyze the evolutionary history of a tumor by extrapolating the temporal sequence in which the driver mutations occurred. Understanding of intratumor heterogeneity of driver mutations and tumor evolution may facilitate more refined management of cancer therapy by using known evolutionary constraints to contain the cancer to slow its progress. However, the existence of intratumor heterogeneity poses a major challenge for targeted therapies of driver mutationsda given therapy may successfully disrupt the major clonal driver mutation, but if cells in the targeted tumor harbor subclonal driver mutations that the therapy does not eradicate, the cancer may recur. The majority of sequencing studies have focused on the cancer genomic landscape of primary tumors. However, several studies have identified intersite heterogeneity between matched primary and metastatic tumors. This is of particular importance if different driver mutations are required for tumor development and for metastatic spread. Furthermore, a metastatic site may be seeded by a cell from the original tumor that harbors subclonal driver mutations. Initial large-scale studies of metastatic tumors, such as the MSK-IMPACT and the Michigan Oncology Sequencing (MI-ONCOSEQ) Program, have begun assess the metastatic genomic landscape. The MSK-IMPACT results were similar when compared to the primary tumors from the TCGA. However, important differences between the two patient cohorts were identified. For example, TP53 was significantly enriched for mutations in several tumor types. However, these studies did not profile matched primary and metastatic tumors from individual patients. Sequencing of matched samples will elucidate further intersite heterogeneity of driver mutations and tumor evolution.

Number of Driver Mutations Cancer is driven by the acquisition of driver mutations that disrupt normal biological processes and allow somatic cells to form detectable lesions. This raises the question, how many driver mutations are typically required for cancer to develop? Early work on this problem used the relationship between cancer incidence and age to attempt to estimate the number of “rate limiting” events that are required before the formation of various cancers, estimating this number in the range of six to seven events. More recent work has attempted to estimate the number of driver mutations required in lung and colorectal cancer by comparing the increased incidence in cancer due to certain risk factors (smoking in lung and genetic predisposition in colorectal) with the increased mutational burden caused by those risk factors. This method estimates that approximately three driver mutations may be required in these cancers. It remains unclear exactly how many driver mutations are required in different cancer types, though cancers of the blood require fewer drivers than solid tumors. Despite this uncertainty, it appears likely that most if not all cancers require more than one driver mutation to proliferate. Additional drivers may be required for metastasis and for relapse after therapy, although subclonal mutations could also be selected for in response to treatment. In addition to their number, there is considerable interest in when mutations occur during the evolution of a tumor. Driver mutations may occur throughout the development of the tumor. The time during tumorigenesis at which certain drivers tend to emerge can give clues about the role the driver plays in tumor development (e.g., the stepwise accumulation of mutations in APC, KRAS, and subsequently genes including PIK3CA, SMAD4, and TP53 in certain colorectal tumors). In this way some drivers may only be selected once other driver mutations have already occurred or during a particular phase of tumor development. The presence and frequency of driver mutations may also change throughout tumor development as certain subclones are selected and increase in frequency within the tumor. The analysis of whole genome sequences from projects such as the pan-cancer analysis of whole genomes and the application of phylogenetic methods to sequencing data from multiple regions of tumors is beginning to elucidate some of the early and late driver mutations that occur during tumor evolution. Indeed, initial analysis of the pan-cancer analysis of whole genomes dataset has suggested that each tumor carries on average 4.6 driver events.

Oncogenomic Resources Interest in cancer genomics in recent years has driven the creation of many publically available resources aimed at facilitating better understanding of the cancer genome.

International Cancer Genome Consortium The International Cancer Genome Consortium (ICGC, http://icgc.org) is a large collaborative project which has produced substantial amounts of genomic data from many countries and cancer types. ICGC provides access to genomic data such as somatic mutations, RNA expression, methylation, and clinical data from large numbers of cancer patients. ICGC provides several ways of accessing data easily and conveniently, including an interactive data portal.

Mutations: Driver Versus Passenger

561

The Cancer Genome Atlas The Cancer Genome Atlas (TCGA, https://cancergenome.nih.gov) is a large collaboration initiated by the National Institutes of Health similar to ICGC which hosts a large set of -omics data. Like ICGC, TCGA also allows access to a vast array of data through the TCGA data portal.

Catalogue of Somatic Mutations in Cancer The Catalogue of Somatic Mutations in Cancer (COSMIC, http://cancer.sanger.ac.uk/cosmic) is a database of cancer genomic information organized by the Wellcome Trust’s Sanger Institute. It includes databases of cancer somatic mutations, as well as several other components, such as curated lists of cancer genes and cancer mutations (e.g., the Cancer Gene Census) and the cancer cell lines project, which provides data on cancer cell lines. COSMIC also provides resources on mutational signatures in cancer (http://cancer.sanger.ac.uk/cosmic/signatures).

Pan-Cancer Analysis of Whole Genomes The pan-cancer analysis of whole genomes (PCAWG, http://docs.icgc.org/pcawg/) is a large collaborative effort to assemble a set of uniformly processed cancer whole genomes from multiple cancer types available in an easily accessible manner. The data from PCAWG are available through the ICGC as well as several other data portals and analysis utilities. PCAWG provides the research community with access to an international sample of over 2800 donors and vast numbers of somatic mutations from a wide array of cancer types.

cBioPortal for Cancer Genomics The cBioPortal for Cancer Genomics (http://www.cbioportal.org/) provides a visualization and analysis platform for cancer genomics data sets, including the TCGA and many smaller studies that have used a variety of sequencing strategies to profile tumors. Results from DNA and/or RNA sequencing of > 45,000 samples can be queried through a user-friendly web interface.

Cancer Genomics Software Analysis of cancer genomic data often makes use of specialized software. Publically available software exists for many important tasks in cancer genomics, including identification of driver genes, mutation annotation, and analysis of intratumor heterogeneity. Several research groups and institutions provide central locations for multiple different cancer genomic software applications, including the McDonnell Genome Institute at Washington University in St. Louis (http://gmt.genome.wustl.edu/index.html), The Broad Institute (https://www.broadinstitute.org/data-software-and-tools), and the Sanger Institute (http://www.sanger.ac.uk/ science/tools). Many other publically available software tools also exist outside these institutional repositories.

Variant Interpretation Several resources also seek to link cancer variants to information that may give insights into the clinical or therapeutic relevance of mutations, such as the Cancer Genome Interpreter (https://www.cancergenomeinterpreter.org/home), OncoKB (http://oncokb. org), and CIViC (https://civic.genome.wustl.edu/about). OncoKB is a precision oncology database, curated by a team of clinicians and investigators at Memorial Sloan Kettering Cancer Center, containing information about the effects of and treatment options for driver mutations or alterations. The database provides information about alterations in > 400 cancer genes and classifies treatment options according to clinical actionability. Clinical Interpretation of Variants in Cancer (CIViC) is an open access, open source database providing information on the prognostic, diagnostic and predisposition of inherited and somatic variants. CIViC has been designed to enable precision medicine and address many of the issues inherent in the clinical interpretation of cancer genome alterations. CIViC currently contains several thousand curated interpretations of clinical relevance for > 1700 variants in 330 cancer genes.

Prospective Vision We have witnessed a genomics revolution as advances in next generation sequencing technologies have resulted in dramatic decreases in sequencing costs. As a consequence, NGS has moved from research and development spheres into the clinical arena to enhance precision medicine for patients. Indeed, over the last decade oncology has been at the forefront of the application of clinical genomics to diagnosis and treatment. Genomic profiling has increasingly become common in many cancer types and clinical trials have been instigated to match patients to targeted therapies based on shared genomic features. With the significant decrease in sequencing costs it has been predicted that millions of cancer patients will have their tumors sequenced over the next decade. One of most the significant and immediate challenges in the field of cancer genomics is the clinical interpretation

562

Mutations: Driver Versus Passenger

of mutational data. The refinement of methods to identify driver mutations and genes, to interpret the clinical significance of specific mutations, and to match patients to therapies based on these mutations and on their genomic profiles, is imperative over the coming years. Finally, the identification of driver mutations promoting recurrence and resistance to therapy will be of significant interest for the foreseeable future.

See also: Chromosome Rearrangements and Translocations. Copy Number Variations in Tumors. Mutational Signatures and the Etiology of Human Cancers.

Further Reading Alexandrov, L.B., et al., 2013. Signatures of mutational processes in human cancer. Nature 500 (7463), 415–421. Kandoth, C., et al., 2013. Mutational landscape and significance across 12 major cancer types. Nature 502 (7471), 333–339. McGranahan, N., Swanton, C., 2017. Clonal heterogeneity and tumor evolution: Past, present, and the future. Cell 168 (4), 613–628. Piraino, S.W., Furney, S.J., 2016. Beyond the exome: The role of non-coding somatic mutations in cancer. Annals of Oncology 27 (2), 240–248. Vogelstein, B., Papadopoulos, N., Velculescu, V.E., Zhou, S., Diaz Jr., L.A., Kinzler, K.W., 2013. Cancer genome landscapes. Science 339 (6127), 1546–1558.

Relevant Websites cBioPortal, n.d. http://www.cbioportal.org/dcBioPortal. CIViC, n.d. https://civic.genome.wustl.edudCIViC. COSMIC, n.d. http://cancer.sanger.ac.uk/cosmicdCOSMIC. ICGC, n.d. http://icgc.orgdInternational Cancer Genome Consortium (ICGC). Mitelman Database, n.d. https://cgap.nci.nih.gov/Chromosomes/MitelmandMitelman Database. TCGA, n.d. https://cancergenome.nih.govdThe Cancer Genome Atlas (TCGA).

Myelodysplastic Syndromes: Mechanisms, Diagnosis, and Treatment Eric Solary, Gustave Roussy, Villejuif, France; and Paris-Sud University, Le Kremlin-Bicêtre, France William Vainchenker, Gustave Roussy, Villejuif, France © 2019 Elsevier Inc. All rights reserved.

Glossary Myelodysplastic syndrome (MDS) Clonal hematopoietic stem-cell disorders that associate cell dysplasia and ineffective hematopoiesis. Myelodysplastic syndrome/myeloproliferative neoplasm (MDS/MPN) Independent category of diseases that associate MDS feature with overproduction of at least one category of myeloid cells. Spliceosome Intracellular multiprotein complex involved in the maturation of messenger RNAs.

Nomenclature ARCH Age-related clonal hematopoiesis ASCT Allogeneic stem cell transplantation CHIP Clonal hematopoiesis of indeterminate potential CMML Chronic myelomonocytic leukemia EPO Erythropoietin ESA Erythropoiesis supporting agents HMA Hypomethylating agents IPSS International Prognostic Scoring System MDS Myelodysplastic syndrome MDS/MPN Myelodysplastic syndrome/myeloproliferative neoplasm MDS-RS Myelodysplastic syndrome with ring sideroblasts sAML Secondary acute myeloid leukemia

Introduction Myelodysplastic syndromes (MDS) are a heterogeneous group of clonal hematopoietic stem-cell disorders characterized by cell dysplasia and ineffective hematopoiesis. These acquired bone marrow failures lead to blood cytopenias with a variable propensity to evolve into acute myeloid leukemia (AML). They most commonly develop in older adults (median age at diagnosis is 65–70 years and less than 10% MDS patients are younger than 50) with a slight male predominance, except for MDS with isolated 5q deletion in which women predominate. The age-adjusted annual incidence rate of these malignancies is estimated to be 4 per 100,000 individuals (reaching at least 75 cases per 100,000 and possibly more after 65 years of age). Risk factors include rare but increasingly recognized inherited genetic predisposition, to be explored in pediatric and young adult cases as well as families with other cases of MDS. Environmental factors include exposure to chemotherapy, radiation, and other toxic insults such as benzene and derivatives and tobacco smoking, but 85% of patients do not have any well-identifiable cause. The pathophysiology of MDS combines sequentially acquired chromosome aberrations and somatic mutations with microenvironmental factors. Disease diagnosis remains mostly based on blood and bone marrow cytological examination. The number and extent of cytopenias, percentage of blast cells in the marrow, and nature of genetic alterations provide prognostic information. Treatment varies from careful monitoring to allogeneic stem cell transplantation. Two therapies, hypomethylating agents and lenalidomide, were approved for these patients. Median overall survival ranges from a few months to almost a decade, depending on age, degree and number of cytopenias, blast percentage, and cytogenetic and genetic aberrations.

MDS Diagnosis An MDS diagnosis is suspected when peripheral blood cytopenias are identified on a routine analysis or a blood exam performed because of nonspecific clinical symptoms: fatigue and destabilization of a preexisting cardiovascular disease because of anemia, repeated infections because of neutropenia and neutrophil dysplasia, bleeding because of thrombocytopenia and platelet dysfunction. Cytopenias can be detected also while exploring an immune disorder including vasculitis, polychondritis,

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65300-X

563

564

Myelodysplastic Syndromes: Mechanisms, Diagnosis, and Treatment

and polyarthritis, which are commonly associated with MDS and often diagnosed simultaneously. Anemia, which is the most frequent cytopenia (90% of MDS), is usually nonregenerative and typically macrocytic. Neutropenia and thrombocytopenia are less frequent (30%), and usually combined with anemia. The presence of circulating blast cells over 1% is used for disease classification while the presence of immature granulocytes is rare. A monocyte count higher than 1000/mm3 and 10% of white blood cell count reclassifies the disease as chronic myelomonocytic leukemia (CMML), which belongs to a distinct category of myeloid malignancies according to the World Health Organization (WHO) and associates myelodysplastic and myeloproliferative features (MDS/MPN). Of note, a fraction of patients diagnosed with an MDS secondarily develop a monocytosis, joining the MDS/MPN category. MDS diagnosis requires morphologic inspection of bone marrow aspiration and sometimes biopsy. The aspirate allows for detailed evaluation of dysplastic features, including ring sideroblasts (erythroblasts with abnormal accumulation of iron in perinuclear mitochondria), and for precise evaluation of the percentage of marrow blasts (to be assessed on 500 nucleated cells). The bone marrow trephine biopsy, whose utility is more controversial, allows for a more accurate determination of bone marrow cellularity and detection of marrow fibrosis. Because morphological assessment of dysplasia can be subjective, especially in the absence of blast excess, interobserved diagnostic discrepancies were observed in up to 20% of patients, especially those with a low grade MDS. Therefore, additional techniques are used to support MDS diagnosis. The most important additional test remains the analysis of bone marrow cell karyotype, as a cytogenetic abnormality is found in 40%–50% of MDS, with partial or complete loss or gain of chromosomes (5q-, 7q-,
7, þ 8, 20q-, 17p-) being the most frequent findings. These heterogeneous abnormalities, whose identification supports the diagnosis when morphologically uncertain, have a strong prognostic significance. For example, complex karyotype is usually associated with an excess of blast cells and a poor outcome. Cytogenetics can also guide the therapeutic choice, for example, lenalidomide is used preferentially to treat MDS with isolated 5q-. The routine 20 metaphase cytogenetic analysis can be completed by fluorescence in situ hybridization (FISH) with probes targeting the most-frequently altered chromosomes when fewer than 20 mitoses have been obtained by routine cytogenetics. The occurrence of additional chromosomal aberrations with evolution can precede transformation to AML. Flow cytometry with increasingly standardized combinations of markers helps recognizing minimal dysplasia by identifying abnormal phenotypic patterns, thus could enter the standard workup in the next future. Specifically, identification of an accumulation of classical monocyte subset in the peripheral blood is a strong argument to predict evolution of an MDS into a CMML. Finally, DNA sequencing has established that MDS arises through the sequential acquisition of somatic mutations in a set of recurrently involved genes. New generation sequencing (NGS) analyses are increasingly included in the routine work-up of patients with MDS as they inform on diagnosis, prognostic, and potential therapeutic targets. In patients with minimal or no diagnostic evidence of dysplasia and no blast excess, additional tests exclude other causes of cytopenia that include aplastic anemia, paroxysmal nocturnal haemoglobinuria clone, toxic exposure, vitamin or iron deficiency, hypersplenism, auto-immune cytopenias, viral infection, hereditary context, and others. When another cause of cytopenia has been excluded, patients with a cytopenia but no dysplasia are considered in the subset of idiopathic cytopenia of unknown significance (ICUS). A fraction of them may have clonal somatic mutations and cytogenetic abnormalities, and their natural history is still poorly known. The 2016 WHO classification of myeloid malignancies still defines MDS subtypes on the basis of the number of dysplastic and cytopenic lineages, the prevalence of blasts, the percentage of ring sideroblasts (RS), and the presence of specific cytogenetic abnormalities. A limited role is given to genetics as the mere presence of recurrent somatic mutations cannot be considered as definitive evidence of MDS diagnosis, for example, can be detected in patients with ICUS. In certain context however, genetic data provide diagnostic utility in MDS, for example, mutations in the spliceosome gene SF3B1 define a subgroup of patients with ring sideroblasts and favorable outcome. Detection of an SF3B1 mutation therefore establishes a diagnosis of MDS-RS in the presence of cytopenias and dysplasia, even when RSs make up as few as 5% of all nucleated erythroid cells (the traditional cutoff is 15%).

MDS Pathophysiology Haploinsufficiency Through Chromosomal Deletion A characteristic event is the occurrence, in a hematopoietic stem cell, of an interstitial deletion in the long arm of chromosome 5, known as del(5)(q31q33), leading to gene haploinsufficiency. The size of the deletion varies but, in MDS patients with an isolated 5q-, a common deleted region that encompasses about 40 genes has been identified, defining the 5q- syndrome. This deletion induces haploinsufficiency for the ribosomal subunit RPS14, which triggers a p53-dependent block in erythroid proliferation and differentiation, and haploinsufficiency of CSNK1A1 (CSK1) and two micro-RNAs, mir-145 and mir-146a, which may account for dysmegakaryopoiesis and thrombocytosis. MDS with complex karyotypes have larger 5q deletions with haploinsufficiency of tumor suppressor genes such as CTTNA1 or EGR1. The selective clonal suppression of del(5q) cells by lenalidomide preserves del(5q) hematopoietic stem cells. A similar haploinsufficiency model might apply also for MDS with del(7q), del(20q), and other recurrent interstitial deletions.

Myelodysplastic Syndromes: Mechanisms, Diagnosis, and Treatment

565

Recurrently Mutated Genes A comprehensive picture of the mutational landscape in MDS has emerged over the past 10 years. With an approach focusing on a set of  50 genes recurrently mutated in myeloid malignancies, a somatic mutation in at least one gene is identified in 90% of patients. These genes encode RNA splicing factors, epigenetic regulators, cohesin components, transcription factors, DNA damage response and signal transduction molecules. These mutations are sequentially acquired, suggesting a disease initiation step with founder mutations driving asymptomatic clonal expansion within the hematopoietic compartment (the so-called age-related clonal hematopoiesisdARCdor clonal hematopoiesis of indeterminate potentialdCHIP), followed by acquisition of mutations that both synergize with the founder mutation on hematopoietic stem cells/progenitors and impair later stages of hematopoiesis and alter blood counts to generate an overt MDS, leading, in 30%–40% of patients, to eventual secondary AML (sAML). The spliceosome is a multiprotein complex that mediates intron excision and exon ligation in the generation of mature messenger RNA. Genes encoding components of the spliceosome, most commonly SF3B1, SRSF2, U2AF1, and ZRSR2, are the most frequently mutated genes detected in MDS (60% of patients). Mutant splicing factors result in altered patterns of premessenger RNA splicing. The splicing factor mutations are generally heterozygous and mutually exclusive, suggesting that either cells cannot tolerate two mutations or these changes have a redundant role in disease pathogenesis. SF3B1 mutation, the most frequent one, is strongly associated with ringed sideroblasts. A second common class of mutations are those affecting genes involved in DNA methylation and histone modification. Recurrent mutations in DNMT3a, a DNA methyltransferase, TET2, an enzyme that hydroxylates methylated cytosines to initiate the process of DNA demethylation, and isocitrate dehydrogenase gene 1 (IDH1) and IDH2 alter the DNA methylation pattern. IDH gene mutations result in the generation of 2-hydroxyglutarate, an oncometabolite that inhibits the activity of numerous targets, including TET2. Components of histone modification complexes are also recurrently mutated in MDS, most commonly ASXL1 and EZH2, which are affected by loss-of-function mutations in approximately 20% and 5% of cases, respectively. Another commonly altered multiprotein complex is cohesin, which is composed of STAG1, STAG2, SMC1A, SMC3, and RAD21. Mutually exclusive loss of function mutations in these components are found in  10% and  20% of low- and high-risk MDS, respectively. These mutations may disrupt DNA loops involved in enhancer-promoter interactions, thus driving aberrant transcriptional programs and complementing loss-of-function mutations in core transcription factors. Mutations in transcription factors such as GATA2 and RUNX1 usually occur somatically. However, they can also be inherited in the germline and cause familial bone marrow failure syndromes with a propensity to evolve into myeloid malignancies. Missense mutations in the tumor suppressor TP53 are highly prevalent among patients with MDS who have undergone chemotherapy. They are frequently associated with the loss of the second TP53 allele via deletion of the short arm of chromosome 17 and their presence correlates with a poor outcome, even with most aggressive treatment regimens.

Pre-MDS Clonal Mutations Recurrent somatic mutations in myeloid malignancy-associated genes, such as DNMT3a, TET2, and ASXL1 can be identified in the peripheral blood of healthy individuals with normal blood counts. These CHIP or ARCH increase the risk of malignant transformation (0.5%–1% per year) through the acquisition of additional mutations (even though the principal cause of morbidity is vascular diseases). Accordingly, mutations in the above-mentioned epigenetic regulators are founder events in MDS pathogenesis. Germline polymorphisms and cell extrinsic factors may promote the rare progression of CHIP into overt MDS or AML. For example, treatment with chemotherapy could enable the preferential expansion of clones carrying mutations in TP53.

Post-MDS Clonal Mutations Evolution to sAML can be considered the final stage of disease progression through the acquisition of characteristic genetic changes, such as activating mutations in signaling molecules such as FLT3 and N-RAS, as well as inactivating mutations in CEBPA. At the time of leukemic transformation, the antecedent MDS clone is outcompeted by aggressive subclones. Of note, mutations in SRSF2, SF3B1, U2AF1, ZRSR2, ASXL1, EZH2, BCOR, and STAG2 identify a subset of patients with AML who have no history of preceding MDS and who share the aggressive clinical behavior of sAML.

Epigenetic Changes Genetic alterations are frequently associated with epigenetic changes arising in hemopoietic stem cells and including aberrant DNA methylation, which was most studied in the promoter region of the tumor suppressor gene P15 INK4B. These alterations are only partly related to gene mutations in epigenetic regulators. Hypomethylating drugs appear to modulate the methylation of DNA without eradicating the mutated cells, which can be sufficient to get a more balanced hematopoiesis.

Ineffective Hematopoiesis The occurrence of cytopenias despite a generally hypercellular marrow indicates ineffective hematopoiesis, which was shown to result from an increased susceptibility of clonal myeloid progenitors to apoptosis. This excessive apoptosis can result from intrinsic

566

Myelodysplastic Syndromes: Mechanisms, Diagnosis, and Treatment

stresses due to the above-described genetic and epigenetic alterations. For example, iron retention in SF3B1 mutated erythroblasts induce mitochondrial depolarization and ribosomal defects in 5q- syndrome lead to reticulum endoplasmic stress. One of the hallmarks of MDS is activation of the NLRP3 inflammasome, which drives clonal expansion and pyroptotic cell death, independently of genotype, Pyroptosis is triggered by the alarmin S100A9 that is found in excess in MDS HSPCs and bone marrow plasma excessive apoptosis can also result from external insults triggered by the microenvironment such as death receptor ligands, including Fas-L and tumor necrosis factor, and myelosuppressive cytokines, such as transforming growth factor b. Importantly, a shift of from apoptosis to proliferation that emerges in a subclonal progenitor, for example, as a consequence of NFkB pathway activation, leads to disease progression to acute myeloid leukemia.

Microenvironment and Immune Cells The hematopoietic stem-cell niche contributes to the pathogenesis of MDS. In the recent years, several mouse models have shown that genetic alterations in specific cells of this microenvironment could promote the emergence of a myelodysplastic clone. Clonal T-cell expansion is commonly detected, especially in patients with a hypoplastic syndrome, inflammatory Th17 cells and myeloid-derived suppressive cell may contribute to ineffective hematopoiesis whereas regulatory T cells contribute to evasion from antitumoral immunity in higher-risk disease. Several of the drugs commonly used in MDS may have an effect on this pathological environment and immune reaction, beyond their specific effect on the malignant cells. It remains unknown if these abnormalities are secondary events that modify the environment of the leukemic clone and facilitate its dominance on polyclonal hematopoiesis or they are primary events that favor the emergence of mutated clones able to escape this altered environment. Analysis of the genetic predisposition to MDS and other myeloid malignancies may help to distinguish the egg and the chicken: if the environment defect is the first to appear, it is expected that some hereditary cases will be the consequence of mutations altering the environment. Among myeloid malignancies, MDS are at the crossroad of several disorders: (1) CHIP and ICUS that we have previously described (2) aplastic anemia, including those of auto immune origin and (3) BCR-ABL-negative MPNs in which primary myelofibrosis could be considered as an MDS. One interesting hypothesis is that somatic mutations identified in these diseases have been selected as they provide HSCs an advantage on aging or immunologically altered HSCs. These mutations provide some advantage to early stages of hematopoiesis but some of them are deleterious for terminal hematopoiesis, leading to cell dysplasia. These mutations can induce a genetic instability that favors secondary mutations and clonal evolution, or behave as rescue mutations such as JAK2V617F mutation whose acquisition in the context of SF3B1-driven refractory anemia with ring sideroblasts, induce some correction of anemia and a thrombocytosis, improving the disease outcome.

Risk Stratification Because the prognosis of patients with MDS is highly heterogeneous, accurate prognostic systems are used to guide the therapeutic choices and timing. A number of scoring systems have been developed, the most widely used being the International Prognostic Scoring System (IPSS) and the Revised IPSS (IPSS-R). The original IPSS, established in 1997, assigned to each patient a risk score that is based on bone marrow blast percentage, cytogenetics, and number of cytopenias (Table 1). Patients with low or intermediate-1 risks are considered to have low-risk disease, those with intermediate-2 or high scores to have high-risk MDS. Median OS ranges from approximately 5 months in untreated

Table 1

International prognostic scoring system for myelodysplastic syndromes (IPSS-R)

Define first the cytogenetic risk as follows Good Normal, -Y, del(5q), del(20q) Intermediate All other situations Poor Complex (3 or more abnormalities) or chromosome 7 abnormalities Then, include the cytogenetic risk in the scoring system Points 0 0.5 Cytogenetic Good Intermediate BM blast 3–4.5 >4.5–6 >6

2 Intermediate 5%–10%

3 Poor >10%

4 Very poor

Very low risk Low risk Intermediate risk High risk Very high risk

patients with high risk to almost 6 years in patients with low-risk disease. This very simple and easy to use system has limitations, especially in patients with low risk disease. The IPSS-R, which was developed from a data set of over 7000 patients, uses different cut-off points of cytopenias, places a higher weight on cytogenetics, and refines the bone marrow blast categories (Table 2). Again, this score separates MDS patients with lower risk (very low or low risk scores) from those with higher risk (high or very high risk scores). The main limitation of IPSS-R is that no drug therapy has been approved yet using this scoring system. A new molecular IPSS system is now expected as somatic mutations may impact the therapeutic choice. It is now clear that patient outcomes progressively worsen as the number of oncogenic mutations increases. In addition, specific somatic mutations reproducibly predict patient outcomes in univariate analysis, including the better clinical outcome of MDS with SF3B1 mutations and the worse outcome of those with TP53, EZH2, ETV6, RUNX1, ASXL1, and SRSF2 mutations. However, genetics only modestly improves scoring systems such as R-IPSS and their precise role in clinical practice has still to be better delineated. Specific mutations can help in specific settings, for example, TP53 and DNMT3A mutations are associated with a poor outcome after allogeneic stem cell transplantation, patients with TET2 mutations better respond to azacitidine and those with IDH1 or IDH2 mutations may deserve targeted drugs. Nevertheless, morphologic and clinical criteria still play a central role in evaluating MDS prognosis and these systems could be further refined by introducing comorbidities that affect the natural history of MDS patients while specific scoring systems could improve the prognostication in hypocellular and therapy related MDS.

Treatment of MDS Therapeutic decision-making is guided by individual risk assessment performed using the IPSS, R-IPSS or similar tools. MDS patients with an indolent disease course and a relatively long life expectancy do not necessarily warrant therapy, for example, those without significant cytopenias can be monitored closely without treatment. Therapy is introduced to reduce transfusion needs and improve quality of life. A common practice is to start with growth factor support and consider lenalidomide or an azanucleoside secondarily. Patients that fail are candidates for clinical trials or allogeneic stem cell transplantation (ASCT). Ideally, treatment may delay transformation to AML and improve survival. Earlier therapeutic intervention is proposed to lower risk patients with a less favorable prognosis, with several innovative therapeutics currently tested, including transforming growth factor b superfamily ligand traps to treat anemia, oral azanucleosides, proteasome inhibitors, and antagonists of toll-like receptor signaling. Because higher-risk MDS patients have a higher rate of progression to AML and shorter survival, the goal of treatment is to alter the natural history of these diseases, prevent progression to AML and improve overall survival. Current available therapies include hypomethylating agents, intensive chemotherapy, and ASCT. Additional supportive care measures include the use of prophylactic antibiotics and iron chelation. There are no approved interventions for patients with progressive or refractory disease after these treatments. The management of patients with an intermediate score must be individualized, taking into consideration age, presence of bone marrow fibrosis and other relevant prognostic data such as gene mutations.

Supportive Care Prophylactic antibiotics are usually used in the context of cytotoxic or immunosuppressive therapy and not in untreated patients as there is not major increase in the risk of infection in those with isolated neutropenia. Iron accumulation, as a consequence of red cell

568

Myelodysplastic Syndromes: Mechanisms, Diagnosis, and Treatment

transfusions, could increase the risk of infection as well as that of AML transformation, and induce liver cirrhosis or cardiomyopathy in lower risk, long living patients, but the use of chelation therapy is currently limited to patients with very high ferritin levels, above 2500 ng/mL.

Growth Factor Support The vast majority of MDS patients develop some degree of anemia during the course of the disease, approximately 40% become transfusion-dependent, and severe anemia decreases health-related quality-of-life and increases cardiovascular risk. There is no consensus on the optimal hemoglobin level threshold for transfusions. Erythroid stimulating agents (ESA) are commonly used in these patients to prevent or correct anemia. A 3-month treatment is needed before evaluating the response. Recommended doses are higher than those used for patients with anemia due to chronic renal insufficiency. Treatment is generally well tolerated, and the incidence of cardiac and thrombotic events is low. Response rates range from 30% to 50% with a median response duration of 18– 24 months. Algorithms have been developed, for example, by the Nordic MDS group, to predict response to ESAs, which is much higher in patients with a low endogenous serum erythropoietin (EPO) level EPO levels of < 500 U/L and with low transfusion requirement (< 2 units of packed red blood cells each month). Substantial improvements in well-being correlate with erythroid response. ESA are discontinued when the transfusion effect is lost. The addition of granulocyte-colony-stimulating-factor (G-CSF) has been proposed in patients failing to respond to ESAs alone. ESAs may improve the natural history of the disease but questions on their potential tumorigenic effect have resulted in increased scrutiny of their use. Pooled analyses and large retrospective studies suggest a small OS benefit with these agents in the MDS population responding to treatment. In patients with thrombocytopenia, the use of eltrombopag remains cautious as there is a fear that, like romiplostim, the use of the MPL agonists expose to disease transformation and marrow fibrosis.

Immunomodulatory Drugs (IMiDs) Thalidomide initially demonstrated some activity in reducing red blood cell transfusion needs in patients with MDS that had previously failed ESAs. Then, lenalidomide was developed as an orally bioavailable analog of thalidomide with immunomodulatory, antiangiogenic, and antiproliferative properties. In patients with anemia and del(5q), lenalidomide (10 mg for 21 days every 4 weeks) has become the standard of care. del(5q) is the most common cytogenetic abnormality in MDS (10%–15%). The socalled “5q-syndrome” includes refractory anemia, isolated del(5q), female predominance, normal to increased monolobulated megakaryocytes, < 5% bone marrow myeloblasts and relatively indolent course. In patients with a 5q-syndrome and a good platelet count, lenalidomide results in lower transfusion requirements and, in 50% of them, cytogenetic complete remission. Response is fast (median time 5 weeks). The median rise of hemoglobin is 5 g/dL. The most common adverse events are early neutropenia and thrombocytopenia, which generally occur within the first two cycles. Their occurrence is associated with a higher probability of response as they correspond to suppression of the del(5q) clone. While deep venous thrombosis is a well-known adverse event associated with the use of lenalidomide in multiple myeloma, it was reported in only 3% of MDS patients. This treatment does no increase the incidence of AML transformation and increases survival for responding patients, indicating that the drug can change the natural history of this specific MDS. A mechanism of action of lenalidomide has been deciphered. This drug binds to an E3 ubiquitin ligase, alters its substrate affinity and induces the selective degradation of a kinase named casein kinase 1A1. This kinase is encoded by a gene in the commonly deleted region of chromosome 5. Complete loss of this kinase triggers p53-mediated apoptosis. This explains why mutations in TP53 are associated with poor response to lenalidomide and TP53-mutant subclones expand in patients treated with this drug. Lenalidomide has been used also in transfusion-dependent MDS patients without chromosome 5 alterations: 25% achieved transfusion independence after a median of 5 weeks of therapy, with median response duration of 41 weeks. The median hemoglobin increase was less robust than that observed in patients with isolated del(5q), cytogenetic responses were infrequent, and the median duration of transfusion independence was shorter than in patients with 5q-syndrome. The drug may not be used in patients with thrombocytopenia.

DNA Methyltransferase Inhibitors Aberrant DNA methylation may be an important component of MDS pathogenesis. DNA methyltransferases (mainly DNMT3A/B) are the main enzymes that catalyze the addition of methyl groups to the 50 position of cytosine nucleotides into DNA. Two drugs that inhibit DNMTs are the standard of care for most patients with high risk MDS, namely 5-azacitidine and 5-aza-20 -deoxycitidine (decitabine). These azanucleosides are also known as hypomethylating agents (HMAs). Azacitidine is a ribonucleoside that is phosphorylated by uridine cytidine kinase to be incorporated into RNA. Decitabine is a deoxyribonucleoside that is directly incorporated into DNA. About 10% of azacitidine is also incorporated into DNA through enzymatic conversation by ribonucleotide reductase. Decitabine requires phosphorylation by a deoxycytidine kinase to become biologically active. Both drugs incorporate into nucleic acids during the S-phase and irreversibly bind to and deplete DNMTs. Demethylation of DNA was associated with clinical response in chronic myelomonocytic leukemia.

Myelodysplastic Syndromes: Mechanisms, Diagnosis, and Treatment

569

Although response rates to these two drugs are similar, only 5-azacitidine (75 mg/m2 subcutaneously daily for 7 days every 4 weeks) has been demonstrated to improve survival in higher risk MDS patients, thus is the standard front line treatment in these patients. Patients respond slowly, in 90% of cases by six cycles, suggesting that therapy should not be abandoned prior to six cycles unless there is documented disease progression. Fifty to sixty percentage of these patients experience hematologic improvement after six cycles, which improves their quality of life compared to best supportive care only. The median duration of response is 15 months. Therapy should be given indefinitely in the absence of severe toxicity or disease progression. The most common toxicities are hematologic, grade 3 or 4 neutropenia is most common and may be more prominent with the initial cycles whereas nonhematologic toxicity is dominated by injection-site reactions to subcutaneous injections. Loss of response after discontinuation can be rapid and retreatment results in inferior quality and duration of responses compared to initial treatment. Achievement of CR is not a prerequisite for improved survival. Power of mutational data to predict HMA response is modest. Nevertheless, early clonal mutations in TET2 gene, encoding a protein involved in the demethylation of 5-methylcytosine, have been associated with an increased response rate. A set of differentially methylated regions, mostly localized in nonpromoter regions, also predicts response to HMAs. Low dose HMAs can be considered in patients with lower risk disease that are transfusion dependent and have failed or are not candidates for growth factor support or lenalidomide. An oral formulation of 5-azacitidine is being studied for patients with lower risk MDS.

AML-Like Induction Chemotherapy The classical anthracycline- and cytarabine-based chemotherapy used to treat AML is less efficient in treating high-risk MDS. In addition to being hard to use because of the advanced age of MDS patient, this treatment results in lower CR rates (40%–60%), higher treatment-related mortality and morbidity, shorter CR duration ( 12 months) and longer periods of aplasia. In particular, patients with a complex karyotype or an alteration of chromosome 7 have a low CR rate and a short duration of response. Therefore, AMLlike therapy is now restricted to younger patients with favorable karyotype, usually as a bridge to allogeneic stem cell transplantation. A large retrospective study comparing azacitidine to induction chemotherapy or both prior to allogeneic stem-cell transplant found that OS, event-free survival, relapse rate and nonrelapse mortality are similar between azacitidine and induction chemotherapy.

Immune Therapy Immune-modulatory agents can be used in hypocellular MDS patients, which are a difficult group of patients. A deregulation of both cellular (abnormal CD8þ cytotoxic T-lymphocytes) and innate immunity is identified in a subset of MDS patients, autoimmunity is commonly associated to these diseases, and there is overlap between MDS and immune-mediated bone marrow failure syndromes such as aplastic anemia. Equine antithymocyte globulin (ATG, 15 mg/kg intravenously for 5 days) combined with oral cyclosporine for 6 months, but also growth factors and steroids, which are all used in aplastic anemia, are considered in younger MDS patients with severe hypocellular bone marrow and short transfusion dependency when allogeneic stem cell transplantation is not feasible. Responses are slow as they may take up to 6 months to manifest. Response rates are around 30%. The impact of ATG based therapy on transformation-free and overall survival is still questioned. Among the new agents, the activity of alemtuzumab, an antibody against CD52, and eltrombopag, which was tested as salvage therapy, is currently investigated.

Allogeneic Stem Cell Transplantation Allogeneic stem cell transplantation is the only curative therapeutic strategy for MDS. Timing of the procedure, optimal conditioning regimen and maintenance therapies are still debated issues. Because of its high morbidity and mortality, allogeneic stem cell transplantation is usually not recommended in patients with lower risk MDS at initial presentation, even if they are young. Because of the time required for donor identification, it is important, nevertheless, to search rapidly for an appropriate donor while evaluating the risk of this therapeutic approach. Young patients with higher risk disease or hypoplastic MDS at diagnosis and MDS patients who have been exposed to multiple therapies should be considered for transplantation. Early transplantation is associated with a prolonged survival obtained in 30%– 50% of the patients. It is important to identify which patients will benefit the most from transplant through accurate pretransplant risk evaluation. Risk factors include cytogenetics (monosomy 7 predicts a poor outcome), elevated serum ferritin levels and some genetic alterations as mutations in TP53, DNMT3A, RUNX1, and ASXL1 mutations are potential markers of poor outcome. Different transplant modalities and donor sources are currently evaluated, which may allow considering older patients for transplantation in the future. Retrospective studies suggest that MDS patients between ages 60 and 70 with a suitable donor can benefit from transplant following a reduced intensity conditioning as long as they are medically fit. Questions include whether or not disease debulking with chemotherapy or hypomethylating agents is recommended prior to transplant when marrow blasts > 10% in order to decrease the very high relapse risk posttransplant. Most of these patients receive azanucleosides as a donor search is initiated and pretransplant testing is completed. Posttransplant hypomethylating use could also

570

Myelodysplastic Syndromes: Mechanisms, Diagnosis, and Treatment

improve outcome, especially in patients who showed decreased donor chimerism (< 80%) while still in morphologic and hematologic complete response.

Investigational New Therapeutic Options All the currently available agents for the treatment of MDS have considerable shortcomings and every patient with a relapsed or refractory disease should be considered for an investigational clinical trial if allogeneic stem cell transplantation is not feasible. Novel therapeutic targets have emerged from recent advances in our understanding of the molecular pathophysiology of these diseases. Targeted molecules include small-molecule inhibitors of either mutant isocitrate dehydrogenase enzymes, or spliceosome, which are currently evaluated in early phase clinical trials. Kinases inhibitors, including those targeting Flt3 and the Ras pathway could also take place in the precision medicine approach. Another approach, whose development has been limited by toxicity, is the combination of azanucleoside with other drugs such as lenalidomide, anthracyclines or histone deacetylase inhibitors. An oral formulation of azacitidine and second-generation azanucleosides such as SGI-110, a dinucleotide form of decitabine and deoxyguanosine that protects it from deamination, are also tested. Since thrombocytopenia can be difficult to manage in MDS patients, thrombopoietin receptor agonists, including romiplostim and eltrombopag, have been tested. Concerns regarding an increase in the rate of progression to AML have led to stop the development of romiplostim. Preliminary results indicated that eltrombopag increases platelet count in lower risk MDS patients. Agents targeting a variety of pathways are currently under clinical investigation for the treatment of anemia, including inhibitors of p38 MAPK and Tie2 (ARRY-614), PI3K and PLK1 (rigosertib), and the TGF-b superfamily (LY2157299, sotatercept). Rigosertib, clofarabine and immune checkpoint inhibitors are also tested in MDS. Finally, high doses of vitamin C could partially revert the phenotypic effects of TET2 mutations but the potential interest of these observations deserve to be tested in clinical trials.

Prospective Vision A number of genetic alterations that drive MDS have been identified but how these somatic aberrations occur and the conditions in which they generate the disease phenotype remain open questions: the role of germline factors, microenvironment alterations and host factors such as immune response and microbiota composition in disease emergence and progression may be clarified in the coming years. Lenalidomide and hypomethylating agents have a major place in the current armamentarium against MDS, as targeted therapies have demonstrated limited efficacy so far, but coming years may define the potential interest of IDH and spliceosome inhibitors and, hopefully, alternative strategies will emerge from ongoing pathophysiological studies.

Further Reading Arber, D.A., Orazi, A., Hasserjian, R., et al., 2016. The 2016 revision to the World Health Organization classification of myeloid neoplasms and acute leukemia. Blood 127, 2391–2405. Canaani, J., Nagler, A., 2016. Established and emerging targeted therapies in the myelodysplastic syndromes. Expert Review of Hematology 9, 997–1005. de Witte, T., Bowen, D., Robin, M., et al., 2017. Allogeneic hematopoietic stem cell transplantation for MDS and CMML: Recommendations from an international expert panel. Blood 129, 1753–1762. Deininger, M.W.N., Tyner, J.W., Solary, E., 2017. Turning the tide in myelodysplastic/myeloproliferative neoplasms. Nature Reviews. Cancer 17, 425–440. Delhommeau, F., Dupont, S., Della Valle, V., et al., 2009. Mutation in TET2 in myeloid cancers. The New England Journal of Medicine 360, 2289–2301. Greenberg, P., Cox, C., LeBeau, M.M., et al., 1997. International scoring system for evaluating prognosis in myelodysplastic syndromes. Blood 89, 2079–2088. Greenberg, P.L., Tuechler, H., Schanz, J., et al., 2012. Revised international prognostic scoring system for myelodysplastic syndromes. Blood 120, 2454–2465. Hellström-Lindberg, E., van de Loosdrecht, A., 2013. Erythropoiesis stimulating agents and other growth factors in low-risk MDS. Best Practice & Research. Clinical Haematology 26, 401–410. Inoue, D., Bradley, R.K., Abdel-Wahab, O., 2016. Spliceosomal gene mutations in myelodysplasia: Molecular links to clonal abnormalities of hematopoiesis. Genes & Development 30, 989–1001. Malcovati, L., Cazzola, M., 2015. The shadowlands of MDS: Idiopathic cytopenias of undetermined significance (ICUS) and clonal hematopoiesis of indeterminate potential (CHIP). Hematology. American Society of Hematology. Education Program 2015, 299–307. Olnes, M.J., Sloand, E.M., 2011. Targeting immune dysregulation in myelodysplastic syndromes. Journal of the American Medical Association 305, 814–819. Raaijmakers, M.H., 2014. Disease progression in myelodysplastic syndromes: Do mesenchymal cells pave the way? Cell Stem Cell 14, 695–697. Sallman, D.A., Davila, M.L., 2017. Is disease-specific immunotherapy a potential reality for MDS? Clinical Lymphoma, Myeloma & Leukemia 17S, S26–S30. Sperling, A.S., Gibson, C.J., Ebert, B.L., 2017. The genetics of myelodyplastic syndrome: From clonal haematopoiesis to secondary leukaemia. Nature Reviews. Cancer 17, 5–19. Talati, C., Sallman, D., List, A., 2017. Lenalidomide: Myelodysplastic syndromes with del(5q) and beyond. Seminars in Hematology 54, 159–166. Wong, T.N., Ramsingh, G., Young, A.L., et al., 2015. Role of TP53 mutations in the origin and evolution of therapy-related acute myeloid leukaemia. Nature 2518, 552–555.

Relevant Websites https://www.mds-foundation.org/. https://www.cancer.gov/types/myeloproliferative/patient/myelodysplastic-treatment-pdq.

Nasopharyngeal Carcinoma: Diagnosis and Treatment ska, Science to the Point, Archamps Technopole, Archamps, France Katarzyna Szyman © 2019 Elsevier Inc. All rights reserved.

Abbreviations 18

F-FDG PET/CT 2-Deoxy-2-[fluorine-18]fluoro-D-glucose PET integrated with CT 5-FU 5-Fluorouracil CT Computed tomography EA Early antigen (of the Epstein-Barr virus) EBV Epstein-Barr virus EGFR Epidermal growth factor receptor FNA Fine-needle aspirates HDACi Histone deacetylase inhibitors HLA Human leukocyte antigen IARC International Agency for Research on Cancer (WHO) ICD-O International Classification of Diseases for Oncology IMRT Intensity-modulated radiotherapy LMP Latent membrane protein (of the Epstein-Barr virus) MRI Magnetic resonance imaging NPC Nasopharyngeal carcinoma PD-L1 programmed death ligand 1 PET Positron emission tomography RT Radiotherapy RT-PCR Real-time polymerase chain reaction VCA Virus capsid antigen (of the Epstein-Barr virus) VEGF Vascular endothelial growth WHO World Health Organization

Definition and Classification Nasopharyngeal carcinoma (NPC) is a carcinoma which arises in the nasopharyngeal mucosa and shows light microscopic or ultrastructural evidence of squamous differentiation. The term encompasses three distinct entities: nonkeratinizing, keratinizing, and basaloid squamous cell carcinoma. Nonkeratinizing squamous cell carcinoma (ICD-O code: 8072/3) is composed of large atypical cells, without morphological evidence of keratin production, whereas keratinizing squamous cell carcinoma (8071/3; also called large cell squamous cell carcinoma, keratinizing squamous cell carcinoma, and keratinizing epidermoid carcinoma) is a squamous cell carcinoma with prominent production of keratin. Basaloid squamous cell carcinoma (8083/3) is histologically characterized by the presence of cells with hyperchromatic nuclei, scant amount of cytoplasm, and peripheral nuclear palisading.

Presentation and Diagnosis NPCs most commonly originate in the pharyngeal recess (fossa of Rosenmüller), ordless frequentlydin the superior posterior wall of the nasopharynx. The tumor can present as a smooth bulge in the mucosa, a discrete raised nodule (in some cases with surface ulcerations), or an infiltrative fungating mass. In some cases, the lesions are not grossly visible. NPC is highly malignant, with extensive locoregional infiltration and lymphatic spread occurring early in the course of the disease. Most patients present with locoregionally advanced disease, commonly with cervical lymph node metastases. The presenting symptoms are related to the presence of a mass in the nasopharynx (e.g., epistaxis, obstruction, and blood-stained postnasal drip), Eustachian tube dysfunction (such as hearing impairment, tinnitus, and serous otitis media), and skull base involvement with impairment of the cranial nerves, most frequently the fifth and the sixth (e.g., headache, diplopia, facial pain, numbness, and paresthesia). Painless neck mass appears due to lymph node metastases. About 10% of patients are asymptomatic. Fiberoptic endoscopic examination of the nasopharynx following complete physical examination including palpation of the head and neck for adenopathy is routinely used to elicit the tumor. The jugulodigastric node is the most commonly palpable

Encyclopedia of Cancer, 3rd edition, Volume 2

https://doi.org/10.1016/B978-0-12-801238-3.65363-1

571

572

Nasopharyngeal Carcinoma: Diagnosis and Treatment

node at presentation. Computed tomography (CT), positron emission tomography (PET)-CT scan, or magnetic resonance imaging (MRI) is used to evaluate the extent of primary tumor and adenopathy as well as the skull base invasion. MRI is often more helpful than CT as it provides a better resolution for assessing parapharyngeal spaces, marrow infiltration of the skull base, intracranial disease, and deep cervical nodes. Neuro-ophthalmological and audiological evaluations allow to evaluate the degree of impairment of the cranial nerve function. Diagnosis is made based on the pathological evaluation of the nasopharyngeal mass from the primary tumor site. Determining the Epstein-Barr virus (EBV) infection status helps to confirm nonkeratinizing NPC diagnosis. Detecting Epstein-Barr virus (EBV) by hybridization in situ for the Epstein-Barr encoding region (EBER) is particularly helpful in the evaluation of the cervical lymph nodes harboring undifferentiated or poorly differentiated squamous cell carcinoma, with the positive result being strongly suggestive of the NPC diagnosis (see “Biomarkers” section). A cytological evaluation of fine-needle aspirates (FNA) of the neck followed by immunohistochemistry is useful for diagnosing metastatic keratinizing and nonkeratinizing NPC.

Epidemiology and Risk Factors Burden Even though NPC is the most common neoplasm of the nasopharynx, it is relatively uncommon compared to other cancer types. However, it has a very unique geographical distribution overall, with remarkable variations in incidence. About 70% of cases are diagnosed in east and south-east Asia, and those occurring in south-central Asia, and east and north Africa roughly account for the remaining 30%, while the annual incidence in North America is only 0.3–0.7 cases per 100,000 population. In 2012, almost 86,700 new NPC cases were reported worldwide (about 0.6% of all cancers), of which over 21,200 occurred in the WHO South-East Asia region. The highest incidence rates were reported for Malaysia, followed by Singapore before Indonesia and Vietnam (Fig. 1). The peak incidence is between 40 and 60 years of age. The age-standardized incidence of NPC has declined over the past decades, particularly among Hong Kong Chinese, and the NPC-related mortality rates are also going down. Still, 120,000 new NPC cases and over 20,000 NPC-related deaths are predicted for the year 2030.

Etiology and Risk Factors Men are two to three times more likely to develop NPC than women. In contrast to Caucasians in whom NPC is rare, NPC is quite common in some specific ethic groups, including the Inuits in the North Pole, the Bidayuh in Borneo, and the Nagas in northern India, suggesting that they may have a genetic predisposition to develop the disease. The etiology of NPC seems to be multifactorial, with viral infections playing an important but not exclusive role in the tumorigenesis. Epstein-Barr virus (EBV), a very common herpesvirus, is strongly associated with the development of nonkeratinizing NPC, in particular in endemic NPC regions, but not with that of the keratinizing type. Conversely, infection with high-risk types of the human papilloma virus (HPV) has been shown to be associated with NPC from nonendemic regions and in particular with the keratinizing type, even though the evidence is not yet so strong as for EBV. Consumption of salted and fermented food has been implicated in the etiology of nonkeratinizing NPC in populations where this type is endemic. In particular, consumption of traditional salted fish in China has been shown to be a risk factor. These foods have a high content of N-nitrosamine. The carcinogenicity of N-nitrosamine compounds has been clearly shown for several other cancer sites and it is also hypothesized to be the carcinogen associated with the NPC risk. Tobacco smoking and heavy alcohol consumption have also been proposed as putative risk factors, however the evidence for the latter remains controversial. Among other environmental factors, occupational exposure to wood dust, heat, smoke, formaldehyde, and chemical fumes have been identified as potentially contributing to the risk of NPC, and two of these exposures have been classified as carcinogenic for the nasopharynx by the IARC monographs (Table 1). The risk of developing NPC is also associated with genes coding for certain tissue antigens (human leukocyte antigen (HLA) complex). For example, high-resolution genotyping studies have consistently shown an association between NPC and the HLAA*0207 allele which is common in the Chinese population. Genetic polymorphisms in genes coding for metabolic and DNA repair enzymes have also been associated with an increased risk of NPC. Familial clustering of NPC is well documented. However, the clinical characteristics of familial and sporadic NPCs do not differ.

Pathology and Genetics Deletions on chromosomes 3p and 9p (spanning the CDKN2A gene on chromosome 9p21) have been identified to be early events in NPC carcinogenesis. Most frequent chromosome gains are on chromosome 12. Gene fusions have been reported in about 10% of cases. Somatic mutations in the TP53 gene are common (15%–20%). The most remarkable characteristic of the NPC genome is the presence of multiple alterations targeting chromatin remodeling pathway, with three genes having been identified as significantly mutated by whole-exome sequencing: BAP1, MLL2, and TSHZ3. The most frequently altered gene in this category is ARID1A. Alterations in the chromatin remodeling pathway have been shown to

Nasopharyngeal Carcinoma: Diagnosis and Treatment

573

be associated with both EBV burden and poor overall survival in univariate but not in multivariate analysis. Interestingly, suppressing endogenous ARID1A in vitro results in an increased anchorage-independent colony formation, cell migration and xenograft growth. Another altered pathway includes receptor tyrosine kinases (ERBB2, ERBB3) and their downstream effectors (PI3KCA, KRAS). Patients with alterations in this pathway seem to have shorter survival and more advanced stage disease. Finally, a number of alterations are observed in genes regulating autophagy (Table 2).

Biomarkers Epstein-Barr Virus (EBV) EBV is prevalent in NPC, with almost all cases of nonkeratinizing NPC being associated with EBV infection. EBV testing has been shown to be useful for early detection and differential diagnosis of NPC as well as for prognostication and monitoring response to treatment (Table 3). The most reliable method to detect EBV in tumor tissue is in situ hybridization for EBV-encoded RNA (EBER) or immunohistochemical staining for latent membrane protein (LMP), even though the reliability and specificity of this method is questioned by some sources. EBV DNA load in blood serum or plasma may be quantified by real-time polymerase chain reaction (RT-PCR)

Fig. 1 Incidence and mortality of nasopharyngeal carcinoma (NPC) worldwide. (A) Age-standardized incidence and mortality rates (ASR) by gender and geographical area. (B) Incidence distribution worldwide. From Ferlay, J. et al. (2013). GLOBOCAN 2012 v1.0, Cancer incidence and mortality worldwide: IARC CancerBase No. 11 [Internet]. Lyon, France: International Agency for Research on Cancer. Available from: http://globocan.iarc.fr (accessed February 20, 2018).

574

Fig. 1

Nasopharyngeal Carcinoma: Diagnosis and Treatment

(continued).

Table 1

Risk factors for nasopharyngeal carcinoma (NPC)

Carcinogenic agents classified by IARC Monographs on the Evaluation of Carcinogenic Risks to Humans (volumes 1–120) Agents of sufficient evidence for nasopharynx carcinogenicity in humans Epstein-Barr virus (EBV; association shown for NK-NPC) Formaldehyde Chinese-style salted fish consumption (NK-NPC) Tobacco smoking (K-NPC) Wood dust Other carcinogenic agents and lifestyle risk factors High-risk human papillomaviruses (HPV; K-NPC) Heavy alcohol consumption (putative risk factor) Occupational exposure to heat, smoke, or chemical fumes (putative) Phenotypic and genetic risk factors Male sex Ethnicity Family history of NPC K-NPC, keratinizing NPC; NK-NPC, nonkeratinizing NPC.

targeting such genomic EBV sequences as BamHI-W, EBNA, or LMP. However, the sensitivity and specificity of PCR-based assays varies and given the plethora of commercially available assays used in clinical settings (e.g., from Viracor Eurofins and Roche Molecular Diagnostics), there is an urgent need to harmonize and validate reliable EBV detection and quantification protocols. In patients who present solely with cervical adenopathy, finding EBV in tissue is strongly suggestive of a primary NPC and indicates a necessity of a more in-depth diagnostics to this effect. Testing for EBV also helps to make a differential diagnosis in case of nonkeratinizing or undifferentiated histology, the latter of which for example being sometimes difficult to distinguish from lymphoma at a histological level. Several studies have also shown a diagnostic value of EBV detection in transoral brushing specimens and in nasopharyngeal swabs for detecting both primary and recurrent NPC. A number of studies have shown elevated titers of IgA antibodies against EBV antigens (virus capsid antigen (VCA) and EBV early antigen (EA)) in NPC patients compared to healthy individuals, supporting the use of EBV serology as a screening method. High titers persisting after therapy have also been associated with a worse prognosis. Other studies suggested that EBV DNA load in blood serum or plasma is a more reliable diagnostic, monitoring, and prognostic marker. In particular, it has been validated as an

Nasopharyngeal Carcinoma: Diagnosis and Treatment Table 2

575

Gene loci that are most frequently altered in nasopharyngeal carcinoma

Pathway

Gene (locus)

– Chromatin remodeling

Deletions on chromosomes 3p and 9p (spanning CDKN2A on 9p21) ARID1A (1p36.11) Deletions and point mutations; likely associated with EBV and poor survival BAP1 (3p21.1) Point mutations PIK3CA (3q26.32) Activation by hot-spot mutations and amplifications ERBB2 (17q12) and ERBB3 (12q13.2) Point mutations KRAS (12p12.1) Amplifications and point mutation AKT2 (19q13.2) Amplifications PTEN (10q23.31) Deletions ATG2A (11q13.1), ATG7 (3p25.3), and ATG13 (11p11.2) Deletions and point mutations CDKN2A (9p21.3) Deletions and point mutations TP53 (17p13.1) Deletions and point mutations

ERBB/P13K signaling (cell metabolism)

Autophagy regulation G1/S transition (cell-cycle) control

Table 3

Alteration type

Biomarkers for the detection and management of nasopharyngeal carcinoma (NPC)

Biomarker Diagnostic and monitoring biomarkers EBV in tissue EBV serology EBV DNA load in blood serum/plasma

Prognostic biomarkers Tumor stage at presentation Tumor size/increasing tumor volume EBV serology EBV DNA load in blood serum/plasma

Predictive biomarkers EBV DNA load in blood serum/plasma

Characteristics and clinical utility Detected in tumor tissue, strongly suggestive of primary nonkeratinizing NPC; useful for differential diagnosis in case of nonkeratinizing or undifferentiated histology, and for early detection in patients with cervical adenopathy; detected in brushing biopsies/swabs, strongly specific for detecting primary or recurrent NPC Elevated anti-EBV IgA titers in NPC patients compared to healthy individuals Elevated EBV DNA levels in NPC patients compared to healthy individuals; also a potential monitoring biomarker allowing early detection of local and distant treatment failures; variable detection sensitivity issues remain to be solved Inversely correlated with prognosis Increasing tumor volume associated with an increased risk of local treatment failure High titers persisting after treatment associated with poor prognosis EBV DNA levels inversely correlated with overall survival in early-stage NPC patients; persistent EBV detection after definitive radiotherapy associated with an increased risk of tumor recurrence and with adverse prognosis; detection methods require standardization and validation (cut-offs, timing, sensitivity) Persistent high posttreatment levels may be a criterion to select patients most likely to benefit from adjuvant chemotherapy after chemoradiation (currently being tested)

EBV, Epstein-Barr virus.

independent prognostic marker which inversely correlates with overall survival in early-stage NPC patients. Several studies have confirmed that persistent EBV detection after definitive radiotherapy, with or without chemotherapy, is associated with an increased risk of tumor recurrence and with adverse prognosis. It has also been associated with distant metastases. Monitoring EBV DNA levels can therefore complement traditional imaging techniques in surveillance for local and distant treatment failures. However, the detection sensitivity of EBV DNA in plasma remains an issue and standardized cut-off values for prognostication have to be defined and validated. Moreover, the detection periods or time-points are not clearly defined (e.g., what is the period of posttreatment recurrence predicted by elevated EBV DNA levels, or else at what time points and for how long these levels should be monitored). The use of EBV DNA load as a biomarker to select patients that are most likely to benefit from adjuvant therapy following chemoradiation has also been suggested. The effectiveness of posttreatment patient stratification based on EBV DNA load into low-risk (undetectable posttreatment plasma EBV DNA) and high-risk as a strategy to select patients for adjuvant therapy is currently being tested in phase II and III clinical trials. The goal of this approach is to create a treatment decision tree allowing to eliminate unnecessary adjuvant chemotherapy for low-risk patients and maximize progression-free survival for those of the high-risk patients who are most likely to benefit from such a regimen. Conversely, monitoring EBV DNA levels during the initial chemoradiation (midtreatment monitoring) has been suggested as a potential tool to identify cases with a good response to treatment and a favorable prognosis. It has been hypothesized that it may also allow to personalize treatment regimens, with decreasing EBV DNA levels indicating a possibility to de-intensify the treatment. However, this approach has not been validated and remains controversial.

576

Nasopharyngeal Carcinoma: Diagnosis and Treatment

Other Prognostic and Predictive Biomarkers The most powerful prognostic biomarker for NPC is tumor stage at presentation, with the overall survival decreasing with increasing tumor stage (Table 3). Tumor size, and in particular increasing tumor size, is a negative prognostic factor associated with an adverse patient outcome, with an estimated 1% increase in the risk of local treatment failure per 1 cm3 increase in tumor volume. Also the involvement of neck lymph nodes adversely affects the outcome. Other factors that have been linked to diminished survival by some (but not all) studies include patient age, male sex, cranial nerve palsy and ear symptoms at presentation, and a long interval between biopsy and the beginning of radiotherapy. The relationship between NPC histological type (keratinizing versus nonkeratinizing) and prognosis remains unclear, with conflicting reports. The significance of high-risk HPV types is not yet elucidated. EBV and HPV infections are usually mutually exclusive. Several studies have suggested that HPV-related tumors are associated with worse prognosis than those related to EBV but may have a better outcome than tumors in patients negative for both EBV and HPV. Many putative molecular and immunohistochemical prognostic markers have been studied but so far only serum/plasma EBV DNA load tests have been applied in routine clinical practice. Developing reliable screening biomarkers for early detection of NPC in high-risk patients, predictive biomarkers which would allow to predict response to treatment and apply personalized treatment regimens, as well as monitoring biomarkers for early detection of recurrence would be invaluable in the management of NPC.

Management and Therapy NPC is usually not curable by surgical resection due to the tumor localization as tumor-free margins cannot be obtained at the base of the skull. Therefore, radiotherapy (RT) with high-energy x-rays, often combined with chemotherapy is the treatment of choice, with different modalities depending on the tumor clinicopathological features. Small (T1) early-stage tumors with no lymph node involvement and no distant metastases are highly curable by RT alone (definitive RT), while a combination of RT and chemotherapy is used for more advanced tumors. Induction chemotherapy prior to RT has been shown to give high response rates but the overall survival benefit is not clear, even though some studies have suggested that using a combination of cisplatin, 5-fluorouracil (5-FU), and a taxane (usually docetaxel) for the induction therapy may significantly improve survival. So far, RT with concurrent cisplatin-based chemotherapy is the standard treatment in Western countries. It may be followed by adjuvant chemotherapy (either cisplatin with 5-FU, or carboplatin with 5-FU). However, studies on the associated clinical benefits give conflicting results. In particular, it is worth to bear in mind that adjuvant therapy after RT is often poorly tolerated. Validating biomarkers which would allow to select patients most likely to benefit from adjuvant therapy would add much value to the current treatment schemes. EBV DNA levels in blood plasma offer some hope to this effect (see “Biomarkers” section). The RT with concurrent chemotherapy regimen appears to increase the 5-year survival rates up to two-fold. However, many patients do not complete the planned therapy because of its toxicity. Given the NPC localization and high total radiation doses, RT alone may result in severe sequelae including local ulceration, occasionally with necrosis, retinopathy, fibrosis of soft tissues in the neck, and middle ear changes. Chemoradiotherapy is associated with an even higher rate of hematological and nonhematological acute toxic effects. Late adverse effects from RT may be limited by using modern radiation technology, for example intensitymodulated radiotherapy (IMRT) which is the recommended option for NPC RT wherever available. It allows to minimize irradiation of normal tissue adjacent to the tumor, thus substantially limiting the radiation-induced adverse effects. A comprehensive assessment of the primary tumor response to treatment includes clinical examination, endoscopic examination of the nasopharynx (with or without biopsy) to assess superficial lesions, radiological imaging for the assessment of deeper lesions within the skull base, and EBV DNA titer measurement. Of note, differentiating between postradiation changes and residual disease on radiological imaging may be challenging. To this effect, PET with 2-deoxy-2-[fluorine-18]fluoro-D-glucose PET integrated with CT (18F-FDG PET/CT) has emerged as the most sensitive and specific method. However, some characteristics captured by MRI are also useful to assess early treatment response. In case of primary treatment failure (local recurrence), re-treatment is the most frequent choice and it has higher success rates than re-treatment of other head and neck cancers. Still, high total dose irradiation of the primary site is needed. This may be done by IMRT or brachytherapy. Limited post-RT failures in the neck may be controlled by surgery (neck dissection). Patients with metastatic NPC have heterogeneous outcomes and long-term survivorship is possible. Cisplatin-containing doublet regimens remain the standard for the first-line systemic treatment of metastatic disease, with the most “classic” choice being cisplatin with 5-FU (response rates of 70%–80%). Other cisplatin-containing combinations for doublet and even triplet regimens, for example with gemcitabine, capecitabine, paclitaxel, and docetaxel, have also been tested. However, the reports on their efficacy (measured by tumor response rates, progression-free survival, and/or overall survival) compared to the conventional treatment remain contradictory. In case of platinum-resistant disease, a monotherapy with gemcitabine, capecitabine, or docetaxel is used, with reported response rates of approximately 30%–40%. Several doublet combinations have been tested to this effect but none of them has been shown to yield better response rates. In the era of personalized medicine, targeted molecular therapies also hold promise for treatment of recurrent and metastatic NPC. In particular, inhibitors of epidermal growth factor receptor (EGFR), such as cetuximab, and of vascular endothelial growth factor (VEGF) have shown clinical efficacy in patients with platinum-refractory disease. Of note, VEGF-targeting drugs increase the

Nasopharyngeal Carcinoma: Diagnosis and Treatment

577

risk of bleeding and are therefore not suitable for patients who have a recurrent tumor within a previously irradiated field or a tumor invading major vascular structure. The high prevalence of EBV in NPC patients and its relation to patient outcome opens the door to EBV-specific T-cell-based immunotherapies. Several vaccines targeting EBV in recurrent and metastatic NPC patients are currently under clinical trials, with promising results. Moreover, EBV-associated tumors express programmed death ligand 1 (PD-L1). A phase II trial testing nivolumab, a PD-L1 inhibitor, in recurrent and metastatic NPC is currently ongoing. Positive results would support the strategy to combine non-antigen-specific immune checkpoint inhibitors with immunogenic vaccination against tumor-specific antigens, which could become standard for the treatment of metastatic NPC in the future.

Prospective Vision Many cancer genomic studies have focused on prognostic biomarkers. However, while identifying patients at higher risk of recurrence and unfavorable prognosis is important to define optimal treatment regimens, identifying new targets for treatment of metastatic NPC remains the major urgent unmet need. Indeed, metastases are frequent in NPC patients and systemic therapy is key for a successful treatment, with personalized approach being the choice of the future. Among others, mutations in the genes of the chromatin remodeling pathway, linked to EBV burden, seem to be potentially interesting surrogate targets as tumors with these mutations may be sensitive to histone deacetylase inhibitors (HDACi). Several clinical trials testing the safety, tolerability and efficacy of different HDACi in combination with other drugs are currently underway. Moreover, given its apparently frequent involvement in NPC, targeting the autophagy pathway may represent another, yet untested, opportunity.

See also: Oral and Oropharyngeal Cancer: Pathology and Genetics.

Further Reading Chmielowski, B., Territo, M., 2017. Manual of clinical oncology, 8th edn. Wolters Kluwer. Chua, M.L.K., et al., 2016. Nasopharyngeal carcinoma. The Lancet 387 (10022), 1012–1024. International Agency for Research on Cancer, 2017. In: El Naggar, A.K., Chan, J.K.C., Grandis, J.R., Takata, T., Slootweg, P.J. (Eds.), WHO classification of head and neck tumors, 4th edn. IARC, Lyon, France. Kim, K.Y., et al., 2017. Current state of PCR-based Epstein-Barr virus DNA testing for nasopharyngeal Cancer. Journal of the National Cancer Institute 109 (4). Kim, K.Y., et al., 2017. Clinical utility of Epstein-Barr virus DNA testing in the treatment of nasopharyngeal carcinoma patients. International Journal of Radiation Oncology, Biology, Physics 98 (5), 996–1001. National Comprehensive Cancer Guidelines (NCCN). NCCN clinical practice guidelines in oncology. Head and neck cancers. Version 1.2018 – February 15, 2018: https://www.nccn. org/professionals/physician_gls/default.aspx#site. Lee, S.C., et al., 2014. The genomic landscape of nasopharyngeal carcinoma. Nature Genetics 46 (8), 866–871.

This page intentionally left blank

ENCYCLOPEDIA OF CANCER THIRD EDITION

This page intentionally left blank

ENCYCLOPEDIA OF CANCER THIRD EDITION EDITORS IN CHIEF

Paolo Boffetta Tisch Cancer Institute Icahn School of Medicine at Mount Sinai New York, NY, United States

Pierre Hainaut Cancer Biology, Faculty of Medicine, University Grenoble-Alpes Grenoble, France Institute for Advanced Biosciences, Inserm, CNRS, UGA Grenoble, France

VOLUME 3

Amsterdam • Boston • Heidelberg • London • New York • Oxford Paris • San Diego • San Francisco • Singapore • Sydney • Tokyo Academic Press is an imprint of Elsevier

Academic Press is an imprint of Elsevier The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK 50 Hampshire St, 5th Floor, Cambridge, MA 02139, USA Copyright Ó 2019 Elsevier Inc. unless otherwise stated. All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Details on how to seek permission, further information about the Publisher’s permissions policies and our arrangements with organizations such as the Copyright Clearance Center and the Copyright Licensing Agency, can be found at our website: www.elsevier.com/permissions. This book and the individual contributions contained in it are protected under copyright by the Publisher (other than as may be noted herein). Notices Knowledge and best practice in this field are constantly changing. As new research and experience broaden our understanding, changes in research methods, professional practices, or medical treatment may become necessary. Practitioners and researchers may always rely on their own experience and knowledge in evaluating and using any information, methods, compounds, or experiments described herein. In using such information or methods they should be mindful of their own safety and the safety of others, including parties for whom they have a professional responsibility. To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume any liability for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein. Library of Congress Cataloging-in-Publication Data A catalog record for this book is available from the Library of Congress British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library ISBN 978-0-12-812484-0

For information on all publications visit our website at http://store.elsevier.com

Publisher: Oliver Walter Acquisition Editor: Sam Crowe Content Project Manager: Kate Miklaszewska-Gorczyca Designer: Matthew Limbert

Printed and bound in the United States

EDITORS IN CHIEF Paolo Boffetta, MD, MPH, graduated in Medicine from the University of Turin and obtained a Master in Public Health from Columbia University. He worked at the American Cancer Society, the International Agency for Research on Cancer, a specialized agency of the World Health Organization located in Lyon, France, and at the German Cancer Research Center in Heidelberg, Germany. In 2010 he moved to Icahn School of Medicine at Mount Sinai, New York, where he is Professor of Medicine, Global Health, Oncological Sciences, and Preventive Medicine, and Associate Director for Global Oncology of the Tisch Cancer Institute, a NCI-designated Cancer Center. Dr. Boffetta also holds a full professorship at the University of Bologna and adjunct appointments at Harvard School of Public Health, Vanderbilt University, Catholic University of Rome, and University of South Carolina. Since 2017 he is Senior Advisor for Research at Vinmec Health System. His main fields of research are cancer epidemiology and cancer prevention, with emphasis on modifiable risk factors (environmental exposure and personal behaviors), gene–environment interactions, molecular epidemiology, and evidence integration. He is the initiator and coordinator of several large-scale international consortia of molecular cancer epidemiology studies, including ILCCO (lung cancer), INHANCE (head and neck cancer), PANC4 (pancreatic cancer), StoP (stomach cancer), and ILCEC (liver cancer). Dr. Boffetta is the editor or associate editor of 5 scientific journals and member of the editorial board of 10 additional journals; he is a member of review panels of NIH and several medical research agencies in Europe. He has edited 13 books and is Editor in Chief of the new edition of Elsevier’s Encyclopedia of Cancer. He has published over 1250 peer-reviewed publications; his publications have been quoted more than 90000 times; his h-index is 148.

Pierre Hainaut is Professor of Exceptional Class in Cancer Biology at University Grenoble-Alpes, France and Director of the Institute of Advanced Biosciences (IAB), a joint research center of Institut National de la Santé et de la Recherche Médicale (Inserm), Centre National de la Recherche Scientifique (CNRS), and University Grenoble-Alpes. He also heads the IAB research team on Molecular Biology and Biomarkers. Pierre Hainaut holds a PhD in Biology (Zoology) from University of Liège, Belgium (1987). After postdocs in Nice (France, 1988–90), Cambridge, and York (United Kingdom, 1990–94), he joined the International Agency for Research on Cancer (IARC, World Health Organization) in 1995, where he held the post of Head of Molecular Carcinogenesis from 1999 to 2011. In 2012, he joined the International Prevention Research Institute (Lyon, France) and became Professor at the Strathclyde Institute of Pharmacy and Biomedical Science (Glasgow, United Kingdom). In 2014, he was awarded a Chair of Excellence in Translational Research from University Grenoble-Alpes (Grenoble, France). His research focuses on TP53 mutations and p53 protein regulation in cancer and chronic diseases. From 1994 to 2011, he has led the development of the international IARC TP53 database, a source of information on the causes and consequences of mutations affecting the TP53 suppressor gene in cancer. His work addresses the mechanisms of TP53 mutagenesis as well as the prognostic and predictive significance of TP53 mutations in lung, liver, and oesophageal cancers. His studies on p53 regulation have focused on the role of environmental mutagens in TP53 mutagenesis, on the biochemical mechanisms of p53 control by oxidation-reduction and by metabolism, and on the identification of p53 isoforms as factors acting as dominant inhibitors of p53 functions in cancers without TP53 mutations. His current activities focus on germline TP53 mutation and on the diversity of genetic and nongenetic factors that modulate the penetrance of the Li–Fraumeni Syndrome, as well as on the mechanisms that maintain optimal p53 protein balance in cells and tissues over lifetime. He is the author of over 450 publications and 50 book chapters. He is Editor of the Cancer Biology section of Current Opinion in Oncology. He has co-edited books on p53 (25 Years of p53 Research, 2005, 2007, p53 in the Clinics, Springer), a textbook on molecular epidemiology (Molecular Epidemiology: Principle and Practice, IARC Press, 2011) and two-volume textbook on human biobanking (Human Biobanking, Principle and Practice, 2017, 2018).

v

This page intentionally left blank

EDITORIAL BOARD Fred T. Bosman Institute of Pathology University of Lausanne Medical Center (CHUV) Lausanne, Switzerland

Graham A. Colditz Alvin J. Siteman Cancer Center and Institute for Public Health Washington University School of Medicine St. Louis, MO, United States Barnes Jewish Hospital St. Louis, MO, United States Division of Public Health Sciences, Department of Surgery Washington University School of Medicine St. Louis, MO, United States

Carlo La Vecchia Department of Clinical Sciences and Community Health University of Milan Milan, Italy

Gerd P. Pfeifer Center for Epigenetics Van Andel Research Institute Grand Rapids, MI, United States

Marco Pierotti IFOM/Cogentech Milan, Italy

Thomas Tursz University of Paris-Sud Orsay, France Institut Gustave Roussy Villejuif, France

vii

This page intentionally left blank

SECTION EDITORS Fred T. Bosman MD PhD, studied Medicine at the University of Leiden (MD 1971), where he also earned his PhD degree (in cytogenetics, 1976), and trained as a pathologist. He was staff pathologist at the University of Leiden, Professor and Chair of Pathology at the Faculty of Medicine of the University of Maastricht, Faculty of Medicine of the Erasmus University in Rotterdam, Director of the University Institute of Pathology, and Professor of Pathology at the University Medical Center (CHUV) of Lausanne in Switzerland, now emeritus. He is Honorary Fellow of the Royal College of Pathologists (United Kingdom) and foreign correspondent of the Royal Netherlands Academy of Sciences. Fred Bosman’s research activities (combining diagnostic and experimental pathology) focused on the biology of digestive tract cancer, notably Barrett’s esophagus and colorectal cancer, with a strong emphasis on the development of molecular diagnostics. He has written over 350 original publications and over 50 book chapters. Fred Bosman was Series co-editor of the 4th edition of the WHO Series Classification of Human Tumours, the international standard for tumor classification, and co-editor of the Volume on Tumours of the Digestive Tract.

Graham A. Colditz, MD, DrPH, FAFPHM, is an internationally recognized leader in cancer prevention. As an epidemiologist and public health expert, he has a longstanding interest in the preventable causes of chronic disease, particularly among women. He focuses his research on early life and adolescent lifestyle, growth, and breast cancer risk. He is also interested in approaches to speed translation of research findings into prevention strategies that work. Dr. Colditz developed the award-winning Your Disease Risk website (www.yourdiseaserisk.wustl.edu) which communicates tailored prevention messages to the public. He has published over 1100 peer-reviewed publications, six books and six reports for the Institute of Medicine, National Academy of Sciences. His h-index is over 220. In October 2006, on the basis of professional achievement and commitment to public health, Dr. Colditz was elected to membership of the National Academy of Medicine, an independent body that advises the US government on issues affecting public health. In 2011, he was awarded the American Cancer Society Medal of Honor for cancer control research. In 2012 he received the AACR-American Cancer Society Award for Research Excellence in Cancer Epidemiology and Prevention. He also received awards in 2014 for cancer prevention research from ASCO and from AACR. During 2016 he served on the Implementation Science Work Group of the Blue-Ribbon Panel to advise the National Cancer Moonshot. He received the 2018 Daniel P. Schuster Award for Distinguished Work in Clinical and Translational Science, Washington University School of Medicine. He was also elected as a Fellow, American Association for the Advancement of Science

Carlo La Vecchia received his MD from the University of Milan and a Master of Science degree in Medicine (epidemiology) from Oxford University. Presently, he is Professor of Medical Statistics and Epidemiology at the Faculty of Medicine at the University of Milan. Dr. La Vecchia serves as an editor for numerous clinical and epidemiologic journals. He is among the most renowned and productive epidemiologists in the field with over 2040 peer-reviewed papers in the literature and is among the most highly cited medical researchers in the world, according to ISI HighlyCited.com, the developer and publisher of the Science Citation Index (2003, 2017, H index 153, H10 index 1543, second Italian in Clinical Medicine). Dr. La Vecchia is Adjunct Professor of Medicine at Vanderbilt Medical Center and the Vanderbilt-Ingram Cancer Center (2002-18).

ix

x

Section Editors Gerd. P. Pfeifer received a PhD degree in biochemistry from Goethe University in Frankfurt, Germany. After postdoctoral training, he became a faculty member at the Beckman Research Institute of the City of Hope in Duarte, California, where he spent much of his career working on cancer research. In 2014, Dr. Pfeifer joined the new Center for Epigenetics at the Van Andel Research Institute in Grand Rapids, MI, United States, as a Professor of Epigenetics. Dr. Pfeifer has authored more than 300 publications, has held an NIH MERIT award, and was elected Fellow of the American Association for the Advancement of Science in 2015. Research in Dr. Pfeifer’s laboratory has been concerned with genetic and epigenetic mechanisms of human carcinogenesis, with emphasis on DNA methylation and genetic toxicology.

Marco Alessandro Pierotti graduated in 1973 in Biological Sciences at the University of Milan, Italy, and started working at the Fondazione IRCCS Istituto Nazionale dei Tumori (INT) in Milan. From 1978 to 1980 he was Visiting Investigator at the Laboratory of Chemical Cancerogenesis of the NCINIH Bethesda (MD, United States) and Postdoctoral Research Fellow at the Laboratory of Viral Oncology of the Memorial Sloan-Kettering Institute in New York. In 2006, Dr. Pierotti was appointed Scientific Director of the Fondazione IRCCS Istituto Nazionale dei Tumori in Milan, where, since 1970, he had already held various positions, including Director of the Department of Experimental Oncology. Since 1988, he has been Professor of Molecular Genetics of Cancer at the Postgraduate School of Oncology, University of Milan Medical School and co-director of the Laboratory of Molecular Diagnosis at the INT. In September 2014 he resigned from his position at INT to take the position of President and CEO of Nerviano Medical Sciences (NMS) srl, one of the biggest oncological pharma companies in Europe. In April 2015 he left the company and took the position of Scientific Coordinator of the Institute of Pediatric Researches (IRP) in Padua, Italy, devoted to study molecular aspects of the main pediatric diseases with particular focus on pediatric onco-hematology. The Institute was created by a private Charity, The City of Hope Foundation of Monte Malo (Vi). Since 2000 he is Senior Group leader of the Molecular Genetics of Cancer group at the Institute FIRC of Molecular Oncology (IFOM, Milan). Past President (2006–08) of the Italian Cancer Society, Dr. Pierotti is a member of the American Association for Cancer Research and of its Advisory Board and the Laboratory Research Awards Selection Committee. He has also been President (2006–08) of the European Association for Cancer Research (EACR) and in this role was among the founders of the European CanCer Organisation (ECCO) where he was appointed as member of the Policy Committee. In recent years, he was the Italian Representative at the Scientific Committee of the International Agency for Research on Cancer (IARC), Lyon. From 2006 to 2014 he was Scientific Secretary of Alleanza Contro il Cancro (ACC), promoted by the Italian Ministry of Health. In 2008 he was Member of the Evaluation of the Research Program Functional and Structural Genomics for DKFZ. He was an expert for the Oncology Research Projects of the European Community and was a consultant in oncology research for the Ministries of Research of different Countries. From 2006 to 2014 he was Chair of the regional Project, The Region Lombardy Oncological Network (ROL), selected in 2014 by the EC as one of the best examples of oncological network. His appointments in the Organisation of European Cancer Institutes (OECI) started in 2007 when he took the position of Vice-President. From 2008 he was the President-elect and then the President of the Organization. Finally, in 2014 he was appointed OECI Executive Secretary. Over the years, Dr. Pierotti has been Principal Investigator or Head of several national and international research grants, funded by both private and public bodies. His authorship includes over 470 publications that deal with various aspect of experimental oncology including studies on immunology, biochemistry, and molecular biology using both experimental and human tumors. In addition, since its fifth edition he is the first author of the chapter on “Oncogenes” in the most reputed textbook Cancer Medicine (Holland-Frei). The metrics of his scientific activity is summarized by an H index of 96 and total citations of 37.256 (Google scholar June 2018).

Section Editors

xi

Professor Thomas Tursz, born in Kraków, Poland, in 1946, died in Paris, France, on April 27 2018. He was Professor of Oncology at the Faculty of Medicine Paris-Sud since 1986 and General Director of the Institut Gustave Roussy (1994–2010). He was the leader of the French Doctoral School of Oncology which he founded in 1999, and President of the French Federation of Comprehensive Anticancer Centres (FNCLCC) from 2004 to 2010. He was highly involved in the European Organization for the Research and Treatment of Cancer (EORTC) as both Chairman of the Scientific Advisory Committee (2003–06) and Vice President of the Board (2006–09). His experience as President of the FNCLCC was crucial for the Organization of European Cancer Institutes (OECI) when he acted as President from 2002 to 2005. His scientific interests included the biology of virus-induced tumors, as well as immunological responses including the role of thioredoxin in lymphocytes infected by Epstein–Barr virus. In the clinical research area, he conducted a number of important clinical trials in breast cancer, lung cancer, and soft-tissue sarcoma. He had a particular interest in cytokines and gene therapy, and his clinical research activities were further disseminated to the European level when he was the Chairman of the Sarcoma Group of the EORTC (1993–96). Prof. Tursz received several prestigious awards, such as the Prix de Cancérologie from the French National League Against Cancer (1979), the Bernard Halpern Immunology Award (1983), the Rosen Oncology Award (1989), the Grand Prix in Oncology from the Academy of Medicine (1992), the Hamilton Fairley Award for clinical research (1998), and the Prix de Rayonnement Français (2001). He was the author of 350 international scientific publications. He was also an esteemed member of the Editorial Board of Molecular Oncology ever since its creation in 2007. Modified from Ulrik Ringborg and Julio E. Celis. Thomas Tursz (1946–2018) in: Molecular Oncology (2018). Published by FEBS Press and John Wiley & Sons Ltd. https://doi.org/10.1002/1878-0261.12361

This page intentionally left blank

CONTRIBUTORS TO ALL VOLUMES Rachel Abbotts University of Maryland School of Medicine, Baltimore, MD, United States

Christina Bamia National and Kapodistrian University of Athens, Athens, Greece

Ludmil B Alexandrov University of California San Diego, San Diego, CA, United States

Thomas D Bannister The Scripps Research Institute, Jupiter, FL, United States

Geneviève Almouzni Institut Curie, PSL Research University, CNRS, Equipe Labellisée Ligue contre le Cancer, Paris, France; and Sorbonne University, UPMC Univ Paris 06, CNRS, Paris, France Fatima Ameer University of the Punjab, Lahore, Pakistan Volker M Arlt MRC-PHE Centre for Environment and Health, King’s College London, London, United Kingdom; and NIHR Health Protection Research Unit in Health Impact of Environmental Hazards at King’s College London in partnership with Public Health England, London, United Kingdom Elena Arrigoni University of Pisa, Pisa, Italy Sylvia L Asa University Health Network and University of Toronto, Toronto, ON, Canada Caroline Aspord Institute for Advanced Biosciences, Immunobiology and Immunotherapy in Chronic Diseases Team, Inserm U 1209, CNRS UMR 5309, University Grenoble Alpes, Grenoble, France Livia S A Augustin National Cancer Institute IRCCS G. Pascale Foundation, Naples, Italy; and Clinical Nutrition and Risk Factor Modification Centre, St Michael’s Hospital, Toronto, On, Canada Rameshwar N K Bamezai National Centre of Applied Human Genetics, School of Life Sciences, Jawaharlal Nehru University, New Delhi, India

Michael Baumann German Cancer Research Center (DKFZ), Heidelberg, Germany; and Technical University of Dresden, Dresden, Germany Daniel Baumhoer University Hospital Basel, Basel, Switzerland Aditya Bele University of Florida, Gainesville, FL, United States Richard L Bennett University of Florida, Gainesville, FL, United States Anton Berns Oncode Institute, Division of Molecular Genetics, The Netherlands Cancer Institute, Amsterdam, The Netherlands Venkaiah Betapudi US Army Medical Research Institute of Chemical Defense, Aberdeen Proving Ground, MD, United States Joydeep Bhadury The University of Tokyo, Tokyo, Japan Justin A Bishop University of Texas Southwestern Medical Center, Dallas, TX, United States Stefania Boccia Institute of Public Health, Catholic University of the Sacred Heart, Fondazione Policlinico “Agostino Gemelli”, Rome, Italy Stefan K Bohlander University of Auckland, Auckland, New Zealand Lubor Borsig University of Zurich and Zurich Center for Integrative Human Physiology, Zurich, Switzerland

xiii

xiv

Contributors to All Volumes

Fred T Bosman Institute of Pathology, University of Lausanne Medical Center (CHUV), Lausanne, Switzerland Ekaterina Bourova-Flin University of Grenoble Alpes, Grenoble, France

Dhyan Chandra Pharmacology and Therapeutics, Roswell Park Comprehensive Cancer Center, Buffalo, NY, United States

Mariana Brandão University of Porto, Porto, Portugal; and Institute Jules Bordet, Brussels, Belgium

Laurence Chaperot Institute for Advanced Biosciences, Immunobiology and Immunotherapy in Chronic Diseases Team, Inserm U 1209, CNRS UMR 5309, University Grenoble Alpes, Grenoble, France

Kristen D Brantley Harvard T.H. Chan School of Public Health, Boston, MA, United States

Adrian K Charles Sidra Medicine, Doha, Qatar

Thomas Brenn University of Calgary, Calgary, AB, Canada; and Alberta Health Services, Calgary, AB, Canada

Harvey Checkoway University of California San Diego, La Jolla, CA, United States

Jan Buckner Mayo Clinic, Rochester, MN, United States

Sai-Juan Chen State Key Laboratory of Medical Genomics, Shanghai Institute of Hematology, Rui Jin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China

Januario B Cabral-Neto Federal University of Rio de Janeiro, Rio de Janeiro, Brazil Jean Cadet University of Sherbrooke, Sherbrooke, QC, Canada Laia Caja Uppsala University, Uppsala, Sweden Ilaria Calabrese University of Naples Federico II, Naples, Italy Elías Campo University of Barcelona, Barcelona, Spain Kaixiang Cao Northwestern University, Chicago, IL, United States Fátima Carneiro São João Hospital, Porto, Portugal; and University of Porto (FMUP), Porto, Portugal Francesca Carozzi ISPO, Cancer Prevention and Research Institute, Florence, Italy

Zhu Chen State Key Laboratory of Medical Genomics, Shanghai Institute of Hematology, Rui Jin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China Nai-Kong V Cheung Memorial Sloan-Kettering Cancer Center, New York, NY, United States Jennifer Choe Duke University School of Medicine, Durham, NC, United States Edward Chow Odette Cancer Centre, Sunnybrook Health Sciences Centre, Toronto, ON, Canada; and University of Toronto, Toronto, ON, Canada Ying-Hsia Chu The University of Wisconsin School of Medicine and Public Health, Madison, WI, United States

France Carrier University of Maryland, Baltimore, MD, United States

Fabio Ciccarone University of Rome “Tor Vergata”, Rome, Italy

Carla Carrilho Hospital Central of Maputo, Maputo, Mozambique; and University Eduardo Mondlane, Maputo, Mozambique

Maria Rosa Ciriolo University of Rome “Tor Vergata”, Rome, Italy

Nathalie Cassoux Institut Curie, Paris, France

Graham A Colditz Alvin J. Siteman Cancer Center and Institute for Public Health, Washington University School of Medicine, St. Louis, MO, United States

JoAnn C Castelli University of California San Francisco, San Francisco, CA, United States

Laura C Collopy University College London, London, United Kingdom

Contributors to All Volumes

Antonio Cossu University Hospital Health Unit/Azienda OspedalieroUniversitaria (AOU), Sassari, Italy Sarah E Coupland Royal Liverpool University Hospital, Liverpool, United Kingdom; and University of Liverpool, Liverpool, United Kingdom Alix Coysh University of Auckland, Auckland, New Zealand Ana Cuenda Centro Nacional de Biotecnología/CSIC (CNB-CSIC), Madrid, Spain Romano Danesi University of Pisa, Pisa, Italy Rachel Dankner Sheba Medical Center, Tel Hashomer, Israel; Sackler Faculty of Medicine, School of Public Health, Tel Aviv University, Tel Aviv, Israel; and The Feinstein Institute for Medical Research, Manhasset, NY, United States Ben Davidson Oslo University Hospital, Oslo, Norway; and University of Oslo, Oslo, Norway Carlos E de Andrea University of Navarra, Pamplona, Spain Carlo DeAngelis Odette Cancer Centre, Sunnybrook Health Sciences Centre, Toronto, ON, Canada; and University of Toronto, Toronto, ON, Canada Vincenzo De Giorgi University of Florence, Florence, Italy Vincenzo De Iuliis Gabriele D’Annunzio University of Chieti, Chieti, Italy Olivier Delattre Institute Curie Hospital, Paris, France; and Inserm U830, SIREDO Oncology Center, Paris Sciences and Letters (PSL) Research University, Paris, France Laurence de Leval Lausanne University Hospital (CHUV), Lausanne, Switzerland

Marzia Del Re University of Pisa, Pisa, Italy Marco Demaria European Institute for the Biology of Aging (ERIBA), University Medical Center Groningen (UMCG), Groningen, The Netherlands; and University of Groningen, Groningen, The Netherlands Catherine de Martel International Agency for Research on Cancer, Lyon, France Evangelina López de Maturana Spanish National Cancer Research Center (CNIO), Madrid, Spain; and Biomedical Research Networking Centre in Oncology, Madrid, Spain Laurence Desjardins Institut Curie, Paris, France Elizabeth E Devore Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, United States Luca Di Leo University of Rome “Tor Vergata”, Rome, Italy Antonello Di Paolo University of Pisa, Pisa, Italy Emanuela D’Angelo   University of LAquila, LAquila, Italy Nadja Ebert German Cancer Research Center (DKFZ), Heidelberg, Germany; and Technical University of Dresden, Dresden, Germany A Heather Eliassen Harvard T.H. Chan School of Public Health, Boston, MA, United States; Brigham and Women’s Hospital; and Harvard Medical School, Boston, MA, United States Iman El Sayed Medical Research Institute, Alexandria University, Alexandria, Egypt Meira Epplein Duke University, Durham, NC, United States

Julio Delgado University of Barcelona, Barcelona, Spain

Iñigo Espinosa Autonomous University of Barcelona, Barcelona, Spain

Stefan Delorme German Cancer Research Center (DKFZ), Heidelberg, Germany

Irene Esposito Heinrich Heine University, University Hospital, Duesseldorf, Germany

xv

xvi

Contributors to All Volumes

Manel Esteller Bellvitge Biomedical Biomedical Research Institute (IDIBELL), Barcelona, Catalonia, Spain; University of Barcelona, Barcelona, Catalonia, Spain; and Catalan Institution for Research and Advanced Studies (ICREA), Barcelona, Catalonia, Spain Shereen Ezzat University Health Network and University of Toronto, Toronto, ON, Canada Chuannan Fan Oncode Institute, Leiden University Medical Center, Leiden, Netherlands

Preethi S George Regional Cancer Centre, Trivandrum, India Katherine A Giles UNSW Sydney, Sydney, NSW, Australia Thomas J Giordano University of Michigan Health System, Ann Arbor, MI, United States Nicolas Girard Curie Institute, Paris, France; and Lyon University, Lyon, France

Lynnette Fernandez-Cuesta International Agency for Research on Cancer, Lyon, France

Abhiram Gokhale Pharmacology and Therapeutics, Roswell Park Comprehensive Cancer Center, Buffalo, NY, United States

Giustina Ferone Oncode Institute, Division of Molecular Genetics, The Netherlands Cancer Institute, Amsterdam, The Netherlands

Tyler Golato National Institute on Aging, Intramural Research Program, National Institutes of Health, Baltimore, MD, United States

Stefano Fiori European Institute of Oncology, Milan, Italy

Paulina Gomez-Rubio Spanish National Cancer Research Center (CNIO), Madrid, Spain; and Biomedical Research Networking Centre in Oncology, Madrid, Spain

Martin Fischer Leibniz Institute on Aging e Fritz Lipmann Institute (FLI), Jena, Germany Jean-François Fléjou Saint-Antoine Hospital, AP-HP, Faculty of Medicine, Sorbonne University, Paris, France Filipa Fontes University of Porto, Porto, Portugal; and Portuguese Institute of Oncology Francisco Gentil, Porto, Portugal Simon J Furney Royal College of Surgeons in Ireland, Dublin, Ireland Nina Gale Faculty of Medicine, University of Ljubljana, Ljubljana, Slovenia Sara Gandini European Institute of Oncology, Milan, Italy Vithusha Ganesh Odette Cancer Centre, Sunnybrook Health Sciences Centre, Toronto, ON, Canada; and University of Toronto, Toronto, ON, Canada Mengqing Gao Shanghai Jiao Tong University School of Medicine, Shanghai, China

M M Gottesman Laboratory of Cell Biology, National Cancer Institute, National Institutes of Health, Bethesda, MD, USA Bernardo H L Goulart Hutchinson Institute of Cancer Outcomes Research (HICOR), Seattle, WA, United States; Fred Hutchinson Cancer Research Center, Seattle, WA, United States; and University of Washington, Seattle, WA, United States William J Graham V Ludwig Institute for Cancer Research, San Diego Branch, La Jolla, CA, United States Emmanuel Griessinger INSERM, C3M, Nice, France; and University of Nice Sophia, Nice, France John D Groopman Johns Hopkins University, Baltimore, MD, United States Irene Gullo São João Hospital, Porto, Portugal; and University of Porto (FMUP), Porto, Portugal Meiyun Guo University of British Columbia, Vancouver, BC, Canada

Contributors to All Volumes

Lena Haeberle Heinrich Heine University, University Hospital, Duesseldorf, Germany Pierre Hainaut Cancer Biology, Faculty of Medicine, University Grenoble-Alpes, Grenoble, France; and Institute for Advanced Biosciences, Inserm, CNRS, UGA, Grenoble, France Susan E Hankinson University of Massachusetts Amherst, Amherst, MA, United States Leonie Hartl University of Gothenburg, Gothenburg, Sweden Christian Hartmann Medical School Hannover, Hannover, Germany Dana Hashim International Agency for Research on Cancer, Lyon, France; and Icahn School of Medicine at Mount Sinai, New York, NY, United States Dan He Drexel University, Philadelphia, PA, United States

xvii

Janusz A Jankowski University Hospitals of Morecambe Bay NHS Foundation Trust, Lancaster, United Kingdom; National Institute of Health and Care Excellence, Manchester, United Kingdom; and Royal College of Surgeons in Ireland, Dublin, Ireland Daniel Jeffery Institut Curie, PSL Research University, CNRS, Equipe Labellisée Ligue contre le Cancer, Paris, France; and Sorbonne University, UPMC Univ Paris 06, CNRS, Paris, France Albert Jeltsch Stuttgart University, Stuttgart, Germany Seung-Gi Jin Center for Epigenetics, Van Andel Research Institute, Grand Rapids, MI, United States Ivo Julião University of Porto, Porto, Portugal; and Portuguese Institute of Oncology Francisco Gentil, Porto, Portugal Kunal C Kadakia Levine Cancer Institute, Carolinas HealthCare System, Charlotte, NC, United States

Andrew D Higham University Hospitals of Morecambe Bay NHS Foundation Trust, Lancaster, United Kingdom

Santosh U Kafle Kathmandu University Birat Medical College and Teaching Hospital, Morang, Nepal; and Biratnagar Eye Hospital, Biratnagar, Nepal

Pancras C W Hogendoorn Leiden University Medical Center, Leiden, the Netherlands

Russel Kahmke Duke University School of Medicine, Durham, NC, United States

Wan Hua Oncode Institute, Leiden University Medical Center, Leiden, Netherlands

Purvi M Kakadiya University of Auckland, Auckland, New Zealand

Tao Huang Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, Shanghai, China Kent Hunter National Institutes of Health, Bethesda, MD, United States Joseph Inigo Pharmacology and Therapeutics, Roswell Park Comprehensive Cancer Center, Buffalo, NY, United States

Sudhakar Kalakonda University of Maryland School of Medicine, and Greenebaum Cancer Center, Baltimore, MD, United States Emarene Kalaw The University of Queensland, Brisbane, QLD, Australia Dhananjaya V Kalvakolanu University of Maryland School of Medicine, and Greenebaum Cancer Center, Baltimore, MD, United States

Thergiory Irrazábal University of Toronto, Toronto, ON, Canada

Heidrun Karlic Ludwig Boltzmann Cluster Oncology, Hanusch Hospital, Vienna, Austria

Corbin D Jacobs Duke University School of Medicine, Durham, NC, United States

Stefan Kempa Berlin Institute for Medical Systems Biology at the MDC Berlin-Buch, Berlin, Germany

xviii

Contributors to All Volumes

Saadi Khochbin University of Grenoble Alpes, Grenoble, France Jakub Khzouz King Hussein Cancer Center, Amman, Jordan Edward S Kim Levine Cancer Institute, Carolinas HealthCare System, Charlotte, NC, United States Peter S Klein University of Pennsylvania, Philadelphia, PA, United States Richard D Kolodner Ludwig Institute for Cancer Research, San Diego Branch, La Jolla, CA, United States Annette M Krais Lund University, Lund, Sweden Paul Krimpenfort Oncode Institute, Division of Molecular Genetics, The Netherlands Cancer Institute, Amsterdam, The Netherlands Ivana Kulhánová Section of Cancer Surveillance, International Agency for Research on Cancer, Lyon, France Rahul Kumar Pharmacology and Therapeutics, Roswell Park Comprehensive Cancer Center, Buffalo, NY, United States Razelle Kurzrock UC San Diego Moores Cancer Center, San Diego, CA, United States Sunil R Lakhani The University of Queensland, Brisbane, QLD, Australia; and Pathology Queensland, Herston, QLD, Australia Sylvie Lantuejoul Léon Bérard Centre, Lyon, France; and Grenoble University, La Tronche, France Stefano La Rosa Institute of Pathology, Lausanne University Hospital, Lausanne, Switzerland Heinz Läubli University Hospital Basel, Basel, Switzerland Matteo Lazzeroni European Institute of Oncology, Milan, Italy Christopher J Lengner University of Pennsylvania, Philadelphia, PA, United States

Jay A Levy University of California San Francisco, San Francisco, CA, United States Jonathan D Licht University of Florida, Gainesville, FL, United States Ricardo V Lloyd The University of Wisconsin School of Medicine and Public Health, Madison, WI, United States Leendert H J Looijenga Erasmus University Medical Center, Rotterdam, The Netherlands Estibaliz Lopez-Rodrigo Memorial Sloan-Kettering Cancer Center, New York, NY, United States Fernando López Álvarez Asturias Central University Hospital, University of Oviedo, University Institute of Oncology of Asturias (IUOPA), ISPA, CIBERONC, Oviedo, Spain Nuno Lunet University of Porto, Porto, Portugal Eamonn R Maher University of Cambridge, Cambridge, United Kingdom Sayantan Maji University of Florida, Gainesville, FL, United States Sharan Malagobadan University of Malaya, Kuala Lumpur, Malaysia Leila Malek Odette Cancer Centre, Sunnybrook Health Sciences Centre, Toronto, ON, Canada; and University of Toronto, Toronto, ON, Canada David Malkin The Hospital for Sick Children, Toronto, ON, Canada; and University of Toronto, Toronto, ON, Canada Olivier Manches Institute for Advanced Biosciences, Immunobiology and Immunotherapy in Chronic Diseases Team, Inserm U 1209, CNRS UMR 5309, University Grenoble Alpes, Grenoble, France Mario Mandalà Papa Giovanni XXIII Hospital, Bergamo, Italy Salomon Manier University of Lille, INSERM UMR-S 1172, Lille, France; University of Lille, CHU Lille, Lille, France; and Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA, United States

Contributors to All Volumes

Alberto Mantovani Humanitas Clinical and Research Center, Milan, Italy; Humanitas University, Milan, Italy; and Queen Mary University of London, London, United Kingdom Alberto Martin University of Toronto, Toronto, ON, Canada Daniela Massi University of Florence, Florence, Italy Alexandre Matet Curie Institute, Paris, France Aleyamma Mathew Regional Cancer Centre, Trivandrum, India Ashley G Matusz-Fisher Levine Cancer Institute, Carolinas HealthCare System, Charlotte, NC, United States Christopher A Maxwell University of British Columbia, Vancouver, BC, Canada Amy E McCart Reed The University of Queensland, Brisbane, QLD, Australia Simon S McDade Queen’s University Belfast, Belfast, United Kingdom Lin Mei University of British Columbia, Vancouver, BC, Canada Carlos F M Menck University of São Paulo, São Paulo, Brazil

Rimsha Munir University of the Punjab, Lahore, Pakistan Matthew Murtha Bellvitge Biomedical Biomedical Research Institute (IDIBELL), Barcelona, Catalonia, Spain Rebecca L Myers University of Pennsylvania, Philadelphia, PA, United States Ferran Nadeu IDIBAPS, Barcelona, Spain Noor Hasima Nagoor University of Malaya, Kuala Lumpur, Malaysia Jean Nakhle IRMB, INSERM, CNRS, University of Montpellier, Montpellier, France Shreeram C Nallar University of Maryland School of Medicine, and Greenebaum Cancer Center, Baltimore, MD, United States Charlotte K Y Ng University Hospital Basel, Basel, Switzerland Mina Nikanjam UC San Diego Moores Cancer Center, San Diego, CA, United States

Jian-Qing Mi Shanghai Jiao Tong University School of Medicine, Shanghai, China

Stephanie Nishi National Cancer Institute IRCCS G. Pascale Foundation, Naples, Italy; and University of Toronto, Toronto, ON, Canada

Orli Michaeli The Hospital for Sick Children, Toronto, ON, Canada; and University of Toronto, Toronto, ON, Canada

J W Oosterhuis Erasmus University Medical Center, Rotterdam, The Netherlands

Concetta Montagnese National Cancer Institute IRCCS G. Pascale Foundation, Naples, Italy

Peter O’Donovan Royal College of Surgeons in Ireland, Dublin, Ireland

Michael J Moravan Duke University School of Medicine, Durham, NC, United States Yvonne Mowery Duke University School of Medicine, Durham, NC, United States

xix

Jordan O’Malley Pharmacology and Therapeutics, Roswell Park Comprehensive Cancer Center, Buffalo, NY, United States Giuseppe Palmieri National Research Council (CNR), Sassari, Italy

Maciej M Mrugala Mayo Clinic, Scottsdale, AZ, United States

Elvira Palumbo National Cancer Institute IRCCS G. Pascale Foundation, Naples, Italy

Karl Munger Tufts University School of Medicine, Boston, MA, United States

Thomas G Papathomas University of Birmingham, Birmingham, United Kingdom

xx

Contributors to All Volumes

Palak R Parekh University of Maryland, Baltimore, MD, United States Sophie Park Grenoble-Alpes University, Grenoble, France; and Institute for Advanced Biosciences, Grenoble, France Yikyung Park Washington University School of Medicine, St. Louis, MO, United States Roberta Pastorino Institute of Public Health, Catholic University of the Sacred Heart, Rome, Italy Alpa V Patel American Cancer Society, Atlanta, GA, United States Päivi Peltomäki University of Helsinki, Helsinki, Finland Francesco Pezzella University of Oxford, Oxford, United Kingdom Gerd P Pfeifer Center for Epigenetics, Van Andel Research Institute, Grand Rapids, MI, United States Thierry Philip Curie Institute, Paris, France Stefano A Pileri European Institute of Oncology, Milan, Italy; and Bologna University School of Medicine, Bologna, Italy Scott W Piraino Royal College of Surgeons in Ireland, Dublin, Ireland Salvatore Piscuoglio University Hospital Basel, Basel, Switzerland Fiona J Pixley The University of Western Australia, Crawley, WA, Australia Martyn Plummer International Agency for Research on Cancer, Lyon, France Katrina Podsypanina Curie Institute, PSL Research University, CNRS, Equipe Labellisée Ligue contre le Cancer, Paris, France; and Sorbonne University, UPMC Univ Paris 06, CNRS, Paris, France

Giuseppe Porciello National Cancer Institute IRCCS G. Pascale Foundation, Naples, Italy Gopinath Prakasam National Centre of Applied Human Genetics, School of Life Sciences, Jawaharlal Nehru University, New Delhi, India Jaime Prat Autonomous University of Barcelona, Barcelona, Spain Christopher D Putnam Ludwig Institute for Cancer Research, San Diego Branch, La Jolla, CA, United States Shuo Qie Medical University of South Carolina, Charleston, SC, United States Kavitha J Ramchandran Stanford University, Stanford, CA, United States Elizabeth G Ratcliffe University Hospitals of Morecambe Bay NHS Foundation Trust, Lancaster, United Kingdom Tibor A Rauch Rush University Medical Center, Chicago, IL, United States Itedale N Redwan University of Gothenburg, Gothenburg, Sweden Erika Rees-Punia University of Georgia, Athens, GA, United States Andrew Reid Pathology Queensland, Herston, QLD, Australia Giuliana Restante University of Pisa, Pisa, Italy J Richard Wagner University of Sherbrooke, Sherbrooke, QC, Canada Juan Pablo Rodrigo Asturias Central University Hospital, University of Oviedo, University Institute of Oncology of Asturias (IUOPA), ISPA, CIBERONC, Oviedo, Spain Isabelle Romieu National Institute of Public Health, Cuernavaca, Mexico; and Emory University, Atlanta, GA, United States

Igor P Pogribny National Center for Toxicological Research, Jefferson, AR, United States

Christina Ross National Institutes of Health, Bethesda, MD, United States

Sergey D Popov University Hospital of Wales, Cardiff, United Kingdom

Paolo Giorgi Rossi AUSL Reggio Emilia, IRCCS, Reggio Emilia, Italy

Contributors to All Volumes

Sophie Rousseaux University of Grenoble Alpes, Grenoble, France Nadine Royla Berlin Institute for Medical Systems Biology at the MDC Berlin-Buch, Berlin, Germany Joseph K Salama Duke University School of Medicine, Durham, NC, United States Lorenzo Salvati University of Florence, Florence, Italy Nianli Sang Drexel University, Philadelphia, PA, United States; and Sidney Kimmel Cancer Center of Philadelphia, Philadelphia, PA, United States Alain Sarasin Paris-Sud University, Villejuif, France Eva S Schernhammer Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, United States; and Medical University of Vienna, Wien, Austria Gudrun Schleiermacher Curie Institute, Paris, France Neil J Sebire Great Ormond Hospital for Sick Children, London, United Kingdom Veena Shankaran Hutchinson Institute of Cancer Outcomes Research (HICOR), Seattle, WA, United States; Fred Hutchinson Cancer Research Center, Seattle, WA, United States; and University of Washington, Seattle, WA, United States Niva Shapira Ashkelon Academic College, Ashkelon, Israel Akanksha Sharma Mayo Clinic, Scottsdale, AZ, United States Ossie Sharon Ashkelon Academic College, Ashkelon, Israel Ali Shilatifard Northwestern University, Chicago, IL, United States Masahiro Shuda University of Pittsburgh Medical Center (UPMC) Hillman Cancer Center, Pittsburgh, PA, United States; and University of Pittsburgh, Pittsburgh, PA, United States

xxi

C Simon Herrington University of Edinburgh, Edinburgh Cancer Research Centre, Institute of Genetics and Molecular Medicine, Western General Hospital, Edinburgh, United Kingdom Stina Simonsson University of Gothenburg, Gothenburg, Sweden Rajinder Singh University of Leicester, Leicester, United Kingdom Pieter J Slootweg Radboud University Medical Center, Nijmegen, The Netherlands Egbert F Smit Netherlands Cancer Institute, Amsterdam, The Netherlands Joshua W Smith Johns Hopkins University, Baltimore, MD, United States Eric Solary Gustave Roussy, Villejuif, France; and Paris-Sud University, Le Kremlin-Bicêtre, France Mingyang Song Harvard Medical School, Boston, MA, United States; Massachusetts General Hospital, Boston, MA, United States; and Harvard T.H. Chan School of Public Health, Boston, MA, United States Claus Storgaard Sørensen University of Copenhagen, Copenhagen, Denmark Xiao-Jian Sun State Key Laboratory of Medical Genomics, Shanghai Institute of Hematology, Rui Jin Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai, China Katarzyna Szyma nska Science to the Point, Archamps Technopole, Archamps, France Valentina Tabanelli European Institute of Oncology, Milan, Italy Phillippa C Taberlay University of Tasmania, Hobart, TAS, Australia Hideyuki Takeshima National Cancer Center Research Institute, Tokyo, Japan E-Jean Tan Uppsala University, Uppsala, Sweden

xxii

Contributors to All Volumes

Peter ten Dijke Oncode Institute, Leiden University Medical Center, Leiden, Netherlands Luigi M Terracciano University Hospital Basel, Basel, Switzerland Valentina Thomas Royal College of Surgeons in Ireland, Dublin, Ireland Mangesh A Thorat Centre for Cancer Prevention, Wolfson Institute of Preventive Medicine, Queen Mary University of London, London, United Kingdom; and Breast Services, Division of Surgery and Interventional Science, Whittington Hospital, London, United Kingdom Andrew Thorburn University of Colorado School of Medicine, Aurora, CO, United States Erik Thunnissen VU University Medical Center, Amsterdam, The Netherlands Falk Tillner Technical University of Dresden, Dresden, Germany Arthur S Tischler Tufts University School of Medicine, Boston, MA, United States Alessia Tognetto Institute of Public Health, Catholic University of the Sacred Heart, Rome, Italy Kazunori Tomita University College London, London, United Kingdom Christina G Towers University of Colorado School of Medicine, Aurora, CO, United States Lawrence D True University of Washington, Seattle, WA, United States Silvia Uccella University of Insubria, Varese, Italy Toshikazu Ushijima National Cancer Center Research Institute, Tokyo, Japan

Funda Vakar-Lopez University of Washington, Seattle, WA, United States Franz Varga Hanusch Hospital of WGKK and AUVA Trauma Centre Meidling, Vienna, Austria Matthew G Varga University of North Carolina, Chapel Hill, NC, United States Michael C Velarde University of the Philippines Diliman, Quezon City, Philippines Marie-Luce Vignais IRMB, INSERM, CNRS, University of Montpellier, Montpellier, France Neus Villamor University of Barcelona, Barcelona, Spain Rocío C Viñuelas University of Gothenburg, Gothenburg, Sweden Sara Vitale National Cancer Institute IRCCS G. Pascale Foundation, Naples, Italy Jamie Von Roenn American Society of Clinical Oncology, Alexandria, VA, United States Gordan M Vujanic Sidra Medicine, Doha, Qatar Jean Y J Wang University of California San Diego, La Jolla, CA, United States Jin Wang Shanghai Jiao Tong University School of Medicine, Shanghai, China Sarah Watson Institute Curie Hospital, Paris, France; and Inserm U830, SIREDO Oncology Center, Paris Sciences and Letters (PSL) Research University, Paris, France

Salvatore Vaccarella Section of Infections, International Agency for Research on Cancer, Lyon, France

Elisabete Weiderpass Cancer Registry of Norway, Oslo, Norway; Karolinska Institute, Stockholm, Sweden; UiT The Arctic University of Norway, Tromsø, Norway; and Folkhälsan Research Center, Helsinki, Finland

William Vainchenker Gustave Roussy, Villejuif, France

Sara Weirich Stuttgart University, Stuttgart, Germany

Contributors to All Volumes

Pieter Wesseling VU University Medical Center/Brain Tumor Center Amsterdam, Amsterdam, The Netherlands; Princess Máxima Center for Pediatric Oncology, Utrecht, The Netherlands; and University Medical Center Utrecht, Utrecht, The Netherlands Katharina Wiedemeyer University of Heidelberg, Heidelberg, Germany Sean R Williamson Henry Ford Cancer Institute, Henry Ford Health System, Detroit, MI, United States; and Wayne State University School of Medicine, Detroit, MI, United States David M Wilson, III National Institute on Aging, Intramural Research Program, National Institutes of Health, Baltimore, MD, United States Frank Winkler University Hospital Heidelberg, Heidelberg, Germany; and German Cancer Consortium (DKTK), German Cancer Research Center (DKFZ), Heidelberg, Germany Matthias Wirth Heinrich Heine University, University Hospital, Duesseldorf, Germany Tejas Yadav Curie Institute, PSL Research University, CNRS, Equipe Labellisée Ligue contre le Cancer, Paris, France; and Sorbonne University, UPMC Univ Paris 06, CNRS, Paris, France

xxiii

Marie Yammine University of Lille, INSERM UMR-S 1172, Lille, France Maryam Yousefi University of Pennsylvania, Philadelphia, PA, United States Nousheen Zaidi University of the Punjab, Lahore, Pakistan Jing Zhang Oncode Institute, Leiden University Medical Center, Leiden, Netherlands Yin Zhang Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, United States Maria Zhivagui International Agency for Research on Cancer (WHO), Lyon, France Nina Zidar Faculty of Medicine, University of Ljubljana, Ljubljana, Slovenia

This page intentionally left blank

HOW TO USE THE ENCYCLOPEDIA Structure of the Encyclopedia All articles in the encyclopedia are arranged alphabetically as a series of entries. There are four features to help you easily find the topic you are interested in: an alphabetical contents list, cross references, a full subject index, and contributors. 1. Alphabetical contents list: The alphabetical contents list, which appears at the front of each volume, lists the entries in the order that they appear in the encyclopedia. So that they can be easily located, entry titles generally begin with the key word or phrase indicating the topic, with any generic terms following. For example, “Multiple Myeloma: Pathology and Genetics” is the entry title rather than “Pathology and Genetics of Multiple Myeloma”. 2. Cross references: Virtually all the entries in the encyclopedia have been extensively cross-referenced. The cross references which appear at the end of an entry, serve three different functions: i. To draw the reader’s attention to related material on other entries ii. To indicate material that broadens and extends the scope of the article iii. To indicate material that covers a topic in more depth Example The following list of cross-references appears at the end of the entry “CarcinogendDNA Adducts”. See also: Cancer Risk Reduction Through Lifestyle Changes. Cell Responses to DNA Damage. Genetic Instability. Hereditary Risk of Breast and Ovarian Cancer: BRCA1 and BRCA2. Molecular Epidemiology and Cancer Risk. Role of DNA Repair in Carcinogenesis and Cancer Therapeutics.

3. Index: The index appears at the end of volume 3 and includes page numbers for quick reference to the information you are looking for. The index entries differentiate between references to a whole entry, a part of an entry, and a table or figure. 4. Contributors: At the start of each volume there is a list of the authors who contributed to all volumes.

xxv

This page intentionally left blank

SUBJECT CLASSIFICATION Causes of Cancer Aflatoxins Aging and Cancer Cancers as Ecosystems: From Cells to Population Diabetes and Cancer Dietary Factors and Cancer Helicobacter Pylori-Mediated Carcinogenesis HIV (Human Immunodeficiency Virus) Obesity and Cancer: Epidemiological Evidence Opisthorchis Viverrini, Clonorchis Sinensis, and Cholangiocarcinoma Papillomaviruses Physical Inactivity and Cancer Radiation Therapy-Induced Metastasis and Secondary Malignancy Sleep Disturbances and Misalignment in Cancer

Diagnosis and Therapy Acute Lymphocytic Leukemia: Diagnosis and Treatment Acute Myelogeneous Leukemia: Diagnosis and Treatment Bladder Cancer: Pathology, Genetics, Diagnosis, and Treatment Bone and Soft Tissue Sarcoma: From Molecular Features to Clinical Applications Cancer Vaccines: Dendritic Cell-Based Vaccines and Related Approaches Cholangiocarcinoma: Diagnosis and Treatment Chromatin Dynamics and Cancer: Epigenetic Parameters and Cellular Fate Chronic Myelogenous Leukemia: Pathology, Genetics, Diagnosis, and Treatment Colorectal Cancer: Diagnosis and Treatment End of Life Support Esophageal Cancer: Diagnosis and Treatment Glioblastoma: Biology, Diagnosis, and Treatment Interferons: Cellular and Molecular Biology of Their Actions Kidney Cancer: Diagnosis and Treatment Laryngeal Cancer: Diagnosis and Treatment Malignant Tumors of the Eye, Conjunctiva, and Orbit: Diagnosis and Therapy Myelodysplastic Syndromes: Mechanisms, Diagnosis, and Treatment Nasopharyngeal Carcinoma: Diagnosis and Treatment Neuroblastoma: Diagnosis and Treatment New Rationales and Designs for Clinical Trials in the Era of Precision Medicine Non-Hodgkin Lymphoma: Diagnosis and Treatment Oncology Imaging Oral Cavity Cancer: Diagnosis and Treatment Ovarian Cancer: Diagnosis and Treatment Pancreatic Cancer: Diagnosis and Treatment P-Glycoprotein-Mediated Multidrug Resistance Pituitary Tumors: Diagnosis and Treatment

xxvii

xxviii

Subject Classification

Prostate Cancer: Diagnosis and Treatment Radiation Oncology Squamous Cell and Basal Cell Carcinoma of the Skin: Diagnosis and Treatment Symptom Control Thyroid Cancer: Pathology, Genetics, Diagnosis, and Treatment Uterine Cervix Cancer: Diagnosis and Treatment

Hallmarks of Cancer Anoikis Autophagy and Cancer Cancer-Related Inflammation in Tumor Progression Cell Adhesion During Tumorigenesis and Metastasis Cell Responses to DNA Damage DNA Mismatch Repair: Mechanisms and Cancer Genetics Epithelium to Mesenchyme Transition Genetic Instability Glutamine Metabolism and Cancer Induced Pluripotent Stem Cells and Yamanaka factors Inhibitors of Lactate Transport: A Promising Approach in Cancer Drug Discovery Lipid Metabolism Metastatic SignaturesdThe Tell-Tale Signs of Metastasis Mevalonate Pathway Mitogen-Activated Protein Kinases (MAPK) in Cancer Pyruvate Kinase Role of DNA Repair in Carcinogenesis and Cancer Therapeutics Senescence and Cellular Immortality Telomeres, Telomerase, and Cancer TGF-b in Cancer Progression: From Tumor Suppressor to Tumor Promotor TP53 Tumor-Associated Macrophages Tumors and Blood Vessel Interactions: A Changing Hallmark of Cancer Tunneling Nanotubes (TNTs): Intratumoral Cell-to-Cell Communication

Mechanisms Animal Models of Cancer: What We Can Learn From Mice Ataxia Telangiectasia Syndrome CarcinogendDNA Adducts Carcinogenesis: Role of Reactive Oxygen and Nitrogen Species Chromosome Rearrangements and Translocations Copy Number Variations in Tumors Defective 5-Methylcytosine Oxidation in Tumorigenesis DNA Methylation Changes in Cancer: Cataloguing DNA Methylation Changes in Cancer: Mechanisms Enhancers in Cancer: Genetic and Epigenetic Deregulation Environmental Exposures and Epigenetic Perturbations Epigenetic Therapy Genome Wide Association Studies (GWAS) Hereditary Risk of Breast and Ovarian Cancer: BRCA1 and BRCA2 Hormones and Cancer Li–Fraumeni Syndrome Lynch Syndrome Microbiota and Colon Cancer: Orchestrating Neoplasia Through DNA Damage and Immune Dysregulation Mod Squad: Altered Histone Modifications in Cancer Mutational Signatures and the Etiology of Human Cancers Mutations in Chromatin Remodeling Factors

Subject Classification Mutations in DNA Methyltransferases and Demethylases Mutations in Histone Lysine Methyltransferases and Demethylases Mutations: Driver Versus Passenger Polyomaviruses in Human Cancer Systems Biology Approach to Study Cancer Metabolism TCA Cycle Aberrations and Cancer Unprogrammed Gene Activation: A Critical Evaluation of Cancer Testis Genes Wnt Signaling in Intestinal Stem Cells and Cancer Xeroderma Pigmentosum: When the Sun Is the Enemy

Pathology and Genetics of Specific Cancers Adrenal Glands Tumors: Pathology and Genetics Anal Cancer: Pathology and Genetics Bladder Cancer: Pathology, Genetics, Diagnosis, and Treatment Bones and Joints Cancer: Pathology and Genetics Breast Cancer: Pathology and Genetics Chronic Lymphocytic Leukemia: Pathology and Genetics Chronic Myelogenous Leukemia: Pathology, Genetics, Diagnosis, and Treatment Colorectal Cancer: Pathology and Genetics Endometrial Cancer: Pathology and Genetics Esophageal Cancer: Pathology and Genetics Eye and Orbit Cancer: Pathology and Genetics Gastric Cancer: Pathology and Genetics Germ Cell Tumors: Pathology and Genetics Glioblastoma: Pathology and Genetics Hepatocellular Carcinoma: Pathology and Genetics Hodgkin Lymphoma: Pathology and Genetics Integrative Molecular Tumor Classification: A Pathologist’s View Jaws Cancer: Pathology and Genetics Larynx Cancer: Pathology and Genetics Malignant Skin Adnexal Tumors: Pathology and Genetics Melanoma: Pathology and Genetics Multiple Myeloma: Pathology and Genetics Neuroblastoma: Pathology and Genetics Non-Hodgkin Lymphoma: Pathology and Genetics Nonsmall-Cell Cancers of the Lung: Pathology and Genetics Oral and Oropharyngeal Cancer: Pathology and Genetics Ovarian Cancer: Pathology and Genetics Pancreatic Cancer: Pathology and Genetics Parathyroid Cancer: Pathology and Genetics Pituitary Tumors: Pathology and Genetics Prostate Cancer: Pathology and Genetics Renal Cell Cancer: Pathology and Genetics Small-Cell Cancer of the Lung: Pathology and Genetics Thyroid Cancer: Pathology, Genetics, Diagnosis, and Treatment Uterine Cervix Cancer: Pathology and Genetics Wilms Tumor: Pathology and Genetics

Prevention and Control Aspirin and Cancer Cancer Disparities Cancer in Populations in Transition Cancer in Sub-Saharan Africa Cancer in the Middle East Cancer Risk Reduction Through Lifestyle Changes

xxix

xxx

Subject Classification

Cancer Survival and Survivorship Cervical Cancer: Screening, Vaccination, and Preventive Strategies Chemoprevention of Cancer: An Overview of Promising Agents and Current Research Chemoprevention Trials Diet and Cancer Environmental and Occupational Exposures Financial Burden of Cancer Care Hereditary Cancer Syndromes: Identification and Management Metformin Molecular Epidemiology and Cancer Risk Prevention and Control: Nutrition, Obesity, and Metabolism

PREFACE Cancer holds a special status in biology, medicine, and society. It offers formidable challenges to basic, clinical, and population science; it ranks at the top of medical research priorities and healthcare costs in most countries. The general public is continuously exposed to news of discoveries promising to defeat the disease in the near future. Indeed, ground-breaking advances are being made, from preventive vaccines to genome-guided personalized medicine, sophisticated imaging and surgical technologies, and systemic therapies aimed at awakening natural immune responses against cancer. These novel therapeutic approaches make cure a real possibility for a growing number of patients. They also launch cancer care into a new era of maintaining the disease under control for an indefinite period of time, turning it into a form of chronic disease. At the same time, evidence-based prevention and early detection strategies have opened a new window on the natural history of the disease, enabling effective intervention well ahead of diagnosis. As a result, the first two decades of this millennium have witnessed a marked decrease in the mortality and, in some instances, the incidence of cancers that have dominated the death toll in more developed countries during the second half of the 20th century. A turning point in our understanding of cancer is the deciphering of the human genome and its spin-off endeavors aimed at exploring the genomic landscape and architecture of human cancers. These discoveries are causing a major overhaul of our vision of cancer as a dynamic, rapidly evolving, and heterogeneous disease at the individual level. Harnessing this complexity requires mastering increasingly complex sources of data at molecular, cellular, systemic, personal, environmental, and societal level, heralding the emergence of big-data science in cancer diagnosis and treatment. However, this exceptional acceleration in knowledge and solutions cannot hide the fact that cancer remains a global scourge that exerts a massive burden on humankind and societies worldwide, in particular in societies in transition and in low-resource contexts. Today, cancer crystallizes many of the major societal challenges pertaining to lifestyles, sustainable development and environmental policies, demography and population aging, access to education and healthcare, sharing of resources and knowledge, and protection of persons and personal information. The information on cancer available at a fingertip is overwhelming in volume, complexity, veracity, and velocity. We worked on the Third Edition of Elsevier’s Encyclopedia of Cancer with this rapidly changing background. Rather than aiming at developing a comprehensive framework encompassing all aspects, we attempted to address the literal meaning of the greek terms ἐgkύklιo2 paιdείa, which means “general education”. While we retained some articles from the previous edition, which, at the time of the publication, represented an exceptional achievement of Dr. J. Bertino, we largely modified the structure and the list of chapters, and the possibility of continuous update of the articles has been a great incentive for us and for the authors of the chapters. This new edition of the Encyclopedia consists of six major parts: (i) mechanisms of cancer, (ii) hallmarks of cancer, (iii) causes of cancer, (iv) cancer prevention and control, (v) diagnosis and treatment of specific cancers, and (vi) pathology and genetics of specific cancers. This repartition is necessarily artificial and is complemented by the extensive cross-references between articles. The repartition, however, reflects our effort to identify discrete topics that would best address the needs of a wide community of readers. The primary target readership of the Encyclopedia comprises medical and other health science students, as well as non-specialized physicians and other health practitioners. Cancer researchers, oncologists, and other cancer professionals may find the articles pertaining to their specific field to be too short, over-simplistic, and perhaps obsolete; they too, however, may benefit from articles on topics other than their own. The Encyclopedia also offers an easy way to navigate across concepts and topics that should be appealing to readers from other communities, including social sciences or stakeholders in public decision-making.

xxxi

xxxii

Preface

We were fortunate to work with a formidable team of section editors, including Fred Bosman, Graham Colditz, Carlo La Vecchia, Gerd Pfeifer, and Marco Pierotti. An additional section editor was Professor Thomas Tursz, who passed away prematurely during the preparation of the Encyclopedia. Thomas was a great colleague and mentor, and a major figure in oncology in France and internationally. We had the privilege to involve him in the last project of his long career, and we want to dedicate this work to him. We wish to thank the many article authors, who agreed to contribute to the success of this international endeavor, and in particular Dr. Katarzyna Szyma nska, who drafted several cancer-specific articles. Finally, we want to thank the staff at Elsevier, whose patience and perseverance helped us bringing the project to the final stage. All these individuals are responsible for the many strengths of the new edition of the Encyclopedia of Cancer, while weaknesses are mainly ours. Paolo Boffetta Pierre Hainaut

CONTENTS OF ALL VOLUMES

VOLUME 1 Acute Lymphocytic Leukemia: Diagnosis and Treatment Katarzyna Szyma nska and Sophie Park

1

Acute Myelogeneous Leukemia: Diagnosis and Treatment Katarzyna Szyma nska and Sophie Park

9

Adrenal Glands Tumors: Pathology and Genetics Thomas G Papathomas, Thomas J Giordano, Eamonn R Maher, and Arthur S Tischler

18

Aflatoxins Joshua W Smith and John D Groopman

30

Aging and Cancer Mingyang Song

44

Anal Cancer: Pathology and Genetics Fred T Bosman

53

Animal Models of Cancer: What We Can Learn From Mice Giustina Ferone, Paul Krimpenfort, and Anton Berns

60

Anoikis Sharan Malagobadan and Noor Hasima Nagoor

75

Aspirin and Cancer Mangesh A Thorat

85

Ataxia Telangiectasia Syndrome Palak R Parekh and France Carrier

101

Autophagy and Cancer Christina G Towers and Andrew Thorburn

112

Bladder Cancer: Pathology, Genetics, Diagnosis, and Treatment Katarzyna Szyma nska, Fred T Bosman, and Pierre Hainaut

122

Bone and Soft Tissue Sarcoma: From Molecular Features to Clinical Applications Sarah Watson and Olivier Delattre

134

xxxiii

xxxiv

Contents of All Volumes

Bones and Joints Cancer: Pathology and Genetics Pancras C W Hogendoorn and Carlos E de Andrea

153

Breast Cancer: Pathology and Genetics Amy E McCart Reed, Emarene Kalaw, Andrew Reid, and Sunil R Lakhani

170

Cancer Disparities Ivana Kulhánová and Salvatore Vaccarella

184

Cancer in Populations in Transition Aleyamma Mathew and Preethi S George

191

Cancer in Sub-Saharan Africa Mariana Brandão, Ivo Julião, Carla Carrilho, Filipa Fontes, and Nuno Lunet

212

Cancer in the Middle East Iman El Sayed

225

Cancer Risk Reduction Through Lifestyle Changes Christina Bamia

243

Cancer Survival and Survivorship Dana Hashim and Elisabete Weiderpass

250

Cancer Vaccines: Dendritic Cell-Based Vaccines and Related Approaches Laurence Chaperot, Olivier Manches, and Caroline Aspord

260

Cancer-Related Inflammation in Tumor Progression Alberto Mantovani

275

Cancers as Ecosystems: From Cells to Population Graham A Colditz

278

CarcinogendDNA Adducts Annette M Krais, Rajinder Singh, and Volker M Arlt

282

Carcinogenesis: Role of Reactive Oxygen and Nitrogen Species Jean Cadet and J Richard Wagner

296

Cell Adhesion During Tumorigenesis and Metastasis Lubor Borsig and Heinz Läubli

307

Cell Responses to DNA Damage Jean Y J Wang

315

Cervical Cancer: Screening, Vaccination, and Preventive Strategies Paolo Giorgi Rossi and Francesca Carozzi

326

Chemoprevention of Cancer: An Overview of Promising Agents and Current Research Elizabeth G Ratcliffe, Andrew D Higham, and Janusz A Jankowski

345

Chemoprevention Trials Kunal C Kadakia, Ashley G Matusz-Fisher, and Edward S Kim

352

Cholangiocarcinoma: Diagnosis and Treatment Katarzyna Szyma nska

367

Chromatin Dynamics in Cancer: Epigenetic Parameters and Cellular Fate Daniel Jeffery, Katrina Podsypanina, Tejas Yadav, and Geneviève Almouzni

372

Contents of All Volumes

xxxv

Chromosome Rearrangements and Translocations Stefan K Bohlander, Purvi M Kakadiya, and Alix Coysh

389

Chronic Lymphocytic Leukemia: Pathology and Genetics Julio Delgado, Neus Villamor, Ferran Nadeu, and Elías Campo

405

Chronic Myelogenous Leukemia: Pathology, Genetics, Diagnosis, and Treatment Katarzyna Szyma nska and Sophie Park

418

Colorectal Cancer: Diagnosis and Treatment Katarzyna Szyma nska

425

Colorectal Cancer: Pathology and Genetics Fred T Bosman

428

Copy Number Variations in Tumors Tao Huang

444

Defective 5-Methylcytosine Oxidation in Tumorigenesis Gerd P Pfeifer and Seung-Gi Jin

452

Diabetes and Cancer Rachel Dankner

459

Diet and Cancer Livia S A Augustin, Concetta Montagnese, Ilaria Calabrese, Giuseppe Porciello, Elvira Palumbo, Sara Vitale, and Stephanie Nishi

471

Dietary Factors and Cancer Isabelle Romieu

501

DNA Methylation Changes in Cancer: Cataloguing Tibor A Rauch and Gerd P Pfeifer

512

DNA Methylation Changes in Cancer: Mechanisms Hideyuki Takeshima and Toshikazu Ushijima

520

DNA Mismatch Repair: Mechanisms and Cancer Genetics William J Graham V, Christopher D Putnam, and Richard D Kolodner

530

End of Life Support Kavitha J Ramchandran and Jamie Von Roenn

539

Endometrial Cancer: Pathology and Genetics Ben Davidson

549

Enhancers in Cancer: Genetic and Epigenetic Deregulation Kaixiang Cao and Ali Shilatifard

559

Environmental and Occupational Exposures Harvey Checkoway

569

Environmental Exposures and Epigenetic Perturbations Igor P Pogribny

574

VOLUME 2 Epigenetic Therapy Richard L Bennett, Aditya Bele, Sayantan Maji, and Jonathan D Licht

1

xxxvi

Contents of All Volumes

Epithelium to Mesenchyme Transition Laia Caja and E-Jean Tan

14

Esophageal Cancer: Diagnosis and Treatment Katarzyna Szyma nska

24

Esophageal Cancer: Pathology and Genetics Jean-François Fléjou

32

Eye and Orbit Cancer: Pathology and Genetics Sarah E Coupland, Santosh U Kafle, and Jakub Khzouz

47

Financial Burden of Cancer Care Bernardo H L Goulart and Veena Shankaran

61

Gastric Cancer: Pathology and Genetics Irene Gullo and Fátima Carneiro

77

Genetic Instability Meiyun Guo, Lin Mei, and Christopher A Maxwell

99

Genome Wide Association Studies (GWAS) Giuliana Restante, Elena Arrigoni, Marzia Del Re, Antonello Di Paolo, and Romano Danesi

110

Germ Cell Tumors: Pathology and Genetics J W Oosterhuis and Leendert H J Looijenga

121

Glioblastoma: Biology, Diagnosis, and Treatment Akanksha Sharma, Maciej M Mrugala, and Jan Buckner

154

Glioblastoma: Pathology and Genetics Christian Hartmann and Pieter Wesseling

165

Glutamine Metabolism and Cancer Nianli Sang, Dan He, and Shuo Qie

179

Helicobacter Pylori-Mediated Carcinogenesis Matthew G Varga and Meira Epplein

187

Hepatocellular Carcinoma: Pathology and Genetics Luigi M Terracciano, Salvatore Piscuoglio, and Charlotte K Y Ng

198

Hereditary Cancer Syndromes: Identification and Management Roberta Pastorino, Alessia Tognetto, and Stefania Boccia

211

Hereditary Risk of Breast and Ovarian Cancer: BRCA1 and BRCA2 Claus Storgaard Sørensen

214

HIV (Human Immunodeficiency Virus) Jay A Levy and JoAnn C Castelli

218

Hodgkin Lymphoma: Pathology and Genetics Stefano A Pileri, Stefano Fiori, Vincenzo De Iuliis, and Valentina Tabanelli

227

Hormones and Cancer Kristen D Brantley, Susan E Hankinson, and A Heather Eliassen

237

Induced Pluripotent Stem Cells and Yamanaka factors Stina Simonsson, Rocío C Viñuelas, Leonie Hartl, Itedale N Redwan, and Joydeep Bhadury

254

Contents of All Volumes

xxxvii

Inhibitors of Lactate Transport: A Promising Approach in Cancer Drug Discovery Thomas D Bannister

266

Integrative Molecular Tumor Classification: A Pathologist’s View Fred T Bosman

279

Interferons: Cellular and Molecular Biology of Their Actions Dhananjaya V Kalvakolanu, Shreeram C Nallar, and Sudhakar Kalakonda

286

Jaws Cancer: Pathology and Genetics Pieter J Slootweg and Daniel Baumhoer

313

Kidney Cancer: Diagnosis and Treatment Katarzyna Szyma nska

325

Laryngeal Cancer: Diagnosis and Treatment Fernando López Álvarez and Juan Pablo Rodrigo

332

Larynx Cancer: Pathology and Genetics Nina Zidar and Nina Gale

346

Li–Fraumeni Syndrome Orli Michaeli and David Malkin

356

Lipid Metabolism Fatima Ameer, Rimsha Munir, and Nousheen Zaidi

369

Lynch Syndrome Päivi Peltomäki

374

Malignant Skin Adnexal Tumors: Pathology and Genetics Katharina Wiedemeyer and Thomas Brenn

389

Malignant Tumors of the Eye, Conjunctiva, and Orbit: Diagnosis and Therapy Laurence Desjardins, Nathalie Cassoux, and Alexandre Matet

402

Melanoma: Pathology and Genetics Lorenzo Salvati, Giuseppe Palmieri, Vincenzo De Giorgi, Antonio Cossu, Mario Mandalà, and Daniela Massi

414

Metastatic SignaturesdThe Tell-Tale Signs of Metastasis Christina Ross and Kent Hunter

426

Metformin Matteo Lazzeroni and Sara Gandini

435

Mevalonate Pathway Heidrun Karlic and Franz Varga

445

Microbiota and Colon Cancer: Orchestrating Neoplasia Through DNA Damage and Immune Dysregulation Alberto Martin and Thergiory Irrazábal

458

Mitogen-Activated Protein Kinases (MAPK) in Cancer Ana Cuenda

472

Mod Squad: Altered Histone Modifications in Cancer Matthew Murtha and Manel Esteller

481

Molecular Epidemiology and Cancer Risk Paulina Gomez-Rubio and Evangelina López de Maturana

487

xxxviii

Contents of All Volumes

Multiple Myeloma: Pathology and Genetics Marie Yammine and Salomon Manier

494

Mutational Signatures and the Etiology of Human Cancers Ludmil B Alexandrov and Maria Zhivagui

499

Mutations in Chromatin Remodeling Factors Katherine A Giles and Phillippa C Taberlay

511

Mutations in DNA Methyltransferases and Demethylases Xiao-Jian Sun, Zhu Chen, and Sai-Juan Chen

528

Mutations in Histone Lysine Methyltransferases and Demethylases Sara Weirich and Albert Jeltsch

538

Mutations: Driver Versus Passenger Scott W Piraino, Valentina Thomas, Peter O’Donovan, and Simon J Furney

551

Myelodysplastic Syndromes: Mechanisms, Diagnosis, and Treatment Eric Solary and William Vainchenker

563

Nasopharyngeal Carcinoma: Diagnosis and Treatment Katarzyna Szyma nska

571

VOLUME 3 Neuroblastoma: Diagnosis and Treatment Gudrun Schleiermacher and Thierry Philip

1

Neuroblastoma: Pathology and Genetics Nai-Kong V Cheung and Estibaliz Lopez-Rodrigo

15

New Rationales and Designs for Clinical Trials in the Era of Precision Medicine Mina Nikanjam and Razelle Kurzrock

30

Non-Hodgkin Lymphoma: Diagnosis and Treatment Katarzyna Szyma nska and Sophie Park

44

Non-Hodgkin Lymphomas: Pathology and Genetics Laurence de Leval

48

Nonsmall-Cell Cancers of the Lung: Pathology and Genetics Erik Thunnissen and Egbert F Smit

76

Obesity and Cancer: Epidemiological Evidence Yikyung Park

88

Oncology Imaging Stefan Delorme and Michael Baumann

98

Opisthorchis viverrini, Clonorchis sinensis, and Cholangiocarcinoma Catherine de Martel and Martyn Plummer

119

Oral and Oropharyngeal Cancer: Pathology and Genetics Pieter J Slootweg and Justin A Bishop

124

Oral Cavity Cancer: Diagnosis and Treatment Corbin D Jacobs, Michael J Moravan, Jennifer Choe, Russel Kahmke, Yvonne Mowery, and Joseph K Salama

131

Contents of All Volumes

xxxix

Ovarian Cancer: Diagnosis and Treatment Katarzyna Szyma nska

169

Ovarian Cancer: Pathology and Genetics Jaime Prat, Iñigo Espinosa, and Emanuela D’Angelo

173

Pancreatic Cancer: Diagnosis and Treatment Katarzyna Szyma nska

193

Pancreatic Cancer: Pathology and Genetics Irene Esposito, Lena Haeberle, and Matthias Wirth

199

Papillomaviruses Karl Munger

212

Parathyroid Cancer: Pathology and Genetics Ying-Hsia Chu and Ricardo V Lloyd

219

P-Glycoprotein-Mediated Multidrug Resistance M M Gottesman

228

Physical Inactivity and Cancer Alpa V Patel and Erika Rees-Punia

235

Pituitary Tumors: Pathology and Genetics Stefano La Rosa and Silvia Uccella

241

Pituitary Tumors: Diagnosis and Treatment Sylvia L Asa and Shereen Ezzat

257

Polyomaviruses in Human Cancer Masahiro Shuda

266

Prevention and Control: Nutrition, Obesity, and Metabolism Niva Shapira and Ossie Sharon

278

Prostate Cancer: Diagnosis and Treatment Katarzyna Szyma nska and Pierre Hainaut

292

Prostate Cancer: Pathology and Genetics Funda Vakar-Lopez and Lawrence D True

299

Pyruvate Kinase Gopinath Prakasam and Rameshwar N K Bamezai

311

Radiation Oncology Nadja Ebert, Falk Tillner, and Michael Baumann

321

Radiation Therapy-Induced Metastasis and Secondary Malignancy Jordan O’Malley, Joseph Inigo, Abhiram Gokhale, Venkaiah Betapudi, Rahul Kumar, and Dhyan Chandra

337

Renal Cell Cancer: Pathology and Genetics Sean R Williamson

347

Role of DNA Repair in Carcinogenesis and Cancer Therapeutics Rachel Abbotts, Tyler Golato, and David M Wilson, III

363

Senescence and Cellular Immortality Marco Demaria and Michael C Velarde

386

xl

Contents of All Volumes

Sleep Disturbances and Misalignment in Cancer Eva S Schernhammer, Yin Zhang, and Elizabeth E Devore

395

Small-Cell Cancer of the Lung: Pathology and Genetics Lynnette Fernandez-Cuesta, Sylvie Lantuejoul, and Nicolas Girard

403

Squamous Cell and Basal Cell Carcinoma of the Skin: Diagnosis and Treatment Katarzyna Szyma nska

412

Symptom Control Vithusha Ganesh, Leila Malek, Carlo DeAngelis, and Edward Chow

417

Systems Biology Approach to Study Cancer Metabolism Stefan Kempa and Nadine Royla

426

TCA Cycle Aberrations and Cancer Fabio Ciccarone, Luca Di Leo, and Maria Rosa Ciriolo

429

Telomeres, Telomerase, and Cancer Kazunori Tomita and Laura C Collopy

437

TGF-b in Cancer Progression: From Tumor Suppressor to Tumor Promotor Chuannan Fan, Jing Zhang, Wan Hua, and Peter ten Dijke

455

Thyroid Cancer: Pathology, Genetics, Diagnosis, and Treatment Katarzyna Szyma nska and Fred T Bosman

471

TP53 Simon S McDade and Martin Fischer

483

Tumor-Associated Macrophages Fiona J Pixley

496

Tumors and Blood Vessel Interactions: A Changing Hallmark of Cancer Francesco Pezzella and Frank Winkler

504

Tunneling Nanotubes (TNTs): Intratumoral Cell-to-Cell Communication Marie-Luce Vignais, Jean Nakhle, and Emmanuel Griessinger

513

Unprogrammed Gene Activation: A Critical Evaluation of Cancer Testis Genes Sophie Rousseaux, Ekaterina Bourova-Flin, Mengqing Gao, Jin Wang, Jian-Qing Mi, and Saadi Khochbin

523

Uterine Cervix Cancer: Diagnosis and Treatment Katarzyna Szyma nska

531

Uterine Cervix Cancer: Pathology and Genetics C Simon Herrington

535

Wilms Tumor: Pathology and Genetics Gordan M Vujanic, Adrian K Charles, Sergey D Popov, and Neil J Sebire

542

Wnt Signaling in Intestinal Stem Cells and Cancer Rebecca L Myers, Maryam Yousefi, Christopher J Lengner, and Peter S Klein

553

Xeroderma Pigmentosum: When the Sun Is the Enemy Alain Sarasin, Carlos F M Menck, and Januario B Cabral-Neto

562

Index

573

PERMISSION ACKNOWLEDGMENTS The following material is reproduced with kind permission of American Association for the Advancement of Science Figure 3 Neuroblastoma; Pathology and Genetics www.aaas.org The following material is reproduced with kind permission of Oxford University Press Table 2 Environmental and Occupational Exposures Figure 1 Driver Versus Passenger Mutations in Tumors Figure 3 Driver Versus Passenger Mutations in Tumors Figure 4 Driver Versus Passenger Mutations in Tumors Figure 5 Driver Versus Passenger Mutations in Tumors Figure 1 Financial Burden of Cancer – Therapies www.oup.com The following material is reproduced with kind permission of Nature Publishing Group Figure 5 Neuroblastoma; Pathology and Genetics Table 2 Neuroblastoma; Pathology and Genetics Figure 13 Neuroblastoma; Pathology and Genetics Table 1 Neuroblastoma; Pathology and Genetics Table 2 Neuroblastoma; Pathology and Genetics Table 3 Neuroblastoma; Pathology and Genetics Table 4 Neuroblastoma; Pathology and Genetics Figure 2 Oesophageal Cancer; Diagnosis and Treatment Figure 15 Germ Cell Tumors: Pathology and Genetics Figure 2 Chemoprevention of Cancer: an Overview of Promising Agents and Current Research http://www.nature.com

i

Neuroblastoma: Diagnosis and Treatment Gudrun Schleiermacher and Thierry Philip, Curie Institute, Paris, France © 2019 Elsevier Inc. All rights reserved.

Glossary Image-defined risk factors in neuroblastoma (IDRF) Guidelines for imaging and staging of neuroblastic tumors as defined per consensus in the International Neuroblastoma Risk Group Project. International neuroblastoma pathology classification (INPC) Guidelines for the pathological characterization and staging of peripheral neuroblastic tumors. International neuroblastoma risk groupdstaging system (INRG) The International Neuroblastoma Risk Group (INRG) classification system was developed to establish a consensus approach for pretreatment risk stratification for patients with peripheral neuroblastic tumors, taking into account clinical, radiological, pathological, and biological features. By defining homogenous pretreatment patient cohorts, the INRG classification system will greatly facilitate the comparison of risk-based clinical trials conducted in different regions of the world and the development of international collaborative studies. International neuroblastoma staging system (INSS) International consensus for the diagnosis and staging of neuroblastoma. Iodine-123-metaiodobenzylguanidine scintigraphy (MIBG scintigraphy) A scintigraphy method using a radiolabeled molecule, Iodine-123-metaiodobenzylguanidine, which is similar to norepinephrine; based on its uptake by neuroblastic cells through the norepinephrine transporter this scintigraphy method is useful for the diagnosis and metastatic work up of peripheral neuroblastic tumors.

Introduction Neuroblastoma is the most common extracranial solid tumor of early childhood. It is an embryonic tumor derived from the developing sympathetic nervous system corresponding to the cells of the neural crest that form the adrenal medulla and the sympathetic ganglia. The large clinical heterogeneity, with a spectrum ranging from the possibility of spontaneous regression to threatening progression despite all treatment, as well as the important biological heterogeneity, has led to intense research on clinical and biological prognostic factors with an aim of improving the definition of risk defined patient subgroups and of developing new therapeutic approaches.

Epidemiology of Neuroblastoma Neuroblastoma represents 8%–10% of all pediatric cancers, with a mean annual incidence of 7–12 cases per million children in Western countries, and a prevalence of 1 case per 8000–10,000 births. Neuroblastoma is a disease of early childhood, with a median age at diagnosis of 18 months, and 90% of patients diagnosed before the age of 5 years. Age at diagnosis is an important prognostic factor, as patients under 18 months usually have a better prognosis, whereas in adolescents and adults is very rare it generally shows a more indolent clinical course with de novo resistance to chemotherapy treatments. Neuroblastoma is slightly more frequent in males than in females (M/F ratio 1.1–1.2). As neuroblastoma can be detected following an increase of urinary catecholamine metabolites (see below), screening programs were put in place in several countries including Japan, Germany, France, and Canada. An increase in the prevalence of neuroblastoma in infants was observed in these countries; however, there was no decrease neither in the prevalence or nor in the mortality of neuroblastoma in patients older than 1 year. This suggests that spontaneous regression of symptomless neuroblastoma is at least as frequent as neuroblastoma leading to clinical symptoms.

Etiology of Neuroblastoma Developmental Origin During embryonal development, the neural crest arises from the neural tube after its closure. Neural crest development involves epithelial to mesenchymal transition, and cells will migrate from the neural tube for the further development of numerous anatomic structures, giving rise to various cell types, including neural, pigmented, craniofacial, and conotruncal cardiac cells, during embryogenesis. This complex process is regulated by various epigenetic programs based on histone modification and DNA methylation, and transcriptional programs involving multiple transcription factors which further activate downstream transcriptional regulators of proliferation and differentiation. The neural crest of the trunk contributes to the development of the peripheral nervous system, including sympathetic ganglia and the adrenal gland. It is thought that neuroblastoma arises from these embryonal sympathoadrenal progenitors.

Encyclopedia of Cancer, 3rd edition, Volume 3

https://doi.org/10.1016/B978-0-12-801238-3.65325-4

1

2

Neuroblastoma: Diagnosis and Treatment

To date, the precise etiology of neuroblastoma is unknown, and unlike many adult malignancies, environmental factors are not thought to play a major role, although predisposing effects of prenatal exposures to potentially toxic substances warrant further investigation. However, genetic factors, at both a constitutional and a somatic level, are thought to play a major role in neuroblastoma development.

Hereditary Genetic Factors Several observations support the hypothesis of a major role of underlying hereditary genetic factors in the etiology of neuroblastoma. First, rare familial cases have been described, accounting for < 1% of all neuroblastoma cases. Gain-of-function mutations in the tyrosine kinase domain of the anaplastic lymphoma kinase gene ALK have been detected in the majority of familial cases associated with an autosomal dominant pattern of inheritance with incomplete penetrance. Second, neuroblastoma can arise in a context of syndromic associations. Neural crest-derived developmental disorders associated with an increased risk to develop neuroblastoma have been linked to inactivating mutations of the PHOX2B gene, a master regulator of neural crest development, which was identified as the first predisposition mutation in neuroblastoma. Whereas expansions of the second polyalanine stretch of PHOX2B are mainly observed in patients presenting with Ondine’s curse (also called CCHS, for Congenital Central Hypoventilation Syndrome) associated with a low risk of peripheral neuroblastic tumors, nonpolyalanine repeat expansion mutations occur preferentially in patients with Hirschsprung’s disease which is associated with a higher risk to develop neuroblastoma. Further associations between neuroblastic tumors and cancer predisposition syndromes include neurofibromatosis type 1 (NF1), characterized by constitutive activation of the RAS–MAPK pathway, as well as Noonan syndrome with involvement of the PTPN11 gene. Third, recent powerful genome-wide association studies have demonstrated that neuroblastoma is a disease of complex underlying genetic factors as at least a dozen common polymorphic alleles, occurring at various loci throughout the genome, have been shown to influence neuroblastoma oncogenesis. Of low individual impact on disease initiation (relative risks of 1.5–2.0), these polymorphic alleles can cooperate in an individual patient to promote malignant transformation during neurodevelopment, and some genes targeted by these polymorphic alleles play potential roles in the pathogenesis of neuroblastoma, including BARD1, LMO1, DUSP12, DDX4, HACE1, and LIN28B. The described polymorphic alleles show a correlation with either highrisk or low-risk disease, indicating that favorable and unfavorable forms of neuroblastoma may represent distinct entities in terms of the genetic events that initiate tumorigenesis. Furthermore, rare cases with various abnormal constitutional karyotypes have been described in neuroblastoma patients, including constitutional copy number anomalies, balanced and unbalanced translocations, as well as specific chromosome deletions. Altogether, there are probably as yet undiscovered additional genes that predispose to neuroblastoma when altered in the germline. However, to date there are no clinically validated guidelines for who should be screened for germline mutations, nor how to perform surveillance of patients or families with known predisposition alleles.

Somatic Genetic Alterations With only 1%–2% of neuroblastoma occurring in a familial or predisposition context, over 98% of all cases occur sporadically. A large number of recurrent somatic genetic alterations have been described in neuroblastoma, the most frequent of which concern quantitative genomic alterations with gains or losses of genetic material. These genetic abnormalities are closely linked to the distinct biological and clinical subgroups of the disease. Amplification of the MYCN oncogene, mapping to 2p24.1, occurs in approximately 25% of all neuroblastomas and 40% of high-risk tumors. It remains one of the most important genetic alterations associated with advanced stages of disease, an aggressive phenotype, and poor outcome, and is the first genetic marker to be used in clinical practice for risk stratification and adaptation of treatment intensity. Closely associated with poor survival in localized disease and in infants, its prognostic impact in metastatic disease of older children with an overall poor outcome is less pronounced. At a cytogenetic level, amplification of the MYCN oncogene occurs either as double-minute chromosomes or homogeneously staining regions, which contain between 10 to over 100 ectopic copies of the MYCN oncogene. The oncogenic role of MYCN has been clearly demonstrated as its ectopic expression in the neural crest is sufficient to drive neuroblastoma tumorigenesis in zebrafish and mice models. Its oncogenic role is based on an enhancement of the expression of genes involved in cell proliferation, and on the repression of genes involved in differentiation and apoptosis. Other recurrent amplifications concern the ALK gene on chromosome 2p23, as well as amplicons of chromosome 12q13–14 encompassing, among others, the MDM2 and CDK4 genes. Preliminary data indicate that NB with focal amplifications other than MYCN might present with atypical clinical features and a poorer outcome. Other recurrent structural alterations recurrently observed in neuroblastoma concern segmental chromosome alterations (SCA), including deletions of chromosome arms 1p, 3p, 4p, and 11q, and gains of 1q, 2p, or 17q. Deletion of 1p36 is observed in 20%– 35% of cases, predicts survival in multivariate analyses, and is significantly associated with aggressive disease markers. Deletions of 11q in a consensus region at 11q23 occur in approximately 40% of cases and are inversely correlated with MYCN amplification, identifying a molecularly distinct high-risk patient subset, characterized by advanced stage, older age, and a higher genomic instability with a higher number of chromosome breakpoints. Gains of chromosome 17q21-qter represent the most frequent genetic

Neuroblastoma: Diagnosis and Treatment

3

alteration in neuroblastoma, occurring in 70% of tumors. Numerous studies have reported that 17q gain is significantly associated with advanced stage of disease, increased patient age, MYCN amplification, and other unfavorable genetic parameters. The overall genomic profile has been shown to be of prognostic impact in neuroblastoma. Whereas numerical chromosome alterations, consisting of gains or losses of whole chromosomes, are associated with a favorable outcome, segmental chromosome alterations of any chromosome region, without or with numerical chromosome alterations, are associated with a higher risk of relapse. Although intense decade-long research has focused on the identification of hypothetical tumor suppressor genes or oncogenes in recurrently altered regions of chromosome loss or gain, the smallest regions of overlap remain quite extensive, and to date do not point to single gene candidates as tumor suppressors or oncogenes. This suggests that an overall imbalance of copy number regions is of importance in neuroblastoma oncogenesis. Recent next generation sequencing approaches have indicated that most neuroblastoma harbor only few mutations, with an average of 10–20 predicted nonsynonymous variations in coding regions per genome, indicating an exonic mutation frequency of 0.2–0.4 per Mb. The frequency of somatic events strongly correlates with tumor stage, lower-stage tumors harboring a lower number of mutations. The most frequent recurrent somatic mutation in neuroblastoma concerns the gene ALK (anaplastic lymphoma kinase), with mutations activating the tyrosine kinase domain in approximately 10% of all cases at diagnosis. The somatic ALK-1174 mutation appears to contribute to a more aggressive phenotype, but unlike ALK-1275 mutations, these specific mutations are not found in familial neuroblastoma and are not tolerated in the germline. The oncogenic role of activating ALk mutations in neuroblastoma has been demonstrated in vitro and in vivo in both zebrafish and mouse models, with co-expression of ALK-F1174L and MYCN producing a synergistic effect for neuroblastoma tumorigenesis in mice. Small molecule inhibitors targeting the activated kinase domain of ALK are now available, making this a promising target for molecular therapy but still requiring more specific development. Other recurrent mutations in neuroblastoma target distinct cellular pathways and include PTPN11 mutations (in 3% of cases), as well as genes involved in cytoskeleton maintenance, neuritogenesis, and other regulators of the RAC/Rho pathway. Interestingly, genes involved in chromatin remodeling have been found to be targeted in a significant number of cases, either by mutations or by structural variations, including mutations in the ARID1A/ARID1B genes. Somatic alterations of ATRX are associated with an increase in telomere length, and with an absence of the ATRX protein in the nucleus. ATRX alterations appear to be more frequent in older children and occur in mutually exclusive fashion with MYCN amplifications. ATRX mutations are associated with activation of a telomere maintenance mechanism termed alternate lengthening of telomeres (ALT), which may be associated with primary chemotherapy resistance. Recurrent genomic rearrangements of the promotor region of the telomerase reverse transcriptase (TERT) gene on chromosome 5p15.33 have been described in > 10% of neuroblastoma cases, defining a further subgroup of high-risk disease and occurring in mutually exclusive fashion with MYCN amplification and ATRX mutations. Thus, a large number of high-risk neuroblastomas are affected by genetic alterations of either MYCN, TERT, or ATRX, all of which converge to an activation of telomere lengthening mechanisms either by direct activation or by ALT, leading to a capacity of nearinfinite cell proliferation. Advances in the development of inhibitors of these pathways and their evaluation in clinical trials will lead to new treatment opportunities. Studies in relapsed neuroblastoma have suggested that genetic alterations can evolve over time and that clonal evolution is common, resulting in the acquisition of somatic alterations in known oncogenic pathways, some of which are targetable. Early evidence suggests that activation of the MAPK pathway and other signaling pathways of epithelial-mesenchymal transition processes might emerge at relapse and might represent promising targets for molecular targeted treatment approaches. Altogether, both spatial and temporal genetic heterogeneity plays an important role in neuroblastoma (Fig. 1). In addition to genetic changes, neuroblastoma can also be characterized by specific expression profiles. A large number of studies have focused on the analysis of differential expression patterns, seeking to identify expression patterns that might enable to distinguish patients with different clinical courses and thus define different prognostic subgroups in high-risk disease, and potentially to identify new therapeutic targets. However, these approaches have not yet found a routine application in a clinical setting. The paucity of recurrent genetic mutations as compared to adult tumors in adults indicates that additional mechanisms such as epigenetic alterations may play an important role in the molecular pathogenesis of these developmental tumors. Alterations in DNA methylation represent one of the most common molecular events in neoplasia, and CpG-island hypermethylation of gene promoters is a frequent mechanism for functional inactivation of relevant tumor-associated genes in neuroblastoma.

Cellular Identity In neuroblastoma, two distinct cellular identities corresponding to either a sympathetic noradrenergic identity, or to a neural crest cell-like, mesenchymal identity can be evidenced. These cellular identities are governed by distinct core regulatory circuits corresponding to different super-enhancer profiles, including the transcription factors PHOX2B, HAND2, and GATA3 for the sympathetic noradrenergic and the AP-1 transcription factors for the neural crest cell-like identity. The committed adrenergic cells and undifferentiated mesenchymal cells can coexist in a tumor and, importantly, can interconvert. This represents an additional important aspect of tumor heterogeneity with an important role with regards to differential sensitivity to chemotherapies.

4

Neuroblastoma: Diagnosis and Treatment

Fig. 1 Macroscopical and microscopical heterogeneity of neuroblastoma. This tumor was histologically composed of a stroma-rich component, and of several stroma-poor nodules corresponding to neuroblastoma, stroma-poor, poorly differentiated. Different copy number profiles can be observed: numerical chromosome alterations (NCA) only, versus segmental chromosome alterations (SCA) with (in this case) chromosome 1p and 11q deletion, and chromosome 2p and 17q gain.

Biological Pathways A number of biological pathways corresponding to important hallmarks of cancer appear to be affected in neuroblastoma. These include abnormal patterns of gene or protein expression involved in tumor differentiation, apoptosis, drug resistance, angiogenesis, invasion, and metastasis. Neurotrophin signaling plays a central role in normal neuronal development and may be involved in both differentiation and regression of neuroblastoma. Differential expression of tropomyosin receptor kinase (Trk) receptors has been shown to influence both the biological characteristics and clinical behavior of neuroblastoma, with TrkA expression (with the ligand NGF) observed in favorable neuroblastoma, and TrkB expression (with the ligand BDNF) observed in unfavorable neuroblastoma. Delayed activation or disruption of normal apoptotic pathways may also be an important phenomenon involved in spontaneous regression, as well as therapy resistance of neuroblastoma. Major elements of the apoptotic signaling cascade with abnormal expression or activation patterns include the BCL2 family, survivin, and caspase-8. Acquired resistance to chemotherapeutic agents may also be conferred by enhanced drug efflux via overexpression of classical multidrug resistance proteins, including the multidrug resistance gene 1 (MDR1) and the gene for the multidrug resistance-related protein (MRP). Increased tumor angiogenesis and expression of pro-angiogenic factors, such as vascular endothelial growth factor, are both correlated with an aggressive phenotype in neuroblastoma. Angiogenesis inhibitors represent a potential treatment option that is currently being evaluated in clinical trials.

Immunological Aspects Immune factors and the tumor microenvironment contribute to the variable natural history of neuroblastoma. As age at diagnosis is of strong prognostic impact in patients with neuroblastoma, it is likely that age-related differences in the host immune reaction to a developing tumor might influence outcome. Further evidence of the possible role of disease control by the immune system derives from observations of the neuroblastomaassociated paraneoplastic syndrome OMS, in which lymphoid infiltrates are frequently observed in the tumor; these patients typically have an excellent oncological prognosis. In metastatic neuroblastoma, higher levels of CD163 þ M2 macrophages have been demonstrated, a finding associated with adverse outcomes in several other cancers. Age-dependent differential expression of inflammatory genes associated with tumor associated macrophages has also been observed in neuroblastoma. Cancer-associated fibroblasts in the neuroblastoma microenvironment have been shown to play a role in the immunosuppression which might favor tumor development in neuroblastoma.

Neuroblastoma: Diagnosis and Treatment

5

Altogether, in high-risk neuroblastoma infiltration with suppressive myeloid cells can decrease antitumor immune responses and favor tumor growth. Both active and passive immunotherapies, including antibody-based therapies, are now being explored in neuroblastoma. Further advances in the understanding of the biology of neuroblastoma are currently based on the use of deep-sequencing approaches for comprehensive molecular characterization of the tumor cells and the host, also at a single cell level to account for tumor heterogeneity, with the goal not only to define precise patient risk groups, but also to identify key therapeutic targets.

Diagnosis Clinical Presentation Clinical signs and symptoms of neuroblastoma depend on the localization of the primary tumor and its metastatic sites. Localized disease often presents as an incidental finding, whereas metastatic disease is linked to systemic symptoms. Primary tumors can arise anywhere along the sympathetic nervous system, but the distribution varies across the sites with age. Cervical neuroblastomas, observed in 4%–5% of cases, may involve the cervical ganglion (ganglion stellatum), causing Horner’s syndrome (ptosis, miosis, enophthalmus, and anhidrosis). Thoracic neuroblastomas, localized in the upper, middle, or lower mediastinum and representing 15% of all cases, may be asymptomatic or more rarely cause respiratory distress. Over half of all primary tumors have their origin in the medulla of the adrenal gland, a localization associated with poorer survival. Bilateral primary adrenal tumors occur in < 1% of all cases, possibly occurring in a context of genetic predisposition. Altogether abdominal localization is observed in 80% of cases, and large abdominal tumors can cause hypertension, abdominal distention, and pain. Pelvic tumors are uncommon and either present as an asymptomatic mass, or cause constipation or urinary symptoms. Tumors arising in any paraspinal sympathetic ganglia can grow along the nerve roots, penetrate into the intervertebral foramina, and lead to spinal cord compression, resulting in neurological symptoms. In very rare cases, no primary tumor can be identified. Metastatic disease is present in 50% of patients at diagnosis, with regional or distant lymph nodes, bone marrow, and bone involved most frequently, whereas liver and skin metastasis can be observed especially in young infants with favorable outcome. Metastasis to other sites such lung or CNS occurs more rarely. Metastatic involvement can cause systemic symptoms such as fever, weight loss, bone pain, or signs of bone marrow involvement such as anemia and thrombocytopenia. Specific metastatic sites can lead to localized bone pain and limping, while periorbital metastasis can cause periorbital ecchymoses (Hutchinson’s syndrome). Other clinical symptoms such as respiratory distress linked to compression by a large mass, coagulation disorders due to liver involvement, or lower body edema due to vena cava compression occur more rarely. Associated paraneoplastic syndromes can be observed in approximately 5% of all patients with neuroblastoma. A hypersecretion of the vaso-intestinal peptide (VIP) can lead to abundant intractable watery diarrhea, requiring treatment with somatostatine or octreotide in rare cases if the symptoms persist after resection of the tumor. Opsoclonus-myoclonus syndrome (OMS, also called Kinsbourne Syndrome or Dancing Eye Syndrome) is a rare neurological syndrome characterized by the association of opsoclonus (rapid, multidirectional eye movements), myoclonus, and cerebellar ataxia, most likely as a result of autoimmune process. This syndrome is observed in 2%–3% of children with neuroblastoma, while a neuroblastoma can be detected in 50%–80% of all children with OMS. These symptoms may precede the detection of a neuroblastoma, and may improve following surgical resection, but in many cases patients require specific immunosuppressive treatment for the neurological symptoms. Symptoms may reappear during infectious episodes, and although the oncological outcome of these patients is favorable, the majority of these children have long-term neurological sequelae.

Laboratory Parameters and Tumor Markers Elevation of serum markers such as LDH, NSE, or ferritine are associated with unfavorable prognosis, but these markers are not specific to neuroblastoma. An increase in urinary secretion of catecholamines or their metabolites, including dopamine, homovanillic acid (HVA), and/or vanillylmandelic acid (VMA), can be detected in 90% of patients. The profile of urinary catecholamines reflects the degree of cellular maturation with a high dopamine level or HVA/VMA ratio associated with an unfavorable prognosis. An elevation of norepinephrine or epinephrine occurs more rarely and can lead to arterial hypertension. With catecholamine metabolites routinely measured in urine samples at diagnosis and during treatment and follow-up, plasma levels of normetanephrine, metanephrine, and methoxytyramine might represent a convenient alternative to urine markers.

Radiological Local and Metastatic Assessment Disease evaluation at diagnosis requires radiologic assessment of the primary tumor as well as an assessment of possible osteomedullary, soft tissue, or other metastatic sites. Although ultrasound is often the first imaging exam because of its high availability and non-invasive nature, local assessment should include an exploration either by CT or MRI, with a high preference now given to MRI due to a higher contrast resolution and an absence of exposure to ionizing radiation, despite the longer acquisition times and sedation requirements in young children (Fig. 2). Neuroblastic tumors are often heterogeneous, and might present calcifications and involvement of regional lymph nodes.

6

Neuroblastoma: Diagnosis and Treatment

Fig. 2 Imaging of a neuroblastoma at diagnosis in a 2-year-old girl. An extensive retroperitoneal latero-aortic mass with microcalcifications shows intraforaminal invasion between D12 and L2 with spinal cord compression. The mass encases the aorta. MRI imaging reveals mediastinal lymph nodes confirmed my MYBG SPECT, which furthermore shows multiple osteomedullary lesions.

The precise analysis of local extension and involvement of adjacent organs includes the identification of image defined risk factors (IDRFs) in relation to possible surgical excision, taking into account perivascular disease, infiltration of tissues and adjacent organs, intraforaminal infiltration, and the epidural space of the spinal canal. The extent of local disease is classified according to the absence or presence of IDRFs. The classification currently proposed by the INRG staging system categorizes tumors as L1 (limited tumor to a body compartment and the absence of any IDRF) or L2 (presence of IDRFs). In addition to an evaluation of local disease extent, metastatic soft tissue, and bone marrow disease are evaluated by a metadibenzylguanidine scintigraphy (mIBG), preferably using iodine 123 (123I), rather than 131I due to lower exposure to ionizing radiation, improved quality of images, and lower thyroid toxicity. mIBG uptake depends on expression of the norepinephrine transporter, and about 90% of neuroblastomas express this transporter and thus are mIBG-avid. mIBG imaging has an estimated sensitivity and specificity of 90% and 99%, respectively. New imaging modalities using multimodal camera systems and integration of single photon emission computed tomography (SPECT) with CT combine the contrast provided by tumor-avid radioactive drugs with the anatomic precision of CT. Different scores (SIOPEN score, Curie score) have been developed to quantify metastatic involvement based on mIBG imaging, also enabling comparisons across different clinical trials in different cooperative groups. For mIBG non-avid neuroblastoma, disease extent should be evaluated by an additional technique, either with a technetium-99 bone scan, or 18 F-fluorodeoxyglucose positron emission tomography–computerized tomography (PET-CT). Recent PET techniques using 18 F-DOPA or 68Ga-DOTATATE are now also being evaluated. Comprehensive metastatic assessment also requires an assessment of bone marrow involvement. This investigation requires bilateral bone marrow aspirates and trephine bone marrow biopsies, obtained from bilateral iliac crest sites, with histological examination and immunohistochemistry for a quantitative approach to detect metastatic disease. Approaches measuring neuroblastoma

Neuroblastoma: Diagnosis and Treatment

7

specific gene transcripts such as PHOX2B or tyrosine hydroxylase (TH) with qRTPCR can provide additional prognostic information, but are not yet incorporated into routine clinical evaluation. New approaches for detection and monitoring of disease might also rely on study of liquid biopsies such as analysis of cell-free circulating tumor DNA (Table 1).

Pathology of Neuroblastic Tumors A pathological analysis of a tumor sample obtained either by surgical resection at diagnosis, or by surgical or percutaneous needle biopsy, allows confirmation of the histological diagnosis and a more precise classification according to the criteria of the International Neuroblastoma Pathology Committee (INPC). Peripheral neuroblastic tumors show different degrees of differentiation, enabling one to distinguish ganglioneuroma tumor cells show maturation with terminal differentiation, ganglioneuroblastoma (some tumor cells show neuronal differentiation, others correspond to ganglion cells), and neuroblastoma (mostly immature small round tumor cells) with differentiating, poorly differentiated or undifferentiated features. A new histopathological classification of neuroblastoma combined biological characteristics including patient’s age and a cellular index of mitosis and karyorrexis (MKI). The International Neuroblastoma Pathology Committee (INPC) defined a modification of the initial classification, taking into account the component of stromal Schwannian cells to classify neuroblastic tumors into four categories: neuroblastoma (Schwannian stroma-poor ¼ SP); ganglioneuroblastoma, intermixed (Schwannian stroma-rich ¼ SR); ganglioneuroblastoma, nodular (composite SR/SD and SP); and ganglioneuroma (Schwannian stroma-dominant ¼ SD). The INPC defines a histoprognostic classification taking into account these categories, the index of mitosis and karyorrexis (MKI), and the age at diagnosis.

Staging and Prognostic Classification In neuroblastoma, disease staging based on the extent of the disease is crucial in order to define prognosis and to inform treatment intensity. The previously used INSS definitions are based on the local and distant tumor extension and the resectability of the tumor (Table 2). However, this staging system is based on peroperative findings, and cannot easily be used in a context of unresectable disease. The International Neuroblastoma Risk Group (INRG) classification system was then designed to identify homogeneous risk groups prior to any treatment to permit uniform therapeutic stratification and a comparison of patients groups in clinical trials conducted by different cooperative groups. Table 1

Analyses for diagnostic confirmation and risk stratification of neuroblastoma

Tumor biopsy: pathology and biology

Bilateral bone marrow aspirate and biopsy Imaging Laboratory

Table 2

Pathological, immunohistochemical analysis and INPC classification FISH for MYCN, aCGH, SNPa or other techniques for genomic copy number profile DNA index (Ploidy) Optional: ALK mutations, NGS sequencing panel, expression profile Storage of tumor tissue for further translational studies recommended Cytology, pathology, IHC Ultrasound, CT and/or MRI of primary site (neck, chest, abdomen, pelvis) Cranial CT or MRI (if clinically involved) 123 I-mIBG scan; if tumor is not mIBG avid, 18FDG-PET scan Full blood count, blood coagulation studies Blood electrolytes, creatinine, uric acid, liver function tests Ferritin, LDH Urine VMA, HVA, dopamine

INSS staging system

Stage

Description

1

Localized tumor with complete gross excision, with or without microscopic residual disease. Representative ipsilateral lymph node microscopically negative for tumor Localized tumor with incomplete gross excision. Representative ipsilateral non-adherent lymph nodes microscopically negative for tumor Localized tumor with or without complete gross excision, with ipsilateral non-adherent lymph nodes positive for tumor. Enlarged contralateral lymph nodes must be microscopically negative for tumor Unresectable unilateral tumor, infiltrating across the midline, with or without regional lymph node involvement; or localized unilateral tumor with contralateral lymph node involvement; or midline tumor with bilateral extension by infiltration or lymph node involvement Any primary tumor with dissemination to distant lymph nodes, bone, bone marrow, liver and/or other organs (except as defined for Stage 4S) Localized primary tumor (as defined for Stage 1, 2A, or 2B), with dissemination limited to liver, skin, and/or bone marrow ( 50%) includes those that progress and metastasize to distant organs (Graeme Eisenhofer and Cheung, 2017). Recent discoveries have helped understand the genetic basis of NB and Omics data should one day be integrated into the diagnosis and prognostication for accurate risk stratification and therapy selection for these patients. It is now estimated that approximately 10% of NB patients carry germline mutations in genes that might confer increased susceptibility to NB development (Tolbert et al., 2017). Many of the more highly penetrant mutations in NB predisposition, inherited or de novo, may affect the risk for other malignancies in survivors (Matthay et al., 2016). Here we will summarize the more relevant findings, and how they may relate to the biology and therapy of NB.

POSTERIOR

ANTERIOR

RIGHT

LEFT

Fig. 3 I123MIBG SCAN 24h. Retroperitoneal mass and lymphadenopathies (red arrow), lymph node involvement (green arrow), multiple sites of bone involvement: skull, scapula, humerus, bilateral iliac and femurs (blue arrows).

20

Neuroblastoma: Pathology and Genetics

Familial NB Familial NB cases are rare, accounting for < 2% of all diagnosis (Tolbert et al., 2017; Cheung and Dyer, 2013). These patients do not seem to present at younger age, but may have multifocal primary disease (Ahmed et al., 2017; Tolbert et al., 2017; Matthay et al., 2016; Cheung and Dyer, 2013; Kamihara et al., 2017). Two genes have been identified within familial neuroblastoma cases: the paired-like homeobox 2b (PHOX2B) and the anaplastic lymphoma kinase (ALK) gene (Tolbert et al., 2017; Matthay et al., 2016; Cheung and Dyer, 2013; Cao et al., 2017). The disease is inherited in an autosomal dominant fashion, though pedigrees show considerable heterogeneity in the biology of tumors that arise in these families: benign ganglioneuromas and malignant NBs have been reported in the same family, and some pedigrees show incomplete penetrance (Tolbert et al., 2017; Kamihara et al., 2017). A small proportion of patients have a clinically recognizable genetic syndrome: (a) most cases are related to abnormalities in neural crest development, like in congenital central hypoventilation syndrome (CCHS), aganglionosis of the colon (Hirschsprung disease), ROHHAD syndrome (rapid-onset obesity, hypothalamic dysfunction, hypoventilation, and autonomic dysfunction); (b) in the context of RASopathies such as Costello syndrome, Noonan syndrome, neurofibromatosis type 1; (c) epigenetic syndromes such as Beckwith–Wiedemann syndrome (BWS, including hemihypertrophy, or 11p overgrowth) (Kamihara et al., 2017). Germline mutations in TP53 (in particular R337H), SDHB, PTPN11 and APC have been reported to occur rarely in NB patients (Ahmed et al., 2017; Kamihara et al., 2017; Varan et al., 2016). Inactivating mutations in PHOX2B were the first found in the study of syndromic NB cases, along with Hirschprung disease and/ or CCHS (Matthay et al., 2016; Cheung and Dyer, 2013). Mutations in this gene account for a minority of familial NB cases (Tolbert et al., 2017; Matthay et al., 2016; Cheung and Dyer, 2013; Cao et al., 2017). It is a transcription factor that promotes cell cycle exit and neuronal differentiation of neural crest derived autonomic precursors (Cheung and Dyer, 2013). PHOX2B has two polyalanine repeat sequences (PARMS), and mutations in these are associated with a 1%–2% risk of developing tumors (Cheung and Dyer, 2013; Kamihara et al., 2017). On the other hand, nonpolyalanine repeat expansion mutations relate to the more severe phenotype including the NB-Hirschprung disease-CCHS association (Cheung and Dyer, 2013), and an increased risk (45%) of developing a neural crest tumor. Activating mutations in ALK are accountable for up to 80% of familial NB cases (Tolbert et al., 2017; Matthay et al., 2016; Cheung and Dyer, 2013). It encodes for a receptor tyrosine kinase member of the insulin receptor superfamily (Ahmed et al., 2017). It is expressed in developing sympathoadrenal lineage of neural crest cells regulating multiple pathways, including the mitogen activated protein kinase (MAPK) ans Ras related protein 1 (Rap1) pathways (Cheung and Dyer, 2013). PHOX2B can directly regulate ALK expression, connecting both pathways found to be affected in familial NB cases (Cheung and Dyer, 2013). The overall penetrance of germline ALK mutations is estimated around 50% (Kamihara et al., 2017). There is a correlation between mutation type and penetrance: the R1275Q, the most common mutation seen in up to 45% of familial NB (Cao et al., 2017), leads to near complete penetrance, whereas G1128A renders a weaker activation and is correlated with a 25% risk of developing NB (Tolbert et al., 2017). Mutations in ALK can also be found in up to 6%–10% of sporadic NB tumors, while another 3%–4% carry amplifications, and overexpression is found in up to 33% of cases (Ahmed et al., 2017; Cheung and Dyer, 2013). The two most highly activated somatic hot-spot mutations (F1174 and F1245) were each observed in the germline once, associated with severe neurocognitive defects and brain stem abnormalities together with NB (Tolbert et al., 2017). ALK is a pharmacologic target: crizotinib, an ALK small molecule inhibitor, and newer agents for crizotinib resistant cases are being developed (Esposito et al., 2017). A specific ALK mutation (F1174L) has been used to develop a NB murine model (Cheung and Dyer, 2013; Graeme Eisenhofer and Cheung, 2017). ALK gain of function mutations in sympathetic precursor cells (using dopamine-b-hydroxylase (Dbh) promoter) require overexpression of MYCN (using tyrosine-hydroxylase (Th) promoter) to develop NB both in zebrafish and in a murine Th driven model (Matthay et al., 2016). The reason why ALK alone is enough to develop NB in F1174L transgenic animal model and not in others is unclear. The search for additional familial NB genes continues: GALNT14 was detected in three individuals from two families. Aberrant function of galactosaminyltransferases has been associated with tumor aggressiveness in various cancers, suggesting that GALNT14 may be involved in NB predisposition or pathogenesis, yet further studies are required (Kamihara et al., 2017). Genetic testing for mutations in genes known to be associated with NB predisposition should be considered in any individual diagnosed with NB who has a family history of NB, features suggesting predisposition such as bilateral or multifocal primary tumors, or clinical features associated with these other predisposition genes. Recently published guidelines recommend to start screening for tumor development once the mutation is identified, especially during the first decade of life (Kamihara et al., 2017).

Sporadic NB In sporadic NB, MYCN amplification, defined as 10 or more copies per genome or fourfold larger signal relative to chromosome 2, is found in up to 20%–30% of cases, conferring poor prognosis (Ahmed et al., 2017; Cheung and Dyer, 2013) (Fig. 4). In normal development, several signaling pathways regulate its expression, which leads to proliferation, growth, differentiation and survival of neurons in developing central and peripheral nervous system (Cheung and Dyer, 2013). Intratumor heterogeneity of MYCN amplification has been described with coexisting amplified and nonamplified cells within tumors (Mlakar et al., 2017). Concomittant ALK mutation/amplification in addition to MYCN amplification is associated with highly lethal disease. The MYCN locus also encodes an antisense transcript: MYCNOS (encoding N-CYM), which is always coamplified and coexpressed with MYCN, and its

Neuroblastoma: Pathology and Genetics

21

Fig. 4 Fluorescent in situ hybridization (FISH) image to detect MYCN amplification. The green dots represent MYCN and the red signals are for the centromere region of chromosome 2 and blue is DAPI counterstain labeling cell nuclei. (A) Shows a non MYCN amplified NB specimen with two copies for MYCN (green dots) and two centromeric regions in chromosome 2 (red dots), one per allele. (B) Shows a MYCN amplified NB specimen with over 10 copies of MYCN (green dots) per chromosome 2 centromeric region (red dots) per cell nuclei. Pictures courtesy of Dr. Yanming Zhang, Director of Cytogenetics Laboratory, Department of Pathology, MSKCC.

expression is associated with poor clinical outcome (Matthay et al., 2016). N-CYM stabilizes N-MYC by inhibiting its degradation, and mice transgenic for both MYCN and MYCNOS show frequent metastasis (Matthay et al., 2016). However, transgenic mice for MYCNOS alone do not develop NB (Matthay et al., 2016). MYC is a master transcription factor of cell proliferation and is considered a compelling target, though the knowledge of mechanisms affecting its expression, function and downstream targets has opened the possibility of an indirect therapeutic approach (Cheung and Dyer, 2013; Esposito et al., 2017) (Table 1). The alpha thalassemia/mental retardation syndrome X-linked (ATRX) gene loss-of-function mutations are the most commonly found in older children and young adults, and its presence is mutually exclusive with MYCN amplification (Matthay et al., 2016; Cheung and Dyer, 2013). No ATRX mutations have been identified in patients < 18 months, in contrast to 17% of children between 18 months and 12 years and 44% over 12 years with stage 4 disease at diagnosis, who have a poor prognosis (Cheung and Dyer, 2013). ATRX encodes for a SWI/SNF chromatin remodeling ATP dependant helicase (Cheung and Dyer, 2013). The role of ATRX in sympathoadrenal development or NB differentiation remains unknown (Cheung and Dyer, 2013). Mutations have also been associated with X-linked mental retardation and alpha-thalassemia, though in these patients no increased incidence of NB is seen (Cheung and Dyer, 2013). Most of ATRX mutated NBs have evidence of alternative lengthening of telomeres possibly related to a defect in histone H3.3 deposition at telomeres (Cheung and Dyer, 2013; Graeme Eisenhofer and Cheung, 2017). This highlits the importance of telomere content and length in NB, where in up to 30% of cases high telomerase activity is found, associated with adverse outcome (Cheung and Dyer, 2013). ATRX mutations have not been modeled although indirect targeted therapies are being evaluated. Telomerase reverse transcriptase (TERT) rearrangements, have recently been detected in around 20%–30% of NB (Ahmed et al., 2017; Matthay et al., 2016; Graeme Eisenhofer and Cheung, 2017). These rearrangements are mutually exclusive from ATRX mutations and MYCN amplifications (Matthay et al., 2016). As TERT is also an N-MYC target, this may mean that all NB require a way to activate TERT or lengthening of telomeres, through MYCN amplification (TERT expression is elevated in MYCN amplified tumors), enhancer hijacking, or through ATRX mutations (Matthay et al., 2016). By WGS, recurrent lesions have also been reported for genes involved in the Rac/Rho pathway and chromatin remodeling genes (AT-rich interactive domain 1A and 1B: ARID1A and ARID1B) (Matthay et al., 2016; Cheung and Dyer, 2013). The role of these genes in NB initiation, progression and prognosis is yet to be determined (Cheung and Dyer, 2013). The PI3K/AKT/mTOR pathway is involved in the development and progression of NB, though the activating molecular mechanisms of this pathway remain to be elucidated (Esposito et al., 2017). RAS pathway mutations are rare at diagnosis: NF1 and PTPN11 have been found mutated in 6 and 3% of patients (Holzel et al., 2010). Increased incidence of activating mutations in the MAPK pathway is seen in relapsed specimens; mutations in ALK as well as NRAS, KRAS, HRAS, PTPN11 and FGFR have been described (Matthay et al., 2016). In NB, P53 pathway (p53/MDM2/p16) is rarely mutated at diagnosis, but is frequently aberrant at relapse contributing to chemoresistance (Esposito et al., 2017). Targeting P53-MDM2 interactions, to prevent MDM2 mediated P53 inhibition, could be a therapeutic approach (Esposito et al., 2017). In NB this has shown to depend on MYCN status, as overexpression of MYCN sensitizes NB cells to MDM2 inhibition (Esposito et al., 2017).

Genetic Susceptibility for Sporadic NB GWAS have helped pinpoint additional mutations/single nucleotide polimorphisms (SNPs) in genes that, when combined in the germline, may influence the probability of NB occurrence (Cheung and Dyer, 2013). DNA alleles that have been significantly

22

Indirect MYCNdNMYC targeting strategies

Interaction

Pathway

Function

MYCN expression inhibition N-MYC function/ degradation

Retinoic acid PI3K/AKT/mTOR Histone DeACelytales (HDAC) Aurora kinases (AURK)

MYCN promoter binding

N-MYC/MAX interaction

Bromodomain and extraterminal (BET)

Preclinical evidence

Trials Variable efficiency

Interaction with Sirt proteins (Marshall et al., Favor N-MYC degradation Vorinostat 2011) In clinical trials, AURKA inhibition in NB Highly conserved serine/threonine kinases AURKA inhibition leads to N-MYC degradation, as showed low efficacy, particularly in MYCN AURKA has been found to sequester N-MYC away important in mitosis and its overexpression is amplified patients. Newer drugs targeting associated with genetic instability and from ubiquitin mediated proteolytic degradation. protein-protein interaction are being AURKB is a direct transcriptional target of N-MYC, aneuploidy. Overexpressed in different human developed (Esposito et al., 2017) and its increased expression is associated with tumors. Its inhibition leads to cell cycle exit poor outcomes (Esposito et al., 2017). In NB cell and apoptosis (Bavetsias and Linardopoulos, lines and murine models with MYCN amplification/ 2015) overexpression, the use of a pan-Aurora inhibitor decreased N-MYC expression and tumor growth (Bavetsias and Linardopoulos, 2015) N-MYC requires the formation of an Blockers of N-MYC/MAX interaction have been heterodimer with the MAX protein for the tested in vitro leading to apoptosis and neuronal activation of transcription (Esposito et al., differentiation. Yet to test in vivo efficacy (Esposito 2017) et al., 2017) Current open clinical trials exclude children BET proteins recognize and bind acetylated One of the BET family members, BRD4, acts as younger than 16 years of age. Concerning lysine residues on histone tails and are a transcriptional coactivator of many genes potential side effects include defective therefore considered “readers” (Varan et al., including MYC. BRD4 inhibitors have been shown spermatogenesis, viral reactivation and 2016) to reduce MYC expression by releasing BET altered bone formation, which could impact proteins from MYC locus, indicating that BET growth (Varan et al., 2016) proteins directly regulate MYC gene expression. In preclinical studies in NB cell lines, BET inhibition induced cell cycle arrest and apoptosis as well as downregulation of MYC expression. In murine NB models they decreased growth and significantly increased survival regardless of MYC amplification status. In addition to antitumor activity, neuronal differentiation was seen both in vitro and in vivo in MYCN amplified tumors (Varan et al., 2016)

Neuroblastoma: Pathology and Genetics

Table 1

Ornitine decarboxylase 1 (ODC1)

Tropomyosin Receptor Kinase (TRK) family of neurotrophin receptors

ALK

Rate limiting enzyme in polyamine synthesis, and its level is increased in highly metabolic active cells

Known target of N-MYC. Difluoromethylornithine In a phase II study on high risk patients in (DFMO) is an anti-protozoal drug that causes remission after standard chemotherapy, the irreversible inhibition of ODC1, and is being addition of DFMO maintenance was well investigated as maintenance therapy in NB tolerated (Esposito et al., 2017) patients. TrkA and its ligand, nerve growth factor (NGF), TrkA (coded by NTRK1) expression is increased in are required for neuronal differentiation NB with favorable biology and spontaneous regression. NTRK1 polymorphisms have been have been associated with reduced survival. In high risk patients, TrkA expression is low or absent. TrkB (coded by NTRK2) and its ligand, Brain-derived neurotrophic factor (BDNF), are highly expressed in biologically unfavorable NB and associated with MYCN amplification, drug resistance and increased proangiogenic factors. No mutations or rearrangements have been found in NTRK2. Entrectinib, a novel ALK inhibitor, has shown to be effective for TrkB dependent NB in preclinical models. Preferential TrkB inhibitors are currently being developed with good in vitro activities (Esposito et al., 2017) ALK and MYCN can be co-amplified, given its close Targeted therapeutic interventions using ALK ALK activation of the phosphoinositide-3and MAPK pathway inhibition in NB, currently localization in chromosome 2, associated with kinase (PI3K) signaling, stabilizes N-MYC. underway ALK also signals through RAS, with advanced stage at presentation and poor downstream activation of the MAPK pathway. prognosis (Ahmed et al., 2017; Matthay et al., 2016). The synergy may be related to downstream activation of the MAPK pathway, which is frequently activated in relapsed NB

Neuroblastoma: Pathology and Genetics 23

24

Neuroblastoma: Pathology and Genetics

associated with the development of high and low risk NB predisposition include CASC15, BARD1, LMO1, LIN28B, HACE1, DUSP12, DDX4, IL31RA, HSD17B12, NEFL, TP53 and NBPF23 (Cheung and Dyer, 2013). A few germline mutations affecting DNA repair genes including CHEK2, PINK1, BARD1 and PALB2 have been found in NB patients (Cao et al., 2017). A GWAS study of high risk NB identified up to six SNPs in BRCA1-associated RING domain 1 (BARD1), all affecting three different N-terminal introns (Tolbert et al., 2017). In depth studies into BARD1 have found an isoform, BARD1b, which lacked the RING domain for BRCA1 binding, as preferentially expressed in NB cell lines homozygous for the risk alleles (Tolbert et al., 2017). Knockdown of this isoform inhibits cell growth, whereas overexpression increases proliferation. Additionally, BARD1b was found to stabilize the Aurora family of kinases in NB cell lines, suggesting a possible mechanism of action and therapeutic strategy (Tolbert et al., 2017). Four SNPs on LMO1 gene, were also shown to be associated with high risk disease and decreased survival (Tolbert et al., 2017). This gene encodes for a cysteine-rich transcriptional cofactor preferentially expressed in the nervous system. Both BARD1 and LMO1 risk alleles seem to be more prevalent in the African–American population, which may underlie the more aggressive disease observed in these patients (Cheung and Dyer, 2013). LIN28B encodes an RNA binding protein developmentally regulated that blocks the expression of let-7 family of micro-RNAs (Matthay et al., 2016). LIN28B and let-7 are involved in stem cell differentiation with opposing effects, inhibiting and promoting differentiation, respectively. LIN28B was expressed at significantly higher levels in NB lines homozygous for the risk allele, which correlated with low levels of let-7, and its inhibition decreased cell growth (Matthay et al., 2016). In a murine model, forced expression of LIN28B under the Dbh promoter drives development of NB with high levels of N-MYC (Matthay et al., 2016). Mechanistic studies have shown that besides depleting let-7 family of micro-RNAs, LIN28B modulates the activity of the GTP-binding nuclear protein RAN and the stability of Aurora Kinase A (AURKA), pointing at potential therapeutic benefits on using AURKA inhibitors in these patients (Matthay et al., 2016).

Chromosomal Instability Ploidy is one of the strongest prognostic determinants in NB (Cheung and Dyer, 2013). In general, low/intermediate and 4S NB have whole chromosomal gains (hyperdiploidy) and high risk have intrachromosomal segmental rearrangements (Cheung and Dyer, 2013). It is thought that together with oncogene amplifications, large scale genomic alterations may deregulate messenger RNAs, micro RNAs and other non coding RNAs, that interfere with apoptosis, differentiation and immune surveillance (Cheung and Dyer, 2013). The incidence of segmental chromosomal aberrations increases with age at diagnosis (from infants to children) and is a predictor of outcome (Cheung and Dyer, 2013). Markers of poor prognosis include losses of 1p, 11q and gain of 17q, present in up to 35%, 40% and 50% respectively (Ahmed et al., 2017). Loss of 11q is associated with older age (30%–60% in adolescents and young adults), no MYCN amplification and more chromosomal breaks (commonly associated with 17q gain and 3p loss) (Mlakar et al., 2017). It is reported in 20%–45% of all high risk NB, depending on detection method used (Matthay et al., 2016). Loss of heterozygosity has been detected in 34%–44% of cases, also associated with high risk features (Mlakar et al., 2017). 11q23 is the region most commonly deleted (Mlakar et al., 2017). Chromosomal instability is one of the main features of 11q deleted tumors (Mlakar et al., 2017). Mapping studies are trying to determine the candidate coding genes (CCND1,CHK1), and miRNAs within that region; none alone has to date mimicked 11q loss (Mlakar et al., 2017). Chromothripsis, that occurs in 2%–3% of human tumors, has been described in up to 18% of stage 4 tumors though there is variability within different series (Cheung and Dyer, 2013). Some recurrent segmental alterations also occur at relapse, including 1p and 6q deletions (Matthay et al., 2016). Presence of candidate tumor suppressor genes in deleted regions of 1p have been identified: CHD5, CAMTA1, KIF1B, CASZ1 and mir-34A (Matthay et al., 2016).

NB Methylome The NB genome is epigenetically distinct, being largely hypomethylated (Decock et al., 2016). Differences at an epigenomic level cluster in groups that correlate with clinicobiological features (Decock et al., 2016). NB methylation changes affect all chromosomes, with hypomethylated regions being enriched in chromosome 6, and hypermethylated regions being more abundant in chromosomes 1 and 5 (Decock et al., 2016). DNA methylation of cytosines in the context of CpG dinucleotides is a major regulatory mechanism implicated in cellular processes critical for mammalian development (Gomez et al., 2015). At a CpG level, high risk NB tumors show higher levels of hypermethylation compared to low risk tumors, targeting mainly enhancers and polycomb repressed regions, affecting genes involved in differentiation, neuron development and biosynthetic processes. Low risk tumors show limited number of de novo CpG hypermethilated sites, mostly in promoter and enhancer related regions of genes, including LIN28B. Hypomethylation in low risk tumors is seen in genes related to development (FOXP1, SOX13, RARRES3) and in high risk tumors, in genes related to regulation of gene expression and RNA processing. The methylome of human embryonic stem cells as well as adult human and mouse brain cortex is enriched for methylation in non-CpG sites (Gomez et al., 2015). Similar to CpG methylation, non-CpG methylation goes through a process of reconfiguration during development (Gomez et al., 2015). Significant non-CpG methylation changes have been found to correlate with NB clinicobiologic features: higher non-CpG methylation was found to be associated with favorable biology, wheras in

Neuroblastoma: Pathology and Genetics

25

unfavorable tumors, non-CpG methylation was very low or absent (Gomez et al., 2015). ALK is one of the genes found to be regulated by non-CpG methylation in clinically favorable tumors rendering low ALK expression levels, in contrast to unfavorable tumors (Gomez et al., 2015). In pre- and post chemotherapy coupled samples where treatment related histological signs of differentiation were seen, higher levels of non-CpG site methylation were also found, showing that non-CpG methylation is modulated during treatment induced NB differentiation, and may be associated with changes in ALK gene expression(Gomez et al., 2015).

Tumor Regression Several mechanisms have been proposed to explain the phenomenon of spontaneous regression of NB, including neurotrophin deprivation, humoral or cellular immune response, loss of telomerase activity or alterations in epigenetic regulation (Matthay et al., 2016; Ratner et al., 2016). High TrkA, a neurotrophin receptor, is associated with favorable clinical and biological features. These tumors in vitro undergo neuronal differentiation when exposed to nerve growth factor (NGF, TrkA ligand), and these same cells die within a week when deprived from NGF. These culture conditions recapitulate the behavior of high TrkA tumors which either undergo differentiation or regression depending on the presence of NGF in the microenvironment. Migrating neural crest precursors (and favorable NB with high TrkA) can survive despite a lack of NGF in the microenvironment upon the expression of TrkAIII, a constitutively active alternatively spliced isoform. Thus, the conversion from TrkA, NGF independent to an NGF dependent tumor, could be a consequence of a developmentally programmed switch from TrkAIII to TrkA. NGF depletion in developing sympathetic neurons induces proapoptotic gene expression: TP53, TP63, E2F1, UNC5D, KIF1B and PRUNE2, which are also expressed at high levels in favorable NB tumors, including 4S (Matthay et al., 2016). High risk disease evades the attack from the immune system by downregulating the expression of HLA class I molecules (Cheung and Dyer, 2013). On the contrary, patients with 4S NB express normal levels of class I HLA, which may augment immune surveillance and response (Ratner et al., 2016). Telomeres are important for chromosome replication and stability, and its length is partly controlled by telomerases (Ratner et al., 2016). Telomerase activity is low in 4S NB, consistent with absence of TERT expression, suggesting that telomere crisis may also have a role in tumor regression (Ratner et al., 2016). Finally, epigenetic developmental changes affecting DNA methylation, histone modifications and chromatin remodeling, may impact the survival of these cells and lead to spontaneous regression (Ratner et al., 2016).

Risk Stratification Age and stage at diagnosis together with the presence of MYCN amplification, are the three strongest determinants of outcome (Cheung and Dyer, 2013). Those features, combined with loss of 11q, histological properties and ploidy, are the foundations of the current risk stratified classification of NB. There are two different staging systems currently in use, with the main difference being presurgical (International Neuroblastoma Risk Group, INRG) vs post-surgical (Intenational Neuroblastoma Staging System, INSS) tumor extent evaluation (Table 2). The benefit of the INRG classification is that it makes pretreatment patient cohorts comparable among different groups, with no surgical bias. Extent of disease is stablished upon the absence (L1) or presence (L2) of image defined risk factors (IDRFs) for locoregional tumors (Cohn et al., 2009). IDRFs refer to involvement of anatomical structures that increase surgical risk (Monclair et al., 2009). Stage M refers to widespread metastatic disease. Stage MS (S for special) refers to distant disease in children 0–18 months, limited to skin, liver and < 10% bone marrow with no cortical bone involvement (Cohn et al., 2009). Besides staging evaluation, the INRG classification expands the infant group up to 18 months of age (Cohn et al., 2009), a major shift in risk classification that spares the 12–18 month age group of high dose chemotherapy and radiation. The INRG is a more granular definition of pretreatment risk group classification among the very low, low, intermediate and high risk categories (Cohn et al., 2009).

Table 2 Stage 1 Stage 2A Stage 2B Stage 3 Stage 4 Stage 4S

International neuroblastoma staging system Localized NB not crossing midline. Complete gross surgical excision. No metastasis to ipsilateral lymph nodes that were not attached to the tumor Localized NB not crossing midline but incomplete surgical resection Localized NB not crossing midline but presence of ipsilateral lymph node involvement NB crossing midline and unresectable, with or without regional lymph node involvement Localized NB not crossing midline with bilateral lymph node involvement Localized NB with epidural extension Any or no primary, with metastasis to distant sites (except 4S distribution) Localized primary (stage 1-2B), 10%) in the peripheral blood. The sudden transformation of CLL into an aggressive diseasedRichter’s syndrome (RS)doccurs in 5%–10% of the cases. Histologically, most cases are represented by DLBCLs. Most RS represent clonal evolution of the preexisting SLL/CLL, but some are clonally unrelated to the CLL and represent secondary malignancies. A higher proportion of TP53 alterations found in clonally related cases, and clonally unrelated RS are associated to a significantly longer survival compared to clonally related RS who do very poorly.

B-Cell Prolymphocytic Leukemia (B-PLL) This extremely rare disease (< 1% of B-cell leukemias) is a malignancy of B-prolymphocytes (medium-sized round lymphoid cells with a single, prominent nucleolus) affecting blood (and accounting for > 55% of peripheral blood lymphoid cells), bone marrow and spleen. Patients typically present with very high white blood counts (> 100  109) and their median survival is 30–50 months. Cases of CLL with increased prolymphocytes or prolymphocytoid transformation, as well as lymphoproliferations associated to a t(11;14) translocation involving CCND1 are by definition excluded. Numerical or structural abnormalities of the MYC gene are recurrently found in B-PLL.

Lymphoplasmacytic Lymphoma ( and Waldenström’s Macroglobulinemia) (LPL) Lymphoplasmacytic lymphoma (LPL), a rare lymphoma occurring in older adults is a neoplasm of small lymphocytes, plasmacytoid, lymphocytes and plasma cells, usually involving the bone marrow and sometimes lymph nodes and spleen. A subset of cases occur in association with chronic HCV infection. The majority of LPL patients have Waldenström’s macroglobulinemia, (WM), which is defined as bone marrow involvement by LPL and presence of an IgM serum paraprotein of any concentration, with or without hyperviscosity syndrome. More than 90% of LPL have MYD88 L265P mutation, but this abnormality, which leads to NF-kappaB signaling, is not specific as it is also encountered in other small B-cell lymphomas and in a subset of DLBCLs. About 30% of the patients harbors somatic mutations in CXCR4 (most commonly the nonsense truncating S338X, or frameshift mutations in the C-terminal domain) in addition to the activating MYD88 L265P mutation, and this genotype tends to correlate with higher disease activity.

58

Non-Hodgkin Lymphomas: Pathology and Genetics

Annexin1

BRAFV600E

Fig. 5 Hairy cell leukemia. This bone marrow section shows a diffuse infiltrate by small lymphoid cells, which by immunochemistry are positive for Annexin1, and show expression of the BRAF V600E variant, as demonstrated by this mutation-specific antibody.

Hairy Cell Leukemia (HCL) HCL (Fig. 5) is a rare lymphoid leukemia in older adults that has a distinct cytology (small lymphoid cells with an oval nucleus and abundant cytoplasm with “hairy” projections) and immunophenotype (expression of CD103, CD25, CD11c and annexin A1). Leukemic cells are present in low numbers in the peripheral blood but show a diffuse bone marrow and splenic red pulp infiltration, responsible of patients’ pancytopenia and susceptibility to infections. Almost all cases carry a BRAF V600E kinase-activating mutation, which aberrantly activates the RAF-MEK-extracellular signal-regulated kinase (ERK) signaling pathway, leading to uncontrolled proliferation. Accompanying mutations of the KLF2 transcription factor or the CDKN1B (p27) cell cycle inhibitor are recurrent in 16% of patients with HCL and likely cooperate with BRAF-V600E in HCL pathogenesis.

Marginal Zone Lymphomas (MZL) Marginal zone lymphomas (MZL) of splenic, nodal and MALT type correspond to postgerminal center memory cells of marginal zone type that derive from and proliferate in splenic, nodal and extranodal tissues (Fig. 6).

Splenic marginal zone lymphoma (SMZL) Splenic marginal zone lymphoma (SMZL) is a rare, indolent B-cell neoplasm involving the spleen by replacing the normal B-cell follicles and showing a marginal zone differentiation, which may have circulating neoplastic cells with short “villous” projections. Patients typically present with splenomegaly, bone marrow involvement and lymphocytosis, usually without peripheral lymphadenopathy. SMZL cells express IgM, IgD, B-cell antigens (CD19, CD20, CD22), BCL2 are usually CD5-negative, and typically lack CD10, CD43, CD23, BCL6, cyclin D1, CD11c and CD25. Allelic loss of 7q21-32 allelic is detected in up to 40% of the cases. Other cytogenetic abnormalities include gains of 3q or 12q and 6q deletion. Recurrent mutations disrupting genes involved in the normal process of marginal zone differentiation, NOTCH2 and KLF2, are found in 10%–25% and in up to 44% of the cases, respectively. These mutations, as well as infrequent TP53 mutations, have been associated to adverse clinical outcome.

Extranodal marginal zone B-cell lymphoma of mucosa-associated lymphoid tissue type (MALT lymphoma) Extranodal marginal zone B-cell lymphoma of mucosa-associated lymphoid tissue type (MALT lymphoma), the third most common type of B-cell lymphoma, occurs in organs that normally lack organized lymphoid tissue. About half involve the gastrointestinal tract, and they represent 40%–60% of lymphomas in the ocular adnexa, thyroid, lung, breast, and salivary gland; skin or soft tissue may also be the primary site. The majority of patients present with localized (stage I or II) extranodal disease. MALT

Non-Hodgkin Lymphomas: Pathology and Genetics

59

Fig. 6 Marginal zone lymphomas. Splenic marginal zone lymphoma (upper panels); gross picture of a splenectomy specimen showing increased size of the spleen, and micronodular pattern on the cut section; a low-power view of the splenic parenchyma shows a nodular expansion of the white pulp, with prominent marginal zones at their periphery. Gastric MALT lymphoma (lower panels): on a low-power view of a gastric section specimen, a diffuse lymphoproliferation obliterating the mucosa is seen, with a pale appearance. The middle panel shows a case with marked plasmacytic differentiation, and the right panel shows a mucosal infiltrate by lymphoma cells showing epitheliotropism with the formation of lymphoepithelial lesions.

lymphomas run an indolent natural course. Dissemination or recurrence may occur, often in other localized extranodal sites, with long disease-free intervals. “Acquired MALT” secondary to autoimmune disease, in particular Sjögren syndrome or Hashimoto thyroiditis, or infection such as H. pylori gastritis, represents the substrate for lymphoma development. MALT lymphoma reproduces the morphologic features of normal MALT, with a polymorphous infiltrate of small lymphocytes, marginal zone (centrocyte-like) B cells, and plasma cells, admixed with scattered blast cells, occupying the marginal zone of reactive follicles and in some cases “colonizing” and disrupting these follicles. In epithelial tissues, neoplastic cells typically infiltrate the epithelium, forming so-called lymphoepithelial lesions. In MALT lymphomas, when clusters or sheets of blasts are present, a separate diagnosis of large B-cell lymphoma is made and these cases are associated with a worse prognosis. MALT lymphoma cells express sIg (M > G > A), lack IgD, and may show plasmacytoid differentiation (40% of the cases). They express B-cell antigens (CD19, CD20, CD22, CD79a), may rarely be positive for CD5, and do not express CD10 and BCL6 and cyclinD1. The most common numerical chromosomal abnormality is trisomy 3 (about 60% of cases), but is not specific for this lymphoma type. Three main recurrent translocations are encountered (see Table 2). The t(11;18)(q21;q21) fusing BIRC3(API2) and MALT1 is the most common, found in roughly 20% of the cases with the highest prevalence in lung (40%) and gastric (25%) MALT lymphomas. The t(11;18)(q21;q21) is a marker for cases of gastric MALT lymphomas that do not respond to H. pylori eradication, but these cases are less likely to undergo transformation to diffuse large B-cell lymphoma. The t(14;18)(q32;q21) leading to deregulated expression of the MALT1 gene mainly occurs in a small proportion of MALT lymphomas of the ocular adnexa and lung and is rare in other localizations. The t(1;14)(p22;q32) causing overexpression of BCL10, is rare (< 5% of the cases, mostly gastric and pulmonary). These mutually exclusive translocations deregulate the activation of the canonical and/or noncanonical NFkappaB pathways. Inactivation of TNFAIP3 (A20) by 6q23 deletion and deleterious mutations, mainly seen in MALT lymphomas of the ocular adnexa, thyroid and salivary glands where translocations are infrequent, impairs the repression of NF-kappaB normally induced by TNFAIP3.

Nodal MZL Nodal MZL is the least frequent form of MZL encompassing an adult type and a pediatric-type. It morphologically resembles nodal involvement by MZL of MALT type or splenic type, but without evidence of extranodal or splenic disease. The lymphoma cells frequently coexpress CD43. Some genetic features are shared with splenic and MALT lymphomas, including NOTCH2 and KLF2 mutations, but translocations associated to MALT lymphomas and 7q deletion are not seen.

60

Non-Hodgkin Lymphomas: Pathology and Genetics

Follicular Lymphoma (FL) FL, the most common form of indolent lymphoma and second most common lymphoma in western countries, affects predominantly older adults. FL remains essentially an incurable disease; moreover, subsets of patients may progress toward transformation to a high-grade B-cell lymphoma and a subset of patients may progress or relapse early after receiving first-line therapy. Transformation of FL usually occurs by a nonlinear divergent evolution from a common progenitor cell, and is associated to a poor outcome. FL is a neoplasm of follicular center B cells composed of a mixture of centrocytes and centroblasts, which usually has, at least partially, a follicular pattern (Fig. 7). Histological grading along a three-grade scale is based on the abundance of centroblasts by microscopic counting per high-power field (HPF). Grade 1 cases have 0–5 centroblasts per HPF, grade 2 cases have 6–16 centroblasts per HPF and grade 3 have > 15. Grade 3 is further subdivided into 3a, when centrocytes are still present, and 3b, composed of centroblasts only, biologically closer to diffuse large B-cell lymphoma. The majority of cases are grade1–2. Several studies have shown a more aggressive course for grade3 FL, but the difference may be alleviated by the use of anthracycline-containing chemotherapy and/or rituximab. Most cases have a predominantly follicular pattern, and the presence of diffuse areas in grade 1–2 FL is thought to not be clinically significant, but diffuse areas of grade 3 qualify for a diagnosis of diffuse large B-cell lymphoma. Rare cases are predominantly diffuse, a variant frequently occurring as large inguinal tumors. FL cells are usually sIg þ (IgM > IgG > IgA). The tumor cells express pan-B-cell associated antigens (CD19, CD20, CD22, CD79a), are usually CD10 þ, BCL6þ, BCL þ, CD5 and CD43 . CD23 may be positive, in particular in the purely diffuse variant. Grade 3 FL may be negative for BCL2 and CD10. Meshworks of follicular dendritic cells (FDCs) are present in follicular areas and may be highlighted by CD21or CD23. IG heavy and light-chain genes are rearranged, with extensive and ongoing somatic mutations, similar to normal germinal center cells, and consistent with a germinal center derivation. The pathogenesis of FL is driven by the t(14;18)(q32;q21) translocation which is the hallmark and most recurrent feature of FL, detected in 85%–90% of the cases and inducing ectopic BCL2 protein overexpression; and the genetic aberrations of epigenetic modifiers. The BCL2 translocation is an early event occurring in the bone marrow in pre-B cells and is considered the initiating oncogenic hit, providing a survival advantage by activating an antiapoptotic program and favoring the acquisition of additional genetic lesions during repeated transits through germinal centers. As for alterations of epigenetic modifiers in FL, mono- or biallelic inactivation of the KMT2D (MLL2) histone H3K4 methyltransferase represents the most highly recurrent lesion (about 80% of the cases). Alterations in other chromatin modifiers include inactivating mutations of the histone acetyltransferases CREBBP and EP300, in about 60% and 10% of the cases, respectively, and activating mutations of EZH2 histone H3K27 methyltransferase (about 25% of the cases). The deregulated epigenetic mechanisms have been implicated in predicting

CD10

Grade 1-2

BCL6

Grade 3a

BCL2

Fig. 7 Follicular lymphoma. At low-power view, lymph node parenchyma is replaced by multiple small nodular structures; this lymphoproliferation extends beyond the capsula into the adjacent fat. A Giemsa stain and hematoxylin & eosin stain illustrate to the high-power view of two cases, one grade 1–2 follicular lymphoma, comprising 15 large cells/hpf). The typical immunophenotype of follicular lymphoma is CD10þ, BCL6þ, BCL2þ.

Non-Hodgkin Lymphomas: Pathology and Genetics

61

patient outcome, and a prognostic model, named m7-FLIPI, that links seven mutations including five with epigenetic functions, to patient prognosis has been proposed. A small proportion of FL do not have BCL2 gene rearrangement and do not express BCL2. In particular grade 3 FL tend to be negative for BCL2 more frequently than grade1–2 cases, and purely diffuse FL are typically negative for BCL2 translocation, while showing recurrent 1p36 deletion (containing TNFRSF14), a feature common to all forms of FL. Various other chromosomal gains and losses affecting multiple chromosomes are frequently encountered as well. Abnormalities of 3q27 and/or BCL6 rearrangement are found in about 5%–15% of the cases.

In situ follicular neoplasia (ISFN) In situ follicular neoplasia (ISFN) designates partial or total colonization of germinal centers by light-chain restricted clonal B cells harboring the BCL2 translocation characteristic of FL, and strongly expressing BCL2 and CD10 in otherwise reactive lymph nodes. It is thought to be the tissue equivalent of circulating FL-like B cells that are detected in the blood of otherwise healthy individuals. ISFN is detected in about 2% of randomly selected reactive lymph nodes; if found incidentally with no evidence of concurrent FL by staging, the risk of subsequent FL is < 5%.

Duodenal-type FL Duodenal-type FL is a variant of FL with very indolent behavior: it involves the duodenum as small polyps and is typically diagnosed incidentally in women undergoing upper gastrointestinal endoscopy for other reasons; the lymphoma is usually purely follicular and grade 1–2, harboring a BCL2 rearrangement, with an immunophenotype similar to that of nodal FL, frequent expression of IgA and of the intestinal homing receptor alpha4beta7 integrin. There is a very low risk of progression to nodal disease.

Testicular FL Testicular FL occurs in children or less commonly in adults, and usually consists of grade3 lesions that are negative for BCL2 and lack BCL2 rearrangement. It is usually diagnosed on surgical excision of a testicular mass and has a very good prognosis.

Pediatric-type FL Pediatric-type FL is an uncommon form of nodal FL occurring in children or young adults, most often presenting as localized disease involving cervical lymph nodes. Cytologically, the lesions are composed of large blastic cells (grade3) and have a high proliferation index, which contrasts with the very indolent behavior. Pediatric FL lacks BCL2, BCL6 or IRF4 translocations, often features deletions of 1p36, and about half of the cases have deletions and mutations of TNFRSF14, and/or MPA2K1 mutations.

Primary cutaneous follicle center lymphoma (PCFCL) Primary cutaneous follicle center lymphoma (PCFCL) is a cutaneous lymphoma composed of centrocytes, centroblasts and large multilobated cells, showing a follicular, follicular and diffuse, or purely diffuse growth pattern. It usually presents as localized skin lesions on the head, forehead or scalp. PCFCL is positive for B-cell antigens and expresses BCL6, but CD10 is variably detected., BCL2 immunohistochemical expression is usually negative or faintly positive, and BCL2 rearrangement is generally absent. Grading is not applied to FCFCL; the proliferation fraction may be high, especially in diffuse cases composed of large cells, and the prognosis is excellent.

Mantle Cell Lymphoma (MCL) Mantle cell lymphoma (MCL) is an aggressive lymphoma of middle-aged to older adults (median survival 3–5 years) with a marked male predominance (75%). It accounts for about 7% B-NHL. Most patients have disseminated disease at diagnosis, with lymphadenopathy, hepatosplenomegaly and bone marrow and blood involvement. Extranodal involvement, especially of the gastrointestinal tract (lymphomatous polyposis), is common. MCL is typically composed of a monotonous population of small to medium-sized lymphoid cells, with slightly irregular or “cleaved,” nuclei which may have a diffuse, nodular, mantle zone, or mixed pattern in lymph nodes (Fig. 8). The blastoid variant, composed of cells resembling lymphoblasts with fine chromatin and a high mitotic rate, and the pleomorphic variant, which comprises many large irregular cells with prominent nucleoli, are considered more aggressive cytological variants. A marginal zone variant may morphologically mimic marginal zone lymphoma. MCL is not graded but the mitotic count or proliferation index have an important prognostic impact. MCL cells strongly express sIgM and IgD, more often lambda than kappa light chains and are positive for B-cell antigens; most cases coexpress CD5, and CD43, and they usually lack CD23, CD10 and BCL6. Nuclear expression of cyclin D1 is present in > 95% of the cases and represents the hallmark of the disease. Expression of SOX11 is detected in most cases including the rare subset of cyclinD1-negative MCL. MCL has unmutated or only minimally mutated IGHV regions, consistent with an origin from naïve B cells. The t(11;14)(q13;q32) (involving CCND1 gene) is present in > 95% of the cases is the primary genetic event and results in overexpression the cyclin D1, a protein involved in the regulation of the cell cycle transition between the G1 to the S phase not normally expressed in lymphoid cells. Deregulated expression of cyclin D1 overcomes the cell cycle suppressive effect of RB and p27 but is not sufficient in itself to induce MCL. Abnormal SOX11 is thought to be another important pathogenic factor. MCL also carries numerous chromosomal aberrationsdgains and lossesdand some of these aberrations may correlate with an aggressive clinical

62

Non-Hodgkin Lymphomas: Pathology and Genetics

CD5

CyclinD1

SOX11

Fig. 8 Mantle cell lymphoma. A medium-power view of a lymph node involved by mantle cell lymphoma (upper left panel) shows a residual germinal center surrounded by a monotonous lymphoproliferation of small lymphoid cells; the upper right panel shows the cytological features of the lymphoid cells, of small size with irregular nuclei (classical variant of mantle cell lymphoma). By immunochemistry, the lymphoma cells are faintly positive for CD5, and most nuclei show expression of cyclin D1 and SOX11.

parameters. Mutations in genes involved in cell cycle and other pathways, including ATM, KMT2D (MLL2), NOTCH1/2 are found in a variable proportion of the cases. Mutations of TP53 and homozygous deletions of CDKN2A are commonly found in highly proliferative cases. A small subset of the cases are negative for cyclinD1 and lack CCND1 translocation; about half of these cases carry an alternative CCND2 or CCND3 translocation. In addition to the usual nodal MCL, two subsets of MCL with an indolent behavior, namely in situ mantle cell neoplasia and nonnodal, leukemic MCL are now recognized.

In situ mantle cell neoplasia In situ mantle cell neoplasia designates the presence of cyclinD1-positive B cells harboring a CCND1 translocation, in the mantle zones of otherwise reactive lymph nodes. This condition is very rare (much less frequent than in situ follicular neoplasia) and carries a very low risk of progression to overt MCL. Compared to classical MCL, in situ lesions tend to be CD5-negative and are variably positive or negative for SOX11.

Leukemic nonnodal MCL Leukemic nonnodal MCL is defined by clinical presentation of MCL with blood and bone marrow involvement, sometimes splenomegaly, and no significant lymphadenopathy. Leukemic cells are usually small, positive for cyclin D1, carry the t(11;14) translocation, but tends to be negative for SOX11 and have hypermutated IGHV regions. Despite a usually indolent clinical course, which can remain stable for many years, these cases may transform, sometimes to a pleomorphic blastoid morphology. Routes of transformation include TP53 mutations or other oncogenic events.

Diffuse Large B-Cell Lymphoma (DLBCL) DLBCL accounts for about one-third of B-NHL and represents the most common type of lymphoma. It is defined as a neoplasm of medium to large B cells more than twice the size of small lymphocytes, with a diffuse growth pattern. DLBCL encompasses marked biological heterogeneity. It may occur de novo or as a high-grade transformation of a small B-cell lymphoma. Median age at presentation is in the 7th decade, but may occur at any age. Patients typically present with a rapidly enlarging mass, with B symptoms in one-third of the cases. About 50% have localized disease involving nodal or in up to 40% of the cases various extranodal sites. Overall, > 30% of the patients will not respond to therapy or will relapse with resistant disease. The International Prognosis Index (IPI) stratifies patients into risk groups and is predictive of outcome.

Non-Hodgkin Lymphomas: Pathology and Genetics

Immunoblastic Fig. 9

Centroblastic

Plasmablastic

63

T-cell/histiocyte-rich

Morphological variants of diffuse large B-cell lymphoma.

Diffuse large B-cell lymphoma, not otherwise specified (DLBCL-NOS) In its usual form, DLBCL is a diffuse proliferation of large lymphoid cells that may be classified as centroblastic (most common variant, may be monomorphous composed entirely of centroblasts, or polymorphous comprising an admixture of centroblasts and immunoblasts), immunoblastic (comprising > 90% immunoblasts, i.e., large lymphoid cells with a centrally located nucleolus), or anaplastic (composed of large pleomorphic cells that are often CD30 þ and tend to form cohesive sheets with a sinusoidal pattern of growth)(Fig. 9). Immunoblast-rich tumors have been found to have a worse prognosis in several studies. Bone marrow involvement in DLBCL seen in about 15% of the cases may appear either as a large-cell infiltrate, or slightly more often as an infiltrate of small atypical B cells (“discordant” marrow involvement); the latter is not associated with a worse prognosis than cases without marrow involvement. DLBCL express CD45, one or more B-cell-associated antigens (CD19, CD20, CD22, CD79a), PAX5, and often surface with or without cytoplasmic Ig (IgM > IgG > IgA). Therapy of B-cell lymphoma with chimeric antibodies against CD20 can result in the loss of CD20 antigen expression. About 5%–10% of de novo DLBCLs (i.e. excluding Richter’s syndrome) express CD5. CD5 þ DLBCLs are reported to occur in elderly women, with a predilection for extranodal involvement, especially bone marrow and spleen and to be associated with shorter survival in comparison with CD5 þ tumors. CD30 expression is seen in 10%–20% of the cases, especially those with anaplastic morphology. Ki67 proliferation index is usually > 40% and may be > 90%. Expression of BCL2 is reported in 50%–80% of the cases, and about 30%–40% of the cases are MYC-positive. Approximately one-third of the cases express CD10, 70% express BCL6, and 50% are positive for MUM1/IRF4; unlike normal B cells, DLBCL frequently coexpresses BCL6 and MUM1. Evaluation of DLBCL gene expression profiles by cDNA microarray techniques identified three molecularly distinct subtypes of DLBCL: the GCB subtype, characterized by the expression of genes normally expressed by GC B cells and showing ongoing somatic hypermutation; the ABC subtype, characterized by the features of BCR-activated B cells entering plasmablastic differentiation, showing activation of the NF-kappa B pathway, upregulation of genes involved in plasmacytic differentiation, and no ongoing hypermutations; and an unclassified group (15%–30% of the cases) (type 3). Patients with GC-like DLBCLs have better outcomes than those with activated B cell-like and type 3 DLBCLs. Different immunohistochemistry-based algorithms based the expression of markers associated to one or the other signature, have been developed as surrogates to microarray-based gene expression profiling in order to define the molecular subtypes; one of the most commonly used is the Hans algorithm, relying on the expression of three markers (CD10, BCL6 and MUM1/IRF4). Compared to other lymphomas, DLBCL shows a higher degree of genomic complexity, harboring between 50 and > 100 lesions per case. The most frequent recurrent cytogenetic aberration consists of rearrangement of BCL6 at the 3q27 locus in about 30% of the cases, both of GCB and ABC subtypes. In addition, BCL6 is the target of somatic mutations abrogating its promoter regulatory sequences in its 50 untranslated region in 70% of the cases, and BCL6 expression is deregulated in many cases as the consequence of mutations in other genes that are normally involved in regulating its function and activity or its degradation. Altogether there is ample evidence that BCL6 deregulation is an essential pathogenic event common to both molecular subtypes. Inactivating mutations and deletions of the histone acetyltransferases CREBBP and EP300, and the histone methyltransferase KMT2D (MLL2) are also highly recurrent in both molecular subtypes and may favor tumor development by reprogramming the cancer epigenome. The immune surveillance pathway is frequently impaired through genetic loss of B2M or HLA-I genes or loss of the gene encoding CD58 ligand, allowing to escape from cell-mediated cytotoxicity. Mutations of TP53 gene are detected in about 20% of DLBCL, in both molecular subtypes, and are associated with clinical drug resistance and poor outcome. Translocations of BCL2 and

64

Non-Hodgkin Lymphomas: Pathology and Genetics

MYC occur in about 20%–25% and 10% of cases respectively, and almost exclusively in GCB-DLBCL. These also harbor gain-offunction mutations of EZH2 in 20%–25% of the cases. Constitutive activation of the NF-kappaB pathway is the hallmark of ABC-DLBCL, the underlying mechanisms being heterogeneous and including gain-of-function mutations in signal transduction components of the BCR, and toll-like receptor pathways (CD79B, CARD11, MYD88), and loss-of-function mutations in TNFAIP3 (A20). ABC-DLBCL is also characterized by defective plasma cell differentiation caused by lesions deregulating BCL6 and inactivating PRDM1/BLIMP1.

T-cell/histiocyte-rich large B-cell lymphoma (T/HRLBCL) This variant of DLBCL comprises < 10% large neoplastic B cells (which may resemble the neoplastic cells of nodular lymphocyte predominance Hodgkin lymphoma or even Reed–Sternberg cells of classical Hodgkin lymphoma) scattered in a background of nonneoplastic T cells with or without histiocytes. T/HRLBCL usually presents with advanced-stage disease with nodal, splenic and bone marrow involvement. Immunophenotypic studies are essential to distinguish this variant from peripheral T-cell lymphoma and Hodgkin lymphoma.

Plasmablastic lymphoma Plasmablastic lymphoma was initially described as a tumor occurring in the oral cavity in HIV-infected patients but since then similar lesions have been reported in other anatomical sites as well as in non-HIV þ individuals. The tumor cells are immunoblastic in morphology, with or without plasmacytoid differentiation, with a plasmacytic phenotype. They are negative for CD45 and CD20, negative or weakly positive for PAX5, may be positive for CD79a, and are positive for MUM1/IRF4 and CD138. EBV is detected overall in 60%–75% of the cases, including the majority of HIV-associated cases. MYC rearrangement, usually to an IG locus, is detected in about half of the cases and translates into strong MYC positivity by immunohistochemistry.

ALK-positive large B-cell lymphoma This a very rare aggressive lymphoma is composed of large immunoblast-like B cells expressing ALK and featuring a plasma cell immunophenotype (weakly positive for CD45, lacking expression of CD20 and PAX5, positive for IRF4/MUM1 and CD138) with expression of EMA (Epithelial Membrane Antigen) and cytoplasmic IgA. CD30 is usually negative. ALK expression is usually granular and cytoplasmic, reflecting expression of the CTCL-ALK fusion protein.

Primary mediastinal (thymic) large B-cell lymphoma (PMBL) PMBL comprises 7% of large B-cell lymphomas and 2.4% of all NHL. This lymphoma has clinical, morphological, molecular and genetic features distinct from usual DLBCL. PMBL tends to occur in young patients (median age of about 35 years), and affects women more commonly than men. Patients present with an anterior mediastinal mass originating in the thymus, with symptoms related to local invasion. Relapses tend to occur in distant extranodal sites such as central nervous system and the gastrointestinal tract. PMBL is associated to a better prognosis than GCB and ABC-DLBCL subtypes. PMBL is cytologically heterogeneous, but clear cells, multilobated nuclei and compartmentalizing sclerosis are common features. The neoplastic cells express the B-cell antigens CD19, CD20, and CD79a but frequently lack surface immunoglobulin (Ig), despite expression of the appropriate transcription factors and have defective expression of HLA class I and/or class II molecules. A minority of cases are positive for CD10; BCL6 staining is found in more than half of the cases, and IRF4/MUM1 is expressed in the majority of cases. Most cases are BCL2 þ. Most cases express CD30 (at variable extent and intensity), CD23, the MAL antigen, and programmed cell death ligands (PD-L1 and PD-L2). PMBL has a unique gene expression signature that differs from that of other DLBCL and partly overlaps with that of classic Hodgkin lymphoma, and reflects activation of the nuclear factor-kB and the Janus kinase-signal transducer and activator of transcription (JAK/STAT) signaling pathways. This not only supports PMBL as a distinct clinicopathologic entity, but also the concept of “gray zone” lymphomas with features overlapping between those of PMBL and classical Hodgkin’s lymphoma. Overexpression of the MAL gene is a characteristic feature of the PMBL signature that is not found in DLBCL arising in other sites. MLBCL has been postulated to derive from medullary thymic asteroid B cell. PMBL cells have isotype-switched IG genes with a high load of somatic mutations without evidence of ongoing mutations. BCL2, BCL6 or MYC rearrangements are extremely infrequent, but half of the cases harbor rearrangements or mutations of the CIITA transactivator resulting in downregulated HLA class II molecules, creating an immune-privileged phenotype. PMBL also exhibits gains at 9p24, including the JAK2/PDCD1LG2/PDCD1LG1 locus in 70% of the cases, an aberration that explains the overexpression of PD-L1 and PD-L2 in PMBL. About half of the cases have gains at 2p16 including REL that encodes a transcription factor of the nuclear factor-kappaB family and BCL11A. Constitutive activation of the JAK/STAT signaling pathway results from JAK2 amplifications, frequent inactivating mutations of SOCS1 and PTPN1, and STAT6B mutations.

Primary central nervous system lymphoma (PCNSL) PCNSL is a rare and aggressive lymphoma that is confined to the brain, eyes, spinal cord, or leptomeninges without systemic involvement. PCNSL can develop in immunosuppressed (HIV/AIDS, organ transplant, immunosuppressive agents) or immunocompetent patients. PCNSL in immunocompetent patients is rare and represents 4% of all intracranial neoplasms. The outcome is worse than for non-CNS DLNCL.

Non-Hodgkin Lymphomas: Pathology and Genetics

65

Pathology reveals highly proliferative large lymphoid cells disposed in an angiocentric growth pattern, diffusely infiltrating the CNS. By immunohistochemistry to identify DLBCL molecular subgroups, most PCNSLs are nongerminal center subtype. Accordingly, frequent recurrent mutations in MYD88 and CD79B are identified in PCNSL. Many cases have defective HLA class I and/ or class II antigen expression. Moreover, PCNSL exhibit frequent 9p24.1 copy-number alterations and infrequent translocations of 9p24.1 and associated increased expression of the programmed cell death protein 1 (PD-1) ligands, PD-L1 and PD-L2, suggesting that immune evasion might play a role in PCNSL.

Primary cutaneous DLBCL, leg type (PCLBCL) PCLBCL represents < 5% of primary cutaneous lymphomas, and arises in older patients, with a median age at presentation > 70 years. Clinically, patients present with solitary or multiple rapidly growing bluish-red nodules or plaques, which may be ulcerated, usually on one or, rarely, both legs. Histologically, the dermis is infiltrated by dense, diffuse sheets of centroblasts and immunoblasts, often extending into the subcutis. PCLBCL is typically positive for pan-B-cell markers, CD10 , BCL6 þ, MUM1 þ, BCL2 þ, IgM þ, and shows an ABC signature by gene expression profiling. Genetic alterations within the NF-kappaB pathway, include mutations in MYD88 in 60% of the cases, or less commonly, mutations in CD79B or CARD11. Translocations of BCL2 do not occur but amplifications of the locus at 18q21 likely explains the usually intense BCL2 expression. Loss of CDK2NA/CDKN2B, by deletion and/or promoter methylation occurs in > 50% of the cases and associates with a poor outcome.

Large B-cell lymphoma with IRF4 rearrangement Large B-cell lymphoma with IRF4 rearrangement is characterized by strong expression of MUM1/IRF4, usually as a consequence of IRF4 rearrangement. This is an overall rare lymphoma, affecting predominantly the pediatric age group, consisting of a follicular and/or diffuse proliferation of large B cells and manifesting as tumors in the tonsils or Waldeyer ring or head and neck lymph nodes. The outcome is favorable.

Intravascular large B-cell lymphoma This entity consists of a disseminated intravascular proliferation of large B cells involving small blood vessels, without an obvious extravascular tumor mass or leukemia, maybe related to defective expression of homing receptors. The central nervous system, kidney, lung, and skin are commonly involved, and patients present with symptoms related to organ dysfunction secondary to vascular occlusion. Many reported cases are diagnosed at autopsy.

EBV-positive large B-cell lymphomas Several EBV-positive DLBCL entities are recognized: EBV-positive DLBCL, not otherwise specified (Fig. 10); EBV-positive mucocutaneous ulcer; lymphomatoid granulomatosis; and DLBCL associated with chronic inflammation. EBV-positive DLBCL-NOS EBV-positive DLBCL-NOS occurs in older patients where age-related immunodeficiency is thought to represent the predisposing conditions, but also in younger patients without known immunodeficiency. Peculiar morphologic features that are frequently encountered include the presence of large very atypical lymphoid cells, resembling those seen in Hodgkin lymphoma, a variable inflammatory background sometimes mimicking T-cell/histiocyte-rich large B-cell lymphoma, and areas of geographic necrosis. The neoplastic cells have an activated immunophenotype, frequent CD30 expression, and express a type II or type III EBV latency program. The disease is aggressive in older individuals but has a good prognosis in younger patients. EBV-positive mucocutaneous ulcer EBV-positive mucocutaneous ulcer manifests as superficial ulcerated lesions in the skin, oral mucosa, gingiva or mucosal sites in patients with age-related or iatrogenic immunosuppression. The course is typically indolent and some lesions may spontaneously regress. Lymphomatoid granulomatosis (LyG) Lymphomatoid granulomatosis (LyG) is an angiocentric and angiodestructive EBV þ large B-cell lymphoma with a T-cell-rich background. Necrosis is common. Patients typically present with extranodal disease, most commonly involving lung, CNS, and/or kidneys. Evidence of past or present immunosuppression may be found. LyG is graded (1 to 3) according to the number of large B cells, which correlates with clinical aggressiveness. DLBCL associated with chronic inflammation DLBCL associated with chronic inflammation occurs in the setting of chronic longstanding inflammation, like pyothorax, osteomyelitis, metallic implant insertion, or chronic skin ulcers. LDBCL associated with chronic inflammation forms tumor masses and is usually localized. The tumor may show plasmacytic differentiation and not uncommonly coexpress T-cell antigens. A type III EBV latency is characteristic. TP53 mutations and MYC amplifications are common.

66

Non-Hodgkin Lymphomas: Pathology and Genetics

CD20

CD30

EBERs

Fig. 10 EBV-positive diffuse large B-cell lymphoma, not otherwise specified. A low-power view of the lymph node (upper left panel) shows a diffuse lymphoproliferation associated to large areas of geographic necrosis. A closer view (upper right panel) shows a polymorphous infiltrate comprising many large lymphoid cells, including some resembling Reed–Sternberg cells, admixed with an inflammatory infiltrate. The atypical lymphoid cells are positive for CD20, coexpress CD30, and EBV infection is demonstrated by EBERs in situ hybridization.

Primary effusion lymphoma (PEL) This rare and aggressive disease originally identified in AIDS patients, may also occur in other clinical settings. Patients present with effusions in serous cavities, generally with no formation of tumor masses. Rare HHV8 þ DLBCLs may occur as solid tumor masses and are designated “extracavitary PELs.” PEL is composed of large, often pleomorphic cells. They are usually CD45 þ but lack expression of B-cell antigens and sIg and are often positive for MUM1 and CD138 as well as for CD30. The tumor cells are characteristically infected by the human herpesvirus-8 (HHV8) and most cases are coinfected with EBV. The gene expression profile of PEL shows features intermediate between those of immunoblasts and plasma cells, suggesting a plasmablastic derivation. The genome of PEL is complex.

Burkitt Lymphoma (BL) Burkitt lymphoma (BL) is a well-defined aggressive B-cell lymphoma with characteristic morphological, immunophenotypical and molecular features with a simple karyotype and a homogeneous gene expression profile. Three distinct forms of BL are recognized: endemic BL (primarily found in Africa), sporadic BL (comprising 30% of pediatric lymphomas in western countries), and immunodeficiency-associated BL (most often affecting HIV þ patients). In all groups, the majority of patients are male. Patients present with rapidly growing often extranodal tumor masses. In endemic cases, the jaws and other facial bones are often involved. In sporadic cases, the majority of patients present with abdominal tumors. Immunodeficiency-related cases more often involve lymph nodes, and both these and sporadic cases may present as acute leukemia. BL is highly aggressive but potentially curable; intensive chemotherapy results in cure rates of up to 90%. BL cells are monomorphic, medium-sized, with round nuclei, multiple nucleoli, and basophilic cytoplasm with small lipid vacuoles that are best seen on cytological smears (Fig. 11). There is a very high rate of proliferation and a “starry-sky” pattern is imparted by macrophages that have ingested apoptotic tumor cells. Some cases may show more pleomorphism or evidence of plasmacytic differentiation. The immunophenotype, with expression of pan B-cell antigens, germinal center markers CD10 and BCL6 and lack of BCL2 protein, is quite constant and part of the definition. Strong MYC protein expression is detectable in the majority of neoplastic cells by immunohistochemistry. The proliferation rate, as assessed by MIB1 staining, should be close to 100% of tumor cells. More than 90% of BL show translocations of the MYC oncogene, usually with the immunoglobulin heavy chain locus, and, less frequently, with the light-chain loci. Immunoglobulin heavy and light-chain genes are rearranged, with somatic mutations in most cases and intraclonal heterogeneity, consistent with their putative derivation from early follicular blasts.

Non-Hodgkin Lymphomas: Pathology and Genetics

67

Ki67

MYC

MYC BA

Fig. 11 Burkitt lymphoma. The histological and cytological features of a typical case are depicted in upper and lower left panels. Burkitt lymphoma is characterized by a very high proliferation fraction (almost 100% nuclei are Ki67 positive). MYC overexpression occurs as a consequence of a MYC rearrangement (inset, FISH using a break-apart probe for the MYC locus, showing a one red, one green and one yellow pattern, indicative of a MYC break).

Most cases have a translocation of MYC from chromosome 8q to either the IGH region on 14qdt(8;14)(q24;q32)dor IGK or IGL loci on 2qdt(2;8)dor 22qdt(8;22). Virtually all endemic cases contain EBV genomes, as do 25%–40% of sporadic and immunodeficiency-associated cases. In EBV-positive cases, the tumor cells contain clonal viral episomal genome, and display a type I latency program of gene expression (EBERs and EBNA1 expression). The exact role of EBV in the pathogenesis of BL is not understood. Next-generation sequencing analyses have revealed mutations of the transcription factor TCF3 or its negative regulator ID3 resulting in activation of the BCR signaling, in about 70% of the cases, and other mutations observed at lower frequency in various genes including CCND3, TP53, and RHOA. In general, cases that are EBV-positive tend to harbor a lesser number of mutations. Recent studies have identified rare cases fulfilling the morphological and phenotypical criteria of BL and showing a similar gene expression profile, but lacking MYC translocations. In some of these cases, recurrent alterations of 11q23, namely high level gains including 11q23.2-q23.3 and telomeric losses of 11q24.1-qter were identified. Their clinico-pathological spectrum is slightly different from BL with more frequent nodal disease, higher morphological variability, lower levels of MYC expression and more complex karyotypes. The characteristic gain/loss pattern of 11q is associated with overexpression of PAFAH1B2 and loss of ETS1.

High-Grade B-Cell Lymphoma High-grade B-cell lymphoma with MYC and BCL2 and/or BCL6 rearrangements, and high-grade B-cell lymphoma, not otherwise specified, were introduced in the most recent WHO classification to encompass a group of highly aggressive lymphomas, that in the past were in part categorized as unclassifiable B-cell lymphoma with features intermediate between diffuse large B-cell lymphoma and Burkitt lymphoma. Patients with high-grade B-cell lymphomas tend to have a very aggressive clinical disease with extensive tumor dissemination, often with bone marrow involvement, elevated LDH, rapid progression and poor response to standard therapy. High-grade B-cell lymphomas may morphologically resemble DLBCL, have features intermediate between DLBCL and Burkitt lymphoma (high proliferation fraction and a starry sky appearance, with cells larger or more pleomorphic than the spectrum of classical Burkitt lymphoma or a morphology resembling Burkitt lymphoma with a discordant immunophenotype, typically strong BCL2 expression), or have a blastoid morphology. A sizeable fraction of high-grade B-cell lymphomas with MYC and BCL2 and/or BCL6 rearrangements, also simply termed “double hit” or “triple hit” lymphomas, represent transformed follicular lymphomas. MYC plus BCL2 double hit lymphomas are usually GCB and those with MYC and BCL6 rearrangement are usually ABC. High-grade Bcell lymphomas NOS by definition do not have double or triple hits, but a significant proportion carries a MYC rearrangement (alone).

Mature T/NK-Cell Neoplasms Neoplasms of mature NK or T cells (PTCLs) show important variations in their geographic distribution: while comprising approximately 10% of all lymphomas in Western world, in Asia, where EBV-associated extranodal NK/T-cell lymphoma (and adult T-cell

68

Non-Hodgkin Lymphomas: Pathology and Genetics

leukemia/lymphoma) are more frequent, they represent up to 20% of lymphomas. Most entities are clinically aggressive with a dismal prognosis. Many distinct T/NK-cell diseases have a broad range of cellular composition, from pleomorphic small cells to anaplastic large cells; immunophenotypic profiles may vary within a disease category, while similar antigen combinations may be found across different entities; and the genetics of many of these diseases is heterogeneous with few diagnostic abnormalities. Thus the delineation of mature NK/T-cell neoplasms into entities along the WHO scheme is not as sharp as for the B-cell tumors, in regard to the numerous T-cell subsets and their functional plasticity, and reflecting the fact that, despite the cell of origin being a major determinant of PTCL biology, the cellular derivation of many PTCL entities is heterogeneous or remains poorly characterized (Fig. 12). There are currently > 25 different T-cell neoplasms (Table 1), which broadly segregate into leukemias (summarized in Table 5), lymphomas with predominant nodal involvement, extranodal, or cutaneous manifestations. Peripheral T-cell lymphoma, not otherwise specified (PTCL-NOS), angioimmunoblastic T-cell lymphoma (AITL), and the anaplastic large-cell lymphomas (ALCL), are the most frequent T-cell lymphoma entities with somewhat varying incidence between the Americas, Asia, and Europe. EBV+ AGGRESSIVE NK CELL LEUKEMIA CHRONIC NK-LPD

T-LGL

INNATE EBV+ EXTRANODAL NK/TCL, NASAL TYPE

NK

NK

BLOOD

MUCOSAE SKIN, SPLEEN

γδ (αβ)

γδ (αβ)

HEPATOSPLENIC TCL MONOMORPHIC EPITHELIOTROPIC INTESTINAL TCL ENTEROPATHY-ASSOCIATED TCL CUTANEOUS γδ TCL

ALK+/- ALCL CD8+ αβ

BLOOD

T-PROLYMPHOCYTIC LEUKEMIA

CD8+ αβ

CD8+ ANTIGEN

CD4+ αβ

LYMPH NODE

CD4+

PERIPHERAL TCL - NOS

CD8+ αβ

TFH

CD4+ αβ

GERMINAL CENTER

Treg

CD8+ αβ

BLOOD

TISSUES

CD4+ αβ

CD4+ αβ

Treg

SUBCUTANEOUS PANNICULITIS-LIKE TCL MYCOSIS FUNGOIDES

ADAPTIVE

HTLV1+ ATLL ANGIOIMMUNOBLASTIC AND OTHER TFH LYMPHOMAS

Fig. 12 Cellular derivation of PTCLs from normal T-cell and NK-cell subsets. Normal mature T cells either express the ab or gd isoforms of the Tcell receptor (TCR) and both express surface and cytoplasmic CD3. Alpha–beta T cells comprise CD4 þ (mainly helper) and CD8 þ (mainly cytotoxic) subsets. Gamma–delta T cells (CD4CD8 or CD4CD8þ) comprise 50% positive; /þ: < 50% positive.

Non-Hodgkin Lymphomas: Pathology and Genetics

Functional features

Entity

69

70

Non-Hodgkin Lymphomas: Pathology and Genetics

Nodal Lymphomas Derived From Follicular Helper T Cells Nodal lymphomas derived from follicular helper T cells (TFH cells) (see Glossary) is an umbrella term to embrace a large group of PTCL that have in common an immunophenotype overlapping with that of normal TFH cells, and share a similar genetic landscape. Angioimmunoblastic T-cell lymphoma (AITL), a disease entity encompassing systemic symptoms, biological abnormalities (such as anemia or hypergammaglobulinemia), usually generalized lymphadenopathy and a polymorphous infiltrate in lymph nodes with prominent proliferation of vessels and follicular dendritic cells, represents the prototypic and most common TFH-PTCL entity (Fig. 13). The two other nodal TFH-derived PTCL entities are: follicular T-cell lymphoma (FTCL), which is rare and defined by a predominantly follicular growth pattern and PTCL of TFH immunophenotype (TFH-PTCL), designating a subset of PTCL “unspecified” by morphology showing expression of a TFH phenotype. Worldwide, AITL is the second most common PTCL after PTCLNOS, with the highest prevalence in North America and Europe. In France, AITL currently represents > 35% of newly diagnosed noncutaneous PTCLs, largely outnumbering PTCL-NOS. AITL usually diffusely involves lymph nodes and comprises a relatively small proportion of neoplastic cells (medium-sized atypical cells with clear cytoplasm), admixed to an abundant reactive microenvironment composed of small lymphocytes, histiocytes or epithelioid cells, B-cell immunoblasts, eosinophils, and plasma cells. The neoplastic cells are mature CD4 þ CD8  T cells, frequently show aberrant loss or reduced expression of one or several T-cell antigens and express several markers of TFH cells (CD10, CXCL13, CXCR5, CD154, CD279/PD-1), CD278/ICOS (inducible T-cell costimulatory), SAP (SLAM-associated protein), BCL6, and/or c-MAF. The presence of large B-cell blasts, often infected by EBV is a feature common to AITL and other nodal TFH lymphomas. Accordingly, the gene expression signature of AITL comprises a strong microenvironment imprint, with overexpression of B-cell- and FDC-related genes, chemokines/chemokine receptors, and genes related to extracellular matrix and vascular biology; conversely the neoplastic cell signature, quantitatively minor, is enriched in genes normally expressed by TFH cells. The TFH phenotype and functionality of AITL neoplastic cells likely explains the peculiar pathological and biological traits of the disease, with the secretion of various soluble factors by TFH cells promoting recruitment, activation and differentiation of other cell types, for example CXCL13 promoting B-cell expansion and plasmacytic differentiation, causing hypergammaglobulinemia and Coombspositive hemolytic anemia commonly found in AITL patients. Broadly, the numerous recurrent genetic aberrations that have been recently identified in AITL and other TFH lymphomas interfere with two major processes: T-cell receptor (TCR) signaling pathway, and DNA methylation. Among the genes related to TCR

CD3

CD20

CXCL13

PD1

CD21

Fig. 13 Angioimmunoblastic T-cell lymphoma. A low-power view of the lymph node (upper left) shows a diffuse obliteration of the architecture, with a somewhat pale appearance. A higher magnification (upper right) shows abundant high endothelial venules, and a polymorphous cellular infiltrate comprising atypical medium-sized lymphoid cells with clear cytoplasm, that represent the neoplastic component. Most cells are CD3 þ while CD20 highlights scattered large lymphoid cells. CXCL13 and PD-1 highlight a TFH immunophenotype in the neoplastic component. CD21 shows an irregular proliferation of follicular dendritic cells which is characteristic of that entity.

Non-Hodgkin Lymphomas: Pathology and Genetics

71

signaling pathway, RHOA, which encodes a small GTPase is the most frequently mutated (60%–70% of the cases). Most mutations are hotspot generating p.Gly17Val RHOA dominant-negative variant which specifically binds to guanine nucleotide exchange factor VAV1, an interaction enhancing the adaptor function of VAV1, and ultimately leading to increased TCR signaling. In addition, activating mutations in genes encoding proximal TCR signaling elements (FYN), costimulatory receptors (CD28), or key intracellular effectors of signal transduction (PLCG1, PI3K, CTNNB1, GTF2I) are detected in up to half of TFH-PTCLs. Oncogenic TCR activation may also result from gene fusions, that is, the rare t(5;9)(q33;q22) translocation producing an ITK-SYK chimeric protein, and overall infrequent gene fusions involving CD28 with CTLA4 or ICOS. The second group of genes frequently altered are involved in DNA methylation: TET2and DNMT3A harbor mono- or biallelic inactivating mutations distributed along their coding sequences in 50%–75% and 20%–30% of TFH-PTCLs, respectively. In addition, about 20%–30% of AITLs have gain-of-function missense IDH2 mutations at the R172 residue which induces the production of an oncometabolite (hydroxyglutarate), which inhibits various deoxygenases including TET2 and histone demethylases. In AITL, DNMT3A and IDH2 mutations almost always occur in association with TET2 mutations, in contrast with myeloid neoplasms where they are usually mutually exclusive. While RHOA and IDH2 mutations are present in tumor cells only, TET2 and DNMT3A mutations have been found also in hematopoietic progenitors. In most cases, RHOA mutations are observed in TET2 mutated tumors, with the allelic burden for TET2 or DNMT3A mutations being higher than for RHOA, suggesting cooperation between impaired RHOA function preceding TET2 loss of function contributing to AITL pathogenesis.

Anaplastic Large-Cell Lymphomas Anaplastic large-cell lymphoma (ALCL) encompasses four disease entities that differ by their clinical presentation and genetic features: anaplastic lymphoma kinase (ALK)-positive and ALK-negative (systemic) ALCLs, each accounting for 7%–8% of PTCL, primary cutaneous ALCL (pcALCL) and breast implant-associated ALCL (BI-ALCL) (Fig. 14). ALCLs share a common core of morphological and immunophenotypical features: they are composed of large anaplastic cells, including so-called “hallmark cells” characterized by an eccentric kidney- or horseshoe-shaped nucleus and a prominent Golgi region; they are homogeneously strongly positive for the activation marker CD30; they often present a cohesive pattern of growth and sinusoidal dissemination in the lymph nodes; they frequently exhibit so-called “null” immunophenotype with defective expression of the TCR/CD3 and of many T-cell antigens, despite a T-cell genotype; and usually express cytotoxic-associated antigens.

(A)

(B)

(C)-CD30

(E)

(F)

(G)

(D)-ALK

Fig. 14 Anaplastic large-cell lymphomas. ALK-positive anaplastic large-cell lymphoma (upper panels): low-power view of the lymph node shows a fibrous thickening of the capsule, and partial obliteration of the architecture (A); at higher magnification, the neoplastic cells are very large, sometimes gigantic, with horseshoe-shaped or multilobulated nuclei (B); the tumor cells are strongly positive for CD30 (C) and show nuclear and cytoplasmic positivity for ALK (D) which is indicative of t(2;5) (NPM–ALK) translocation. Primary cutaneous anaplastic large-cell lymphoma (lower left panels): the lesion is characterized by a diffuse lymphoproliferation occupying the dermis and sparing the epidermis (E); the neoplastic cells are large and grow in diffuse cohesive sheets; nuclei are irregular; mitotic figures are seen (F). Breast implant-associated anaplastic large-cell lymphoma (lower right panel): Giemsa stain of a cytological smear of seroma associated BI-ALCL, comprising large pleomorphic cells with abundant cytoplasm (G).

72

Non-Hodgkin Lymphomas: Pathology and Genetics

ALKþ ALCL ALK þ ALCL preferentially affects children and young adults. Overall ALK þ ALCLs are aggressive neoplasms with good response to therapy and a significantly better survival as compared to ALK-negative ALCL (perhaps as a result of the younger age of ALK þ ALCL patients) and other PTCLs. Among the variety of ALK translocations, which induce the expression of chimeric fusion proteins having constitutive ALK tyrosine kinase activation, the most common fuses ALK gene @2p23 to the nucleophosmin gene (NPM) @5q35. The type of translocation determines the subcellular distribution of ALK but has no prognostic significance. ALK oncoproteins represent the major oncogenic driver in ALK þ ALCL: in vitro, constitutively activated ALK chimeras induce cellular transformation, enhance cell proliferation and survival, and account for the silencing of the T-cell phenotypic markers. These properties result from engagement of multiple signaling pathways, including the JAK/STAT and the PI3K/Akt pathways. No etiologic agent has been linked to ALCL, but there have been case reports of systemic ALK þ ALCL presenting with skin lesions occurring after an insect bite, suggesting the possible role of inflammatory mediators released upon the bite in eliciting lymphoma development. A recent study presented a model of peripheral ALCL pathogenesis where the malignancy is initiated in early thymocytes, before TCR b-rearrangement, which is bypassed in CD4/NPM–ALK transgenic mice following NOTCH1 expression. However, TCR is required for thymic egress and development of peripheral murine tumors, yet TCR must be downregulated for T-cell lymphomagenesis. These observations suggest that children affected by ALCL may harbor thymic lymphoma-initiating cells capable of seeding relapse after chemotherapy.

ALK-negative ALCL ALK-negative ALCL tends to occur in older individuals than ALK þ ALCL, and to have more preserved T-cell immunophenotype, with less frequent expression of cytotoxic molecules. ALK  ALCL by definition contains no ALK translocation and is negative for ALK protein expression. While distinct signatures have been derived from the comparison of ALK þ and ALK  ALCL, other studies have also evidenced much commonality in RNA or protein expression in ALK þ and ALK-negative ALCL, and in ALKnegative ALCL and a subset of PTCL, NOS with strong CD30 expression. ALCL, irrespectively of ALK status, is molecularly distinct from other PTCLs and express a distinct set of transcripts, including CD30 (TNFRSF8), BATF3, and TMOD, and demonstrate low expression of genes associated with TCR signaling. Two types of recurrent translocations were discovered in ALK  ALCL by massive parallel sequencing: (1) the most frequent rearrangements involving the 6p25.3 locus which are rather specific to ALK  ALCLs (both systemic and primary cutaneous), and virtually absent in other PTCL entities, and involve either IRF4 or DUSP22 (which encodes a dual-specificity phosphatase that inhibits TCR signaling) with various partners; (2) the less frequent TP63 rearrangements encoding fusion proteins homologous to DNp63, a dominant-negative p63 isoform that inhibits the p53 pathway, which have been found also in PTCL-NOS. The most common rearrangement resulting from inv(3)(q26q28), fuses TP63 to TBL1XR1. Rearrangements of DUSP22 and TP63 in ALK- ALCL are mutually exclusive and detected in 19%–30% and 7%–8% of the cases, respectively. DUSP22-rearranged ALCLs have a prognosis similar to ALK þ ALCL, while, conversely, TP63 rearrangements are associated with a poor outcome (17% 5-year OS). Cases lacking all three genetic markers have an intermediate prognosis. In addition there are diverse mechanisms convergent to induce constitutive oncogenic activation of the JAK/STAT pathway in a substantial proportion of ALK ALCL including occasionally co-occurring activating mutations of STAT3 and JAK1, and recurrent gene fusions involving a transcription factor (NFKB2 or NCOR2) with tyrosine kinase genes (ROS1 or TYK2). Thus, STAT3 activation represents a shared oncogenic mechanism in both ALK þ and ALK  ALCLs.

Primary cutaneous (pc)ALCL Primary cutaneous (pc)ALCL is the second most common cutaneous T-cell lymphoma (after mycosis fungoides). pcALCL presents as solitary or less commonly multiple skin nodule(s) or tumor(s) that may regress and recur, and usually carries a good prognosis. Its pathological and genetic features overlap with those of systemic ALK  ALCL, including rearrangements of DUSP22 in up to 30% of the cases, rearrangements of TYK2 in 13% of the cases, occasional STAT3 mutations and rearrangements of TP63.

Breast implant-associated ALCL (BI-ALCL) Breast implant-associated ALCL (BI-ALCL) is a very rare form of ALK-negative ALCL that arises in association with various kinds of breast implants with an estimated risk of one case per 500,000 to 3,000,000 women with breast implants. Most cases are confined to the periprosthetic effusion and capsule have excellent outcomes, while a minority of patients present with a breast tumor mass, which is an adverse prognostic factor. Information on the genetics of BI-ALCL is essentially limited. Next-generation sequencing findings in a few primary cases consistently showed either no alteration or a small number of somatic alterations, including activating mutations of STAT3 or JAK1. DUSP22 and TP63 rearrangements were not found in a reported series.

Peripheral T-Cell Lymphoma, Not Otherwise Specified (PTCL-NOS) PTCL-NOS comprises a heterogeneous group of mature T-cell neoplasms, which do not qualify for any other specific entity. PTCLNOS make up to 40% of mature T-cell malignancies in historical series, but their relative frequency in more recent registries is lower now that cases with of a TFH immunophenotype are considered in the group of AITL-related PTCLs. Lymph nodes usually are diffusely involved and cytology is typically pleomorphic, with predominantly smaller or larger cells. Reactive eosinophilia and/or histiocytes may be prominent. Many cases are CD4 þ CD8 , a subset are CD4  CD8 þ, and more rarely tumors are either double-negative or -positive for CD4 and CD8. Most cases derive from a/b T cells, a minority are of gd

Non-Hodgkin Lymphomas: Pathology and Genetics

73

derivation, or TCR-silent. CD30 is often detected in a variable proportion of tumor cells. Up to 50% of PTCL-NOS are EBV-positive usually in a small subset of cells, likely bystander B cells, and this feature has been associated with a poor survival. A subset of PTCLNOS, most commonly CD8 þ, feature a cytotoxic immunophenotype which in general correlates with a poorer prognosis. Gene expression profiling has identified two subgroups of PTCL-NOS characterized by high expression of either GATA3 or TBX21 transcription factors and downstream target genes, associated with different prognosis. These findings can be translated to routine immunohistochemistry: PTCL-NOS with high expression of GATA3 or TBX21/T-bet appear to be essentially nonoverlapping and the high GATA3 group was shown to portend a significantly worse prognosis in two independent series. Conventional cytogenetics and comparative genomic hybridization have shown recurrent genetic aberrations and imbalances, usually complex, but have not allowed to capture specific driver alterations. Very rare recurrent translocations involve breaks in the TCR genes. The t(6;14)(p25;q11.2) translocation involving the IRF4 locus, has been reported in clinically aggressive cytotoxic PTCL. Expression and constitutive activation of PDGFRa and SYK tyrosine kinases appear to represent features common to many PTCLs, and although their role in lymphomagenesis remains poorly understood, they represent novel potential therapeutic targets. The mutational landscape of PTCL-NOS is not fully characterized; targeted sequencing analyses have highlighted a heterogeneous pattern of alterations including recurrent mutations in epigenetic mediators, regulators of signaling pathways, and tumor suppressor genes.

Extranodal NK/T-Cell Lymphoma (ENKTCL) ENKTCL, nasal type (ENKTCL) represent the prototype of EBV-positive NK- or T-cell neoplasm (Fig. 15, upper panels). ENKTCL is not exceptional in western countries, but predominantly affects middle-aged men in Asia, Mexico and South America. It presents as tumors or destructive lesions in the nasal cavity, maxillary sinuses or palate, and despite a localized presentation in most patients, tends to relapse locally or at other extranodal sites, such as the skin, with an overall 40%–50% 5-year survival rate. “Extranasal NK/Tcell lymphomas” otherwise similar to the nasal NKTCL, may present in other localizations, especially in the skin, gastrointestinal tract, or testis, with a more adverse clinical outcome. Aggressive NK-cell leukemia, also EBV-associated and derived from NK cells, is regarded as the systemic form of NKTCL. ENKTCL cells usually express cytoplasmic CD3 (CD3ε þ), are CD2 þ,CD5  CD7 þ CD16 þ/ CD56 þ, with an activated cytotoxic profile. Consistent PD-L1 expression by neoplastic cells was reported in the majority of the cases. Most cases are derived from NK cells and have a germline TCR configuration; however, in some series up to more than half of the cases derived from clonal T cells, with a gd or more rarely ab TCR configuration.

(A)

(E)

(C)-TIA1

(B)

(F)-CD30

(G)

(D)-EBER

(H)-TCRγ

Fig. 15 Extranodal NK/T-cell lymphomas. Extranodal NK/T-cell lymphoma, nasal type (upper panels): this low-power biopsy shows a piece of sinonasal mucosa, with surface ulceration, diffuse necrosis and focal areas of viable tumor (A); the latter comprises an admixture of atypical lymphoid cells, including large cells (B); the tumor cells are positive for TIA1 (C) and EBV (D) as demonstrated by in situ hybridization with EBER probes. Intestinal T-cell lymphomas (lower panels): enteropathy-associated T-cell lymphoma composed of large cells with anaplastic morphology, associated to necrosis (E) and showing strong expression of CD30 (F). Monomorphic epitheliotropic intestinal T-cell lymphoma comprising a monotonous proliferation of small to medium in size lymphoid cells with pale cytoplasm (G) and showing TCR-gamma membrane expression (H).

74

Non-Hodgkin Lymphomas: Pathology and Genetics

By definition, all cases are associated with EBV, with a type II latency program (expression of EBERs, LMP1 and EBNA2). EBV is clonally present in an episomal form and exerts oncogenic effects through the production of cytokines such as IL-9 and IL-10, and the upregulation of IP10/MIP2 chemokines that may contribute to vascular damage and secondary necrosis. TNFa production may explain the common hemophagocytic syndrome. Partial deletion of chromosome 6 (6q21-25) is a recurrent aberration in ENKTCL. Several candidate tumor suppressor genes, such as PRDM1, ATG5, AIM1 and HACE1, are mapping to that region and their inactivation by deletion and/or methylation might be involved in lymphomagenesis. Compared to normal NK cells, ENKTCL is characterized by activation of PDGFRA and of the AKT, JAK/STAT, MYC and nuclear factor-kappaB pathways. While ENKTL consistently shows nuclear localization of p-STAT3, very different results have been published regarding the actual frequency of activating JAK3 mutations in this disease. Based on whole exome or large gene panel sequencing, the most frequent mutations have been found in BCOR (a corepressor acting selectively with the POZ domain of BCL6 (21%, loss-of-function mutations), DDX3X (a RNA helicase, mutated in 16% of the cases, and recurrently methylated as well), TP53 (15%) and STAT3 (13%).

Intestinal T-Cell Lymphomas Primary intestinal T-cell lymphomas derived from intraepithelial T lymphocytes overall account for 5%–8% of PTCLs (Fig. 15, lower panels). These diseases tend to occur in adults, usually manifest with single or multiple lesions/tumors in the small intestine, and have a poor prognosis. They include two entities: enteropathy-associated T-cell lymphoma (EATL), and monomorphic epitheliotropic intestinal T cell lymphoma (MEITL). EATL is most common overall, especially in Northern Europe where celiac disease is more prevalent. EATL occurs as a complication of gluten-sensitive enteropathy, but in up to half of the patients, it is the first manifestation of enteropathy. EATL is composed of pleomorphic, medium to large, occasionally anaplastic lymphoid cells, often comprises necrosis and an inflammatory infiltrate rich in histiocytes and eosinophils. The neoplastic cells are usually CD3þ, CD5 , CD4 , CD8, CD30 þ CD56 , have an activated cytotoxic immunophenotype and usually express TCRab þ and the intraepithelial homing integrin CD103. MEITL lacks the association to celiac disease in the majority of cases and is overall very rare, but represents the more prevalent intestinal T-cell lymphoma type in Asian-Pacific studies. It comprises a monomorphic proliferation of medium-sized cells, without necrosis and inflammation spreading to the adjacent mucosa with marked epitheliotropism. The neoplastic cells are usually CD3 þ, CD5 , CD7 þ, CD4 /CD8 þ, CD30 , CD56 þ, with an activated cytotoxic immunophenotype. Aberrant expression of CD20 or other B-cell markers is seen in some cases. MEITL appears heterogeneous in terms of TCR expression with the majority of cases being of gd origin. EATL and MEITL share common recurrent chromosomal imbalances (chromosome 9q gains with almost mutually exclusive losses of 16q12.1, gain of chromosome 7 and losses of 8p22-23.2, 16q21.1, 11q14.1- q14.2 and 9p21.2-p21.3), and also distinctive genetic alterations. MEITL is characterized by significantly more frequent gains of the MYC oncogene locus and less frequent gains of chromosomes 1q and 5q as compared with EATL. Additionally the loss of 3p21.31 has been reported as recurrent in MEITL but not in EATL. A few studies have examined the mutational landscape of these lymphomas. Whole exome sequencing analysis of MEITL led to the discovery of highly recurrent alterations of the tumor suppressor SETD2 gene encoding a nonredundant H3K36-specific trimethyltransferase, in the majority of the cases (14/15) (93%). SETD2 alterations were often biallelic, mainly by loss-of-function mutations and/or loss of the corresponding locus (3p21.31). SETD2 alterations are found in both gd and ab-expressing tumors and consistently correlated with defective SETD2 expression and H3K36 trimethylation. In a subsequent study, SETD2 alterations were identified in a small percentage of EATL as well. In a T cell-specific SETD2 knockout mouse model, an expansion of gd T cells was observed. Interestingly, SEDT2 is also most frequently mutated gene (about one-third of the cases) in Hepatosplenic T-cell lymphoma, an entity having in common with MEITL to be highly aggressive and often derived from gd T cells. In another study from Asia, GNAI2 was mutated gene in 24% of MEITL cases. Both EATL and MEITL harbor recurrent mutations in both the JAK/STAT and in MAPK (TP53, BRAF and KRAS) signaling pathways with STAT5B and JAK3 most frequently mutated. In MEITL mutation-induced alterations in the JAK-STAT pathway, mainly affecting STAT5B (60%–65%), JAK3 (35%–46%) and SH2B3 (20%), are almost constant.

Hepatosplenic T-Cell Lymphoma (HSTL) HSTL mostly derives from gd T cells of the splenic pool with vd1 gene usage, but there are rare cases with an ab phenotype and similar clinicopathologic and molecular features. HSTL predominantly affects young male adults and may arise in the setting of chronic immune suppression or prolonged antigenic stimulation, particularly in solid organ transplant recipients or in children treated by azathioprine and infliximab for Crohn’s disease. The disease typically presents with hepatosplenomegaly, thrombocytopenia and systemic symptoms, without lymphadenopathy or peripheral blood involvement. HSTL comprises a monotonous infiltrate of atypical medium-sized lymphoid cells, with a prominent sinusoidal pattern in the splenic red pulp, liver and bone marrow. The usual immunophenotype is CD3þ CD5  CD56 þ CD4/CD8  TCRgd þ with a nonactivated cytotoxic profile. Irrespective of TCR cell lineage, HSTL has a unique gene signature, with elevated levels of transcripts of SYK, NK-cell-associated molecules and oncogenes (FOS and VAV3), and low expression of the tumor suppressor AIM1 due to promoter hypermethylation.

Non-Hodgkin Lymphomas: Pathology and Genetics

75

The majority of the cases carry an isochromosome 7q, and trisomy 8 and loss of a sex chromosome seem to be associated with progression of the disease. Deep sequencing analyses have highlighted frequent mutations in epigenetic modifiers in 62% of cases, most commonly SETD2 but also INO80, and ARID1B. Other frequent abnormalities include activating mutations of STAT5B (31%) and STAT3 (9%), and of PIK3CD (9%).

Conclusion Lymphoid neoplasms are a diverse group of tumors arising from cells of the immune system. The normal counterpart of many but not all of these tumors can be recognized. With available techniques, a large number of clinically and biologically distinct tumors can be defined; in daily practice fewer than 10 of these make up the majority of the cases seen by clinicians and pathologists. New genetic data are enhancing our ability to define clinically relevant subsets of heterogeneous groups such as diffuse large B-cell lymphomas. Recognizing distinct diseases is important for future progress in understanding the pathogenesis of diseases and in defining optimal therapies.

See also: Non-Hodgkin Lymphoma: Diagnosis and Treatment.

Further Reading Basso, K., Dalla-Favera, R., 2015. Germinal centres and B cell lymphomagenesis. Nature Reviews. Immunology 15 (3), 172–184. Elenitoba-Johnson, K.S.J., Lim, M.S., 2018. New insights into lymphoma pathogenesis. Annual Review of Pathology 13, 193–217. Fabbri, G., Dalla-Favera, R., 2016. The molecular pathogenesis of chronic lymphocytic leukaemia. Nature Reviews. Cancer 16 (3), 145–162. Ott, G., 2017. Aggressive B-cell lymphomas in the update of the 4th edition of the World Health Organization classification of hematopoietic and lymphatic tissues: Refinements of the classification, new entities and genetic findings. British Journal of Haematology 178 (6), 871–887. Robbiani, D.F., Nussenzweig, M.C., 2013. Chromosome translocation, B cell lymphoma, and activation-induced cytidine deaminase. Annual Review of Pathology 8, 79–103. Rosenquist, R., Bea, S., Du, M.Q., Nadel, B., et al., 2017. Genetic landscape and deregulated pathways in B-cell lymphoid malignancies. Journal of Internal Medicine 282 (5), 371–394. Schreuder, M.I., van den Brand, M., Hebeda, K.M., et al., 2017. Novel developments in the pathogenesis and diagnosis of extranodal marginal zone lymphoma. Journal of Hematopathology 10 (3–4), 91–107. Shaffer 3rd, A.L., Young, R.M., Staudt, L.M., 2012. Pathogenesis of human B cell lymphomas. Annual Review of Immunology 30, 565–610. Smedby, K.E., Ponzoni, M., 2017. The aetiology of B-cell lymphoid malignancies with a focus on chronic inflammation and infections. Journal of Internal Medicine 282 (5), 360–370. Taylor, J., Xiao, W., Abdel-Wahab, O., 2017. Diagnosis and classification of hematologic malignancies on the basis of genetics. Blood 130 (4), 410–423. Vallois, D., Dobay, M.P., Morin, R.D., et al., 2016. Activating mutations in genes related to TCR signaling in angioimmunoblastic and other follicular helper T-cell-derived lymphomas. Blood 128 (11), 1490–1502.

Nonsmall-Cell Cancers of the Lung: Pathology and Genetics Erik Thunnissen, VU University Medical Center, Amsterdam, The Netherlands Egbert F Smit, Netherlands Cancer Institute, Amsterdam, The Netherlands © 2019 Elsevier Inc. All rights reserved.

Glossary APOBEC family Is a family of enzymes with cytidine deaminase function that catalyze the deamination of cytidine to uridine in the single stranded DNA substrate. Beside antiviral function they also have mutagenic effect later in oncogenesis. Comparative genomic hybridization Is a molecular technique to screen chromosomes for discrete chromosomal regions, of which the copy number has increased or decreased; that is, gained or lossed compared to the standard of two alleles (one maternal and one paternal allele in each nucleus). For pathology diagnosis this technique is now replaced by shallow sequencing, providing similar information with less input material. Mutational load of a neoplasm Is the cumulative result of all acquired mutational changes in a tumor, since the first division of the cell. Next generation sequencing Is a term covering high-throughput (multiple parallel) sequencing technologies, enabling the identification of coding-sequence alterations down to single-base-pair resolution with a fairly high level of precision and accuracy in a short period of time (days). Coverage is a term used in next generation sequencing denoting the number of reads in a sample for a given DNA fragment. Synthetic lethality An occurrence in which the disruption of two or more gene products leads to cell death, but the inhibition of either one alone does not. Transcriptome alterations Analyses in tumor specimens have led to several important observations regarding the effects of DNA-sequence alterations on RNA transcripts, splice-site mutations, and gene fusions. Whole exome sequencing Is a transcriptomics technique for sequencing all of the protein-coding genes in a genome (which is called the exome). The size of the exome in humans is approximately 1% of the genome and covers approximately 30 million base pairs. Whole genome sequencing The process of determining the complete DNA sequence of an organism’s genome at a single time. The size of the human genome is 3 billion base pairs.

Pathology of Nonsmall Cell Lung Carcinoma Introduction Until the beginning of the 21st century, the pathological classification of lung cancer for guiding of systemic treatment was quite simple: a division existed in small cell lung carcinoma and the other carcinomas (nonsmall cell lung cancer, NSCLC), each having their own (but different) chemotherapy treatment. NSCLC is a nondescriptive term that comprises the large majority of malignant lung tumors and several histologic subtypes, including squamous cell carcinoma, adenocarcinoma and large cell or undifferentiated carcinoma. Pulmonary carcinoid, mesothelioma, mucoepidermoid carcinoma, and other rare forms of lung tumors are considered separately from this general SCLC and NSCLC classification. The World Health Organization (WHO) classification of 2004 and earlier classifications for decades had been based on resection specimens but this has been influenced by two developments. One was the finding in several clinical studies that in some histological categories (adenocarcinomas (AdC) and large cell carcinomas) a chemotherapeutic agent (pemetrexed) worked better than other chemotherapic agents, but less optimal than in squamous cell carcinomas (SqCC). In parallel, it was discovered that genetic alterations which drive tumorigenesis (driver mutations, like EGFR activating mutations) were predominantly found in adenocarcinomas and their presence predicts the response to small molecule inhibitor treatment. The latter development opened a new dimension in pulmonary pathology, since after the diagnosis of lung cancer an additional (predictive) test was needed that is, a test that, when positive, directs toward alternative treatment. As a result the pathological diagnosis of lung cancer has become a multistep process, beginning with morphology on routine hematoxylin and eosin stain (H&E) and immunohistochemistry (IHC), to discriminate benign from malignant processes and metastases from primary lung cancer. In case of primary nonsquamous NSCLC, this will be followed by molecular characterization. We will first discuss the diagnostic process and subsequently the predictive procedures.

Small Sample Diagnosis Squamous cell carcinoma is recognized by signs of squamous differentiation, that is, keratinization or intercellular bridges. In contrast, adenocarcinomas present different patterns: acinar, micropapillary, rarely papillary, solid component with intracellular

76

Encyclopedia of Cancer, 3rd edition, Volume 3

https://doi.org/10.1016/B978-0-12-801238-3.65083-3

Nonsmall-Cell Cancers of the Lung: Pathology and Genetics

77

mucin, and/or lepidic growth (i.e., growth along alveolar walls). Large cell carcinoma is a diagnosis by exclusion: it is a carcinoma that has no characteristics fitting small cell carcinoma, adenocarcinoma, or squamous cell carcinoma. Thus, in morphological terms large cell carcinoma is an undifferentiated carcinoma. The diagnosis of large cell carcinoma is only made on resection specimens, because resection specimens provide larger fragments of the tumor for examination. If histological criteria characteristic for adenocarcinoma or squamous cell carcinoma are focally present (arbitrarily in > 10%), the tumor is diagnosed accordingly. Biopsies invariably are of small size, and as a consequence the proportion of cases without distinct differentiation by H&E stain is as high as 30%–40%. This morphological pattern in biopsies is called “nonsmall cell carcinoma, not otherwise specified” (NSCLCNOS, a term only used on biopsies). In daily practice approximately 70%–80% of all lung cancer diagnoses are made on small samples, the proportion of NSCLC-NOS is relatively high and this diagnosis does not provide any direction for decisions on systemic treatment. To provide arguments for therapy decisions, additional markers have to be added in the diagnostic process, as elaborated in the 2015 WHO classification of lung cancer. A set of three markers is used, pointing either to squamous cell carcinoma (immunohistochemistry strongly positive for p40 (or p63) in all undifferentiated tumor cells) or adenocarcinoma (immunohistochemistry for TTF-1 positive in most tumor cells in 60%–70% of adenocarcinomas, and histochemical mucin stain, which may be complementary to TTF-1), as shown in Fig. 1. The use of this marker panel reduces the NSCLC-NOS category to < 10%, and reliably can be used for guiding treatment.

Genetics of Nonsmall Cell Lung Carcinoma Introduction Since the discovery of DNA it became gradually clear that acquired (somatic) DNA changes in DNA are fundamental to tumor development. In lung cancer, smoking (local pollution) is the main cause of the acquired (somatic) changes. The somatic driver changes roughly corroborate with the histologic subtypes, but several additional (also called passenger) mutations are responsible for a tremendous diversity. Lung cancer is not one disease, but turns out to be a spectrum of different diseases. In this article the general mutational process in carcinogenesis is discussed and subsequently the somatic changes related to the different histological subtypes.

Fig. 1 For three different lung cancer cases (three rows) of nonsmall cell lung carcinoma not otherwise specified (NSCLC-NOS) the results of additional stains is shown. The top row shows positivity for p40 for a diagnosis of nonkeratising squamous cell carcinoma. The two rows below show adenocarcinoma based on positivity for TTF1 and mucin, respectively. H&E at 100
and additional stains at 200
magnification. In case of lack of morphological clues for the distinction between squamous cell carcinoma and adenocarcinoma on routine histological staining, or when the pathologist is in doubt, additional diagnostic stains are requested. The left panels show staining patterns and the diagnoses with which these are associated. The right panels list the predictive tests per diagnostic category. ESMO guideline indicates that molecular testing is only recommended in adenocarcinomas and NSCLC-NOS and not in patients with SqCC, except for never/former light smokers ( A), also called the “smoking signature.” A similar pattern may also occur in nonsmokers, if exposure happens to similar toxins by, for example, occupational exposure. The second signature is an age related change: C > T transition at CpG sites. The third arises during cancer progression, during which additional mutations from cytosine to guanine or thymidine (C > G or C > T) develop. This pattern is associated with APOBEC enzyme-mediated mutagenesis creating the so called “APOBEC signature.” Interestingly, despite maintained carcinogen exposure, tumors from smokers showed a relative decrease in smoking-related mutations over time compared to the APOBEC related mutations, which indicates that further expansion of the genetic derangement/chromosomal instability occurs. Interestingly, cessation of smoking for a longer period is associated with a reduction of the smoking signature. This implies that the tumor mutation spectra in former smokers reflects not only quantity of smoking exposure, but also time since smoking cessation. The general assumption is that DNA mutations develop stochastically, but nonetheless the distribution of mutations is not random over the whole genome, as the folding of DNA in chromatin varies between cell types. In morphological terms, heterochromatin in a nucleus contains densely packed DNA strands, while euchromatin contains transcriptionally active DNA which is more loosely packed and more vulnerable to DNA adduct formation. Regarding molecular oncogenesis of lung cancers, it is assumed that following early (tabacco related) mutations genome doubling occurs, along with chromosomal instability responsible for gain or loss of chromosomes chromosomal frangments, responsible for so called “copy number alterations.” The pattern of copy number alterations, visualized by comparative genomic hybridization, shows some similarities between adenocarcinomas and squamous cell carcinomas, but also many differences, allowing to distinguish between these two categories in approximately 80% of the cases. In line with genomic instability as one of the hallmarks of cancer, in squamous dysplasia, the morphologically recognizable premalignant precursor lesion of respiratory epithelium, the presence of abundant copy number alterations mark progression to squamous cell carcinoma. Dysplasia without abundant copy number variation remain stable or even regress. Abundant copy number alterations in precursor lesions are not reversible.

Cell of Origin The epithelium of the respiratory tract consists of basal, parabasal, neuroendocrine, ciliated, goblet cells, type I and type II pneumocytes. Depending on the cell type, some genes will be active, while others will be silent. Inhaled toxic components may affect any cell and cause a mutation, which raises the question whether the various cell types of the respiratory tract differ in sensitivity for mutagenesis. Ciliated cells and type II pneumocytes are terminally differentiated cells, which will undergo apoptosis at the end of their

Nonsmall-Cell Cancers of the Lung: Pathology and Genetics

79

life-span and will then be replaced freshly matured cells derived from local stem cells. In such terminally differentiated cells a mutation will not initiate cancer. In contrast, if a mutation develops in a cell with inherent proliferative capacity (e.g., basal, neuroendocrine, alveolar type II, and Clara cells), this may be the initial molecular event responsible for malignant transformation of a cell that might become the cell of origin of a tumor. Oncogenesis, however, requires the accumulation of multiple mutations in genes responsible for vital cellular functions (e.g., regulating cell proliferation and apoptosis). The question of cell of origin of lung cancer has been studied by introducing specific gene mutations in specific cell types in mouse models, including exposure to toxins. Cell-of-origin appeared to be associated with the histopathological characteristics of the resulting tumor. The neuroendocrine cell appears to be the cell of origin of small cell lung cancer. For KRAS mutated adenocarcinomas Clara cells or alveolar type II cells serve as the initiating cell type. Of note, basal cells, Clara cells and alveolar type II pneumocytes may also serve as initiating cell types for squamous cell carcinoma, with SOX2 activation as the event responsible squamous differentiation. Surprisingly, only 10% of the smoking population develops lung cancer and that even after a long latency period, in spite of the enormous daily toxic burden of a regular smoker. This is (at least in part) due to the efficacy of the DNA repair process as well as the fact that many mutations will occur in terminally differentiated cells which will not become cancer initiating cells.

Clonal Development Once the cell has acquired a sufficient number of mutations conferring autonomic growth, autonomous clonal expansion will occur. During this process all cells of the clone will contain the same set of initiating mutations. To reach the volume of a detectable tumor (> 0.5 cm) at least 30 doubling cycles have to be accomplished. With a tumor volume doubling time for malignant tumors varying between 50 and 400 days, the time it takes to develop a detectable tumor is counted in years. During this time frame additional mutations will have developed but not necessarily in all tumor cells. When samples from different regions of a large tumor are extensively studied with whole genome and/or exome sequencing, different parts of the tumor will show different mutational profiles, even though the set of initiating mutations is retained. Mapping these regions analogous to a phylogenetic tree structure, mutations present in all regions of the tumor are called “trunk mutations,” heterogeneous mutations present in only some regions of the tumor are called “branch mutations,” and mutations that are present only in one region of the tumor are called “private branch mutations.” Trunk mutations are acquired early on during carcinogenesis while branch mutations have occurred later. Private branch mutations discovered with high sequencing depth might confer significant growth advantage, resulting in clonal expansion and this will turn them into branch mutations. This concept illustrates ongoing accumulation of somatic genome aberrations due to genome instability, continued background mutagenesis (age-related signature), and genetic drift. Intratumor heterogeneity is manifested in multiple subclones that coexist in different regions of one tumor.

Somatic Genome Changes in Adenocarcinoma Comprehensive analysis of the genomic landscape of adenocarcinomas identified a very high number and a large variety of somatic gene aberrations, including mutations and/or copy number alterations (76%), as listed in Table 1. Some genes are frequently affected, but in most of the listed genes aberrations occur at a frequency < 10%. Affected genes are usually abnormally active in one or more cellular signaling pathways. We will discuss gene aberrations in adenocarcinomas, including those affected in AdC as well as in SqCC, according to functional categories. Some genes fit into more than one functional category, and for reasons of simplicity have been put together under the heading “deregulated proliferation.”

Deregulated Apoptosis TP53 is a tumor suppressor gene, implying that both alleles need to be aberrant for the carcinogenic effect. An important physiological function of p53 is detection of DNA damage which affects transcription of downstream targets, leading to growth arrest and apoptosis. When one allele of TP53 is mutated and the other allele is lost, p53 is inactive which permits continuation of cell growth and cell survival in spite of DNA damage. TP53 is the most frequently mutated gene in NSCLC, and in addition it is most frequently (92%) associated with mutations of other genes.

Deregulated Receptor Tyrosine Kinases EGFR (ERBB1) is a transmembrane protein with receptor tyrosin kinase function. EGFR is activated by ligand binding and subsequent dimerization, which activates autophosphorylating capacity of the intracytoplasmic domain. This results in conformational changes of the protein, which activates downstream signaling via RAS (see KRAS). In AdC, EGFR mutations in the intracellular domain confer driver oncogene properties to the protein, as it is activated without ligand-binding. This results in oncogenic activities largely similar to those of KRAS mutations. ERBB2, previously called her2/neu, is a member of the EGFR growth factor receptor family. Signaling promotes cell proliferation and opposes apoptosis and requires tight control. In AdC deregulation (mutation, amplification, and overexpression) the tight control is lost and oncogenic effect occurs.

80

Nonsmall-Cell Cancers of the Lung: Pathology and Genetics Table 1

Gene TP53 EGFR KRAS Keap 1 STK11 SMARCA4 RBM10 RB1 ARID1 CDKN2A PIK3CA NF1 ALK ROS1 BRAF ERBB2 MET ATM SETD2 FTSJD1 SMAD4 ARID2 NTRK1

Gene aberrations in pulmonary adenocarcinomas, with aberration frequency Also in SqCC Yesa

Yes Yes Yesa Yesa Yes

Aberration M% 54 10–50b 30–10b 15 14 9 6 5 7 4 5 10 2–3 1 1–8 2–4 1–3 8 5 4 3 5 0,1

Lesion type

L L L L F F A

L L F

M%: frequency of gene aberrations, including mutations, small insertions/ deletions, and chromosomal rearrangements. Lesion type other than point mutations: F, fusion; G, gain; L, loss in CNV analysis; A, amplification. a Higher frequency in SqCC than in AdC. b In western countries prevalence for EGFR 10%–15% and for KRAS 25%– 30%, while in eastern countries prevalence for EGFR is 30%–60% and for KRAS 10%.

The MET gene (tyrosine-protein kinase Met or hepatocyte growth factor receptor) encodes for the c-MET protein. The c-MET protein plays a role in embryogenesis and is expressed on epithelial cells, but also some mesenchymal cells (endothelial cells, neurons, hematopoietic cells, melanocytes, and neonatal cardiomyocytes). In cancer two changes are relevant: amplification and deletion of exon 14. This exon contains an ubiquitin binding site. Deletion maintains an active c-MET protein with continuous activation of oncogenic pathways (RAS, PI3K, STAT3, b-catenin) and neoangiogenesis. The NTRK1-3 genes encode for proteins belonging to the tropomyosin kinase receptor family. The proteins have prosurvival and differentiation functions. Fusions occur at low frequency in several cancer types, and lead to constitutive oncogenic tropomysin kinase activity, also via MAPK pathway.

Deregulated Proliferation KRAS is a dominant oncogene. When one allele is mutated it translates into a continuously active protein, resulting in persistent downstream signaling of the RAF/MEK/ERK pathway. This stimulates cell proliferation. It also activates the PI3K/AKT pathway, which stimulates cell survival (e.g., protein synthesis, cell motility). The RB1 gene codes for the retinoblastoma protein, a transcription repressor inhibiting cell growth. When pRB is phosphorylated it becomes inactive, allowing cell cycle progression. Rb1 mutations result in an inactive protein, likewise allowing cell cycle progression. RB1 is mutated at high rates in SCLC, and much lower in squamous cell carcinomas and adenocarcinomas. CDKN2A gene (cyclin-dependent kinase inhibitor 2A) encodes for two proteins, p16 and p14arf, through an alternative reading frame. Both are cell cycle regulating tumor suppressor proteins; p16 inhibits CDK4 and CDK6 and thereby activates pRb, while p14arf activates the p53 tumor suppressor. Mutations will allow cells to continue to cycle. PIK3CA encodes the phosphatidylinositol-3-kinase catalytic subunit alpha, that has a role in cell cycle control, differentiation, cytoskeleton reorganization, and intracellular trafficking. In lung cancer activation of the PI3K pathway allows uncontrolled cell proliferation.

Nonsmall-Cell Cancers of the Lung: Pathology and Genetics

81

The NF1 gene encodes the protein neurofibromin, a negative regulator of the RAS signal transduction pathway. A germline NF1 mutation is responsible for neurofibromatosis. The role of somatic NF1 mutations in lung cancer is not clear; they may be driver mutations, but may also contribute to therapy resistance. In NSCLC the ALK gene is involved in chromosomal translocation, resulting in a fusion gene. There are different fusion partners. The fusion gene is transcribed as fusion protein, containing the kinase domain of ALK. The fusion protein results in abnormal ALK signaling, which stimulates cell growth. Wild type ALK protein is normally not expressed in the lung. Screening for ALK aberrations in NSCLC can therefore be performed by immunohistochemical staining for the ALK protein. The ROS1 gene is also involved in chromosomal translocation in NSCLC. There are several different fusion partners (CD74, EZR, FIG1, CCD6C, KDELR2, LRIG3, SDC4, SLC34A2, TPM3, and TPD52L1). Transcription of the fusion gene usually contains the kinase domain of ROS1. This leads to constitutive activation of MAP kinase ERK, PI3K, mTOR, JAK, and STAT pathways. Like ALK, ROS1 can be detected using immunohistochemistry, but ROS1 staining may be positive in nonmalignant cells (e.g., type II pneumocytes). BRAF is part of the RAS signal transduction pathway. Mutations result in activation of the MAPK pathway which stimulates cell growth. SMAD4 (mothers against decapentaplegic homolog 4) also known as DPC4 (deleted in pancreatic cancer-4) is member of the SMAD protein family, that functions as signaling molecules downstream of the TGFb receptor and so modulate gene transcription. SMAD4 activates the expression of genes that inhibit cell growth. In NSCLC SMAD4 is inactivated by homozygous deletion or mutation.

Deregulated Oxidative Stress KEAP1 encodes the Kelch-like ECH-associated protein 1 (KEAP1). KEAP1 interacts with NRF2 in the cytoplasm and has a repressive effect on NRF2. In lung cancer, mutations in NRF2 and KEAP1 are mutually exclusive. STK11 encodes serine/threonine kinase 11, a cytoplasmic protein that functions as upstream kinase of adenosine monophosphate-activated protein kinase. It is involved in cell polarity and a necessary element in cell metabolism, required for maintenance of energy homeostasis, especially in a low energy state. Mutations in STK11 lead to disorganization of cell polarity and facilitates tumor growth under energetically unfavorable conditions.

Deregulated DNA Repair/Chromatin Modeling The SWI/SNF family is a multisubunit chromatin remodeling complex, including two mutually exclusive ATPase activities (SMARCA2 and SMARCA4). These reposition nucleosomes, which has an impact on gene expression. Mutations have been found most commonly in the SMARCA4 subunit but also in other subunits, and are thought to confer functional specificity (ARID1A, ARID1B, PBRM1, and ARID2). In NSCLC, loss of SMARCA4 can be demonstrated using immunohistochemistry and this is associated with a high mutation rate. ATM is a gene encoding a serine/threonine protein kinase, that senses the presence of oxidative DNA damage, including doublestrand DNA breaks, and facilitates subsequent repair. Insufficient DNA repair increases genome instability and contributes to cancer progression. SETD2 protein is a histone methyltransferase, specific for lysine-36 of histone H3. Methylation of this residue prevents inappropriate transcriptional initiation, supports repair of damaged DNA, and regulates pre-mRNA splicing. Mutations of SETD2, or mutations at or near lysine-36 of histone H3, are associated with cancer development.

Deregulated Transcription RBM10 encodes for an RNA binding protein involved in alternative splicing. Somatic alterations in RBM10 lead to loss of function. FTSJD1 (also known as CMTR2) was recently discovered and encodes an enzyme that methylates the ribose of the second transcribed nucleotide of messenger RNA and small nuclear RNA. In lung cancer frame shift mutations of FTSJD1 have been found.

Somatic Genome Changes in Squamous Cell Carcinoma Comprehensive analysis of the genomic landscape of SqCC identified a very high rate and a large variety of somatic gene alterations, both mutations (74%, with a mutation rate per megabase of 3–10 times more than in other cancers) and/or copy number alterations (76%). Consequently, most SqCCs (95%) have potentially targetable molecular alterations and almost two-thirds (> 70%) have two or more alterations. This high rate of molecular changes reflects the mutagenic effects of tobacco in this strongly smoking-associated histological subtype. Some of the most significant somatic alterations of lung SqCCs identified in the Cancer Genome Atlas Research Network (TCGA) and in several studies are listed in Table 2. The genes carrying aberrations in squamous cell carcinomas only, excluding those that can be mutated in AdC as well as SqCC (which have been discussed earlier), are summarized

82

Nonsmall-Cell Cancers of the Lung: Pathology and Genetics Table 2

Gene TP53 MLL2 PIK3CA CDKN2A NFE2L2 KEAP1 PTEN NOTCH1 RB1 DDR2 FAT1 RASA1 CUL3 PASK SOX2 TP63 FGFR1 F (3) FGFR2 FGFR3 EGFR PDGFRA/KIT CCND1 MET NF1 ARID1A NOTCH1

Gene aberrations in squamous cell carcinomas of the lung with aberration frequency Also in AdC Yesa Yesa Yesa

Yes

Yes Yes Yes

Aberration M%

Lesion type

51–81 20 3–16 9–18 15 12 8–10 8 7 4 7 6 5 4 21 16 16–21

92% co-mutations

2–3 2–3 7–14 8 14 3–21 8 5 5

G (34–43) L (22–29) Meþ(21) L (17)

G (21) SNP (16) G (16–22) F (3) F (3) G (7–14) G (8) G (14) G (3–21)

M% ¼ frequency of gene aberrations, including point mutations, small insertions/deletions, chromosomal rearrangements. Lesion type if not (only) point mutations (frequency in % of other lesion types) F, fusion; G, gain; L, loss in CNV analysis; SNP, single nucleotide polymorphism; Me þ, gene silencing due to promoter hypermethylation. a Higher frequency in SqCC than in AdC.

according to functional categories. Genes that fit into more than one functional category have been combined where appropriate under “deregulated proliferation.”

Deregulated Receptor Tyrosine Kinases FGFR 1-3 are proto-oncogenes and encode fibroblast growth factor receptor proteins, that belong to the fibroblast growth factor receptor family. FGFR proteins are involved in several signaling pathways including phospholipase C/PI3K/AKT, Ras subfamily/ ERK, protein kinase C, and STAT. FGFR alterations as a rule are gain-of-function changes (amplification/fusion), but the exact mechanism is not always clear. PDGFRa gene encodes platelet-derived growth factor receptor, a cell surface tyrosine kinase receptor. PDGFR proteins are involved in several signaling pathways including phospholipase C/PI3K/AKT, Ras subfamily/ERK, protein kinase C, and STAT3. Alterations of PDGFR in cancer include amplification, mutations, and rearrangements making the receptor constitutively active, and alternative PDGFR splicing. NOTCH1 Notch homolog 1, translocation-associated (Drosophila), also known as NOTCH1, is a human gene encoding a singlepass transmembrane receptor, that belongs to the NOTCH family. In lung embryogenesis NOTCH1 is expressed in pluripotent progenitor cells in human bronchial epithelium. NOTCH1 suppresses multiple pathways, and when functionally disturbed it impacts on cancer-associated processes including proliferation, cell survival, epithelial–mesenchymal transition, metastasis, and angiogenesis. PASK encodes the PAS domain-containing serine/threonine-protein kinase, which is involved in nutrient signaling. PASK mutations have recently been reported in lung cancer. The downstream oncogenic mechanism is unclear. DDR2 discoidin domain-containing receptor 2 (also known as CD167b) encodes a receptor tyrosine kinase. Mutations are oncogenic, but the downstream mechanism is unclear.

Nonsmall-Cell Cancers of the Lung: Pathology and Genetics

83

Deregulated Proliferation The CCND1 gene encodes the cyclin D1 protein that belongs to the cyclin family, which show periodicity in protein abundance during the cell cycle. Cyclin D1 is required for progression through the G1 phase of the cell cycle. Constitutive overexpression of cyclin D1 leads to increased growth. NFE2L2 (nuclear factor erythroid-derived 2-like 2, also known as Nrf2), encodes a basic leucine zipper (bZIP) protein. This transcription factor regulates the expression of antioxidant proteins that protect against oxidative damage triggered by injury and inflammation. Aberrant activation of the NRF2 signaling pathway has an oncogenic effect, because of crucial functions in cell survival and proliferation. PTEN is a tumor suppressor gene which encodes phosphatase and tensin homolog protein, and functions as tumor suppressor through negative regulation of the Akt/PKB signaling pathway. Its inactivation in cancer leads to increased cell proliferation and reduced cell death. RASA1 encodes p21 protein activator 1 or RasGAP (Ras GTPase activating protein), which is part of the GAP1 family of GTPaseactivating proteins. RASA1 is located in the cytoplasm and stimulates GTPase activity of normal RAS p21, acting as suppressor of RAS function, thereby allowing control of cellular proliferation and differentiation. FAT1 a tumor suppressor gene encoding a protein that is a member of the cadherin superfamily. FAT1 is expressed in fetal epithelium and has a supporting role in controlling cell proliferation.

Deregulated Oxidative Stress CUL3 is a tumor suppressor gene encoding a core scaffolding protein in an E3 ubiquitin ligase complex that includes KEAP1 and RBX1, which plays a role in the oxidative stress response pathway. Loss of PTEN and CUL3 synergize to activate NFE2L2 signaling. The oncogenic mechanism is unclear, possibly decreased production of superoxide might be protective for tumor cells.

Deregulated Migration P63 encodes the transformation-related protein 63, that is part of the p53 gene family. P63 is expressed in the basal layer of most stratified epithelia and plays a role in epithelial development and differentiation. Even though p63 expression is used in the diagnostic process of NSCLC to make a diagnosis of SqCC, strictly speaking p63 is not a marker of squamous differentiation. TP63 is frequently amplified and overexpressed of in a variety of epithelial cancers, which has led to the assumption that this protein behaves as an oncogene. Reduction of p63 expression levels has been associated with epithelial mesenchymal transition and metastasis. SOX2 encodes a protein (sex determining region Y-box 2, SRY), that is a member of the Sox family of transcription factors. SOX2 is essential for maintaining self-renewal, or pluripotency, of undifferentiated embryonic stem cells. SOX2 is frequently amplified in SqCC and its overexpression activates cellular migration and anchorage-independent growth.

Deregulated DNA Repair/Chromatin Modeling/Transcription MLL2 is a tumor suppressor gene encoding histone H3 lysine 4 (H3K4) mono-methyltransferase, that colocalizes with lineage determining transcription factors on transcriptional enhancers and is essential for cell differentiation and embryonic development. Mutations occur in several cancer types. The oncogenic mechanism is not clear. One possibility is that in situations of transcriptional stress RNA polymerase II malfunctions (pausing, stalling, and backtracking), resulting in abnormalities in early replicating fragile sites within the chromosome. In NSCLC, most genes are affected in < 10% of the cases except for TP53, which is frequently mutated in NSCLC. Development of malignancy does not require aberrations in all of these genes. Genes such as TP63, SOX2, and PIK3CA reside in the same chromosomal region (chromosomal region 3q) and co-amplification of these three genes with FGFR1 amplification (chromosomal region 8p12), may lead to cross-activation.

Predictive Biomarkers in Nonsmall Cell Lung Carcinoma Once the diagnostic process has been completed, predictive testing is a second equally important step that needs to be performed in patients with advanced lung cancer. The outcome of predictive testing is essential for the choice of drug treatment. This may also include information on xenobiotic metabolism, which influences drug dosage. For practical purposes this article will only discuss NSCLC testing for the choice of drug. Predictive testing does not only assess acquired genetic changes (Table 3), but also includes expression of specific proteins. The prevalence of aberrations in many of the relevant genes is low and therefore many tests (10 to almost 100) need to be performed to have even a single positive result, directing the choice for a patient toward a specific drug. This makes predictive testing rather expensive. Most genome alterations occur in nonsmokers. Rearrangements of EGFR, ALK, and ROS1 are predominantly found in young patients. ALK and ROS1 alterations occur in adenocarcinomas with a solid or cribriform growth pattern and/or the presence of signet ring cells. KRAS mutations are most frequent (they occur in nearly 30% of NSCLC), predominantly in smokers, but they

84

Nonsmall-Cell Cancers of the Lung: Pathology and Genetics Table 3

Genes examined for predictive testing in NSCLC

Gene

Histology

Alteration

Frequency (%)

Method(s)

EGFR ALK ROS1 BRAF ERBB2 MET MET PD-L1

Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma Adenocarcinoma NSCLC

Mutation Rearrangement Rearrangement Mutation Mutation Amplification Mutation Expression

10–60 2–5 1 1–3 2–4 1–3 1–3 20–30

Sanger sequencing/NGS IHC, FISH, NGS IHC, FISH, NGS Sanger sequencing/NGS Sanger sequencing/NGS ISH Sanger sequencing/NGS IHC

IHC, immunohistochemistry; FISH, in situ hybridization; NGS, next generation sequencing.

are not (yet) targetable. An important role of the pathologist is to characterize the tumor and choose the tumor sample for testing. This includes testing for responsiveness to immune checkpoint inhibitors. As in about 80% of NSCLC patients only small cell/tissue samples are available (small biopsies or cytology specimens), the pathologist needs to make most efficient use of the tumor specimens and prioritize the analyses. The pathologist will select tissue for DNA extraction to be used for next-generation-sequencing (NGS), will assess chromosomal rearrangements using fluorescence in situ hybridization (FISH) and will also assess PD-L1 expression, to select patients for treatment with PD1/PD-L1 inhibitors. NSCLC tissues are heterogeneous and composed of a mix of nonneoplastic stromal cells (immune cells, fibroblasts, endothelial cells) and neoplastic cells. This implies that tumor DNA is “diluted” with nontumor DNA from the stromal compartment. Any molecular test needs to be sufficiently sensitive (analytical sensitivity) to detect a mutation: when a molecular test can detect a minimum of 10% of mutated DNA, assuming that the mutation is heterozygous this will require the presence of 20% of malignant cells in the sample. When the fraction of neoplastic cells is lower, macro- or microdissection can be used to arrive at a higher fraction of tumor DNA. Morphological assessment of the percentage tumorcells in a sample, on H&E stained sections on both sides of the series of sections used for DNA extraction, is mandatory for reliable results. For most PCR-based NGS techniques, a minimum of 5–10 ng of total DNA (including DNA from neoplastic and nonneoplastic cells) is required, which is equal to approximately 1000 cells in formalin-fixed and paraffin-embedded tissue material. Hybrid capture-based NGS technologies, necessary for RNA sequencing, require 100–200 ng of RNA or DNA, which may not be attainable in up to one third of small biopsy samples.

EGFR The first responses to EGFR tyrosin kinase inhibitors in NSCLC were observed in metastatic patients who no longer responded to conventional treatment. Sequencing of the EGFR gene in these tumors led to the discovery that the gene sequence coding for the intracytoplasmic domain of EGFR contained activating mutations in exons 18–21, notably 15–18 base pair deletions in exon 19 and point mutations in exon 21 (p. L858R). Progression free survival improved, compared to standard chemotherapy, which resulted in approval of gefitinib, erlotinib, and afatinib for first-line treatment of EGFR-mutated advanced NSCLC. However, after a few months to a few years progression occurred, indicating resistance to treatment. Second and third generation EGFR inhibitors have been developed (such as osimertinib, designed against a resistance mutation T790M).

ALK Rearrangement Anaplastic lymphoma kinase (ALK) gene rearrangements in NSCLC result in expression of ALK fusion-proteins with strong oncogenic properties. Screening for ALK rearrangement can be done by immunohistochemistry, which requires limited equipment and is available in pathology laboratories worldwide. The method allows detection of the presence of the fusion protein even in a small number of tumor cells. Interpretation of ALK stained sections is quite straight forward, as ALK protein is not expressed in normal epithelial lung tissue. Other nonneoplastic cells will stain, such as light dot-like cytoplasmic staining in alveolar macrophages, nerve and ganglion cells, epithelial cells of bronchial glands, extracellular mucin, and necrotic tumor areas. In case of doubt, Fish can be used to confirm a rearrangement. ALK positive NSCLC is treated with ALK-inhibitors crizotinib and ceritinib and the results are superior to those of standard chemotherapy. Recently, alectinib has been shown to be superior to crizotinib.

ROS1 Rearrangement Rearrangements of the ROS1 gene have been discovered recently and can be found in 1%–2% of lung cancer patients. Treatment with Crizotinib has been approved for ROS1 rearranged NSCLC patients. Data on the predictive value of immunohistochemical staining for ROS1 are still evolving. ROS1 rearranged tumors invariably show diffuse cytoplasmic staining, varying significantly in intensity. At low staining intensity a ROS1 rearrangement might not be found by FISH as more cases stain positive by immunohistochemistry than are amplified by FISH. An additional issue is expression of ROS1 in normal reactive pneumocytes and

Nonsmall-Cell Cancers of the Lung: Pathology and Genetics

85

macrophages. Current guidelines suggests screening for ROS1 by immunohistochemistry and subsequent confirmation of immunohistochemically positive cases by FISH or any molecular method.

PD-L1 PD-L1 expression on tumors cell is one of the mechanisms of immune evasion, since this inhibits functional activity of cytotoxic lymphocytes which will then not attack tumor cells. For NSCLC, PD-L1 expression by immunohistochemistry is the first predictive biomarker for checkpoint blocking immunotherapy. Interpretation of PD-L1 immunohistochemistry requires familiarity with expression patterns, since PD-L1 is expressed on dendritic cells, macrophages, mast cells, T and B lymphocytes, endothelial and tumor cells. Currently, there are no universally accepted test protocols. Results may be different depending on the used antibody, thresholds for calling a test positive have not been universally validated and to what extent immunostaining of immune cells rather than tumor cells needs to be taken into consideration has not been clarified. This is exemplified by the observation that of patients with negative PD-L1 staining by immunohistochemistry who received anti-PD-1/PD-L1 treatment, as many as 9% will still have a clinical response. This may be due to heterogeneity in PD-L1 expression, with some subclones completely negative while others strongly positive. A form of sampling bias may also be involved: PD-L1 immunostaining will as a rule be performed on the sample on which the initial diagnosis was made. Chemotherapy may have affected PD-L1 expression, which might be resolved by taking a biopsy of the recurrent tumor. This is, however, not always feasible.

Other Emerging Biomarkers New targeted therapies gave encouraging results in early phase clinical trials for NSCLC patients with BRAF, HER2, or exon 14 MET mutations, and RET or NTRK rearrangements.

Development of Resistance Although treatment targeting EGFR, ALK, and ROS1 has proven a major step forward in lung cancer management, in most patients at some point in time a relapse occurs. Biopsy of the recurrent lesion, especially after tyrosine kinase inhibitors targeting EGFR, has provided insight into resistance mechanisms. Apparently, under the selection pressure of targeted therapy, drug sensitive tumor cells undergo apoptosis or at least become quiescent, while other tumor cells survive and continue to proliferate. The mechanisms determining resistance are not yet fully understood but three mechanisms have been proposed. Firstly, the target gene of the inhibitor may acquire an additional mutation, which reduces the binding affinity of the drug to the target, or may be amplified. After treatment with EGFR tyrosine kinase inhibitors (gefitinib, erlotinib, and afatinib), the most frequent resistance mechanism is a second mutation in exon 20 (p. T790M, occurring in  50% of secondary resistant cases). This mutation changes protein conformation at the ATP binding site, which is also the EGFR TKI binding site, resulting in constitutive activation. Remarkably, the proportion of tumor DNA carrying T790M is usually lower than that of the initial driver mutation. When the fraction of tumor cells is close to the detection threshold and the initial driver mutation was barely detectable, in such a situation it may be difficult to exclude a T790M. The relevance of rebiopsy after a recurrence is underscored by the development of a drug (osimertinib) specifically against EGFR with the T790M mutation. Rebiopsy should be included as standard of care in a setting of targeted therapy. Secondly, activation of alternative pathways such as mutation or amplification of ERBB2 or MET may bypass the originally targeted pathway. Thirdly, histological transformation, for example, of NSCLC to small cell lung cancer or epithelial mesenchymal transition may develop (Fig. 2). As ALK targeting treatment is more recent, less is known about the mechanism(s) involved in the development of secondary ALK resistance. Treatment with the first generation ALK inhibitor crizotinib has resulted in secondary resistance and in about 30% of the cases a second acquired mutation or occasional amplification have been found (Fig. 3). Even against second generation ALK inhibitors resistance develops and in such cases in 45% additional ALK resistance mutations have been reported. Further elucidation of mechanisms involved in development of resistance is important. To this end it is important to obtain tissue samples at the time of disease progression. Liquid biopsies and ever more sensitive sequencing methods will help to better document mechanisms of resistance to targeted therapies in lung adenocarcinoma.

Prospective Vision Molecular analysis of a tumor tissue sample has traditionally been highly informative but to obtain a tissue sample usually requires an invasive procedure. A less invasive method is taking a blood sample (what is now customarily called a liquid biopsy) as in the circulation tumor cells but also free circulating DNA and RNA are present. Analysis of a blood sample is rapid, precise, easy to interpret and might provide information relevant for therapeutic management of the patient. The quantity of circulating free DNA is low, which currently limits extensive testing and the liquid biopsy momentarily is used to obtain information on specific molecular events such as EGFR T790M mutation, which when found has direct impact on therapy choice. Isolation of circulating tumor cells requires sophisticated equipment and low numbers of circulating cells render clinical application less informative. It remains to be seen which position the liquid biopsy will ultimately will take in the molecular analysis toolbox.

86

Nonsmall-Cell Cancers of the Lung: Pathology and Genetics

EGFR TKI RESISTANCE MECHANISMS

Gene amplification T790M Tertiary mutation Mutation

PI3CA

PTEN

AKT mTOR

NF-Kβ

Cell membrane

Cytoplasm

Proliferation, Survival

Nucleus

Off target down stream

On target

Off target by pass track

Morphologic change SCLC EMT Fig. 2

EGFR resistance.

Gene amplification

ALK TKIRESISTANCE MECHANISMS, possibly drug related

One mutation Two mutations Mutation downstream Uncertain relevance

PI3CA

PTEN

AKT mTOR

Cell membrane

Cytoplasm

Nucleus

On target (in ~30-45%)

TP53

Proliferation, Survival Off target down stream

Off target by pass track

Morphologic change SCLC EMT

Fig. 3

ALK resistance.

In an immunotherapeutic perspective, testing for expression of PD-L1 is gradually entering into clinical practice. Although this immunohistochemical test currently provides the best information, it is likely that more discriminative testing modalities will be developed. As the mutation rate in NSCLC is high, many neo-antigens will be generated driving an antitumor immune response. A surrogate marker for response to PD-L1 blocking therapy may be the tumor mutational burden, which can also be determined on a liquid biopsy. It is likely that more markers will emerge to provide a more comprehensive and specific appreciation of the antitumor immune status of individual patients, to be used in decisions on modalities of immunotherapy. In the future, pathologists will be required to provide detailed three dimensional assessment of tumor morphology and the host response to the tumor but also information on the mutation status of the tumor tissue, allowing clinicians to target molecular aberrations with high specificity as well as choosing immunotherapeutic modalities with the highest likelihood of response.

Nonsmall-Cell Cancers of the Lung: Pathology and Genetics

87

Further Reading Bubendorf, L., Lantuejoul, S., de Langen, A.J., et al., 2017. Nonsmall cell lung carcinoma: Diagnostic difficulties in small biopsies and cytological specimens. European Respiratory Review 26, 17007. Campbell, J.D., Alexandrov, A., Kim, J., et al., 2016. Distinct patterns of somatic genome alterations in lung adenocarcinomas and squamous cell carcinomas. Nature Genetics 48, 607–616. Gainor, J.F., Dardaei, L., Yoda, S., et al., 2016. Molecular mechanisms of resistance to first- and second-generation ALK inhibitors in ALK-rearranged lung cancer. Cancer Discovery 6, 1118–1133. Jamal-Hanjani, M., Wilson, G.A., et al., 2017. Tracking the evolution of non-small-cell lung cancer. The New England Journal of Medicine 376, 2109–2121. Macintyre, G., Ylstra, B., Brenton, J.D., 2016. Sequencing structural variants in cancer for precision therapeutics. Trends in Genetics 32, 530–542. Sacher, A.G., Komatsubara, K.M., Oxnard, G.R., 2017. Application of plasma genotyping technologies in non-small cell lung cancer: A practical review. Journal of Thoracic Oncology 12, 1344–1356. Schallenberg, S., Merkelbach-Bruse, S., Buettner, R., 2017. Lung cancer as a paradigm for precision oncology in solid tumors. Virchows Archiv 471, 221–233. Swanton, C., Govindan, R., 2016. Clinical implications of genomic discoveries in lung cancer. NEJM 374, 1864–1873. Thunnissen, E., Unger, M., Flieder, D.B., 2013. Epidemiological and clinical aspects of lung cancer. In: Hasleton, P., Flieder, D.B. (Eds.), Spencer’s pathology of the lung, 6th edn. Cambridge University Press, Cambridge, pp. 945–1003. Thunnissen, E., Allen, T.C., Adam, J., 2018. Immunohistochemistry of pulmonary biomarkers: A perspective from members of the Pulmonary Pathology Society. Archives of Pathology & Laboratory Medicine 142, 408–419.

Obesity and Cancer: Epidemiological Evidence Yikyung Park, Washington University School of Medicine, St. Louis, MO, United States © 2019 Elsevier Inc. All rights reserved.

Glossary Abdominal obesity Also known as central obesity, is the status that excessive body fat is accumulated around the stomach and abdomen. Waist circumference is commonly used as an indirect measurement of abdominal obesity. Body mass index (BMI) Defined as body weight in kilograms divided by the square of height in meters (kg/m2). For adult over 20 years old, normal or healthy weight is defined as BMI 18.5–24.9, overweight is BMI 25–29.9, and obesity is BMI  30. The obesity category is further classified as class I (BMI 30–34.9), class II (BMI 35–39.9), and class III (BMI  40) obesity. Obesity Status of abnormal or excessive fat accumulation that may impair health.

Obesity Definition and Measurements Body Mass Index Obesity is defined as excessive fat accumulated in the body. A most commonly used tool to assess obesity is the body mass index (BMI), a simple index of weight-for-height. It is defined as a body weight in kilograms divided by the square of height in meters (kg/ m2). In 1832, Adolphe Quetelet (1796–1874), a Belgian astronomer, mathematician, and statistician, found that normal body weight in kilograms was proportional to the square of the height in meters in adults. In 1972, Ancel Keys found that the Quetelet index (kg/m2) was strongly correlated with body adiposity measured by skinfold thickness and body density measurements in men in Europe, United States, South Africa, and Japan. Later studies also found that BMI was strongly correlated with more direct measures of body fat obtained from bioelectrical impedance, densitometry (underwater weighing), dual energy X-ray absorptiometry, computed tomography, or magnetic resonance imaging. Since Keys renamed the Quetelet index to the BMI, BMI has been the most practical, inexpensive, and easy-to-measure method to screen for weight category. For adult over 20 years old, weight categories recommended by the World Health Organization (WHO) are underweight (BMI < 18.5), normal or healthy weight (BMI 18.5–24.9), overweight (BMI 25–29.9), and obesity (BMI  30). The obesity category is further classified as class I (BMI 30–34.9), class II (BMI 35–39.9) and class III (BMI  40) obesity. Studies found that for the same BMI, percent of body fat differs by sex, age, and race/ethnicity. It has been debated whether the cut-off points for overweight and obesity are lower for Asian populations. For reporting purposes to facilitating international comparisons, it is recommended to use the additional cut-off points of 23, 27.5, 32.5, and 37.5 kg/m2. For children and adolescents, overweight and obese are based on age- and sex-specific growth standard. WHO definition for overweight and obesity in children under 5 years of age is weight-for-height greater than 2 and 3 standard deviations, respectively, above WHO Child Growth Standards median. For children aged between 5 and 19 years, overweight is BMI-for-age greater than 1 standard deviation above the WHO Growth Reference median, and obesity is greater than 2 standard deviations above the WHO Growth Reference median. BMI has been widely used to examine the relationship between body fatness and diseases, including cancer, in numerous epidemiologic studies. Because people tend to over-report their height whereas underreport their body weight, BMI calculated from self-reported height and weight in observational studies may be lower than their actual BMI. However, the difference between self-reported BMI and measured BMI was negligible and did not introduce substantial bias in epidemiologic evaluation of body fatness and cancer association in large epidemiologic studies.

Waist Circumference Waist circumference is also widely used as an indirect measurement of body fatness, particularly for central or abdominal obesity. Waist circumference is typically at the natural waist, just above hipbones. However, some studies measured at umbilicus or naval. People tended to underreport their waist circumference, in part, due to difficulties in measuring circumference around the waist. Waist circumference is positively correlated with visceral adipose tissue that is more metabolically active than subcutaneous fat. A consensus on the cut-off points for high risk for metabolic complications (e.g., diabetes, hypertension, cardiovascular diseases) is > 102 cm in men and > 88 cm in women. However, lower cut-off points have been proposed for Asians.

Imaging-Based Methods There are several imaging methods that assess body fatness more accurately than BMI. Bioelectrical impedance analysis (BIA) measures body fatness as well as lean mass by assessing the impedance or resistance to a small electric current passed across body tissues. The greater the fatty tissue, the greater the resistance to the current. However, the bioelectrical impedance analysis

88

Encyclopedia of Cancer, 3rd edition, Volume 3

https://doi.org/10.1016/B978-0-12-801238-3.65037-7

Obesity and Cancer: Epidemiological Evidence

89

depends on individual’s hydration status. The dual-energy X-ray absorptiometry (DXA) measures bone mineral density as well as fat and fat-free mass. Magnetic resonance imaging (MRI) and computed tomography (CT) can assess total body fat as well as visceral and subcutaneous adipose tissue, separately. However, these imaging-based methods are expensive and not readily available in clinics and studies. Also, there is no standardized visceral adipose tissue assessment method used across studies.

Prevalence of Obesity The global prevalence of obesity tripled between 1975 and 2016 and increased in all age groups and every country. Agestandardized prevalence of obesity (BMI  30) increased from 3.2% in 1975 to 10.8% in 2014 in adult men, and from 6.4% to 14.9% in adult women. Overweight and obesity has increased not only high-income countries but also in low- and middleincome countries, particularly in urban areas. In 2016, 671 million adults were obese, and 1.3 billion adults were overweight while 125 million and 213 million children and adolescents aged 5–19 years were obese and overweight, respectively. If this trend continues, child and adolescent obesity is expected to surpass underweight by 2022, and one out of five adults in the world will be obese by 2025.

Obesity and Cancer In 2016, the International Agency for Research on Cancer (IARC)WHO convened a Working Group to review extensive evidence on the effect of obesity on cancer risk. The Working Group reviewed more than 1000 epidemiologic studies and also experimental studies and concluded that excess body fatness causes cancer of the esophagus (adenocarcinoma), gastric cardia, liver, gallbladder, pancreas, colon and rectum, kidney, thyroid, breast (postmenopausal), endometrium, and ovary as well as meningioma and multiple myeloma. Another independent group of investigators also conducted an umbrella review of 95 meta-analyses that reported the association between body fatness and cancer in cohort studies. After a rigorous evaluation for strength and validity of reported associations in meta-analyses, the umbrella review also made similar conclusions that current evidence supports the positive association of obesity with esophageal adenocarcinoma, multiple myeloma and cancers of gastric cardia, biliary tract system, pancreas, colon, rectum, kidney, ovarian, endometrium (premenopausal women), and breast (postmenopausal) cancer. Similarly, recent reports from the World Cancer Research Fund that evaluate and summarize a body of evidence on obesity in relation to cancer concluded that obesity increased the risk of esophageal adenocarcinoma, gastric cardia, colorectal, gallbladder, pancreas, liver, kidney, thyroid, endometrial, postmenopausal breast, and ovarian cancer and multiple myeloma.

Head and Neck Cancer Head and neck cancer includes cancers in oral cavity, pharynx, and larynx. Earlier case-control studies suggested that obesity was inversely related to the risk of head and neck cancer, particularly among smokers and alcohol drinkers. However, case-control studies have significant methodological limitations such as selection bias and reverse causality, which may lead to an inverse association between obesity and risk of head and neck cancer. Healthier patients were more likely to participate in a case-control study, and obese patients may already have significant weight loss before cancer diagnosis, thus have lower BMI at the study entry and misclassified as having normal weight. Also, given that smoking is a strong risk factor for head and neck cancer, and smokers tended to have lower BMI than nonsmokers, the inverse association observed in smokers suggested residual confounding (i.e., incomplete adjustment) by smoking. Later, a large pooled analysis of 20 prospective cohort studies found that obesity was associated with modestly increased risk of head and neck cancer in never smokers. Every five unit increase in BMI was associated with 15% increased risk of head and neck cancer in never smokers. There was no association between BMI and head and neck cancer in former or current smokers. In addition, abdominal obesity assessed by waist circumference was related to a higher risk of head and neck cancer in never and former smokers.

Esophageal Cancer Esophageal cancer is the six leading cause of cancer death in the world. The most common types of esophageal cancer are squamous cell carcinoma that arises from the epithelial cells lining of the esophagus and adenocarcinoma that arises from glandular cells in the lower third of the esophagus. There is strong evidence that obesity increases the risk of esophageal adenocarcinoma, but not squamous cell carcinoma. Obese individuals had two to threefold increased risk of esophageal adenocarcinoma compared with normal weight individuals. It is hypothesized that obesity increases intragastric pressure, which subsequently relaxes the lower esophageal sphincter, exposing the lower esophagus to gastric acid and increasing the risk of gastro-esophageal reflux disease. Thus, the effect of obesity on esophageal cancer is mediated by gastro-esophageal reflux disease. However, a positive association between obesity and esophageal cancer was found in people with a history of gastro-esophageal reflux disease as well as those without the disease, suggesting an independent effect of obesity on esophageal cancer.

90

Obesity and Cancer: Epidemiological Evidence

In addition, waist circumference and waist-to-hip ratio are related to a significantly increased risk of esophageal cancer. Every 10 cm increase in waist circumference is related to approximately 80% increased the risk of esophageal adenocarcinoma. Increase in waist-to-hip ratio (per 0.1 unit) was associated with a twofold increase in esophageal adenocarcinoma risk.

Stomach Cancer Stomach (gastric) cancer is the third leading cause of cancer death in the world and more common in eastern and central Europe, Asia, and central and South America. The effect of obesity on gastric cancer differs by subtypes defined by the tumor locationdcardia and noncardia. The cardia region is the top of the stomach, close to the esophagus, and the noncardia region is the main area of the stomach. Noncardia stomach cancer is associated with Helicobacter pylori infection. A body of evidence supports that obesity increases the risk of stomach cancer in cardia and is not related to the risk of stomach cancer in noncardia. Obesity is related to about 1.5- to 2-fold increase in the risk of cardia stomach cancer. The positive association between obesity and cardia gastric cancer was consistently found in studies conducted in Asia, Europe, and North America. Interestingly, emerging evidence suggests that adolescent obesity is associated with an increased risk of noncardia stomach cancer in mid-life. Compared with people with normal weight at age 16–19 years, people who were obese at age 16–19 years had 1.8 times higher risk of noncardia stomach cancer in later adulthood in a cohort of over 1.7 million Israeli men and women who underwent health examinations at the age of 16–19 between 1967 and 2002.

Colorectal Cancer An extensive body of literature showed that body fatness was associated with increased risk of colorectal cancer in both men and women. There is a clear dose–response relationship between BMI and colorectal cancer. Compared with normal weight individuals, overweight individuals had 10%–15% increased risk of colorectal cancer, and obese individuals had approximately 30% increased risk of colorectal cancer. The association between obesity and colorectal cancer is stronger in men than in women. The positive association with obesity was found in all subtypes by tumor locationsdproximal and distal colon and rectal cancer. Abdominal (central) obesity measured by waist circumference is also related to significantly increased risk of colorectal cancer. Every 10 cm increase in waist circumference is related to approximately 6%–15% increased risk of colon cancer and 2%–6% increased risk of rectal cancer. Weight gain during adulthood is also associated with higher risk of colorectal cancer. Five kilograms increased in weight gain during adulthood was related to a 6% increased risk of colon cancer. Accumulating evidence also indicates that being overweight or obese at adolescence and young adulthood are associated with an increased risk of colorectal cancer in later life.

Liver Cancer Liver cancer is the second leading cause of cancer deaths and the fifth most common cancer in men and the ninth in women in the world. Hepatitis B or C virus infection and heavy alcohol use are well-known risk factors for liver cancer. Causes of liver cancer differ among populations and geographical locations. However, convincing evidence exists for the harmful effect of obesity on liver cancer in all populations and geographic regions, including Asia, Europe, and America. In general, obesity almost doubles the risk of liver cancer. Compared with normal weight individuals, obese individuals have 1.5- to 2-fold increased risk of liver cancer. The positive association between obesity and liver cancer is stronger in men than in women. A recent large pooled analysis of 14 US based prospective cohort studies found the positive association between obesity and liver cancer in people with sera-negative for hepatitis, but not in people with sera-positive for hepatitis, suggesting the role of obesity in liver cancer may differ by viral hepatitis infection status. Waist circumference, marker of central obesity, was also related to a higher risk of liver cancer. A recent study examining BMI measured at age 17–19 years in Swedish men in 1969–96 found that the late adolescence BMI was related to the risk of liver cancer in later life. Compared with normal weight (BMI 15.5–22.5), overweight and obesity in late adolescence were associated with 1.5and 3.5-fold, respectively, increased risk of liver cancer in later adulthood.

Gallbladder Cancer Although gallbladder cancer is rare, it is the most common tumor among biliary tract cancers. Earlier studies, although small, suggested that body fatness was associated with increased risk of gallbladder cancer. Recently, a large pooled analysis of prospective cohort studies found a clear link between obesity and gallbladder cancer. Compared with normal weight, overweight and obesity were related to 27% and 64%, respectively, increased risk of gallbladder cancer. Central obesity measured by waist circumference was also positively associated with risk of gallbladder cancer. Every 5 cm increase in waist circumference was related to 9% increased risk of gallbladder cancer. Obesity at adolescence or young adulthood (18–21 years) was also associated with higher gallbladder cancer risk in later life.

Obesity and Cancer: Epidemiological Evidence

91

Pancreatic Cancer The increased risk of pancreatic cancer in obese men and women is well established. Obese individuals are more likely to be diagnosed with pancreatic cancer at younger age. There is a clear dose–response relationship between BMI and the risk of pancreatic cancer. Compared with normal weight people, obese people had 1.5–2 times higher risk of pancreatic cancer. Also, abdominal (central) obesity is associated with an increased risk of pancreatic cancer. Every 10 cm increase in waist circumference was related to approximately 23% increased risk of pancreatic cancer. The link between obesity and pancreatic cancer was also confirmed in a study using the Mendelian randomization approach that eliminated confounding by diabetes and other metabolic risk factors. When obesity-associated single-nucleotide polymorphisms (SNPs) found in the large genome-wide association studies were used as an unconfounded marker of obesity (i.e., genetic instrument for BMI), the risk of pancreatic cancer increased by 34% per 4.6-unit increase in BMI. Studies also have found that obesity in late adolescence and young adulthood was associated with increased risk of pancreatic cancer in late life. Every five unit increase in young adulthood BMI was associated with 18% increased risk of pancreatic cancer in later life.

Lung Cancer The role of obesity in lung cancer is inconsistent. Many studies reported that higher BMI was associated with lower risk of lung cancer. However, the inverse association between BMI and lung cancer was likely due to residual confounding (i.e., incomplete adjustment) by smoking in the analysis. Smokers who are at high risk for developing lung cancer tend to have lower BMI than nonsmokers of the same age and sex. Therefore, obesity may appear to be protective against lung cancer. When the obesity and lung cancer relation was examined in smokers and nonsmokers, separately, obesity was not associated with lung cancer in nonsmokers but was related to a lower risk of lung cancer in smokers. A Mendelian randomization study of obesity and lung cancer using a genetic score for BMI as instrumental variable to eliminate reverse causation and reduce confounding found that the genetic risk score for obesity was associated with increased risk of lung cancer. In addition, waist circumferenceda proxy for abdominal obesitydwas positively associated with lung cancer. Each 10 cm increase in waist circumference was associated with a 10% increased risk of lung cancer in both smokers and nonsmokers. Indeed, a genetic risk score for obesity in the GAME-ON consortium was positively associated with lung cancer, suggesting that the observed inverse associations with BMI may be noncausal.

Melanoma and Nonmelanoma Skin Cancer The association between obesity and melanoma is less clear. Some studies found that obesity was related to an increased risk of melanoma, particularly in men, but not in women, while others found no association between obesity and melanoma. Sun exposure and skin type are important risk factors for melanoma, but most studies did not adjust for these factors, raising concern about residual confounding (i.e., incomplete adjustment) in the analyses. It has been suggested that obese people tended to spend less time outdoors and avoid sunbathing compared with normal weight people. Studies that adjusted for sunlight exposure observed a positive association between obesity and risk of melanoma in women. In contrast, although limited, studies suggested that obesity was related to a lower risk of nonmelanoma skin cancer, including squamous cell carcinoma and basal cell carcinoma, even after controlling for sun exposure or UV radiation susceptibility. More studies with measures of sun exposure or sun-seeking behaviors are needed to investigate the role of obesity in melanoma and nonmelanoma skin cancer.

Breast Cancer Obesity is an established risk factor for postmenopausal breast cancer in women. Five unit increase in BMI increases risk of postmenopausal breast cancer by approximately 15%. The effect of obesity on breast cancer is stronger in postmenopausal women who never used the exogenous hormones (i.e., hormone replacement therapy) than those who used. The effect of obesity on breast cancer also differs by tumor hormone receptor status (i.e., estrogen and progesterone receptor). Obesity is associated with 1.5to 2-fold increased risk of both estrogen and progesterone receptors positive breast cancer, but not with both estrogen and progesterone receptors negative breast cancer. Weight gain during adulthood or after menopause also increased risk of postmenopausal breast cancer, especially among women who never used hormone replacement therapy. On the other hand, obesity is related to a lower risk of premenopausal breast cancer. The inverse association with premenopausal breast cancer seems to be restricted to estrogen receptor-positive breast cancer. Triple-negative breast cancer that lacks expression of estrogen and progesterone receptors and human epidermal growth factor receptor 2 (HER2) tended to be diagnosed at younger age and have aggressive disease course. Several studies have shown that higher BMI and greater waist circumference are associated with a two to three times increased risk of triple-negative breast cancer among premenopausal women. Interestingly, accumulating evidence suggests that obesity is related to an increased risk of breast cancer in men as well. Studies found that obesity was associated with approximately 30% higher risk of breast cancer in men. Also, obesity during adolescence or early adulthood was related to an increased risk of breast cancer in men.

92

Obesity and Cancer: Epidemiological Evidence

Cervical Cancer Cervical cancer is the fourth most common cancer in women in the world and is caused by persistent infection of human Papilloma Viruses (HPV). However, the majority of women with HPV do not develop cervical cancer, and other environmental factors are required for cancer to develop. Cervical cancer screening is effective in detecting precancerous lesions that can be treated and cured before they develop into cancer. The association between obesity and cervical cancer is not clear. Some studies found a weak but significantly positive association between obesity and cervical cancer while others found no association. Most of these studies did not have cervical cancer screening information that may have confounded the association. Recently, a large retrospective cohort study of over 900,000 women who underwent cervical cancer screening (i.e., cytology and human papillomavirus DNA testing) found that overweight and obese women had an increased risk of cervical cancer. The positive association between obesity and cervical cancer was consistently found in HPV positive women as well as in HPV negative women. Obesity was also related to an increased risk of both subtypesdglandular and squamous. However, the study found that the increased risk of cervical cancer in obese women was likely due to low efficacy of cervical cancer screening in obese women, resulting in underdiagnosis of precancerous lesions, thus increased risk for cervical cancer.

Endometrial Cancer Obesity is a well-established risk factor for endometrial cancer. There is a clear dose-response relationship between BMI and endometrial cancer. Every five unit increase in BMI is associated with 50% increased risk of endometrial cancer. The positive association between obesity and endometrial cancer was found in premenopausal and postmenopausal women and in hormone replacement therapy users and nonusers. The detrimental effect of obesity on endometrial cancer tended to be stronger in women who never used hormone replacement therapy than those who used. Obesity was also related to an increased risk of type I (mostly endometrioid adenocarcinomas) and type II (nonendometrioid such as serous, clear cell, and carcinosarcoma) endometrial cancer. Mendelian randomization study also found that genetic risk scores for BMI was positively associated with the risk of endometrial cancer. Abdominal (central) obesity assessed by waist circumference is also associated with increased risk of endometrial cancer. Every 5 cm increase in waist circumference was related to a 13% higher risk of endometrial cancer risk. Weight gain in adulthood also increases the risk of endometrial cancer, particularly in women who never used hormone replacement therapy. Obesity at young adulthood (age 18–25 years) is also associated with higher risk of endometrial cancer later in life.

Ovarian Cancer Emerging evidence suggests that obesity is related to a modest increase in the risk of ovarian cancer, particularly in premenopausal women. Every five-unit increase in BMI is associated with approximately 10% increase in ovarian cancer risk. The positive association between obesity and ovarian cancer is stronger in premenopausal women than in postmenopausal women. Some studies suggest that obesity is related to certain subtypes of ovarian cancer such as low-grade serous and invasive endometrioid and mucinous tumors. Abdominal obesity assessed by waist circumference is also related to a small increase in the risk of ovarian cancer. Every 10 cm increase in waist circumference was associated with a 6% increase in ovarian cancer. Weight gain in adulthood seems to increase risk of ovarian cancer, especially in postmenopausal women who never used hormone replacement therapy. Interestingly, studies also found that obesity at young adulthood (age 18–29) was associated with higher risk of ovarian cancer at later life.

Prostate Cancer Findings from earlier studies that examined the association between obesity and the risk of prostate cancer were inconsistent. Studies conducted in North America found no or inverse association between obesity and prostate cancer, whereas European studies found a weak but significantly positive association. It was postulated that the differential rate of prostate-specific antigen (PSA) screening between these regions was attributable to the inconsistent findings. PSA screening that resulted in a diagnosis of earlystage less aggressive cancer was common in North America, but not in Europe. Therefore, European studies tended to have advanced and fatal prostate cancer. When the obesity and prostate cancer relation was examined by prostate cancer aggressivenessdnonadvanced, advanced, and fatal prostate cancer in later prospective cohort studies, a clear pattern emerged. Obesity was associated with an increased risk of advanced or fatal prostate cancer. Every five unit increase in BMI was related to approximately 8% increased risk of advanced prostate cancer. The risk of advanced prostate cancer also increased by 12% per 10 cm increase in waist circumference. In contrast, obesity was associated with a lower risk of nonadvanced prostate cancer. Recent studies suggest that the relation of obesity to prostate cancer differs by molecular subtypes of tumor as well. Two studies found that obesity was inversely associated with risk of TMPRSS2:ERG (T2E)-positive prostate cancer, but not in T2E-negative prostate cancer.

Obesity and Cancer: Epidemiological Evidence

93

Bladder Cancer The association of obesity with bladder cancer is less certain. A meta-analysis of prospective cohort studies found a weak but significantly positive association between obesity and bladder cancer. Given that smoking is an established risk factor for bladder cancer, and smokers tended to have lower BMI than nonsmokers, most studies of obesity in relation to bladder cancer may have suffered from residual confounding (i.e., incomplete adjustment) by smoking. A large prospective cohort study found a positive association between waist circumference and bladder cancer, but it was limited to men.

Kidney Cancer There is convincing evidence that obesity increases risk of kidney cancer (mostly renal cell carcinoma). The effect of obesity on kidney cancer is independent of hypertension that is another established risk factor for kidney cancer. In general, obesity is associated with a 1.5- to 2-fold increased risk of kidney cancer in both men and women. Abdominal obesity assessed by waist circumference is also positively related to the risk of kidney cancer. Every 10 cm increase in waist circumference increased risk of kidney cancer by approximately 11%. Weight gain in adulthood is also associated with increased risk of kidney cancer. On the other hand, several studies that examined the association between obesity and cancer survival in kidney cancer patients found a contradicting result. Patients who were obese at the time of kidney cancer diagnosis tended to have better survival compared with patients with normal weight at cancer diagnosis. It was suggested that obese patients were more likely to have indolent tumors, leading to better survival.

Brain Tumors Meningioma is slow-growing brain tumor that arises in the membranes surrounding the brain and the spinal cord. Although a limited number of studies examined obesity in relation to meningioma, they consistently found that obesity was associated with increased risk of meningioma. Overall, compared with normal weight individuals, obese individuals had a 1.5-fold increased risk of meningioma. In contrast, obesity is not associated with risk of glioma, the most common intracranial tumor that starts in the glial cells of the brain or the spine.

Thyroid Cancer Obesity has been consistently associated with modestly increased risk of thyroid cancer. Every five unit increase in BMI was related to a 6% increase in thyroid cancer risk. The positive association between obesity and thyroid cancer was found in both men and women, but it tended to be stronger in men than in women. Obesity was related to an increased risk of all major histological types of thyroid cancer except medullary thyroid carcinoma. Similarly, abdominal obesity measured by waist circumference was associated with a higher risk of thyroid cancer. Emerging evidence also suggested that young adulthood obesity was related to approximately 13% increased risk of thyroid cancer in later life.

Non-Hodgkin Lymphoma Non-Hodgkin lymphoma is a cancer originated from the lymph system, a part of immune system. The most common non-Hodgkin lymphoma subtypes are diffuse large B-cell lymphoma, follicular lymphoma, and chronic lymphocytic lymphoma/small lymphocytic lymphoma. Earlier studies on obesity and non-Hodgkin lymphoma reported inconsistent results. Some studies found a weak to moderate positive association between obesity and non-Hodgkin lymphoma whereas others found no association. When each subtype of non-Hodgkin lymphoma was examined separately, it became clear that obesity was related to an increased risk of diffuse large B-cell lymphoma, but not other subtypes. Compared with normal weight people, obese people had approximately 25% increased risk of diffuse large B-cell lymphoma. Recent studies also suggest that obesity during adolescence or young adulthood is associated with a higher risk of diffuse large B-cell lymphoma in later life.

Multiple Myeloma Multiple myeloma is a rare but fatal malignancy. Mounting evidence indicates that obesity is related to an increased risk of multiple myeloma in men and women. A large pooled analysis of 20 prospective cohort studies found that obesity was associated with approximately 20% increased risk of multiple myeloma, compared with normal weight. The positive association of obesity with multiple myeloma seems to be stronger for obesity at adolescence or young adulthood (age 18–21 years). Waist circumference was also related to a higher risk of multiple myeloma, suggesting the harmful effect of abdominal obesity.

Leukemia The role of obesity in leukemia is less clear. Limited evidence suggests that obesity is associated with a modestly increased risk of leukemia. The positive association with obesity was found for all leukemia subtypes such as acute myeloid leukemia, acute lymphoblastic leukemia, chronic myeloid leukemia, and chronic lymphocytic leukemia.

94

Obesity and Cancer: Epidemiological Evidence

Obesity at Adolescence or Young Adulthood and Cancer Emerging evidence suggests that excess body weight in early life increases risk of cancer in later life, particularly those known to be related to obesity in late adulthood. Due to the methodological challenges in ascertaining height and body weight in childhood, adolescence, and young adulthood and following participants for several decades for cancer outcomes, a limited number of studies examined the association between obesity in early life and cancer risk in later life. Some studies linked historical mandatory health examination data (e.g., for military service) to a national cancer registry data, and others collected self-reported recalled height and weight at early life period. Although limited, studies consistently found that obesity at adolescence or young adulthood was associated with increased risk of stomach, colorectal, liver, gallbladder, pancreas, and thyroid cancer and diffuse large B-cell lymphoma and multiple myeloma in late adulthood. In contrast, obesity at early life was related to a lower risk of both premenopausal and postmenopausal breast cancer.

Weight Gain During Adulthood and Cancer On average, young adults gain approximately 0.5 kg per year, which is translated into approximately 10 kg weight gain or 3.1–3.5 BMI increase during 18 years. Most people move into a higher BMI category by middle age. People who experienced early and rapid weight gain at young adulthood are more likely to continue to gain weight throughout the adulthood. The weight gain during adulthood is mostly due to increase in body fat, particularly around the waist. It is postulated that weight gain during adulthood, which can be measured easily, is a good surrogate of body fatness. A meta-analysis of studies of adult weight gain and cancer found that adult weight gain was associated with postmenopausal breast, endometrial, and ovarian cancer and cancers in colon and kidney. The increased risk of female cancers tended to be higher among hormone replacement therapy nonusers than among users. In general, every 5 kg weight gain during adulthood was related to approximately 10% increased risk of cancers.

Intentional Weight Loss and Cancer A body of literature provides convincing evidence that obesity increases risk of numerous cancers. However, whether or not intentional weight loss reduces cancer risk is unclear. It is, in part, due to the difficulties in intended weight loss and long-term maintenance of weight loss in large observational studies. Therefore, bariatric surgery patients who had significant and sustained weight loss represent a unique population to examine the effects of long-term intentional weight loss on cancer risk. Studies found that intentional weight loss by bariatric surgery in morbidly obese individuals was related to a lower risk of cancer. Compared with people with severe obesity but no bariatric surgery, people who received bariatric surgery had approximately 30% lower risk of cancer overall. The risk reduction was more pronounced for obesity-related cancers. There was a 40% risk reduction for postmenopausal breast cancer, 60% reduction for endometrial cancer, 40% reduction for colon cancer, and 50% reduction for pancreatic cancer in people who underwent bariatric surgery compared with those who did not.

Obesity Paradox in Cancer Survival In contrast to the harmful effect of obesity on most cancers, some studies found that obesity was related to a lower risk of cancer death among cancer survivors. This obesity paradox (i.e., obese people live longer) was suggested in some studies of survivors of colorectal cancer, renal cell carcinoma, diffuse large B-cell lymphoma, and melanoma. Several methodological limitations in studies were proposed to explain the observed inverse association between obesity and cancer survival. First, residual confounding (i.e., incomplete adjustment) due to inadequate or unmeasured confounders such as smoking, comorbidities, socioeconomic status, and lifestyle factors may cause a spurious positive association between BMI and cancer survival. Second, given that weight loss often precedes cancer diagnosis and is associated with an increased risk of mortality in cancer survivors, reverse causation may have occurred in the analysis. The third is a collider stratification bias that occurs in data analysis stratified by a variable affected by exposure and outcome. For example, obese people may have developed cancer due to obesity or smoking whereas normal weight people developed cancer due to smoking in the absence of obesity. Therefore, obese cancer patients are less likely to be smokers, while nonobese cancer patients are more likely to be smokers. Because smoking is a strong risk factor for mortality, obese patients appear to have lower mortality risk than nonobese patients in the analysis of cancer survivors. In addition, although BMI is the most commonly used measurement to define obesity, BMI may not well represent body fatness in cancer survivors who often experienced changes in body weight and body composition. Studies that measured body fat mass and muscle mass using computed tomography images found that overweight cancer patients tended to have a normal body composition whereas normal weight cancer patients had low muscle mass, similar to underweight. Thus, cancer patients with normal weight at diagnosis may have different body composition compared with normal weight noncancer individuals. The common BMI category may not applicable to cancer patients. Lastly, there are significant heterogeneities in study design and methods, including time of BMI ascertainment (e.g., few months to years before or after cancer diagnosis), adjustments of potential confounders, and

Obesity and Cancer: Epidemiological Evidence

95

duration of follow-up. These limitations in current literature contribute to inconsistent findings. More studies are needed in this emerging field.

Biological Mechanism There are several biological mechanisms that may explain links between obesity and increased risk of cancer.

Sex Hormones Estrogen has been hypothesized to play a role in mostly female cancers such as breast, endometrial, and ovarian cancer. Estrogen is predominantly produced in the ovary, but after menopause, it is produced primarily in adipose tissue through a cellular pathway involving the enzyme aromatase that converts androgens to estrogen. Postmenopausal obese women have higher concentration of total and free circulating estrogen levels compared with normal weight women. Women with higher concentration of circulating various estrogen metabolites and lower concentration of sex hormone binding globulin had an increased risk of postmenopausal breast cancer. Animal and experimental studied showed that estrogen promotes tumor growth; stimulates cellular proliferation; inhibits apoptosis via estrogen receptor-a agonism; and induces vascular endothelial growth factor and angiogenesis. Estrogen is also suggested to promote free-radical mediated DNA damage, genetic instability, and mutation. Selective estrogen receptor modulators (SERMs) such as tamoxifen and raloxifene and aromatase inhibitors are effective breast cancer treatments by blocking the effects of estrogen. In men, obesity was related to high estrogen, low testosterone, and low sex hormone binding globulin levels, which led to greater availability of estrogen. The majority of male breast cancer is estrogen receptor positive.

Insulin and Insulin-Like Growth Factor-1 High levels of insulin and insulin-like growth factor (IGF)-1 have been hypothesized to promote the development of cancer. Obesity is positively associated with circulating levels of insulin and IGF-1. Many obese people are insulin resistance and develop type 2 diabetes. Circulating insulin binds to the insulin receptor on the cell surface and promotes hyperactivity of phosphatidylinositol-3-kinase (PI3K)/Akt/mammalian target of rapamycin (mTOR) and mitogen-activated protein kinase (MAPK) signaling pathways that stimulate cell growth and cell proliferation. Metformin that is a type 2 diabetes medication increases insulin sensitivity and activates the AMP-activated protein kinase (AMPK) pathway, thereby counteracting the PI3K/AKT/mTOR signaling pathway. Metformin has been suggested as a chemopreventive agent for cancer prevention.

Chronic Inflammation Obesity is a common cause of chronic inflammation, both systemically and at the tissue level. Obesity is associated with systemic chronic inflammation markers such as C-reactive protein, tumor necrosis factors (TNFs), interleukin (IL)-1b, IL-6, IL-8, and vascular endothelial growth factor. TNF-a, IL-1b, and IL-6 promote tumor growth in mouse models of obesity, and IL-6 and CRP have been associated with the development and progression of breast tumors. Also, chronic inflammation in the tissue microenvironment such as pancreatitis and gallstone are known risk factors for pancreatic and gallbladder cancer, respectively.

Adipokines Adipose tissues, particularly white adipose tissue, also secrete various cytokines, known as adipokines that modulate the chronic inflammatory state. Leptin and adiponectin are the most studied adipokines in relation to cancer and play opposing roles in cancer. Leptin is a proinflammatory adipokine while adiponectin is an antiinflammatory adipokine. Leptin activates cell proliferation and multiple signaling pathways including the PI3K, MAPK, and Janus kinase/signal transducers and activators of transcription (JAK/ STAT) pathways. Leptin can also induce IL-6 production, suggesting its contribution to the systemic inflammation. On the other hand, adiponectin inhibits cell growth, increases apoptosis, and activates AMPK that induces cell cycle arrest and inhibits mTOR activity, therefore plays a protective role in cancer. Obesity is associated with higher leptin level and lower adiponectin level. It is suggested that adipose tissue in tumor microenvironment is also a rich source of pro-inflammatory mediators, stimulating tumor growth, and is more relevant than total body fatness to local tumor development.

Microbiome Emerging studies suggest that microbiome also play a role in obesity and vice versa. The microbiome is microbes such as bacteria, bacteriophage, fungi, protozoa and viruses that live inside and on the human body. Growing evidence indicates that microbiomes in guts predispose individuals to obesity, and obesity also changes the gut microbiota, creating environment prone to

96

Obesity and Cancer: Epidemiological Evidence

carcinogenesis. Studies showed that microbiomes were related to proinflammatory and immunosuppressive responses. It has been hypothesized that the microbiome in gut affects multiple pathways by which obesity acts on cancer.

Burden of Obesity on Cancer Assuming that obesity causes cancer, it is estimated that approximately 4% of all new cancers in the world in 2012 was due to high BMI ( 25 kg/m2). The cancer burden attributable to obesity varied by cancer site in men and women. The population attributable fraction that is a contribution of obesity to a cancer is approximately 30% for esophageal, 10% for colon, 8% for pancreatic, and 18% for kidney cancer in men and 30% for endometrial, 10% for postmenopausal breast, 8% for colon, 8% for pancreatic, and 25% for kidney cancer in women. Also, approximately 10%–13% of cancers in liver, gallbladder, and gastric cardia and 7%–9% of multiple myeloma were attributable to high BMI in both men and women. It is estimated that the global increase in obesity prevalence since 1980s contributed to 25%–32% of obesity-related cancers in 2012. Obesity also attributed to 120.1 million disability-adjusted life-years in 2015, which made the obesity one of the largest contributors to global disability-adjusted life-years. The cancer burden, including the disability-adjusted life-years, due to obesity is likely to increase if the obesity prevalence continues to rise worldwide.

Future Directions Although convincing evidence indicates that obesity increases risk of numerous cancers, several challenges remain to advance the understanding of body fatness and weight management in the cancer control continuum. Accumulating evidence suggests that body fatness in adolescence or early adulthood is associated with increased risk of several cancers at later life. People who experienced excessive weight gain at early life are more likely to keep their excess weight or gain even more weight throughout adulthood. Therefore, body fatness in early life could be an indicator of life-long exposure to obesity. Even though we do not know the biologically relevant time period for obesity to cause cancer, considering a long latency period of cancer, obesity in early life may play a role in carcinogenesis during adulthood, leading to cancer diagnosis at later life. However, there are gaps in the understanding of the effect of obesity at different life course and duration of obesity on cancer risk. In addition, whether intended weight loss at mid-life lowers cancer risk in later life is less clear. More studies are needed to investigate the time course of weight management-either gain or loss- in a life course and its effect on cancer development as well as cancer survival after diagnosis. The role of obesity in cancer survival is the emerging area of interest. The inconsistencies in the current literature on the effect of obesity on survival among cancer patients indicate the need for a comprehensive examination of body composition, including both body fat mass and muscle mass, in relation to cancer survival. However, there are no practical and standardized methods to measure body composition in large-scale observational studies. A practical and economical method to assess body composition is needed to close a significant gap in understanding of the role of body fatness in cancer survival. Future research on the effect of body fatness and body composition after cancer on patients’ overall health and well-being is warranted. With increasing recognition of tumor heterogeneities in cancer, more and more cancers are reclassified into subtypes according to their molecular characteristics. It is hypothesized that the etiology of one molecular subtype of cancer differs from other subtypes. Further research to understand the role of obesity, including its biological pathways, in the etiology of each molecular subtype of cancer may contribute to find more effective cancer prevention strategies, including development of chemopreventive agents and targets for cancer treatment. Given the increasing trend in obesity prevalence worldwide, the burden of obesity on cancer is likely to increase continuously. As obesity is a preventable and modifiable risk factor for cancer, obesity prevention and management can contribute significantly to cancer prevention. More efforts are needed to develop multilevel but personalized and sustainable approaches for a healthy lifestyle that maintains healthy weight and prevents weight gain throughout individual’s life-course.

See also: Breast Cancer: Pathology and Genetics. Cancer-Related Inflammation in Tumor Progression. Endometrial Cancer: Pathology and Genetics. Esophageal Cancer: Pathology and Genetics. Hormones and Cancer. Prevention and Control: Nutrition, Obesity, and Metabolism.

Further Reading Arnold, M., Pandeya, N., Byrnes, G., et al., 2015. Global burden of cancer attributable to high body-mass index in 2012: A population-based study. Lancet Oncology 16 (1), 36–46. Committee on Accelerating Progress in Obesity Prevention, 2012. Accelerating progress in obesity prevention: Solving the weight of the nation. National Academies Press (US), Washington, DC. Iyengar, N.M., Gucalp, A., Dannenberg, A.J., Hudis, C.A., 2016. Obesity and cancer mechanisms: Tumor microenvironment and inflammation. Journal of Clinical Oncology 34 (35), 4270–4276. Khandekar, M.J., Cohen, P., Spiegelman, B.M., 2011. Molecular mechanisms of cancer development in obesity. Nature Reviews Cancer 11 (12), 886–895.

Obesity and Cancer: Epidemiological Evidence

97

Keum, N., Greenwood, D.C., Lee, D.H., et al., 2015. Adult weight gain and adiposity-related cancers: A dose–response meta-analysis of prospective observational studies. Journal of the National Cancer Institute 107 (2), djv088. Kyrgiou, M., Kalliala, I., Markozannes, G., et al., 2017. Adiposity and cancer at major anatomical sites: Umbrella review of the literature. BMJ 356, j477. Lauby-Secretan, B., Scoccianti, C., Loomis, D., et al., 2016. Body fatness and CancerdViewpoint of the IARC Working Group. New England Journal of Medicine 375 (8), 794–798. NCD Risk Factor Collaboration, 2017. Worldwide trends in body-mass index, underweight, overweight, and obesity from 1975 to 2016: A pooled analysis of 2416 population-based measurement studies in 128.9 million children, adolescents, and adults. Lancet 390 (10113), 2627–2642. Park, Y., Peterson, L.L., Colditz, G.A., 2018. The plausibility of obesity paradox in cancer. Cancer Research 78 (8), 1898–1903. Renehan, A.G., Zwahlen, M., Egger, M., 2015. Adiposity and cancer risk: New mechanistic insights from epidemiology. Nature Reviews Cancer 15 (8), 484–498. Schauer, D.P., Feigelson, H.S., Koebnick, C., Caan, B., Weinmann, S., Leonard, A.C., Powers, J.D., Yenumula, P.R., Arterburn, D.E., 2017. Bariatric surgery and the risk of cancer in a large multisite cohort. Annals of Surgery (Epub ahead of print).

Relevant Website https://www.wcrf.org/int/research-we-fund/continuous-update-project-findings-reportsdWorld Cancer Research Fund Continuous Update Project findings and reports.

Oncology Imaging Stefan Delorme and Michael Baumann, German Cancer Research Center (DKFZ), Heidelberg, Germany © 2019 Elsevier Inc. All rights reserved.

Advances in Technology Computed Tomography Computed tomography (CT) continues to be the workhorse in oncology imaging, since it is fast, extremely reliable, and can be performed in patients who are severely ill, suffer from pain, require monitoring of vital functions, or even mechanical ventilation. Following the first publication of the principle in 1990 (Kalender et al., 1990), spiral CT was rapidly introduced, followed by the implementation of multirow detectors. In contemporary scanners, the beam geometry is cone-shaped to cover an array with 64 to > 300 detector rows. The rotation time of the tube now lies below 1 s, in some scanners at 250 ms. As a result, large ranges can be scanned in few seconds time, and thin slices (< 1 mm thickness) can be obtained without significantly increasing the radiation exposure to the patient. Neighboring primary slices will be merged to generate slices in the 3–5 mm range, which are less degraded by image noise than the primary thin slices and are required for assessing soft tissue. The primary image stack is almost isotropic, that is, the voxels (volume elements) have similar diameters in all three dimensions. Therefore, sagittal or coronal images can be reconstructed from the primary data stack. The speed of acquisition permits to obtain images during a defined phase of contrast enhancement following intravenous injection. Dynamic scanning to quantitatively assess the microcirculation is also possible but used less often, because repeated scans will cause significant radiation doses. Radiation dose is an increasing issue even in oncology, since more than half of patients with cancer can be expected to become long-term survivors. CT now offers numerous options for dose reduction: Dose modulation: The tube current needed to avoid significant image degradation by noise depends on the radiation absorption by the tissue in the gantry. The absorption is typically low for the neck (because it is slim) or the mid-chest (containing the lungs) and high for the shoulder region or the abdomen. For dose modulation along the body axis (in z direction), the image noise is measured in real time, and the tube current is adjusted in order to keep the noise in a target range. This can be combined with dose modulation during tube rotation, where the tube current is changed as the tube rotates around the patient. The rationale for this is that in a given area the absorption will depend on the direction of the beam in the transverse plane. For the shoulder region, for example, it will be lower for a sagittal than for a transversal beam direction, where the shoulders, clavicles, and vertebral bodies lie in one line and will all be passed by the same beam. Obviously, the tube current will not be based on image noise (because the image does not yet exist) but rather on the beam intensity directly at the detector. Adjustment of tube voltage and dual-energy CT: The tube voltage determines the maximal beam energy and is around 120 kV in standard diagnostic protocols. However, the optimal voltage is always a tradeoff: The contrast between either bone (that contains calcium) or iodine-based contrast media and soft tissue will increase if the voltage is lowered to 80 kV or even 70 kV. At the same time the radiation dose will decrease significantly, as long as the tube current remains unchanged, and the image noise will increase. In body regions with little radiation absorption, the increase in image noise will be more than compensated by the gain in contrast, resulting in an increase in the contrast-to-noise ratio. Low-kV protocols can be used for angiographic protocols in the head, neck, chest, or abdomen, but also for nodule detection in the lungs. For abdominal CT, however, where the contrast resolution is at a premium, and the X-ray absorption is about the highest in the body, low-kV protocols still play less a role, except for children and very lean adults. Postacquisition data processing: Users of darkroom software for private digital pictures know how image noise can to some degree be reduced without degrading contrast or resolution, and the same is possible for medical images. Nevertheless, the iterative reconstruction algorithms used on CT platforms do more than that. If applied to image data, they are a way of image postprocessing, but advanced algorithms are applied to raw data and are therefore an entirely different way of reconstructing images than conventional, filtered back-projection. They are powerful tools that should be used carefully, and can compensate for the increase in image noise caused by lowering tube current or voltage (Fig. 1). Many radiologists are still reluctant using them, because iteratively reconstructed images often have an unnatural, plastic-like appearance. Nevertheless, it will have to be learned that image aesthetics are at least not the sole criteria to wisely choose the scanning technique. This unconsidered, raw data processing also permits to reduce artifacts due to metal implants or highly concentrated contrast media, for example, in the subclavian or anonymous vein. Dual-energy CT (DECT) is not a new development, but only in the past 10 years has it become routinely applicable. It consists of scanning the same slice with different tube voltages, fully or almost simultaneously. The way of accomplishing this differs by manufacturerdtwo tubes arranged in a 90 degree angle, rapid switching between voltages in a single tube, or separation of the components of the energy spectrum in different layers of the detector ring. The rationale is that tissues or substances in the body (typically soft tissue, fat, and either calcium or contrast medium) not only differ by X-ray absorption (which is the classical principle behind Xray or CT imaging), but that they are different with regard to which part of the X-ray spectrum is more strongly absorbed. If the same region is passed by beams of different energy, or if two regions of the energy spectrum are retrospectively extracted, the contribution of two substances to the absorption may be calculated. This principle has been used since decades for dual-energy X-ray absorptiometry (DEXA), a method for measuring bone density. Here, the dual-energy principle was used to extract the contribution of bone and nonbone to the X-ray absorption. Since for DEXA, only a sagittal beam was used to scan the lumbar spine, the X-ray absorption

98

Encyclopedia of Cancer, 3rd edition, Volume 3

https://doi.org/10.1016/B978-0-12-801238-3.65427-2

Oncology Imaging

99

Fig. 1 CT of the lung reconstructed with standard filtered back projection (A) and raw data-based iterative reconstruction (B). Note the reduced noise artifacts on the iteratively reconstructed images. Iterative reconstruction can be used either to generate images with less noise, or to lower the tube current (and thereby the radiation dose to the patient), ending up with the same contrast-to-noise ratio as it would be with standard filtered back projection in combination with standard tube current.

Fig. 2 Dual energy CT: CT of the kidneys with i.v. contrast medium (A) shows a small, hypodense lesion in the left kidney (arrow). Whether or not there is any contrast uptake in that lesion is crucial for discriminating a neoplasm from for example, a hemorrhagic cyst. Virtual native contrast (VNC) images (B) are retrospectively calculated from the simultaneously acquired Data obtained with different tube voltages, to exclude the attenuation caused by iodine. In this example, the density values on postcontrast and VNC images are identical, so that the suspicion for malignancy is low. Ultrasound confirmed a cyst.

by soft tissue and fat would otherwise have made bone density measurements impossible. In its first years, DECT has been used mainly for enhancing CT angiography, and only recently for oncological questions, particularly for assessing tissues such as bone marrow, where calcium, fat, and soft tissue are present in one voxel and cannot be resolved in space. A very practical, every-day application is calculating virtual native scans (Fig. 2). This means that only a postcontrast scan is obtained, and the dual-energy data are used to remove the attenuation caused by iodine. This obviates separate unenhanced scans and contributes to reducing the radiation dose. The logical continuation of DECT will be spectral CT, where each photon is measured for its energy. Today, only prototypes exist, and where the future clinical role of spectral CT is difficult to foresee.

Magnetic Resonance Imaging Higher field strengths For long, the field strengths were in the range from 0.2 T (particularly with open permanent magnets) to 1.5 T (closed superconducting magnets), with 1.5 T being most frequently used for oncological applications. 3 T magnets are now widely being used, mainly in hospitals or large outpatient facilities that operate them together with 1.5 T scanners. The advantages of 3 T over 1.5 T lies in a higher signal-to-noise ratio and a better quality of morphological images, which is of particular value in studies of the brain and spinal cord as well as in the pelvis. For prostate imaging at 1.5 T, for example, endorectal coils were used in the past to achieve

100

Oncology Imaging

sufficient image quality, which made it an unpleasant procedure that was not happily accepted by patients. Since the introduction of 3 T scanners (and also fundamental improvements with 1.5 T ones), endorectal coils could be abandoned, which strongly increased the acceptance of the method. Nevertheless, increasing the field strength comes at a price. First, the forces to ferromagnetic implants are higher, and safety is even more an issue. Second, the radiofrequency that is transmitted by the body coil is 128 MHz instead of 64 MHz, and the heat deposition is higher than with 1.5 T, so sequence protocols need to be adapted not to exceed the specific absorption rate (SAR) limit. For both reasons, there are increasing compatibility issues with any kind of implants, be they stents, ports, shunts, bony implants with metal rods, and of course active medicinal products like insulin pumps or cochlea implants. Therefore, patients are expected to provide information about any implants already when making their appointment, so the clarification of compatibility issues (including inquiries, web searches, etc.) can be done beforehand. Beyond safety issues, high field strengths may indeed degrade image quality because susceptibility artifacts are more pronounced than with lower field strengths. They will arise either in presence of metal or blood degradation products (where they may indeed improve the detection of small hemorrhages) or at tissue–air interfaces. Here, magnetic field inhomogeneities, if not compensated by “shimming” (geometrically adjustment of the magnetic field) the scanner, may, for example, cause difficulties in spectral fat suppression. Particularly in the lower neck and upper thoracic aperture such problems are common, because the geometry of the tissue–air interface (the skin) is so complex that it cannot be compensated by the shimming procedure, which is always rigid in all space directions. Ultra-high field (UHF) MRI scanners (with e.g., 7 T or more) are so far only experimental. Technical challenges are paramount, some extrapolated from those with 3 T, some entirely new. For the brain, the results may nevertheless be stunning, but only as a result of meticulous refinements of the applied sequences. The potential of UHF MRI lies, for example, in the exploitation of susceptibility effects, imaging nonhydrogen nuclei, chemical characterization of tissue, all not yet ready for routine use. For the body trunk, the protocols are still in their infancy.

Whole-body MRI Classically, MRI was designed for examining regions of the body, covering a field of view along the body axis (z direction) of about 40–50 cm. Modern scanners now come with an array of receive coils (the radiofrequency is transmitted by the “body coil” that is built into the scanner itself) that can cover the body all the way from the vertex to the feet. With this arrangement, the body can be scanned entirely, usually adding multiple coach positions to another and then composing for example, coronal or sagittal images to display the entire body (Fig. 3). Thanks to rapid MR sequences, this can be achieved in 30–45 min time, depending on which image contrasts are needed, and on whether i.v. contrast agents will be used. Obviously, this is a valuable tool for whole-body staging of metastatic disease or hematological neoplasms. Nevertheless, the images are complex, and the workload for the radiologist reading the entire body scan with several weightings and scan directions without missing relevant findings is much higher than with wholebody CT. Without a systematic workflow, relevant findings may easily be missed.

Functional imaging techniques Dynamic contrast-enhanced MRI (DCE MRI) implies repetitive T1-weighted 3D image stacks being acquired over the same region with high temporal resolution before, during and after pump-controlled infusion of a gadolinium-based contrast agent, in order to assess tissue microcirculation. Using pharmacokinetic models, parameters can be extracted that reflect the relative blood volume in tissue, and the exchange rate of the contrast agent between the central and peripheral compartment, the latter consisting of both the terminal capillary intravascular as well as the extravascular space. The exchange rate is determined by both capillary perfusion and trans-capillary substrate exchange. DCE MRI is most frequently used for MR mammography and prostate imaging for discriminating benign from malignant lesions. Nevertheless, since spatial resolution is at a premium in both instances, the temporal resolution will not permit pharmacokinetic calculations. Therefore, the protocol may be termed multiphasic rather than dynamic, and instead of parameters, the shape of the time-intensity curve (e.g., continuously ascending, early peak followed by either a washout or a plateau) will be visually assessed. DCE MRI must be discriminated from dynamic susceptibility contrast MRI (DSC MRI). Here, a sharp contrast bolus is given, and T2*-weighted images are dynamically acquired to measure a drop in signal intensity which is induced by the presence of gadolinium inside and its absence outside the microvessels. Using indicator dilution models, the tissue perfusion can be calculated directly, provided there is no extravasation of the contrast agent. DCS MRI is standard of care for stroke imaging but can also be used for example, low-grade gliomas, where the blood–brain barrier is not disturbed. With diffusion-weighted MR imaging (DWI), a series of magnetic gradients is applied to induce a signal loss in areas where extracellular water molecules are free to move randomly. The strength and duration of the gradients determine (expressed by the socalled B-value) the degree of diffusion-weighting. Since images are basically T2-weighted, the signal intensity of tissue depends on both its T2-signal intensity and diffusion restriction. Based on two or more signal intensities obtained with different Bvalues, the apparent diffusion coefficient (ADC) can be calculated, often represented in ADC maps generated from images obtained with different B-values (Fig. 4). More advanced techniques will account for the complexity of the interstitial space (e.g., diffusion kurtosis imaging) or the differential contribution of tissue perfusion and water diffusion to the signal (using, e.g., the intravoxel incoherent motion (IVIM) model; Le Bihan et al., 1988). Moreover, assessing the movement of water molecules separately for three dimensions may serve to assess the preferential direction of anisotropic structures, such as neural tracts. As a rule, diffusion is restricted in tissues where cells are densely packed, as is often the case in malignant tumors (Chenevert et al., 2000). Although this is not specific, DWI has become an invaluable adjunct in oncologic imaging and can be used either to assist in the differential diagnosis of unclear lesions (e.g., in MRI of the prostate), or, in combination with a suppression of

Oncology Imaging

101

Fig. 3 Native whole-body MRI in a patient with asymptomatic multiple myeloma. Since the longitudinal field of view of clinical MRI scanners is typically 40–50 cm, image stacks are obtained stepwise, as the couch is advanced into the scanner, and then “stitched” together to generate, for example, coronal T1-weighted images as in this example. This patient has a typical, diffuse infiltration pattern with hypointensity of the bone marrow (owing to the replacement of fat by cell-dense tissue) in the vertebrae, the pelvis, the humeri and femora, sparing the epiphyses.

the background signals (from e.g., muscles or fat tissue), to generate whole-body overview images where only diffusion-restricted structures are displayed (Figs. 9 and 12). Such images resemble those from FDG-PET studies (sometimes they are called “poor man’s PET”) and are helpful to attract the reader’s attention to suspicious finding, and then retrieve and further assess them on morphologic images. Susceptibility-weighted MRI (SWI) uses sequences where signal voids arise in presences of differences in magnetic susceptibility, for example, in presence of contrast agents (in DSC MRI, see above), melanin, or blood degradation products. SWI sensitively detects microhemorrhages arising in brain metastases either spontaneously or as a result of treatment. Very typical intratumor susceptibility signals have been found in high-grade gliomas and helped to discriminate them from primary central nervous system lymphoma (Kickingereder et al., 2014). Gadolinium chelates have been used as contrast agents in MRI since the 1980s. They are injected intravenously and are distributed in the intravascular and extracellular space (except for the brain, where they remain intravascular) and renally eliminated, like iodine-based contrast agents that are used for CT or fluoroscopy. One contrast agent (Gd-EOB-DTPA) is partly taken up by hepatocytes and excreted in the bile, and is therefore approved as a liver-specific contrast agent.

102

Oncology Imaging

Fig. 4 Anaplastic astrocytoma in the left cerebral hemisphere, visible on a T2-weighted MR image with free water suppression (A) (“fluid-attenuated inversion recovery,” FLAIR) and a parameter map showing the apparent diffusion coefficient ADC (B). The tumor is T2-hyperintense and shows increased water diffusion compared to the unaffected brain. Functional imaging techniques like diffusion-weighted imaging may assist in both differential diagnosis of suspicious tumors and in the delineation of the gross tumor volume for radiotherapy planning.

Gadolinium chelates are generally well tolerated, but risks are associated with free gadolinium being liberated from the chelator. In the 1990s, it became obvious that scleroderma-like skin changes (nephrogenic systemic fibrosis, NSF) occurred in patients with renal insufficiency who had received high or repeated doses of gadolinium chelates. Thanks to clear guidelines on the use and exclusion criteria for MRI contrast agents, hardly any NSF cases have occurred since. Only recently, however, gadolinium deposits were found in the dentate nucleus of the brain, visible as hyperintensity on T1-weighted, native images, and also proven in post mortem specimens, and have also to be expected elsewhere. There has been a clear association with both the cumulative amount of contrast agent given and also the type of chelators that have been used. Gadolinium is mainly released from linear chelators, whereas macrocyclic chelates are obviously so stable that the amount of liberated gadolinium is negligible. There is a clear move to using macrocyclic agents only, although no clinical sequelae of CNS deposits are known so fardand will be difficult to prove. The consequences for manufacturers may be serious, since some linear contrast agents that have been very successful until now will probably disappear from the market, except for the liver-specific Gd-EOB-DTPA, for which there is currently no macrocyclic alternative. This in mind, there is continuing interest in techniques for tissue characterization, some being in use since long (e.g., spectroscopy, blood oxygen level-dependent (BOLD) neurofunctional MRI), others newly developed (MRI with hyperpolarized oxygen, chemical exchange saturation transfer (CEST) imaging). Mostly, they still require either specialized equipment such as ultra-high field scanners or an intense support by basic scientists or physicists, but some may evolve to become part of clinical protocols in future, as it has happened with diffusion-weighted imaging.

Ultrasound Ultrasound (US) is one of the most underestimated imaging techniques, for numerous reasons, the most important ones being that it requires much skill and dexterity to be carried out, and that it will usually not yield volume-covering image stacks that can easily be reviewed by a second reader. It is sad to say that its being applied by numerous insufficiently educated users has damaged its reputation, triggering further, potentially costly and invasive measures. In the hand of a skilled and trained examiner, ultrasound is a powerful, versatile and fast imaging tool that is capable of solving many issues or at least guide further imaging studies. In the near field, with high frequencies, its spatial resolution is unsurpassed. Its contrast and resolution have been constantly improved over years, a kind of a “silent revolution”. The most striking development in the past years has been the introduction of ultrasound contrast media, along with specific contrast-specific imaging techniques. They consist of gas-filled microbubbles 1–10 mm in size with a soft shell that are suspended in water and injected intravenously. The most frequently used compound contains sulfur hexafluoride with a soft phospholipid shell. Originally, they were developed as signal enhancer for Doppler applications, but along with the official approval that became unusually protracted, the spectral and color Doppler techniques improved so substantially that the substances became almost obsolete for the originally intended use. However, the physics were deeply investigated, to reveal three different mechanisms of interaction between ultrasound waves and microbubbles, depending on the acoustic pressure: At very low intensities, the microbubbles act as simple scatterers and increase the echo intensity of the blooddas originally intended. At high intensities (i.e., those usually applied for B-mode and Doppler ultrasound), the positive and negative pressure forces cause the bubbles to rupture and emit

Oncology Imaging

103

an acoustic signal, an event that is termed a “stimulated acoustic emission” (SAE). An SAE can be detected by the ultrasound probe and will be encoded either in a B-mode or in a color Doppler image. Thus, SAEs are a way to generate contrast-specific images, because their intensity is comparably high, but since they accompany the bubble’s death, they are very short-lived. SAE imaging cannot be performed as a real-time examination, as would be desirable. At low intensities (about 10% of the usual output power), an interesting interaction occurs: Microbubbles have an own resonance frequency at which they expand and contract, dependent on their diameter. In case of resonance between transmitted ultrasound pulse and microbubble, this will vibrate at that frequency. As a happy coincidence, this is the case for microbubbles between 1 and 10 mm in diameter (i.e., that easily flow in the blood stream) and ultrasound in the usual diagnostic frequency range (between 1 and 20 MHz). The vibration is too low in amplitude to emit a signal, but the backscattered echo of the bubble will be distorted in shape when compared to the transmitted signal. In other terms, it will contain higher-frequency components, that is, second or higher harmonics. This is a peculiarity that not shared by backscatter from tissue and is now being used to separate echoes from microbubbles and tissue, using either high-pass frequency filters, sequences of inverted pulses, or both, and generate images that show the pattern of contrast enhancement. Although low-MI imaging (MI ¼ mechanical index) does go along with some bubble destruction, enough bubbles will survive to enable real-time scanning for about 5 min. Other than gadolinium- or iodine-based contrast media used for MRI or CT, ultrasound contrast agents remain strictly intravascular. Once the bubbles perish, the phospholipids are metabolized, and the gas is exhaled via the lungs. The absolute quantity of gas is in the range of few micro liter. Ultrasound contrast media are extremely well tolerated; allergic reactions may occur but are exceedingly rare, and so are nonallergic toxicities. They are contraindicated in advanced cardiac failure because fatalities had been observed in this context, but in absence of a plausible pathomechanism it can be assumed that these were coincident and caused by the disease itself rather than due to the compound. The main application of contrast-enhanced ultrasound (CEUS) is to work up unclear liver lesions, where it has been shown to be diagnostically equivalent to CT or MRI (Fig. 7). Sonographic access provided, CEUS is the most sensitive method to detect even weak perfusion in tissue in anybody region. Ultrasound is an exquisite method to guide a biopsy needle or an ablation device in real time and verify its correct placement, without exposing the physician to ionizing radiation as would be the case with fluoroscopic and sometimes CT guidance. It permits to directly visualize the progress of thermal ablation, and CEUS can be used to examine whether viable parts of the target have remained. If a lesion that was detected with CT or MRI is not visible in ultrasound, because it is for example, isoechogenic, the relevant image stack from the corresponding modality can be matched, using anatomical landmarks and an electronic positioning system attached to the transducer. The device is then guided via a split screen or an image overlay showing both the ultrasound and the CT or MRI images reconstructed in identical planes. Such systems are available either as additional equipment or readily implemented in high-end ultrasound machines. Such fusion-guided procedures are widely performed for example, liver or prostate lesions. Current ultrasound research is focused on, for example, elastography, motion registration, or molecular imaging. Tissue elasticity can be assessed by analyzing the deformation of tissue upon movement of the transducer. More advanced techniques use a “push” pulse that causes tissue deformation and shear waves, both of which can be analyzed using ultra-fast techniques. Elastography may assist in detecting firm lesions that are isoechogenic to their surroundings, or in the differential diagnosis of otherwise unclear lesions, and is being used clinically for example, for assessing thyroid nodules or for detecting breast lesions. Current research is directed at using ultrasound for motion detection in the context of image-guided radiotherapy (IGRT). Ultrasound also offers possibilities for molecular imaging. Microbubbles can, for example, be attached to specific ligands or antibodies. The signal from stimulated acoustic emissions is so strong that every single bubble can be detected. However, such applications are limited to intravascular targets. Another approach is to make use of Bell’s effect: Short laser pulses directed into the tissue will induce heat-induced, pulsatile tissue expansions and thereby the emission of an acoustic signal. Depending on whether monochromatic or spectral lasers are used, such optoacoustic imaging scan can be tailored to targets with known absorption peaks (e.g., melanin, oxygenated or de’oxygenated hemoglobin) or even dyes that are attached to specific ligands. Ultrasound for IGRT or molecular ultrasound imaging are still in their preclinical phase of development.

Positron Emission Tomography The most important developments in positron emission tomography (PET) have been improvements of detector technology, the introduction of hybrid scanners and the development of specific tracers. The principle of PET lies in the detection of two 511 keV photons emitted in opposing directions during an annihilation event from a collision of a positron (an antimatter particle emitted from isotopes like 18-fluorine, 11-carbon, or 68-gallium) and an electron. Positron-emitting isotopes are either chemically bound or chelated with substrates and are so used to map biochemical processes or specific markers in tissue. The main component of a PET scanner consists of a ring detector array that form its gantry. Whenever two coincident photons are registered, their source will be assumed to lie anywhere on the “line of response” connecting the two detector crystals. In modern “time of flight” scanners, a time lag between the two signals can be discriminated to better locate the annihilation event on the line of response and improve the spatial resolution. To quantify the tracer distribution in a 3D dataset, the raw data have to undergo a correction for attenuation and scatter. Attenuation correction is based on a transmission scan that would originally be acquired using a germanium source rotating around the patient.

104

Oncology Imaging

Today, almost all PET cameras are part of PET/CT hybrid scanners. Only recently, PET/MR combinations are on the market, but only few of them have been installed so far. In the simplest case, the CT component will be used to obtain a low-dose transmission scan as the basis for the attenuation correction and as a map to locate the tracer distribution anatomically. However, the combination has the potential to fully exploit the diagnostic capabilities of both PET and contrast-enhanced multidetector spiral CT. The CT and PET scan are obtained after each other, and anatomical mismatches may be inevitable close to the diaphragm, in the abdominal viscera (in case of vivid peristalsis), or may occur in case of patient movements. Since the PET measurements take several minutes per bed position, the corresponding CT data cannot be acquired in a breathhold as would be done in normal CT examinations. For PET/MR combinations, there are different solutions: (1) Shuttle systems with a PET/CT and an MRI scanners, with a couch system that can be docked to both machines, (2) two separate scanners (one MRI scanner and one PET camera) placed opposed to each other and connected by a common, rotatable couch table, (3) a PET detector insert inside a standard MRI gantry, and (4) a fully integrated, dedicated PET/MR hybrid machine. Besides technical challenges associated with such a construction (replacement of photomultipliers by photodiodes that do not interfere with the influences from the MRI scanner), specific issues need to be solved, since an MRI scan will obviously not contain radiodensity data as they would be needed for attenuation correction. Air and compact bone, for example, are both signal void on MRI scans. Therefore, one or several MRI sequences are used to create an attenuation map, using prior knowledge and empirical radiodensities. Furthermore, additional objects in the gantry like radiofrequency coils or ear protectors will not be contained in the obtained images but have to be accounted for. The developments in this regard are still ongoing. Despite these challenges, PET/MR promises various improvements that are currently being explored. For various anatomical regions, MRI, not CT, is the morphologic imaging modality of choicedfor example, the central nervous system, the pelvis, or parts of the musculoskeletal systemdand is therefore preferable as an anatomic reference for the PET data. Fully integrated PET/MR scanners offer the possibility to measure PET and MRI data simultaneously and thereby perform a motion correction. This will improve both the anatomical location of the tracer uptake and also its quantification. Finally, PET/MR combines two highly versatile functional imaging modalities, which to compare is of high scientific interest. The most widely used and best evaluated tracer is 18-fluorine-labeled deoxyglucose (FDG), a glucose analogue that is being used for staging and therapy monitoring of many tumors like lung cancer, lymphoma, malignant melanoma, multiple myeloma (Fig. 13), and others. Limitations arise from a high FDG uptake in for example, inflammatory lesions or healthy brain (which limits the value of FDG for imaging brain tumors) as well as a low FDG avidity in some common (prostate cancer) or less common (e.g., neuroendocrine) tumors. However, the past years have seen the development of additional tracers that are meanwhile being used clinically. The most important to mention are 68-Ga-DOTA-TOC (a somatostatin analogue taken up by neuroendocrine tumors or meningiomas), radiolabeled PSMA (prostate-specific membrane antigen) ligands for prostate carcinomas and its metastases, or amino acids for brain tumors. In addition to 11-C-methionin, 18-F-ethyl-tyrosine is now increasingly being used. In patient with advanced metastatic disease, DOTA-TOC or PSMA PET can serve to estimate whether sufficient dose may be achieved by radionuclide therapy, using the same substances as carriers.

Medical Image Computing The era of films and lightboxes is definitely past. Even in private offices, images are stored and viewed digitally, and can be transferred to third parties for reading or consultation. There exist companies that have a team of radiologists on their payroll, whose task it is to read imaging studies performed anywhere on the globe. The availability of digital imaging data has triggered a plethora of applications designed to assist the reader, facilitate image handling and storage; some even promise to take over the radiologist’s intellectual task, that is, to interpret the studies. Whether or not this may happen is mere speculation. Among numerous applications for analyzing, viewing and demonstrating images, two have immediate impact on imaging for cancer, automatic lung nodule detection and tools for response evaluation. On plain X-ray films, even experienced readers will only detect nodules 1 cm in diameter or larger, and still miss many, because of overlying structures like ribs, vessels, etc. Therefore, CT is the method of choice for detecting lung nodules. Still, the presence of cross-sectioned vessels makes reading it a challenging task. One way of improving the conspicuity of lung nodules is to obtain thin primary slices (1 mm) and then create maximum intensity projections (MIP) in craniocaudal direction out image stacks of, for example, 10 primary slices. That way, vessel cross-sections will retain their strand-like or oval shape, whereas nodules will stand out as spherical structures. Software solutions exist that analyze thin-section CT series of the lungs to find nodules in the lung and discriminate them from vessels. Some are implemented on the platforms of CT scanners and can be configured to add annotated images and transfer them to the PACS. Others will run on dedicated standalone workstations. Meanwhile, such software is reliable but will also yield false positive findings. So the user will have to cross-check the search results. Important to realize is that a nodule has to be surrounded by ventilated lung in order to be detected. Nodules with contact to the pleura, wedge-shaped sessile densities in the subpleural lung or masses with contact to the mediastinum will not be automatically detected, nor will be hilar masses.

Response evaluation Clinical trials in oncology beyond phase I will often require an objective assessment of the tumor burden, usually based on imaging studies. Various evaluation criteria are being used, depending on the type of tumor and treatment, the most common ones being the Response Evaluation Criteria In Solid Tumors (RECIST), the International Working Group criteria for lymphomas, and the RANO

Oncology Imaging

105

(Response Assessment in Neuro-Oncology) criteria for brain tumors. For immunotherapy, criteria have been adapted to better account for pseudoprogression or delayed response as may be encountered with this type of treatment. Apart from the RANO criteria, a response assessment in metastatic disease implies a heavy workload and requires a meticulous documentation to ensure that the index lesions that are followed are indeed identical. Originally, spreadsheets were used, but turned out to be impractical and error-prone. Software solutions that assist in this task are of invaluable help in any imaging facility involved in clinical trials. Such programs will use a dedicated, PACS-like user interface, provide a database for all measurements, measurement tools for diameters, areas, volumes, and suggest a timepoint response according to guideline’s criteria. Additionally, they provide tools for measuring standardized uptake values in PET studies to support the response assessment. It has to be said clearly that response criteria have been defined to support clinical trials, and so have been all related software tools. Neither were intended to serve as a basis for clinical decisions in routine oncology, and indeed they are too rigid for this purpose. Nevertheless, they are sometimes being used accompanying (rather than influencing) clinical decisions, as part of a quality assurance routine.

Imaging for Treatment Planning and Guidance Both radiotherapy and percutaneous interventions use imaging for planning and guidance. For radiotherapy, the gross tumor volume (GTV) as well as the organs at risk will be defined using the imaging method and windowing on which the tumor is best to discriminate from its uninvolved surroundings, most often CT. Its advantage is that it is unaffected by spatial distortion that may accompany MRI measurements, but for MRI sequences are as well available that are corrected for spatial distortion. Image registration may be used to overlay CT, MRI, or PET images and improve the GTV delineation. PET in particular is a valuable adjunct to assess the presence of tumor in an otherwise unclear structure (i.e., a lymph node level), but for precise delineation of the GTV it should be used with caution, because the apparent borders between normal and abnormal can be strongly influenced by windowing. Besides uncertainties regarding the true tumor extent and positioning (accounted for by the clinical and planning target volume), inter- and intra-fractional motion can cause underdosage to the tumor and overdosage to normal tissue. While issue of intrafractional motion is still far from having been solved, movement or variations in positioning between treatment sessions can be detected by daily imaging in treatment position, using either an in-room CT or a cone-beam CT implemented in the linear accelerator, and corrected by for example, repositioning or alternating treatment plans (image-guided radiotherapy, IGRT). Hybrid installations of MRI scanners and linear accelerators, where image guidance would be feasible based on MR images are about to be installed, so that hardly any experiences exist in this regard. Nevertheless, MR guidance has been shown possible based on scans obtained in a standalone scanner, using a slim, custom-developed fixation device with integrated, MR-visible, stereotactic markers (Bostel et al., 2018). Apart from endovascular interventions (e.g., chemoembolization), percutaneous biopsies or tumor ablations (using radiofrequency, microwaves, cryotherapy, or alcohol) are usually performed under ultrasound or CT guidance, depending on which method best shows the target and is best suited for its locationdand is best mastered by the physician. MRI-guided interventions are only rarely performed, and if at all only in open scanners. For lesions that are best seen on MR images, image fusion is the best option. This has successfully been used for example, patients with elevated serum PSA and previous negative biopsies. Here, MRI including DWI and DCE MRI was used to locate suspicious areas in difficult to reach locations, and then to biopsy them under fusion ultrasound guidance.

Imaging the Primary Tumor The choice of imaging techniques depends on the anatomical site to be imaged, available technology, and local expertise. Following the EU directive 97/43/EURATOM, several European countries, among them the United Kingdom and Germany, have published referral criteria (The Royal College of Radiologists, 2012; Strahlenschutzkommission, 2010) for imaging studies that have meanwhile undergone several revisions. Originally, they served radiation protection purposes, but have in the meantime evolved into up-to-date guidelines that are both scientifically founded and pragmatic in the context of availability and cost-effectiveness. Many of the recommendations in this article will follow the British and German referral criteria as well as the current oncological guidelines, for example, available from the German Arbeitsgemeinschaft Wissenschaftlicher und Medizinischer Fachgesellschaften (AWMF, www.awmf.org).

CNS Tumors The annual mortality from primary CNS tumors is declining since the mid-1990s and in 2012 it was 2.89/100,000 (mortality) for women, and 4.43/100,000 for men (Becker and Wahrendorf, 1998). The treatment is multimodal and includes resection (if possible), radio- and chemotherapy as well as antiangiogenic, targeted or immune-stimulating agents. The primary imaging modality for both primary brain intra- and extraaxial brain tumors as well as brain metastases is clearly MRI, including postcontrast T1-weighted sequences. Compared to CT, MRI is more sensitive and permits a better delineation of the gross tumor volume for radiotherapy planning. CT may be used whenever MRI is contraindicated and in addition to MRI in

106

Oncology Imaging

radiotherapy planning or for detecting calcifications in oligodendrogliomas. T2-weighted sequences in MRI should include fluidattenuated inversion recovery (FLAIR) sequences, where the signal from cerebrospinal fluid is suppressed (Fig. 4). FLAIR images are best suited to depict perifocal edema and low-grade components with an intact blood–brain barrier that will not take up contrast medium. Supplementary MRI investigations may be diffusion-weighted sequences (Fig. 4), dynamic susceptibility contrast (DSC) sequences or 1H spectroscopic imaging that may assist in depicting otherwise occult anaplastic foci in low-grade tumors and guide biopsies. 1H spectroscopy has proven valuable in irradiated patients for discriminating radiation-induced alterations from recurrences. In doubtful cases, amino acid PET/CT (using, e.g., 18-F-ethyl-tyrosine) may be used to assist in the delineation of vital foci, the choice of a biopsy site, and differentiation between treatment-induced changes and recurrences. 68-Ga-DOTA-TOC PET/CT is a specialized investigation for meningiomas and may aid in discriminating them from other meningeal tumors and in planning surgery or radiotherapy. In the follow-up of treated gliomas, MRI remains the primary investigation. Here FLAIR sequences play a pivotal role in detecting recurrences, since treatment with VEGF antibodies (e.g., Bevacizumab) may mask recurrences on contrast-enhanced images (Wen et al., 2010).

Head and Neck Tumors Tumors of the upper aerodigestive tract are typically caused by chronic alcohol and tobacco consumption in combination, human papillomavirus (HPV) being another, independent risk factor. Apart from laryngeal cancer that causes hoarseness in early stages already, head and neck cancer in other locations may cause little or nonspecific symptoms, if at all. Often, a palpable metastatic lymph node will be the first clinical sign. The aggressiveness and metastatic pattern differ widely, depending on the site of the primary tumor. Whereas oropharyngeal cancer often presents with a pattern of rather few, morphologically unambiguous lymph node metastases, carcinoma of the hypopharynx rather may cause numerous micrometastasesdwhich limits the diagnostic accuracy of lymph node imaging. The annual mortality (excluding laryngeal cancer) in 2012 was 5.34/100,000 in men, and 1.28/ 100,000 in women (Becker and Wahrendorf, 1998). Treatment is multimodal and includes chemoradiotherapy alone or preoperatively, resection where feasible, as well as targeted or immune-stimulating drugs, depending on the site of the primary tumor and the TNM stage. For all palpable abnormalities in the neck or salivary glands, ultrasound is the immediate imaging modality of choice to verify the finding, learn which anatomical structure it belongs to, and, if necessary, to plan the next diagnostic steps. Often, the findings are typically benign, and no further measures are necessary. For pharyngeal or laryngeal tumors the role of imaging is restricted to local and regional staging, whereas the diagnosis itself is made by endoscopy and biopsy. For assessing the depth of invasion, contrast-enhanced CT is accurate and reliable. MRI is the method of choice for nasopharyngeal tumors, but not for cancer of the oro- or hypopharynx or the larynx, where motions due to breathing and swallowing often degrade the image quality. For lymph node staging, CT and ultrasound are used in combination. The discrimination between metastatic and inflammatory lymph nodes (which often accompany head and neck cancer) remains a problem with all imaging modalities. Ultrasound, using high-frequency transducers (> 14 MHz) has a superb spatial resolution, compared to CT and MRI, and allows to assess the normal intranodal anatomical structures (hilum, peripheral hyperplastic follicles, vascular architecture) as well as a lymph node’s shape, and should be part of every regional staging for head and neck cancer. FDG PET/CT is of clear value for lymph node staging and being used depending on local standards and resources. It is an important adjunct in assessing the treatment response after neoadjuvant treatment prior to lymph node dissection. If used in radiation therapy planning for head and neck cancer, FDG PET/CT may help to significantly reduce the target volume in the majority of patients and thereby reduce radiation-induced toxicities (Wang et al., 2015).

Thyroid Tumors Nodular goiters are common and endemic in areas with low iodine concentration in drinking water, and most nodules are benign. In fact, the annual incidence of thyroid carcinoma in 2014 was 3.7/100,000 in men, and 8.9/100,000 in women (Gesellschaft der epidemiologischen Krebsregister in Deutschland e.V. (GEKID), 2014), whereas the mortality in 2012 was 0.35/100,000 in men and 0.29/100,000 in women (Becker and Wahrendorf, 1998), which indicates that the cure rate is in the range of 90% or higher. This results from the fact that after thyroidectomy and modified neck dissection, the standard surgical procedure, radioiodine ablation using 131-J is highly effective in eradicating not only local remnants but also distant metastatic disease of follicular and papillary thyroid carcinoma. It will not work, however, if iodine is not (primarily in anaplastic or medullary thyroid carcinoma, MTC) or no longer be taken up, as may occur by dedifferentiation. Hence, the prognosis of anaplastic or secondarily dedifferentiated carcinoma is poor. For the same reason, distantly metastatic MTC is impossible to control definitely, but apart from highly aggressive variants, MTC will progress only slowly, over decades rather than years. The primary task in the workup of thyroid nodules is to spare as many patients as possible invasive measures without missing the relatively few ones who have thyroid carcinoma, with the diagnostic tools ultrasound, 99-m-pertechnetate scintigraphy, cytology from fine needle aspiration, accompanied by serum levels for T3, T4, TSH, and calcitonin for the rare cases of medullary thyroid carcinoma. The tumor marker thyroglobulin is used only to screen for recurrences in patients post thyroidectomy and radioiodine ablation; in all other patients, its serum levels are of no diagnostic use. As a rule, hypoechogenic and scintigraphically cold nodules more than 10 mm in diameter should undergo fine needle aspiration, be followed up with ultrasound if the biopsy is benign, and rebiopsied in case of further growth. Recently, Tc-99m-MIBI scintigraphy has been used for nodules that are cold in pertechnetate

Oncology Imaging

107

scintigraphy, and a persisting MIBI uptake after 120 min may be an additional indicator of malignancy. Still, under which circumstances MIBI scintigraphy should be performed remains to be determined. For preoperative nodal staging, additional ultrasound is the method of choice. Computed tomography plays almost no role for locoregional staging, since iodine-containing contrast agents are contraindicated. After thyroidectomy, radioiodine scintigraphy is the primary method for restaging, in combination with ultrasound. As long as any metastases continue to take up iodine, additional imaging studies will serve for morphologic workup of suspicious scintigraphic findings. Tumors that do not or no longer take up iodine (in case of secondary dedifferentiation, anaplastic, or medullary carcinoma), contrast-enhanced CT or MRI will be used in addition to ultrasound. FDG PET/CT is reserved for secondarily dedifferentiated carcinoma or anaplastic carcinoma. Ten percentage of thyroid cancers are medullary carcinomas. Here, contrast-enhanced CT and ultrasound of the neck and abdomen are usual methods for staging as well as follow-up for patients who continue to have elevated serum levels for calcitonin or CEA. FDG PET/CT may be used in aggressive variants of the disease; slowly progressing tumors, the majority of cases, rather take up 18-F-DOPA, and PET/CT may be used to detect distant metastases. However, PET/CT studies for medullary carcinomas should be reserved for patients in whom positive findings would result in surgical resection. In late stages, this is only rarely the case.

Lung Tumors For no other tumor, the etiology (smoking) is as well understood as for lung cancer, although nonsmokers with lung cancer are not infrequently seen, particularly at older age. There is now increasing awareness that the second-most important cause, exposition to radon (with an average annual dose of 1 mSv, depending on the geographic region), as it is being released for example, from building materials, is far from being under control. Lung cancer is a common and fatal disease, the annual mortality in 2012 being 32.08/100,000 in men and 14.5/100,000 in women (Becker and Wahrendorf, 1998). The treatment depends on whether the patient has small-cell (SCLC) or nonsmall-cell lung cancer (NSCLC). SCLC is considered a systemic disease, even if no metastases are detected at staging. For NSCLC, surgical resection is restricted to patients without contralateral (N3) mediastinal and more distant metastases, often preceded by neoadjuvant chemo- and immunotherapy (in some centers, ipsilateral mediastinal (N2) nodal metastases already will rather trigger nonsurgical treatment). In all other stages, the patients are treated with chemoradiation, and here as well, targeted drugs and immune checkpoint inhibitors play an increasing role. Many patients who are diagnosed with lung cancer have a plain chest X-ray as their first imaging study, being performed for more or less specific symptoms. In case of highly suspicious clinical symptoms (e.g., hemoptysis, dyspnea, recurrent nerve palsy) or X-ray findings (e.g., hilar mass, pneumonia irresponsive to antibiotics) and for staging the disease, cross-section imaging is needed. For diagnosing and staging lung cancer, contrast-enhanced CT remains the primary imaging modality of choice, supplemented by ultrasound of the abdomen, the neck, and the supraclavicular fossa. CT has to include the range between the lung apexes and both adrenal glands, but a complete scan of the abdomen in the portal phase of contrast enhancement will definitely make sense, considering that a baseline scan may be helpful to assess findings that are seen during the later course of the disease. In small cell lung cancer, a contrast-enhanced MRI of the brain is mandatory and may only be replaced by CT in case of contraindications to MRI. MRI of the chest has no clear advantages over CT, with few exceptions. In tumors in the superior sulcus (Pancoast tumors), MRI, particularly in coronal orientation, is indicated to assess its depth of invasion into the cervical structures and its relationship to the cervicobrachial plexus. Moreover, MRI may be the only modality to delineate a tumor against a surrounding atelectasis, which may be important for the GTV or for response assessment. Ultrasound of the chest may supplement CT in suspected chest wall invasion, particularly when MRI is unavailable. Endobronchial ultrasound is a specialized investigation and usually a guidance tool for transbronchial biopsies of suspicious lymph nodes. FDG PET/CT is primarily indicated for characterization of singular pulmonary nodule undetermined in CT and for staging of nonsmall-cell lung cancer, which is highly FDG-avid in almost all cases, whenever radical surgery is intended. It may detect unsuspected mediastinal lymph node or distant metastases and identify patients who are no candidates for surgery. Those patients will usually undergo radiotherapy with selective nodal irradiation, and an available PET/CT study facilitates the delineation of the nodal GTV. If PET is unavailable, a bone scan may be used to screen for skeletal metastases. If they are suspected, their impact on bone stability needs to be assessed, often with X-ray films or CT. Of note, bone scintigraphy may be false negative (or cause a hard to detect “cold lesion”) whenever there are large destructions of bone without reactively increased bone turnover. Pure bone marrow metastases may be negative as well, if they do not cause alterations to the mineralized bone. There is an ongoing discussion on using low-dose CT for early detection of lung cancer in risk populations, that is, in current or former smokers (Fig. 5). Indeed, the largest prospective trial so far, the American National Lung Cancer Screening Trial (NLST) on more than 53,454 participants, comparing low-dose CT against chest X-ray, achieved a 20% reduction of lung cancer, and a 6.7% reduction in all-cause mortality in the screening group (Aberle et al., 2011). At least in some European trials, including the German LUSI trial (Becker et al., 2015), a similar mortality reduction was achieved, but due to smaller sample size, they still need to be metaanalyzed. The results from the largest European trial, NELSON from the Netherlands, are about to be published. While the hazards from radiation exposure are minimal (the effective dose is below 2 mSv and lies in the range of the annual dose from environmental sources), false positive findings are of concern, since they may trigger full-dose CT or PET/CT studies as well as invasive chest procedures. Furthermore, the issue of possible overdiagnosis and overtreatment (i.e., finding and treating lung cancer that, if left alone, would not have caused symptoms or been fatal) is unresolved. Since the NLST lacks a postscreening observation phase, other than some European trials, the question will remain open until the latter have been concluded. Upon a recent European position

108

Oncology Imaging

Fig. 5 Low-dose CT scans of the lung in a participant of the German “Lung Cancer Screening and Intervention” study (LUSI), 6 months apart. The initial scan (A) shows a ground-glass opacification (arrow) that at follow-up (B) has increased in size and density, and would now be termed “partsolid.” Subsequent video-assisted thoracoscopy (VATS) with biopsy revealed lepidic adenocarcinoma with invasive foci.

statement calling for an implementation of lung cancer screening (Oudkerk et al., 2017), concerns have been raised that lung cancer were inadequate, arguing that smoking prevention would the more effective measure to reduce lung cancer mortality. In any case, it should be considered that quitters will remain at increased risk for the next 10–15 years.

Gastrointestinal Tumors Barium-enhanced fluoroscopy, formerly the standard imaging method for tumors of the gastrointestinal tract, has become obsolete, except for few indications, like verifying or ruling out an esophageal stenosis as a cause of dysphagia, or examining the colon proximal to a stenosis that cannot be passed with the endoscope. Contrast-enhanced CT is now the imaging modality of choice, since due to the speed of acquisition peristalsis will no longer impair the image quality. MRI may be used as well, but is more susceptible to motion artifacts, and is inferior to CT in detecting lung metastases. It is, however, preferable in the aftercare for patients with a good long-term prognosis, to spare them a significant radiation dose that would otherwise accumulate over the years of follow-up. It can then be supplemented by native low-dose CT studies of the lung. To assess the wall of hollow organs, some distension should be achieved, since nondistended or contracted bowel segments may easily mimic a mass. Most commonly, 1 to 1.5 L of water are administered either orally for the esophagus or stomach, or rectally for the large bowel, in combination with i.v. butylscopolamineda technique commonly termed “Hydro-CT.”

Esophageal cancer Squamous cell and adenocarcinoma of the esophagus are different in many respects. Whereas squamous cell carcinoma may be located at any level, and its most important cause in the western world is tobacco and alcohol consumption, adenocarcinoma is most frequently induced by gastroesophageal reflux, intestinal metaplasia (Barrett’s esophagus) being the typical precancerous change. The annual incidence for esophageal cancer in 2014 was 9.6/100,000 in men, and 2/100,000 in women (Gesellschaft der epidemiologischen Krebsregister in Deutschland e.V. (GEKID), 2014); the 2012 mortality was 4.78/100,000 in men, and 1.07/100,000 in women (Becker and Wahrendorf, 1998). As with all gastrointestinal tumors, the diagnosis is routinely obtained by endoscopy and biopsy. Nevertheless, patients with dysphagia may in first instance have a barium swallow to verify whether a stenosis is present at all, and if so, where it is located. Not infrequently, esophageal carcinomas may grow mainly in the submucosa, so that mucosal abnormalities may not be very obvious at endoscopy. For staging, both contrast-enhanced CT and endoscopic ultrasound are primarily indicated. With endoscopic ultrasound (EUS), the layers of the esophageal wall can be nicely discriminated, and tumors can be diagnosed to either be confirmed to the mucosa and submucosa (T1), invade the muscular layer (T2), or breech the adventitia (T3 or T4). A discrimination between T3 and T4 (invasion of adjacent organs) is difficult, however, because EUS high-frequency probes are limited in penetration. Lymph nodes directly adjacent to the esophagus can be assessed and biopsied under endoscopic ultrasound guidance, if a suitable device is available. A precise discrimination of T stages lower than T4 may be of limited relevance for planning surgery, but is required prior to and after neoadjuvant therapy to assess the response to treatment. The strength of CT lies in the assessment of the overall extent of the tumor, a possible T4 stage with invasion into adjacent organs (e.g., trachea, left atrium, great vessels, thoracic vertebrae), and lymphatic and distant spread. CT is insensitive, however, in discriminating T1, T2, and T3 stages. Since both downstream and upstream lymph nodes may be involved, the neck, chest, and at least upper (if not the entire) abdomen has to be included. A postoperative CT study should be obtained as a baseline for follow up, because surgically induced alterations may

Oncology Imaging

109

otherwise be hard to interpret and to discriminate from a recurrence. FDG PET/CT is not used routinely but in difficult cases, during staging as well as follow-up. If done, its CT part should be carried out as a full diagnostic protocol, including intravenous contrast, and not only in native, low-dose technique for attenuation correction and anatomical orientation.

Stomach cancer For gastric cancer, the incidence and mortality have been constantly decreasing over the past decades, the 2012 mortality being 5.89/ 100,000 in men, and 3.14/100,000 in women, compared to almost 10-fold values around 1950 (Becker and Wahrendorf, 1998). Possible reasons among others are the consumption of fresh or refrigerated rather than nitrite-pickled food, or more effective treatment for chronic gastritis (including Helicobacter-associated disease). Nevertheless, gastric cancer is still a fatal disease once lymph node or distant metastases are present. According to the Laurén classification, tumors of the intestinal type are to be differentiated from diffuse tumors that lack adhesion molecules and may infiltrate the gastric wall in a single-cell pattern and extend far beyond what is evident clinically or at imaging. If resectable, it is treated by gastrectomy and lymph node dissection. Superficial tumors (T1a) may undergo endoscopic resection. If the resection has been radical, adjuvant chemotherapy is usually not given. In absence of distant disease, definite radiochemotherapy is a possible attempt at a cure. If distant metastases are present at the time of diagnosis, chemotherapy is indicated, and surgical procedures mainly serve to prevent symptoms, mainly obstruction. The diagnosis is established by endoscopy and biopsy. To not miss a diffuse gastric carcinoma, special staining like PAS is needed to detect signet cells. For staging, hydro-CT and abdominal ultrasound, and endoscopic ultrasound are the imaging modalities of choice, CT mainly to detect a T3 or T4 stage, lymph node or distant metastases as well as a peritoneal spread and ascites. Nevertheless, a discrimination of an extension beyond the gastric wall (T3) may be difficult to discriminate from reactive inflammatory changes that often accompany gastric and also other gastrointestinal tumors, and lymph node metastases, especially those directly adjacent to the gastric wall, may be too small to appear suspicious or even be detected. Diffuse gastric carcinoma, not surprisingly, may show no circumscribed mass at all and be missed or grossly underestimated, and even a peritoneal dissemination may occur before a true gastric mass is apparent. CT has to cover the entire abdominal cavity, since peritoneal deposits may be present in the pelvis (a Krukenberg’s tumor is an ovarian metastasis due to stomach cancer), and also the chest and infrahyoid neck, not to miss a Virchow’s gland, a lymph node metastasis in the left venous angle. MRI is a special investigation to resolve open issues after CT and endoscopic ultrasound. FDG PET/CT is not routinely used for local staging but may be applied during aftercare for otherwise unclear lesions. Again, due to its single-cell infiltration pattern, diffuse gastric cancer will easily escape detection with PET/CT, and so will its metastases.

Cancer of the small bowel and appendix Malignant tumors of the small bowel and appendix have in common that they are comparatively rare and usually inaccessible to endoscopy. If suspected clinically, contrast-enhanced hydro-CT in combination with abdominal ultrasound is usually the first imaging modality being used, but MRI is a good alternative, either in hydro technique as with CT, or, as a special investigation, with administration of a methyl-cellulose solution via a nasojejunal tube (“MR Sellink”), as it has been established for example, Crohn’s disease. MR is especially useful if diffusion-weighted sequences are included that may highlight and draw the reader’s attention to small lesions that would otherwise have escaped detection. If PET/CT is being used, it should be carried out with CT as a full diagnostic, contrast-enhanced protocol. It should be considered that tumors of the small bowel or appendix not uncommonly have a neuroendocrine differentiation, in which case they may fail to take up FDG. 68-Ga-DOTA-TOC would then be a possible alternative.

Colorectal cancer Colorectal cancer is among the three most common cancers and cancer-related causes of death in both sexes. Its mortality is on the decline, however. In 2012, it was 13.7/100,000 in men, and 8.2/100,000 in women (Becker and Wahrendorf, 1998). Its diagnosis is based on endoscopy and biopsy. Colon and rectal cancer are fundamentally different regarding treatment protocols, and therefore also the required imaging procedures. If no distant metastases are present, colon cancer is always treated by radical resection. The only imaging studies required preoperatively according to current guidelines are chest X-ray and abdominal ultrasound, to rule out lung or liver metastases. For rectal cancer, neoadjuvant as well as definite radiotherapy play a major role. As a consequence, thin-section MRI of the pelvis (CT in case of MRI contraindications) and endoscopic ultrasound are required to determine the distance of the tumor from the mesorectal fascia and diagnose possible perirectal lymph node metastases (Fig. 6). Their presence or absence is one criterion for the choice duration and fractionation of the neoadjuvant radiochemotherapy protocol. For both colon or rectal carcinoma, abdominal CT is indicated if there is suspicion of metastases, for example, at abdominal ultrasound. CT or MRI colonography are special investigations where the colon is inflated with air (or liquid endoluminal contrast agents with MRI). The tumor may so be seen as mural thickening and mass, and the depth of invasion into the wall or the surrounding be assessed. Additionally, 3D rendering virtual reality techniques can be used to generate colonoscopy-like views, allowing the user to “fly through” the lumen. The added value of CT or MRI colonography is obvious, but, considering the radiation dose, it should be reserved for resolving open questions that might influence clinical decisions, or rare cases where complete colonoscopy is not possible. Virtual colonoscopy may be used as an adjunct, but can neither replace CT or MR colonography or classical endoscopy, nor should it be used as a nonenhanced low-dose

110

Oncology Imaging

Fig. 6 Pelvic high-resolution MRI in a patient with locally advanced rectal cancer. On fat-suppressed, contrast-enhanced T1-weighted images at the time of diagnosis (A), there is a large rectal mass (bold arrows) with streaky infiltration into the mesorectal fat, and a broad extension to the mesorectal fascia (curved arrow), along with multiple enlarged lymph nodes within the mesorectum as well as along the pelvic wall. After neoadjuvant radiochemotherapy (B) there is marked remission of the tumor (bold arrows) as well as the extension to the mesorectal fascia and the lymph nodes. Note the recovery of the mural architecture in the rectal wall, the mucosa now being visible against the less enhancing submucosa (open arrow).

protocol for only visualizing the mucosal surface. Whether or not such techniques may in future be used as an alternative to colonoscopy for colon cancer screening is an open question. Currently, they clearly not being recommended for this purpose. FDG PET/CT may have a peculiar role in colon cancer: Presacral tissue densities after primary treatment for rectal cancer often represent scar tissue, but their differential from presacral recurrences is often difficult. Here, a high uptake of FDG will point toward a recurrence as the cause. Furthermore, unlike in most other tumors, resection of metastases is an option in colorectal cancer, because there is a reasonable possibility that the visible metastases are the only ones. Whenever they are planned to be respected, a whole-body FDG PET/CT is a sensible choice to rule out additional metastases in other locations, that would prohibit such aggressive measures.

Hepatic Tumors Focal hepatic lesions are an extremely common clinical problem, because they are frequent and also benign in most cases. The commonest tasks in which imaging studies are required are:

• • •

To differentiate incidental liver lesions, usually detected by abdominal ultrasound, To stage for liver metastases in patients being diagnosed with for followed up for extrahepatic primary tumors, To diagnose and stage a suspected primary liver tumor.

Most incidentally detected liver lesions are benign (cysts, pseudotumors, hemangiomas, focal nodal hyperplasias, etc.). As a rule, the diagnosis of a simple cyst or a pseudotumor (e.g., focal sparing in fatty liver) should be made with ultrasound alone, without any further imaging or follow-up. Similarly, a lesion with the typical appearance of a hemangioma at ultrasound in a patient without a cancer history or risk factors should be definitely diagnosed with ultrasound, and have a single, final follow-up ultrasound for example, after 3 or 6 months. If one is unsure, it is better to refer the patient to someone more experienced with the technique than to order additional imaging studies. These are reserved for unclear cases, or for patients who have cancer in another location. If an additional CT or MRI would be performed anyhow (e.g., to stage a primary tumor in the abdomen), its protocol may be adapted in order to clarify the liver lesion. If not so, contrast-enhanced ultrasound (CEUS) would be first choice, because it implies no ionizing radiation, and in best case can be performed in the same session without leaving the patient in worries for longer time (Fig. 7). In large trials, CEUS has been found diagnostically equivalent to contrast-enhanced CT (Seitz et al., 2009; Strobel et al., 2009). If CEUS is unavailable or fails to resolve the issue, contrast-enhanced MRI including diffusion-weighted imaging is preferable over CT, at least in the interest of radiation protection. Furthermore, liver-specific contrast media are available in MRI, that may assist in differentiating liver lesions. In unclear cases, nuclear medicine studies (hepatobiliary or blood-pool scans) provide additional information to make the probable diagnosis. When staging for liver metastases, the choice of the imaging method depends on the primary tumor and the a-priori risk for metastases. In many cases, CT or MRI of the liver will be included in the imaging protocol for primary staging, but should be supplemented by an abdominal ultrasounddif possible, after the CT or MRI, to help clarify unclear lesions, such as cysts or hemangiomas. It should be remembered that in cancer patients benign liver lesions are as frequent as in healthy persons. Most primary liver tumors are either peripheral cholangio- or hepatocellular carcinomas (HCC). While HCC is common in Africa or the far east, following the endemic presence of hepatitis B virus and the exposure to aflatoxins, it is less common in Caucasians, mostly associated with alcoholic or nonalcoholic liver cirrhosis and/or chronic hepatitis B or C, or, less frequently, hereditary

Oncology Imaging

111

Fig. 7 Longitudinal section of the right liver lobe during contrast-enhanced ultrasound (CEUS) in a patient with malignant melanoma. Unenhanced B-mode ultrasound of the liver had been normal. In the portal phase of contrast enhancement, there is a hypoenhancing, subdiaphragmal nodule (arrow), that progressed at follow-up and disappeared after immunotherapy, and was therefore diagnosed as a liver metastasis. As a rule, CEUS is equally sensitive and specific to CT or MRI, sonographic access provided.

hemochromatosis. It is a highly vascularized tumor with a strong tendency to invade portal venous branches, giving it the typical pattern of a strong contrast enhancement in the arterial phase of injection (which is uncommon in cholangiocarcinoma and most liver metastases) and a contrast washout in later phases. The primary treatment, depending on the number and size of lesions as well on the remaining liver function, consists of resection, local radiofrequency tumor ablation, or, in eligible patients, liver transplantation. In transplant-eligible patients, chemoembolization via a supraselective angiographic catheter or radiofrequency ablation may serve to prevent further tumor growth until a donor organ becomes available (“bridging”). In clinical trials, photon or particle radiation therapy are being evaluated. In case of irresectability or recurrence, sorafenib is effective for palliation and prolonging survival. Cholangiocarcinoma usually shows a strong fibroplasia with only scarce vessels in its center. Cholangiocarcinomas are best imaged and staged with MRI (CT, if MRI is contraindicated) in combination with abdominal ultrasound. Imaging studies should be used to guide the biopsy to the vital periphery and avoid the desmoplastic reaction in the tumor’s center, where it may be false negative. Owing to the aggressiveness of the tumor, both the entire abdomen and the chest (using contrast-enhanced CT) should be imaged. For HCC, valid imaging studies are crucial: The presence of a lesion with typical enhancement pattern in a single imaging modality in a patient at high risk is accepted as proof of HCC, without the need for a biopsy. The modality may be contrastenhanced CT or MRI, or contrast-enhanced ultrasound, all of which are seen as equivalent. Only in unclear cases either an additional imaging modality or biopsy is needed. Imaging also serves to diagnose multifocal lesions as well as portal venous invasion, that may preclude surgical resection.

Pancreatic Tumors Ductal adenocarcinoma of the pancreas is one of the most fatal cancers. Although it is not exceedingly frequent, it is the fourth-most common cancer-related cause of death in men and women, and its frequency is on the slow rise. Smoking is a risk factor, but on the whole, there is no clearly definable risk population, and no effective and acceptable measures are available for early detection. Radical surgery is the only chance for a cure. The role of imaging is to enable a diagnosis as early as possible, but since the symptoms are unspecific at the beginning, ultrasound is the only modality that can be recommended to be used generouslydknowing that its effectiveness in detecting pancreatic cancer in a resectable stage is yet unproven. In unclear findings, suspicious biochemical findings, suspicious and persistent pain (especially a combination of epigastric and middle back pain) in combination with limited sonographic access, contrast-enhanced CT or MRI are to follow. CT is used most frequently, because it will reliably image the entire abdominal cavity, and the chest if needed. MRI is equally capable, but more expensive and less well tolerated by the patient. Whenever the pancreas is well visible at sonography, contrast-enhanced ultrasound is a reliable and safe procedure to prove or rule out a focal pancreatic lesion, because pancreatic adenocarcinoma is almost always hypovascular compared to the normal parenchyma. It would clearly deserve to be used more frequently and could so obviate CT or MRI studies for unclear sonographic findings. Endocrine tumors of the pancreas are an entirely different entity and may be found in patients with type I multiple endocrine neoplasia. Whereas some produce insulin, glucagon, gastrin or vasoactive intestinal peptide (VIP) and are therefore being suspected

112

Oncology Imaging

because of endocrine symptoms (e.g., hypoglycemia, recurrent peptic ulcers, or watery diarrhea), a significant portion will be nonsecreting. The probability of a malignant variant depends on the endocrine type and is highest in gastrinomas. Their morphology is often different from adenocarcinoma; typically they will be sharply delineated and hypervascular. As in other locations in the body, endocrine tumors of the pancreas express somatostatin receptors and can thus be detected by somatostatin analogs, using for example, octreotide scintigraphy or 68-Ga-DOTA-TOC PET/CT.

Tumors of the Kidney, Ureters, and Bladder Renal carcinoma is a less common cause of cancer-related death. Its mortality in 2012 was 5.3/100,000 in men, and 2.2/100,000 in women (Becker and Wahrendorf, 1998). Its growth rate is highly variable, and symptoms like flank pain, or gross hematuria usually indicate advanced local disease. Apart from rare disorders like von Hippel–Lindau disease, there are no known risk constellations. Although no screening program exists, the most renal tumors are nowadays found incidentally on occasion of an abdominal ultrasound for any reason. As a result, most patients with renal cancer will be definitely cured. Apart from metastases to the lymph nodes, lung, and bones, two findings are peculiar: Renal cell carcinomas tends to invade the renal vein, extending along the inferior vena cava until above the diaphragm. Furthermore, pancreatic metastases may occur, even as a late recurrence, long after primary therapy. The treatment is almost exclusively surgical, whenever possible by kidney-preserving procedures, otherwise by nephrectomy. Two kinds of space-occupying lesions in the kidney are definitely benigndcysts and angiomyolipomas. Cysts can be confidently diagnosed at ultrasound if they exhibit typical features. This, however may be difficult in case of limited sonographic access, so that contrast-enhanced CT or MRI may be necessary for clarification. The other entity, angiomyolipoma, has a typical morphology at ultrasound (high echogenicity, sharp delineation), but as state of the art, a second imaging modality (usually native CT) is required to prove the presence of significant amounts of fat in the lesion. As a rule, all focal renal lesions that are neither simple cysts nor angiomyolipoma are suspicious and should be respected, because the majority of them will be malignant, and no imaging modality is capable of excluding this. Cancer of the renal pelvis, ureters, and bladder are entirely different clinically and in etiology. Typically, they are caused by chronic exposition to carcinogens, most importantly smoking, but also for example, aniline dyes. Therefore, as many as 50% of patients may have recurrences or second neoplasms of the urothelium at the time of diagnosis or in the later course. Hematuria is the leading symptom in 75% of patients. The treatment is mainly surgical, but superficial bladder tumors might be locally ablated via cystoscopy. In advanced bladder tumors not undergoing cystectomy, instillation of BCG has successfully been used to temporarily control the disease. CT or MRI may be limited to the kidneys if only carried out to rule out a malignant lesion. Whenever the ultrasound findings are highly suspicious for malignancy, or gross hematuria is the chief symptom, the heart should be included, to detect a possible thrombus in the vena cava, as well as the entire abdomen including the bladder. Furthermore, the urine collecting system has to be assessed for possible neoplasms as a cause for hematuria. Therefore, scan timing in a renal CT or MRI protocol will differ from that in a standard abdominal one. A baseline native scan (either acquired prior to contrast infusion or, if DECT is being used, as a virtual native scan) is needed, to detect a weak contrast enhancement in a focal lesion. An arterial phase is optional and often omitted. The most important will be a scan in the phase when both renal cortex and medulla are enhanced, which is later than the usual portal phase, that is, 120 s post infusion. Finally, a CT urography, with contrast filling of the collecting system is obtained after 5 min. The same timing will apply if MRI is used instead of CT. For bladder cancer, ultrasound as well as cystoscopy are the primary examinations, and contrast-enhanced MRI whenever invasion of the muscular layer or extension beyond the bladder are suspected. As a rule, the entire urine collecting system including renal calices and pelvis as well as ureters needs to be examined with CT or MRI urography, due to the high prevalence of simultaneous second neoplasms, if not plain X-ray films with retrograde contrast filling of the ureters or ureteroscopy are performed on occasion of cystoscopy.

Prostate Cancer Prostate cancer is the most common malignant tumor in men. Although there has been a steep rise in incidence, its annual mortality is slowly declining, being about 11/100,000 in 2012, indicating that the rising incidence is mainly a screening artifact due to the frequent measurement of PSA in serum. This means that there is a significant risk of overdiagnosis and overtreatment, if not measures are taken to identify high-risk and low-risk cancers and manage them adequately. In this process, imaging plays a pivotal role, using four tools: Endorectal ultrasound-guided biopsies, MRI, MRI/ultrasound fusion biopsies, and PET/CT or PET/MRI with radiolabeled PSMA ligands. First, patients with suspected prostate carcinoma will undergo systematic endorectal, ultrasound-guided biopsy and, if positive, nodal and skeletal imaging for staging, and then treatment: Radical prostatectomy, definite intensity-modified or particle radiotherapy, or, in case of metastases, antihormonal therapy. In case of negative biopsies despite elevated or increasing serum PSA levels, or to stage biopsy-proven prostate carcinoma for local extracapsular extension, multiparametric MRI is the method of choice. It combines T1- and T2-weighted imaging with diffusion-weighted imaging (including calculated maps for the apparent diffusion coefficient) and DCE MRI. 1H spectroscopy is no longer standard part of the protocol. Depending on the morphologic and functional features, possible lesions may be classified according to their probability of malignancy, using, for example, the PI-RADS that was published in its 2nd version in 2015 (American college of radiologists, 2015). Using commercial image fusion system,

Oncology Imaging

113

suspicious lesions may be targeted and biopsied under ultrasound guidance from a perineal approach, although they may be invisible at ultrasound. This is particularly useful to reach suspicious lesions that may have been missed by standard systematic biopsies because of their location (e.g., ventrally). The development of a PET ligand to the prostate-specific membrane antigen (PSMA) has been a revolution in prostate imaging, since it enables to locate metastases even in otherwise unsuspicious structures, like normal-size lymph nodes. The first available F18labeled PSMA-compounds and the currently distributed tracers labeled with 68-gallium are excreted in the urine. The excreted tracer in the urinary bladder may hinder the detection of local recurrent tumors in close proximity to the bladder. However, 18-fluorinelabeled tracers that are not primarily renally excreted will soon be available. This might not only facilitate the detection of prostate carcinoma but also support the delineation of the gross tumor volume for radiotherapy planning. Optimally, for imaging the prostate itself using PSMA ligands, PET/MRI hybrid scanners would be preferable over PET/CT machines.

Tumors of the Uterus and Ovaries Uterine tumors may be either endometrial adenocarcinoma or cervical squamous cell carcinoma, with an annual mortality of 2/100,000 and 1.9/100,000, respectively. They may be treated by resection or radiochemotherapy, depending on the local stage and presence or absence of nodal or distant metastases. Cervical carcinomas, if locally advanced or recurrent, can be devastating regarding quality of life. The diagnosis is typically made by either abrasion (corpus) or cone excision (cervix), supported by bimanual palpation and endovaginal or endorectal ultrasound for local staging. MRI is primarily indicated for staging both tumors locally, and should include the entire abdominal cavity to detect spread to retroperitoneal lymph nodesdthat would be considered distant, not regional. The main target feature when imaging corpus tumors will be the proportion of the uterine wall occupied by the tumor. For cervical tumors, particular attention is paid to an infiltration of the parametria, the bladder, and the rectum, and the downwards extension into the vaginal wall. FDG PET/CT is a specialized investigation in cases where the presence or absence of metastases remains unclear. Ovarian cancer accounts for 5/100,000 deaths per year and is often diagnosed in an advanced, incurable stage. Due to the high prevalence of benign, and mostly cystic masses, the confident exclusion of ovarian cancer may be highly challenging, endovaginal ultrasound being the primary imaging modality of choice. Here, the presence of solid or papillary components, thick septa or strong vascularization at color Doppler will argue in favor of malignancy. For MRI as a problem solver, recommendations have recently been released (Forstner et al., 2017), that pay particular attention to DCE MRI and diffusion-weighted imaging for discriminating unclear masses. Nevertheless, laparoscopy and biopsy are indicated early, not to miss the disease in a curable stage. Once the diagnosis of ovarian cancer has been made, staging has to be completed, preferably by abdominal ultrasound and CT of the chest and abdomen. Bone scans are usually not required, because bone metastases due to ovarian cancer are extremely rare.

Breast Cancer Breast cancer is the commonest tumor in women and with an annual mortality of 16.1/100,000 also their leading cause of cancer death. Besides eradicating the disease, all treatment approaches strive at limiting resection as much as possible. This in mind, imaging is directed at enabling a diagnosis as early as possible, and at visualizing the true extent of the tumor inside the breast, in order to guide breast-conserving surgery. Mammography screening programs are in action in most countries, but despite rigorous quality assurance and training of the readers, only about 50% of all women who finally undergo biopsy for a screening-detected lesion will in fact have cancer. Besides the introduction of digital mammography and constant improvement of ultrasound technology, three developments need mentioning that have significantly influenced detection and staging of breast cancer, or can be expected to do so: Digital tomosynthesis, DCE MRI, and diffusion-weighted MR imaging. Digital tomosynthesis is a technique where the breast remains fixed between the detector and the compression plate, and the tube is being moved over an angle of up to 50 degrees while the beam is on. From the data, stacks single slices (e.g., 1 mm thick) can be reconstructed, which permits an improved delineation of nodules and masses against other overlying structures. This is possible without significant radiation dose penalty, and additionally conventional mammography images can be extracted, whose quality is apparently not inferior to that of a normal mammogram (Zuley et al., 2014). DCE MRI has been investigated since the early 1990s, and has now become part or the routine. Its role as a problem solver for unclear lesions is limited, and biopsy under imaging guidance is clearly the best way toward a reliable diagnosis. Nevertheless, it contributes significantly to assessing the extent of even complex shaped tumors, multifocal or multicentric lesions, and even DCIS components are visible owing to their contrast uptake, especially if they are high-grade (Fig. 8). Other clear indications are breasts with implants (where neither mammography nor ultrasound make sense), and suspicious lesions post treatment, where scars have to be differentiated from residual or recurrent tumor. Diffusion-weighted images, finally, are difficult to obtain in good quality, owing to motion and magnetic field inhomogeneity along the complex-shaped tissue–air interface. Nevertheless, when fine-tuning and quality assurance are thoroughly carried out, the images will reliably highlight malignant foci against a signal-void background (Fig. 9). In a prospective study, an abbreviated (< 10 min) protocol applied for women scheduled for biopsy due to screening-detected, suspicious lesions, diffusion-weighted imaging was sensitive and specific enough to possibly spare women without breast cancer a biopsy, with extremely few true cancers being missed (Bickelhaupt et al., 2015). Such protocols do not require any contrast agent and are short enough to be used on large scale. Whether or not they might be used instead of screening mammography in the future, is yet undetermined.

114

Oncology Imaging

Fig. 8 Dynamic contrast-enhanced T1-weighted images in a patient with multifocal and bilateral breast cancer. Note the solid, circumscribed nodules (bold arrows) that correspond to invasive foci. Additionally, there is an extensive in-situ component that is visible as a less intense, inhomogeneous area of contrast enhancement with a segmental configuration (open arrow).

Fig. 9 Craniocaudal projection of axial diffusion-weighted images with background suppression in a participant of the German mammography screening program who had been recalled for a suspicious (BI-RADS 4) mammography finding in the left breast. The projection image permits to check the breasts for lesions with restricted diffusion at a single glance. In this case, the mammography finding corresponds to a lesion less than 10 mm in diameter that was histologically confirmed to be a ductal invasive carcinoma. Courtesy Dr. Sebastian Bickelhaupt, German Cancer Research Cencer (DKFZ), Heidelberg.

Hematologic and Lymphatic Neoplasms Leukemia is being diagnosed based on blood and bone marrow samples. Other than myeloid ones, lymphatic leukemias are often accompanied by lymph node and sometimes extranodal masses that will be seen in imaging studies. In chronic lymphoid leukemia, which is in fact an indolent non-Hodgkin lymphoma, imaging is part of staging and follow-up. For malignant lymphomas, imaging of the neck, chest, and abdomen is mandatory to stage the disease and to assess its response to treatment (Cheson et al., 2007), usually by a combination of abdominal and cervical ultrasound and contrast-enhanced CT. In patients with a favorable longterm prognosis, particularly young ones, it is recommended to replace CT by MRI, at least for the abdomen. Furthermore, FDG PET/CT is recommended in guidelines for Hodgkin lymphoma and those non-Hodgkin lymphomas that are typically FDG-avid. It may supplement primary staging, but its most important role is to assess the presence or absence of viable lymphoma at an interim staging and/or after completion of the primary treatment. Especially when there was bulky disease at diagnosis, macroscopic residuals will persist after treatment that may or may not contain viable lymphoma tissue. Increased FDG uptake in these lesions almost invariable heralds a recurrence and also indicates an unfavorable prognosis. Furthermore, FDG PET/CT may be indicated whenever there is suspicion of recurrence during the later course. Multiple myeloma, formally a non-Hodgkin lymphoma, is a monoclonal plasma cell disorder that arises primarily in the bone marrow and interacts with osteoclasts and osteoblasts in a specific fashion. It may cause destructions to mineralized bone, along with anemia, hypercalcemia, and renal damage, which are indications for systemic treatment. There are, however, asymptomatic precursor stages, namely monoclonal gammopathy of unclear significance (MGUS) and smoldering multiple myeloma (SMM), that are at risk for progression into a symptomatic stage, but do not yet require treatment. Together with biochemical and bone marrow diagnostics, imaging plays a pivotal role for decisions on treatment. For assessing the integrity of mineralized bone, the previously used plain X-ray skeletal survey are now being substituted by noncontrast whole-body CT, because the latter is more

Oncology Imaging

115

Fig. 10 Lateral X-ray plain film of the cervical spine (A) in a patient with smoldering multiple myeloma and so far without end organ damage. A sagittal reconstruction from an unenhanced, low-dose CT obtained within 4 weeks (B) shows a lytic bone lesion (arrow) that is invisible on plain films. The patient is therefore upstaged to symptomatic multiple myeloma, where systemic treatment is indicated.

Fig. 11 Pelvic X-ray film (A) and CT (B) in a patient with multiple myeloma. On CT, a large destructive lesion in the iliac bone (arrow) is easily detected and was originally missed on plain films. Retrospectively, one can discriminate the vertical contour of the lesion (arrow in A), but it was obscured by overlying bowel gas.

sensitive and specific for bone destructions (Figs. 10 and 11), and its findings are of proven prognostic significance (Hillengass et al., 2017). For assessing the bone marrow and possible extraosseous myeloma foci, noncontrast whole-body MRI is now standard and is capable of detecting diffuse or focal bone marrow lesions that still have left the cortical and cancellous bone intact. According to the revised criteria of the International Myeloma Working Group, the presence of more than one focal bone marrow lesion is an indication to treatment, even if the mineralized bone remains intact (Rajkumar et al., 2014). FDG PET/CT is a valid modality for detecting focal myeloma infiltrates, but its sensitivity for diffuse bone marrow involvement is variable and probably dependent on the density of infiltration. If it is used, the CT data should be acquired using a full diagnostic protocol and enable a reliable assessment or the mineralized bonedand so to spare the patient an additional skeletal CT. Whole-body diffusion-weighted imaging is now a standard supplement to MRI in many centers and is extremely useful to facilitate the detection of diffuse and focal lesions, even in unusual locations (Fig. 12).

Tumors of the Bones and Soft Tissue Primary bone and soft tissue malignancies are relatively rare tumors. The modality of choice to approach the diagnosis depends on the location. For primary bone tumors, diagnostic criteria applied to plain X-ray films have been established over many decades and

116

Oncology Imaging

Fig. 12 Diffusion-weighed imaging (DWI) in plasma cell diseases. A transverse section at the level of the pelvis shows a diffuse infiltration pattern (A). A frontal projection image in a patient with a multifocal infiltration pattern, with lesions (B), for example, in the left scapula, left upper chest wall, both humeri, left femur, and numerous others.

remain valid. The first task of the reader is to identify tumor-like lesions where a biopsy should not be performed, because it is either unnecessary or even likely to yield falsely suspicious histologic findings that result in invasive measures (“do not touch lesions”). Of the remaining, some can be confidently diagnosed based on the plain film morphology, location, and patient’s age, but in some the diagnosis will remain unclear until resection. Considering that such tumors are rare, and there are few radiologists who are truly experienced, it is strongly advised to seek an expert’s opinion and to consult an orthopedic surgeon even before a biopsy is attempted. CT and MRI are valuable adjuncts for making the diagnosis, but their most important role is to delineate the true extent of the tumor, beyond what is visible on plain films. For primary soft tissue tumors, MRI, CT, or ultrasound again serve to identify typically benign entities (e.g., lipoma), and to identify the organ or compartment of origin. If not the entity is evident from the lesion’s imaging morphology (e.g., liposarcoma) or location, the contribution of imaging, including PET/CT, to making the diagnosis is limited. Such tumors will be respected, taking care of their integrity, in order not to contaminate the operation bed. Following up an unclear soft tissue mass is definitely inadequate.

Imaging Distant Disease Usually, regional lymph nodes will be assessed when imaging the primary tumor, as outlined above, and some protocols also account for the most common distant metastases. In special situations, imaging is mainly dedicated to detecting distant disease. The most important applications are:

•

• • •

Whole-body imaging: PET/CT is a valid method for lymph node as well as distant metastases, using FDG, PSMA ligands, DOTATOC or other tracers, depending on the histology. Diffusion-weighted MRI is sensitive for distant as well as lymph node metastases, but in the latter it is not specific for malignant involvement. Like metastases, reactively altered lymph nodes have a high signal. For brain metastases, neither FDG PET/CT nor diffusion imaging are valid, because of the brain’s high FDG uptake and physiologic diffusion restriction. Detecting lung metastases: The modality of choice is CT. If for any reason no information about soft tissue is needed, it can be performed without contrast medium and in low-dose technique, which will keep the effective dose below 2 mSv, usually even lower. Detecting lymph node metastases: MRI contrast media dedicated to lymph node imaging have been on the market, but have been unsuccessful there and meanwhile withdrawn. For superficial lymph nodes, high-frequency ultrasound (14 MHz or higher) is clearly the best and cheapest method, but requires a skilled and experienced examiner. Detecting liver metastases: Usually, the liver is included in the primary imaging protocol. If examined alone, contrast-enhanced ultrasound is the first method to suggest, because it is sensitive, specific, and cost-effective. Alternatives are contrast-enhanced CT

Oncology Imaging

• • •

117

or MRI including diffusion-weighted sequences. FDG PET/CT may be limited in value because of the moderately high background uptake of the liver itself. Detecting CNS metastases: The method of choice is clearly contrast-enhanced MRI, which is superior to all competing modalities. If meningeal spread is suspected, longitudinal, contrast-enhanced images of the entire spinal canal are added. Detecting skeletal metastases: 99m-Tc-Bisphosphonate bone scan continues to be the standard but has shortcomings in highly lytic lesions (e.g., due to thyroid or renal cancer) because the bony structures are either completely destroyed or the metastases induce no reactively increased bone turnover. The same may apply to pure bone marrow deposits. PET/CT with NaF-18 is an alternative to bone scans whenever 99m-Tc is unavailable but has the same limitations. Cancer of unknown primary (CUP) is a special indication. Usually, the origin of the resected tumorous tissue will be tried to discriminate from the pathological workup of the biopsy. If this is unsuccessful, the imaging methods are tried to be wisely chosen, along with for example, an endoscopic search. PET/CT is often included but may be unsuccessful if the tracer is not being taken up, or the assumed primary tumor is very small, as for example, in malignant melanoma. By the same token, in widespread disease with a high tumor load, it will be difficult to discriminate the metastases from the tumor of origin.

Response Monitoring A tumor’s response to treatment is generally attributed to a reduction in size, which is the sole criterion in almost all formal response criteria, including those for immune therapy. Powerful software tools are clinically approved that support physicians in formal response assessment in clinical trials. However, reduction in size occurs relatively late in the course of the treatment, with few exceptions like some lymphomas, where one can assume that treatment-induced apoptosis rather than necrosis is the predominant response mechanism. Therefore, functional features have been and are being explored to mirror or predict treatment response. The best evaluated one is the FDG uptake in PET/CT (Fig. 13), which in many tumors has been shown to be reduced much earlier than any changes in tumor size become apparent. The enhancement of contrast agents, that can easily be picked up in

Fig. 13 FDG-PET/CT (A and C) and contrast-enhanced, fat-suppressed T1-weighted MRI (B and D) in a patient with a large, destructive myeloma infiltrate in the right sacral bone, obtained before (A and B) and after (C and D) high-dose chemotherapy with stem cell support. Initially, there is intense contrast enhancement and marked FDG uptake, both being markedly reduced after treatment. Courtesy of Dr. Bettina Beuthien-Baumann, Heidelberg.

118

Oncology Imaging

contrast-enhanced ultrasound, CT or MRI, or can be formally quantified in dynamic MRI, is another criterion and is for example, being used for gastrointestinal stroma tumors (GIST), or hepatocellular carcinomas. Changes in diffusion-weighted imaging are further, possible criteria that are being investigated. Nevertheless, the current state is that the disappearance or non-progression of lesions remains the chief response criterion, and that functional interim tests for deciding upon a preterm change in treatment regimen are rarely in action, with rare exceptions. In this context, FDG-PET/CT is being used in malignant lymphoma for response assessment after primary chemotherapy, and increasingly for interim staging during treatment. Whether or not these studies will contribute to decisions regarding the therapy regime, is still open.

Acknowledgments The author wishes to cordially thank Dr. Bettina Beuthien-Baumann for her support in the field of nuclear medicine, and for her very helpful discussion of this manuscript.

References Aberle, D.R., Adams, A.M., Berg, C.D., Black, W.C., Clapp, J.D., Fagerstrom, R.M., Gareen, I.F., Gatsonis, C., Marcus, P.M., Sicks, J.D., 2011. Reduced lung-cancer mortality with low-dose computed tomographic screening. The New England Journal of Medicine 365, 395–409. American college of radiologists (2015) PI-RADS. Prostate imagingdreporting and data system Version 2. https://www.acr.org/-/media/ACR/Files/RADS/Pi-RADS/PIRADS-V2. pdf?la¼en. Becker, N., Wahrendorf, J., 1998. Krebsatlas der Bundesrepublik Deutschland 1981–1990. Springer, Berlin/Heidelberg/New York. Becker, N., Motsch, E., Gross, M.L., Eigentopf, A., Heussel, C.P., Dienemann, H., Schnabel, P.A., Eichinger, M., Optazaite, D.E., Puderbach, M., Wielputz, M., Kauczor, H.U., Tremper, J., Delorme, S., 2015. Randomised study on early detection of lung cancer with MSCT in Germany: Results of the first 3 years of follow-up after randomisation. Journal of Thoracic Oncology 10, 890–896. Bickelhaupt, S., Laun, F.B., Tesdorff, J., Lederer, W., Daniel, H., Stieber, A., Delorme, S., Schlemmer, H.P., 2015. Fast and noninvasive characterization of suspicious lesions detected at breast Cancer X-ray screening: Capability of diffusion-weighted MR imaging with MIPs. Radiology 278, 689–697. Bostel, T., Pfaffenberger, A., Delorme, S., Dreher, C., Echner, G., Haering, P., Lang, C., Splinter, M., Laun, F., Müller, M., Jäkel, O., Debus, J., Huber, P.E., Sterzing, F., Nicolay, N.H., 2018. Prospective feasibility analysis of a novel off-line approach for MR-guided radiotherapy. Strahlentherapie und Onkologie. Chenevert, T.L., Stegman, L.D., Taylor, J.M., Robertson, P.L., Greenberg, H.S., Rehemtulla, A., Ross, B.D., 2000. Diffusion magnetic resonance imaging: An early surrogate marker of therapeutic efficacy in brain tumors. Journal of the National Cancer Institute 92, 2029–2036. Cheson, B.D., Pfistner, B., Juweid, M.E., Gascoyne, R.D., Specht, L., Horning, S.J., Coiffier, B., Fisher, R.I., Hagenbeek, A., Zucca, E., Rosen, S.T., Stroobants, S., Lister, T.A., Hoppe, R.T., Dreyling, M., Tobinai, K., Vose, J.M., Connors, J.M., Federico, M., Diehl, V., 2007. Revised response criteria for malignant lymphoma. Journal of Clinical Oncology 25, 579–586. Forstner, R., Thomassin-Naggara, I., Cunha, T.M., Kinkel, K., Masselli, G., Kubik-Huch, R., Spencer, J.A., Rockall, A., 2017. ESUR recommendations for MR imaging of the sonographically indeterminate adnexal mass: An update. European Radiology 27, 2248–2257. Gesellschaft der epidemiologischen Krebsregister in Deutschland e.V. (GEKID) (2014) Krebsatlas der Gesellschaft der epidemiologischen Krebsregister in Deutschland e.V. (GEKID). www.gekid.de. Hillengass, J., Moulopoulos, L.A., Delorme, S., Koutoulidis, V., Mosebach, J., Hielscher, T., Drake, M., Rajkumar, S.V., Oestergaard, B., Abildgaard, N., Hinge, M., Plesner, T., Suehara, Y., Matsue, K., Withofs, N., Caers, J., Waage, A., Goldschmidt, H., Dimopoulos, M.A., Lentzsch, S., Durie, B., Terpos, E., 2017. Whole-body computed tomography versus conventional skeletal survey in patients with multiple myeloma: A study of the International Myeloma Working Group. Blood Cancer Journal 7, e599. Kalender, W.A., Seissler, W., Klotz, E., Vock, P., 1990. Spiral volumetric CT with single-breath-hold technique, continuous transport, and continuous scanner rotation. Radiology 176, 181–183. Kickingereder, P., Wiestler, B., Sahm, F., Heiland, S., Roethke, M., Schlemmer, H.P., Wick, W., Bendszus, M., Radbruch, A., 2014. Primary central nervous system lymphoma and atypical glioblastoma: Multiparametric differentiation by using diffusion-, perfusion-, and susceptibility-weighted MR imaging. Radiology 272, 843–850. Le Bihan, D., Breton, E., Lallemand, D., Aubin, M.L., Vignaud, J., Laval-Jeantet, M., 1988. Separation of diffusion and perfusion in intravoxel incoherent motion MR imaging. Radiology 168, 497–505. Oudkerk, M., Devaraj, A., Vliegenthart, R., Henzler, T., Prosch, H., Heussel, C.P., Bastarrika, G., Sverzellati, N., Mascalchi, M., Delorme, S., Baldwin, D.R., Callister, M.E., Becker, N., Heuvelmans, M.A., Rzyman, W., Infante, M.V., Pastorino, U., Pedersen, J.H., Paci, E., Duffy, S.W., de Koning, H., Field, J.K., 2017. European position statement on lung cancer screening. The Lancet Oncology 18, e754–e766. Rajkumar, S.V., Dimopoulos, M.A., Palumbo, A., Blade, J., Merlini, G., Mateos, M.V., Kumar, S., Hillengass, J., Kastritis, E., Richardson, P., Landgren, O., Paiva, B., Dispenzieri, A., Weiss, B., LeLeu, X., Zweegman, S., Lonial, S., Rosinol, L., Zamagni, E., Jagannath, S., Sezer, O., Kristinsson, S.Y., Caers, J., Usmani, S.Z., Lahuerta, J.J., Johnsen, H.E., Beksac, M., Cavo, M., Goldschmidt, H., Terpos, E., Kyle, R.A., Anderson, K.C., Durie, B.G., Miguel, J.F., 2014. International Myeloma Working Group updated criteria for the diagnosis of multiple myeloma. The Lancet Oncology 15, e538–e548. Seitz, K., Strobel, D., Bernatik, T., Blank, W., Friedrich-Rust, M., Herbay, A., Dietrich, C.F., Strunk, H., Kratzer, W., Schuler, A., 2009. Contrast-enhanced ultrasound (CEUS) for the characterization of focal liver lesionsdProspective comparison in clinical practice: CEUS vs. CT (DEGUM multicenter trial). Ultraschall in der Medizin 30, 383–389. Strahlenschutzkommission (2010) Orientierungshilfe für bildgebende Untersuchungen. Empfehlungen der Strahlenschutzkommission http://www.ssk.de/SharedDocs/ Beratungsergebnisse_PDF/2008/Orientierungshilfe.pdf?__blob¼publicationFile. Strobel, D., Seitz, K., Blank, W., Schuler, A., Dietrich, C.F., von Herbay, A., Friedrich-Rust, M., Bernatik, T., 2009. Tumor-specific vascularization pattern of liver metastasis, hepatocellular carcinoma, hemangioma and focal nodular hyperplasia in the differential diagnosis of 1,349 liver lesions in contrast-enhanced ultrasound (CEUS). Ultraschall in der Medizin 30, 376–382. The Royal College of Radiologists, 2012. iRefer. Making the best use of clinical radiology. The Royal College of Radiologists, London. Wang, J., Zheng, J., Tang, T., Zhu, F., Yao, Y., Xu, J., Wang, A.Z., Zhang, L., 2015. A randomized pilot trial comparing position emission tomography (PET)-guided dose escalation radiotherapy to conventional radiotherapy in chemoradiotherapy treatment of locally advanced nasopharyngeal carcinoma. PLoS One 10, e0124018. Wen, P.Y., Macdonald, D.R., Reardon, D.A., Cloughesy, T.F., Sorensen, A.G., Galanis, E., Degroot, J., Wick, W., Gilbert, M.R., Lassman, A.B., Tsien, C., Mikkelsen, T., Wong, E.T., Chamberlain, M.C., Stupp, R., Lamborn, K.R., Vogelbaum, M.A., van den Bent, M.J., Chang, S.M., 2010. Updated response assessment criteria for high-grade gliomas: Response assessment in neuro-oncology working group. Journal of Clinical Oncology 28, 1963–1972. Zuley, M.L., Guo, B., Catullo, V.J., Chough, D.M., Kelly, A.E., Lu, A.H., Rathfon, G.Y., Lee Spangler, M., Sumkin, J.H., Wallace, L.P., Bandos, A.I., 2014. Comparison of twodimensional synthesized mammograms versus original digital mammograms alone and in combination with tomosynthesis images. Radiology 271, 664–671.

Opisthorchis viverrini, Clonorchis sinensis, and Cholangiocarcinoma Catherine de Martel and Martyn Plummer, International Agency for Research on Cancer, Lyon, France © 2019 Elsevier Inc. All rights reserved.

Opisthorchis viverrini (O. viverrini) and Clonorchis sinensis (C. sinensis) are trematode (flatworm or fluke) parasites belonging to different species, but members of the same family, Opisthorchiidae. They are found in specific endemic areas in Eastern Asia and the Russian Federation. Because these flukes establish chronic infection within intrahepatic bile ducts, they are also commonly called “liver flukes”. Both species are similar in morphology, life cycle, and mode of transmission. After years of chronic infection, the damage caused by the parasites may lead to the development of the bile duct cancer cholangiocarcinoma (CCA), which is the second most common primary liver cancer after hepatocellular carcinoma and bears a very poor prognosis. In 1994, the International Agency for Research on Cancer (IARC) classified O. viverrini as carcinogenic to humans (group 1 in the scale used by IARC to assess strength of evidence); C. sinensis was only classified as probably carcinogen to humans (group 2A) due to the limited number of human studies published at that time. In 2009 however, new data had become available and allowed the classification of C. sinensis in group 1 as well, with sufficient evidence in humans to conclude that both parasites cause CCA in endemic regions.

Geographical Distribution and Burden of Infection Within areas of Eastern Asia and the Russian Federation, the geographic pattern of liver fluke infection is very uneven (Fig. 1), but high rates are more likely seen in rural than urban environments, especially in wetlands and agricultural areas. Prevalence of liver

Fig. 1 Prevalence of liver flukes in Eastern Asia. Modified from Opisthorchis Viverrini and Clonorchis Sinensis. IARC Monograph, 100B, 345. http:// monographs.iarc.fr/ENG/Monographs/vol100B/index.php. Courtesy of Dr. Song Liang, Department of Environmental and Global Health, College of Public Health and Health Professions, University of Florida, USA, who did the art work based on data provided by the Working Group.

Encyclopedia of Cancer, 3rd edition, Volume 3

https://doi.org/10.1016/B978-0-12-801238-3.65041-9

119

120

Opisthorchis viverrini, Clonorchis sinensis, and Cholangiocarcinoma

fluke infection depends on the distribution of fresh water, specific intermediary hosts (snail and fish), and local customs of eating traditional dishes made of raw or undercooked fish. Men are more frequently infected with liver flukes than women, especially in areas where C. sinensis infection is endemic: in these areas, the prevalence sex ratio (M/W) is around 2.0 whereas it is only around 1.2 for O. viverrini infection. Precise estimates of the number of infected people are extremely difficult to obtain as epidemiological studies are few and patchy, and the disease occurrence shows high variation within small geographical areas. For example, the results of field surveys in Kampong Chan province, Cambodia for the prevalence of O. viverrini showed that endemic villages (population prevalence of 30%) and non-endemic villages (population prevalence of 0%) coexisted within a distance of only 9 km, whereas the raw freshwater fish consumption rates were very high in all villages. Nationwide surveys prove very difficult to undertake, until the complex human and environmental factors that influence the distribution of liver fluke infection in endemic areas are fully understood.

C. sinensis The world burden of C. sinensis infection is much higher than that of O. viverrini, as the parasite is found in a much wider area. C. sinensis is endemic in Eastern Asia, namely in China, the Republic of Korea, northern Viet Nam and eastern parts of the Russia Federation (see Fig. 1). It is also probably endemic in the Democratic People’s Republic of Korea, and possibly in Malaysia. Although C. sinensis was highly prevalent in Japan decades ago, it is now extremely rarely seen there. Rare cases have also been reported in some other countries or areas as a result of tourism or immigration. In Viet Nam, C. sinensis is endemic in northern provinces, whereas O. viverrini is detected in central provinces. Estimates of the total number of people infected with C. sinensis in the world have risen from 7 million in the 1990s to 15 million in 2004. This increase is due to several factors: population growth in major endemic areas during this period, the expansion of freshwater fish aquaculture, and better performing food distribution networks. One should add to these factors the difficulty for the elderly to modify their cultural habit of consuming undercooked or raw fish, while the young also become progressively accustomed to this habit. More recent estimates advance a total number varying between 12.5 and 35 million infected people worldwide, acknowledging that these numbers might only reflect the tip of the iceberg. China alone bears the main burden of C. sinensis infection (over 80%). The parasite is found in 25 provinces but the population prevalence varies considerably, ranging from sporadic in some provinces (0.1% in Shandong or Sichuan provinces) to highly prevalent in others (70% in some areas from Guangdong province). The two main endemic areas are situated in the Southeast (Guangxi and Guangdong provinces) and in the Northeast (Heilongjiang province) (see Fig. 1). Note that some areas such as Hong Kong or Shanghai are not endemic for the parasite, but infections may still be acquired theredas in other non-endemic areas of Chinadby eating raw fish transported from endemic areas.

O. viverrini O. viverrini is highly endemic in the northeastern region of Thailand, in Cambodia, in Southern regions of the Lao People’s Democratic Republic, and in central Viet Nam (see Fig. 1). The number of people infected in these countries is estimated to be around 8 million in Thailand and 2 million in the Lao People’s Democratic Republic. No national prevalence data are available in Viet Nam, but several hotspots for the infection have been recognized in the center of the country, including Nam Dinh and Ninh Binh provinces, with population prevalence of around 30%. Similarly, four Cambodian provinces (Takaev, Kandal, Kampong Cham, and Kampong Thom) have been identified as endemic areas of O. viverrini infection in a study conducted between 2006 and 2012 in five provinces, but nationwide surveys are still lacking, and other endemic area are yet to be precisely described.

Life Cycle of the Parasites As food-borne trematodes, O. viverrini and C. sinensis have similar life cycles (see Fig. 2). People that have a habit of eating raw or uncooked freshwater fish constitute the final hosts for the worms, but so do a wide range of wild or domestic fish-eating animals such as dogs, pigs and cats. The elaborate life cycle of the parasite involves two separate intermediate hosts living in fresh water. Adult liver flukes may live in their final hosts for years: in absence of treatment, their lifespan in humans can be as long as 25– 30 years. Laid eggs are produced in the biliary tract by hermaphrodite adult worms (up to 4000 eggs per worm per day) and then eliminated with feces. If eggs happen to fall into freshwater reservoirs, they may then be ingested by specific aquatic snails, which constitute the primary intermediate hosts; eggs hatch into the snail as miracidiae, that is, ciliated larvae, and after a 90 days cycle of reproduction, cercariae, that is, free swimming larvae, escape from the snail into the water and adhere to the flesh of susceptible fish species, secondary intermediate hosts. Cercariae penetrate the subcutaneous tissues and muscle of the fish and slowly mature as metacercariae, its final larval form, within 45 days. The life cycle ends when definitive hosts (i.e., humans or other mammals) catch and eat raw or undercooked infected fish. Through gastric acidic digestion, the metacercariae separate from the fish, excyst in the duodenum of the host and migrate into intrahepatic bile ducts where they develop into adult worms.

Opisthorchis viverrini, Clonorchis sinensis, and Cholangiocarcinoma

121

Fig. 2 Life cycle of O. viverrini and C. sinensis. From Centers for Disease Control and Prevention. https://www.cdc.gov/dpdx/opisthorchiasis/index. html and https://www.cdc.gov/dpdx/clonorchiasis/index.html. Courtesy of Alexander J. da Silva, PhD, and Melanie Moser.

Clinical Symptoms and Complications O. viverrini and C. sinensis share the same clinical manifestations. The variations described in different geographical areas are more likely to reflect the duration and intensity of infection or the nutrition and comorbidity affecting the host than differences due to the species. Clinical symptoms are related to the intensity of the infection. Infections involving < 100 adult worms are usually asymptomatic, but even in heavily infected hosts, only 10% show clinical manifestations. Common symptoms are unspecific and intermittent and may include asthenia, insomnia, pain in the right upper abdominal quadrant, nausea, anorexia, headache, dizziness, weight loss and diarrhea. If undertaken, abdominal ultrasound examination may show an enlarged gallbladder, and the presence of sludge and/or gallstones. Periductal fibrosis seen as periportal echoes along the intrahepatic biliary trees is also a common ultrasonographic feature in chronic liver fluke infections. In children, heavy infection may lead to slower growth or developmental retardation. A small percentage of patients might develop complications, ranging from cholecystis, cholangitis, and pancreatitis, to cholangiocarcinoma. More typical symptoms of these complications may include fever, jaundice, hepatomegaly and liver tenderness.

Liver Flukes and Cholangiocarcinoma The most severe complication of chronic liver fluke infection is the development of CCA. Many pathways could be implicated as the pathogenesis of the infection encompasses several factors. These include mechanical injuries of the biliary ducts by the suckers of the feeding worms; biliary obstruction by the adult parasites leading in turn to stasis and subsequent bacterial infections, including possibly colonization by Helicobacter species; chronic inflammation and excessive immune response from the host; and carcinogenic actions of the parasite’s metabolic products such as Ov-GRN-1 secreted by O. viverrini, which can stimulate angiogenesis, suppress apoptosis, and promote tumor invasion. Epidemiological studies (including correlation ecological studies, case series, and case control studies), animal studies (mostly in hamsters), and mechanistic evidence strongly indicate that liver flukes are causally related to CCA. Various attempts to quantify the

122

Opisthorchis viverrini, Clonorchis sinensis, and Cholangiocarcinoma

number of CCA cases attributable to liver flukes in Asia range from 1300 cases to 7000. All the authors who tried this exercise acknowledged that the estimate they provided was most probably well under the true number. Pooled relative risks (RRs) from the available case control studies were estimated to range from 4 to 8 for both parasites. However, the RR of cholangiocarcinoma might be substantially underestimated by some methods of exposure assessment (i.e., detection of eggs in feces, or patient history self-testimony), both in the population, and in case control studies. Some evidence for this underestimate comes from the contrast between analytical and descriptive epidemiology. For example, the RR between 4 and 8 for C. sinensis from pooled case control studies is unable to account for the 60-fold variation in cholangiocarcinoma incidences between geographical regions. It is therefore probable that all statistical methods are underestimating the true burden with currently available data.

Diagnosis of Liver Flukes Infection Liver fluke infection is suspected on the anamnestic description of recurrent clinical symptoms and raw fish consumption; blood cell count showing an elevated count of eosinophilsdwhite cells whose function is to neutralize invading microbes, primarily parasites, and dilatation of the peripheral intrahepatic bile ducts shown by ultrasound, computed tomography or magnetic resonance imaging scans. Confirmation of diagnosis relies on different types of diagnostic techniques. In most areas, the diagnosis is mainly based on detection of eggs in feces by microscopy, usually using the Kato Katz technique. Trained parasitologists are however needed to make the identification. Even then, because the eggs are similar in size and are all oval and operculated, C. sinensis eggs cannot sometimes be differentiated from other Heterophyidae (e.g., Haplorchis, Heterophyes, etc.) or Opisthorchiidae eggs. The World Health Organization (WHO) has recommended the Kato Katz technique in areas with moderate to high endemicity (i.e., where the proportion of infected individuals is > 10%–50%). Where the prevalence is lower, the low specificity of this technique makes it less appropriate; more specific tools should be used, such as immunodiagnostic techniques and molecular methods. Of note, because endemic areas overlap with areas of high prevalence of hepatitis B virus infection, which shares some of its unspecific symptoms, liver fluke infection might be misdiagnosed as viral hepatitis, and coinfection of the flukes and the virus frequently happen in the same individuals. Whether liver flukes and hepatitis B virus co-infection have a synergistic effect on the development of CCA or hepatocellular carcinoma is plausible but has been little studied to date.

Individual Treatment and Population Control Currently, the major strategies for community prevention and control encompasses fecal examination and treatment of individual cases with praziquantel, a highly effective broad-spectrum trematocidal and custodial drug. The recommended dose is 25 mg/kg three times daily for 2–3 days. A single dose of 40 mg/day has been used in preventive chemotherapy programs and resulted in 95% cure rate of light infections, but only 89% and 69% in intermediate and heavy infections, respectively. For the purposes of public health control, WHO recommends carrying out community screening and diagnosis at the district level, and implementing preventive chemotherapy with praziquantel at a dosage of 40 mg/kg, single administration, as indicated in Table 1. When treatment is delivered, information, education, and communication (IEC) should be given to the population to enhance sustainability by avoiding constant reinfection. Education may be a real challenge in areas where raw fish consumption is a strong tradition, or where people wrongly believe that drinking alcohol or consuming spicy food can prevent infection. No vaccine candidate has yet shown efficacy in animal models, and the control of human disease through vaccination is not an option in the near future. Environmental control is equally challenging and may encompass many measures including prevention of fecal contamination of fresh water, or, in the future, innovative techniques such as GIS-Based spatial identification of risk areas for liver flukes, or oral vaccination of the secondary intermediate hosts (fish) using probiotics.

Table 1

WHO recommendations for community diagnosis

District

Prevalence of infection in the sample population a

High risk

 20

Low risk

< 20

Action to be taken Type of intervention and target population

Interval of re-treatment

Universal treatment (MDA)dTreat all individuals in the district Option 1: universal treatment (MDA)dTreat all individuals in the district Option 2: targeted treatmentdTreat all individuals in the district who report habitually eating raw fish

12 months 24 months 12 months

a The Kato-Katz thick smear technique should be used.Abbreviation: MDA, mass drug administration. From http://www.who.int/foodborne_trematode_infections/clonorchiasis/clonorchiasis_more/en/ and http://www.who.int/foodborne_trematode_infections/opisthorchiasis/viverrini_ more/en/.

Opisthorchis viverrini, Clonorchis sinensis, and Cholangiocarcinoma

123

See also: Cancer Disparities. Cancer-Related Inflammation in Tumor Progression.

Further Reading Fürst, T., Keiser, J., Utzinger, J., 2012. Global burden of human food-borne trematodiasis: A systematic review and meta-analysis. The Lancet Infectious Diseases 12, 2010–2021. IARC, 2012. Opisthorchis viverrini and Clonorchis sinensis. In biological agents. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans 100B, 342–370. http:// monographs.iarc.fr/ENG/Monographs/vol100B/index.php. Lai, D.H., Hong, X.K., Su, B.X., et al., 2016. Current status of Clonorchis sinensis and clonorchiasis in China. Transactions of the Royal Society of Tropical Medicine and Hygiene 110, 21–27. Plummer, M., de Martel, C., Franceschi, S., 2017. Global burden of cancers attributable to liver flukes. The Lancet Global Health 5, e140. Qian, M.B., Utzinger, J., Keiser, J., Zhou, X.N., 2016. Clonorchiasis. The Lancet 387, 800–810.

Relevant Website http://www.who.int/foodborne_trematode_infections/en/dWHO webpage on foodborne infection.

Oral and Oropharyngeal Cancer: Pathology and Genetics Pieter J Slootweg, Radboud University Medical Center, Nijmegen, The Netherlands Justin A Bishop, University of Texas Southwestern Medical Center, Dallas, TX, United States © 2019 Elsevier Inc. All rights reserved.

Abbreviation HPV Human papilloma virus

Oral Squamous Cell Carcinoma Definition A carcinoma with squamous differentiation arising from the mucosal epithelium. Recognized subtypes are: verrrucous carcinoma, basaloid squamous cell carcinoma, papillary squamous cell carcinoma, spindle cell squamous cell carcinoma, adenosquamous carcinoma and lymphoepithelial carcinoma. Subtypes only will be discussed regarding their pathologic features as they are with a few exceptions morphological variants without any proven relationship with other clinicopathological variables.

Burden Squamous cell carcinoma represents > 90% of cancers in the oral cavity. With respect to epidemiology, specific geographical regions must be considered separately, because there is a marked variation in incidence. A high incidence of oral cancer is found in southern Asia with age-standardized incidence rates of > 10 cases per 100,000 population per year. Worldwide, oral cancer incidence is higher among males (5.5 cases per 100,000 population per year) than females (2.5 cases per 100,000). Most oral cancers occur in patients aged 50–70 years.

Risk Factors Smoking is by far the most important cause of oral cancer. Alcohol consumption interacts synergistically with smoking, resulting in a more than additive risk. In some parts of the world, the use of areca nut whether or not mixed with other substances represents an important risk factor. In contrast to the oropharynx, oral cavity squamous cancers only rarely harbor high-risk HPV.

Pathology Most cancers of the oral cavity and mobile tongue are moderately or well differentiated; poorly differentiated squamous cell carcinoma is less frequent. Well-differentiated OSCC is characterized by nests, cords, and islands of large cells with pink cytoplasm, prominent intercellular bridges, and round nuclei, which may not be obviously hyperchromatic. Dyskeratotic cells and squamous pearls are prominent. Well-differentiated tumors tend to invade in large islands, whereas less-differentiated tumors tend to have small islands and dispersed individual cells at the invasive front (Figs. 1 and 2).

Pathology of Subtypes Verrucous Carcinoma Verrucous carcinoma consists of blunt projections and invaginations of well-differentiated squamous epithelium, composed of one to several layers of basal cells and an expanded layer of spinous cells that lack cytological atypia. There is marked surface keratinization. Mitoses are rare and confined to the basal cell layer, and there are no abnormal mitoses. Verrucous carcinoma invades the stroma with a well-defined pushing border (Fig. 3). Verrucous carcinoma has a better prognosis than conventional oral squamous cell carcinoma.

Basaloid Squamous Cell Carcinoma Basaloid squamous cell carcinoma consists of rounded nests with smooth borders and peripheral palisading. Gland-like foci with basophilic myxoid or mucoid material are common and mimic true gland formation. A variable degree of nuclear pleomorphism is present, and high mitotic activity, apoptosis, and necrosis are common. Also, comedonecrosis may be present. Stromal hyalinization is characteristic; it can be linear between and around nests and nodular within nests. Keratinization occurs abruptly adjacent to

124

Encyclopedia of Cancer, 3rd edition, Volume 3

https://doi.org/10.1016/B978-0-12-801238-3.65090-0

Oral and Oropharyngeal Cancer: Pathology and Genetics

Fig. 1

Well-differentiated OSCC arises from the surface epithelium and invades the submucosal tissues as large keratinizing tumor nests.

Fig. 2

Poorly-differentiated OSCC lacks significant keratinization and irregularly infiltrates the stroma as small cords or individual cells.

125

basaloid cells (Fig. 4). Whether basaloid squamous cell carcinoma has a worse prognosis than conventional squamous cell carcinoma is an issue that has not yet been definitely settled.

Papillary Squamous Cell Carcinoma Papillary squamous cell carcinoma is characterized by a papillary growth pattern, with thin fibrovascular cores covered with malignant epithelial cells (Fig. 5). Two types of surface epithelium are described: one resembles high-grade keratinizing epithelial dysplasia, and in the other, the epithelial cells are immature and basaloid, with no evidence of maturation or keratinization. This type of squamous cell carcinoma has a better prognosis than conventional squamous cell carcinoma, primarily due to lowstage presentation, with a low metastatic potential.

Spindle Cell Squamous Cell Carcinoma Spindle cell squamous cell carcinoma typically grows as an exophytic mass with a predominantly ulcerated surface, sometimes containing remnants of dysplastic squamous epithelium and frequently showing areas of transition from squamous cells to malignant spindled or pleomorphic tumor cells with hypercellularity, necrosis, atypical mitotic figures, and hyperchromatic nuclei (Fig. 6). Some cases exhibit heterologous mesenchymal differentiation in the form of malignant bone, cartilage, or skeletal muscle.

126

Oral and Oropharyngeal Cancer: Pathology and Genetics

Fig. 3

Verrucous carcinoma exhibits prominent surface keratinization and a broad, pushing pattern of stromal invasion.

Fig. 4

Basaloid squamous cell carcinoma consists of basaloid cells with high nuclear:cytoplasmic ratios demonstrating abrupt keratinization.

Occasionally, spindle cell squamous cell carcinoma may have deceptively bland appearance, mimicking reactive myofibroblastic proliferation or granulation tissue. The prognosis of this subtype does not differ from conventional squamous cell carcinoma.

Adenosquamous Carcinoma Adenosquamous carcinoma shows both squamous and glandular differentiation. The adenocarcinoma consists of cribriform and tubuloglandular structures and tends is mainly restricted to the deeper parts of the tumor (Fig. 7). This subtype is rare in the oral cavity, with larynx and hypopharynx being more common sites. Prognosis is worse than for conventional squamous cell carcinoma.

Lymphoepithelial Carcinoma Lymphoepithelial carcinoma is a squamous cell carcinoma subtype morphologically similar to non-keratinizing nasopharyngeal carcinoma, undifferentiated subtype. The tumor is characterized by large cells with round to oval vesicular nuclei, and large central nucleoli. The neoplastic cells generally have scant amphophilic or eosinophilic cytoplasm. Unlike in the nasopharynx, lymphoepithelial carcinoma in the oral cavity appears to be unrelated to EBV. Lack of data on outcome precludes any firm statements.

Oral and Oropharyngeal Cancer: Pathology and Genetics

Fig. 5

127

Papillary squamous cell carcinoma grows as exophytic fronds of malignant squamous epithelium surrounding fibrovascular cores.

Fig. 6 Spindle cell squamous cell carcinoma has a component of conventional squamous cell carcinoma and a separate component of undifferentiated malignant spindle cells.

Fig. 7

Adenosquamous carcinoma demonstrates clear-cut glandular differentiation, particularly in the deeper portions of the tumor.

128

Oral and Oropharyngeal Cancer: Pathology and Genetics

Molecular Pathology and Genetics Most OSCCs are genetically unstable with chromosomal loss at 3p, 8p, 9p, 17p and gains at 3q and 11q. These changes may extend for some distance from the clinical lesion, underpinning the clinical phenomenon of field cancerization. Genes reported to have a role in OSCC (e.g., TP53, CDKN2A, PTEN, HRAS, PIK3CA, and NOTCH1) are mutated with sufficient frequency to support their potential role as drivers in cancer development.

Microenvironment Including Immune Response Data regarding the immune microenvironment of oral squamous cell carcinoma is limited. A few studies have reported a subset of these cancers expressing the immune checkpoint receptor PD-L1.

Staging and Grading Most cases of oral squamous cell carcinoma are moderately differentiated. There is no clear relationship between grading and prognosis. There are some indications however that evaluation of the invasive tumor front may have some prognostic relevance, with growth in tiny strands being worse than invasion in a pushing border. Staging is done conforming the most recent UICC guidelines (8th edition).

Prognostic and Predictive Biomarkers An important prognostic factor is the status of surgical resection margins. Resection margins clear of tumor are associated with a lower recurrence rate and a better survival. An adequate margin of resection has not been precisely defined, but for oral squamous cell carcinoma margins of 5 mm are generally believed to be adequate. Other significant prognostic factors are tumor size, nodal status, and distant metastasis. Histological risk factors associated with a worse prognosis include a non-cohesive growth pattern at the invasion front, perineural and lymphovascular invasion, bone invasion, and tumor thickness > 4 mm. High-grade dysplasia at a mucosal margin correlates with local recurrence and second primary tumors. Extracapsular spread from metastasis in the neck, two or more positive nodes, and involvement of lower neck worsen prognosis. Conventional histological grading correlates poorly with clinical outcome.

Prospective Vision Current research on oral squamous cell carcinoma has included work on improving early detection through screening saliva or serum for biomarkers, refining risk progression models particularly for low-stage cancers that behave unpredictably, and investigating novel treatment strategies like targeted therapies and immune checkpoint inhibition to improve a prognosis that has largely remained unchanged for decades.

Oropharyngeal Squamous Cell Carcinoma Definition Oropharyngeal squamous cell carcinoma is an epidemiologically, pathologically, and clinically distinct form of head and neck squamous cell carcinoma associated with high-risk HPV.

Burden The incidence of oropharyngeal carcinoma has risen over the past three decades. Most patients are middle-aged males.

Risk Factors Oropharyngeal squamous cell carcinoma is caused by high-risk HPV, with type 16 responsible for > 90% of all cases. Oral sex is an established risk factor for oral HPV infection.

Pathology HPV associated oropharyngeal squamous cell carcinoma generally exhibits a distinctive non-keratinizing morphology (Fig. 8). The tumor originates in the crypt epithelium and grows beneath the surface epithelial lining as nests and lobules. Tumor nests are often embedded in lymphoid stroma, and may be penetrated by lymphoid cells. Similar to squamous cell cancer of the oral cavity, the

Oral and Oropharyngeal Cancer: Pathology and Genetics

129

Fig. 8 HPV associated oropharyngeal squamous cell carcinoma grows as lobules of primitive cells with high nuclear:cytoplasmic ratios and limited keratinization.

morphological spectrum includes variants with papillary, adenosquamous, lymphoepithelioma-like, and spindle cell features. The clinical behavior of these morphological variants is the same as for the cases showing the typical histomorphology.

Molecular Pathology and Genetics HPV oncoproteins E6 and E7 inactivate p53 and RB by targeting them for protein degradation, thus disabling the protective effects of these tumor suppressor genes. Oropharyngeal HPV-driven cancer has a gentic profile that differs from smoking-driven squamous cell carcinomas, with fewer mutations overall and fewer TP53 mutations, but more frequent PIC3CA mutation or amplification and occasional TRAF3 mutations not seen in HPV-unrelated tumors.

Microenvironment Including Immune Response HPV associated oropharyngeal carcinomas arise from the specialized tonsillar crypt epithelium. These crypts have a distinct and complex immune environment that is not fully understood. As one important example, the immune checkpoint receptor PD-L1 is selectively expressed in the tonsillar crypt epithelium, seemingly creating an immune-privileged site where the effector function of virus-specific T cells is downmodulated, facilitating immune evasion by HPV.

Staging and Grading Histologic grading is not currently advocated. Staging is done conforming the most recent UICC guidelines (8th edition).

Prognostic and Predictive Biomarkers HPV associated oropharyngeal squamous cell carcinoma shows a better prognosis than the HPV negative tumors seen at this site but more commonly in the oral cavity. This improved prognosis is diminished in smokers with HPV associated oropharyngeal squamous cell carcinoma.

Prospective Vision Future research directions in HPV associated squamous cell carcinoma research include early detection (made challenging by a lack of pre-malignant epithelial changes), novel treatment strategies like de-escalation radiation therapy or immunotherapy, and the long-term effects of routine HPV vaccination on the incidence of these carcinomas.

See also: Cancers of the Oral Cavity: Diagnosis and Treatment. Nasopharyngeal Carcinoma: Diagnosis and Treatment.

130

Oral and Oropharyngeal Cancer: Pathology and Genetics

Further Reading Brierly, J.D., Gospodarowicz, M.K., Wittekind, C. (Eds.), 2017. TNM classification of malignant tumours, 8th edn. Wiley-Blackwell, Chichester. Gale, N., Zidar, N., Cardesa, A. and Nadal, A. (2016). Benign and potentially malignant lesions of the squamous epithelium and squamous cell carcinoma. In: Cardesa A., Slootweg P.J., Gale N. and Franchi A. (eds.) Pathology of the head and neck. 2nd edn, pp. 1–48. Heidelberg: Springer. Lydiatt, W.M., Patel, S.G., O’Sullivan, B., et al., 2017journal. Head and neck cancers – major changes in the American Joint Committee on Cancer eighth edition cancer staging manual. CA: a Cancer Journal for Clinicians 67, 122–137. Lyford-Pyke, S., Peng, S., Young, G.D., et al., 2013journal. Evidence for a role of the PD-1:PD-L1 pathway in immune resistance of HPV-associated head and neck squamous cell carcinoma. Cancer Research 73, 1733–1741. Ramos-García, P., Ruiz-Ávila, I., Gil-Montoya, J.A., et al., 2017journal. Relevance of chromosomal band 11q13 in oral carcinogenesis: an update of current knowledge. Oral Oncology 72, 7–16. Rivera, C., Oliveira, A.K., Costa, R.A.P., et al., 2017journal. Prognostic biomarkers in oral squamous cell carcinoma: a systematic review. Oral Oncology 72, 38–47. Sloan, P., Gale, N., Hunter, K., et al., 2017book. Tumours of the oral cavity and mobile tongue. In: El-Naggar, A.K., Chan, J.K.C., Grandis, J.R., Takata, T., Slootweg, P.J. (Eds.), WHO classification of head and neck tumours, 4th edn. Lyon, IARC, pp. 105–131. The Cancer Genome Atlas Network, 2015journal. Comprehensive genomic characterization of head and neck squamous cell carcinomas. Nature 517, 576–582. Westra, W.H., Boy, S., el-Mofty, S.K., et al., 2017book. Tumours of the oropharynx (base of tongue, tonsils, adenoids). In: El-Naggar, A.K., Chan, J.K.C., Grandis, J.R., Takata, T., Slootweg, P.J. (Eds.), WHO classification of head and neck tumors, 4th edn. Lyon, IARC, pp. 133–146.

Oral Cavity Cancer: Diagnosis and Treatment Corbin D Jacobs, Michael J Moravan, Jennifer Choe, Russel Kahmke, Yvonne Mowery, and Joseph K Salama, Duke University School of Medicine, Durham, NC, United States © 2019 Elsevier Inc. All rights reserved.

Glossary Alveolar ridge One of the two jaw ridges either on the roof of the mouth between the upper teeth and the hard palate or on the bottom of the mouth behind the lower teeth. Erythroplakia Chronic, red, generally asymptomatic lesion or patch on the mucosal surface that cannot be attributed to a traumatic, vascular, or inflammatory cause. Extranodal extension Extension of metastatic carcinoma from within a lymph node through the fibrous capsule and into the surrounding connective tissue. Interstitial brachytherapy Radioactive material inserted directly into body tissue. Leukoplakia A white patch or plaque on the mucosal surface that cannot be rubbed off and may represent hyperplastic or dysplastic tumors. Retromolar trigone A paired triangular area situated posterior to the third mandibular molars on the left and right sides of the oral cavity.

Nomenclature ABS American Brachytherapy Society AIDS Acquired immunodeficiency syndrome AJCC American Joint Committee on Cancer ARO Arbeitsgemeinschaft Radiologische Onkologie (i.e., Association of Radiological Oncology) ASCO American Society of Clinical Oncology CI Confidence interval CT Computed tomography EGFR Epidermal growth factor receptor EORTC European Organization for Research and Treatment of Cancer ESTRO European Society for Therapeutic Radiology and Oncology GEC Groupe Européen de Curiethérapie (i.e., European Brachytherapy Group) GITR Glucocorticoid-induced tumor necrosis factor receptor-related protein GORTEC Groupe d’Oncologie Radiothérapie Tête et Cou (i.e., Radiotherapy Oncology Group Head and Neck) Gray (Gy) Joule per kilogram HDR High dose rate HIGRT Hypofractionated image-guided radiation therapy (HIGRT) HPV Human papillomavirus HR Hazard ratio HSV Herpes simplex virus IDO Indoleamine-pyrrole 2,3-dioxygenase IGRT Image-guided radiotherapy IMRT Intensity modulated radiotherapy KIR Killer-cell immunoglobulin-like receptor MACH-NC Meta-analysis of chemotherapy in head and neck cancer MARCH Meta-analysis of radiotherapy in carcinoma of the head and neck MRI Magnetic resonance imaging NCCN National Comprehensive Cancer Network NCDB National Cancer Database NRG National Clinical Trials Network group created through the coordinated efforts of the National Surgical Adjuvant Breast and Bowel Project (NSABP), the Radiation Therapy Oncology Group (RTOG), and the Gynecologic Oncology Group (GOG) OCAT Oral Cavity Adjuvant Therapy trial PD-1 Programmed cell death protein-1

Encyclopedia of Cancer, 3rd edition, Volume 3

https://doi.org/10.1016/B978-0-12-801238-3.65316-3

131

132

Oral Cavity Cancer: Diagnosis and Treatment

PET Positron emission tomography PORT Postoperative radiotherapy RTOG Radiation Therapy Oncology Group SABR Stereotactic ablative radiotherapy SBRT Stereotactic body radiation therapy SEER Surveillance Epidemiology and End Results TIGIT T-cell immunoreceptor with Ig and ITIM domains VMAT Volumetric modulated arc therapy

Introduction The oral cavity consists of the oral vestibule and oral cavity proper. Its primary function is to serve as the entrance of the alimentary tract and to initiate the digestive process. It also serves as a secondary respiratory conduit, a site of sound modification for the production of speech, and a chemosensory organ. The development of a primary malignancy within the oral cavity as well as subsequent cancer treatment can affect these normal functions. While cancers of the oral cavity only account for approximately 2% of all new cancer diagnoses in the United States, it is the most common site of all head and neck cancers (Siegel et al., 2017). In India, oral cavity cancer is among the three most common malignancies, accounting for over 30% of all reported malignancies in the country due to the high prevalence of chewing and smoking various carcinogens (Sankaranarayanan et al., 2005; Coelho, 2012). Surveillance Epidemiology and End Results (SEER) program data estimate 32,670 cases of cancer of the oral tongue, mouth, and oral cavity for 2017 in the United States (Siegel et al., 2017). An estimated 300,400 new cases and 145,400 deaths from cancer of the oral cavity and lip occurred worldwide in 2012 (Torre et al., 2015). Oral cavity cancer incidence rates worldwide have increased in many countries with tobacco epidemics that are currently peaking, but incidence rates have declined significantly among both males and females in areas where tobacco use peaked previously (Warnakulasuriya, 2009; Chaturvedi et al., 2013; Franceschi et al., 2000; Simard et al., 2014). Incidence rates for oral cancer sites related to human papillomavirus (HPV) infections are increasing in some countries, hypothesized to be in part due to changes in oral sexual behavior (Chung et al., 2014; Fig. 1).

Incidence ASR Both sexes

Cancer of the lip and oral cavity 5.1+ 3.8-5.1 2.5-3.8 1.9-2.5 2 mm or gross ENE).

Patterns of Spread About 90% of all squamous cell carcinomas of the lip start on the sun-exposed lower lip vermilion (Fitzpatrick, 1984). Basal cell carcinomas, alternatively, are more common on the upper lip. The commissures are rarely sites of disease origin. The disease may spread along the lip to involve the buccal mucosa or spread deeply to involve the gingiva, mandible, and in advanced cases the mental nerve (Baker & Krause, 1980). Lymphatic invasion and nodal positivity occurs in about 5%–10% of lip cancer cases overall. This rate is higher for bulkier and more advanced tumors. Draining lymphatics and therefore sites of nodal spread for upper versus lower lip tumors, as well as all other intraoral primary tumors, are specified in the prior Anatomy section (Fig. 9). Squamous cell carcinomas of the oral tongue tend to originate on the lateral and ventral surfaces of the tongue. Most lateral tumors tend to be located on the middle and posterior thirds of the oral tongue. Dorsal tumors are rare and tend to occur in the posterior midline. Tumors on the tip of the tongue are usually diagnosed early. Local invasion into surrounding structures generally occurs late, but when it occurs, it can spread anteriorly or laterally into the floor of mouth and mandible; posteriorly into the glossotonsillar sulcus, retromolar trigone, tonsillar pillars, and base of tongue; and deep into the oral tongue musculature. Clinical nodal positivity ranges from 15% to 80%, the highest risk of nodal metastases among all oral cancer subsites, and is dependent on the size of the primary tumor and depth of invasion (Rusthoven et al., 2008; Almangush et al., 2015). For patients with clinically positive nodes, the risk of occult contralateral nodal disease is about 30%. Cancers of the anterior oral tongue have been documented as being able to skip levels II–III and spread directly to level IV (Byers et al., 1997; Woolgar, 1997). Carcinomas of the floor of mouth, in contrast to oral tongue cancer, generally invade into surrounding structures early. About 90% of tumors start within 2 cm of the anterior midline floor. The tumors spread anteriorly into the lower gingiva and lower lip; posteriorly into muscles at the root of the tongue; laterally into the gingiva and along the mandibular periosteum or through the cortex in advanced stages; and deep into the genioglossus, geniohyoid, and sublingual gland. The mylohyoid acts as a barrier to deeper extension, until the tumor is more advanced, such that disease spreads along the muscle, escapes the oral cavity, and emerges in the neck near the angle of the mandible mimicking an enlarged lymph node. Lymphatic invasion and clinical nodal positivity range from 10% to 55% and are directly correlated to T-stage. For patients with clinically positive nodes, the risk of occult contralateral nodal disease is about 50%. Most gingival tumors, roughly 80%, arise on the lower gingiva. Carcinomas of the alveolar ridge and retromolar trigone tend to invade bone early. Tumors of the lower alveolar ridge and retromolar trigone also have a high propensity of lymph node metastasis, second only to oral tongue carcinomas among all oral cavity subsites. Tumors of the inferior alveolar ridge may access the mandibular canal and the inferior alveolar nerve, whereas tumors of the superior alveolar ridge may invade into the maxillary antrum or floor of the nose. Tumors of the superior alveolus are less likely to metastasize to the neck than tumors of the inferior alveolus. Carcinomas of the buccal mucosa commonly arise on the lateral walls, adjacent to the lower molars. One route of progression is into the gingivobuccal gutters, and ultimately invasion of the gingiva and bone can arise. Alternatively, they can invade the buccinator muscle, extend to the buccal fat pad, and invade subcutaneous tissues. More advanced tumors can even go on to involve the parotid gland and facial nerve. The likelihood of clinical nodal positivity at the time of diagnosis is approximately 10%–30%, and the risk of subclinical disease is between 5% and 15%. Bilateral neck involvement is rare for this subsite. The risk of nodal involvement increases to 60% for more advanced stage tumors (i.e., T4). The hard palate has a relatively dense mucoperiosteum that is relatively resistant to invasion from hard palate carcinomas. Carcinomas of the hard palate can invade superiorly into the nasal cavity through the incisive fossa where the primary and secondary palates are fused. Furthermore, these tumors may invade into the surrounding gingiva and maxillary alveolar process, posteriorly into the soft palate, and posterolaterally through the greater palatine foramina into the pterygopalatine fossa which communicates with the skull base. The rate of nodal involvement at the time of diagnosis is about 13%–24% with a risk of subclinical involvement of 22% (Fig. 10). The superior deep jugular nodes are the most frequently involved sites of lymphatic metastasis by cancers of the oral cavity. Approximately 25% of patients with carcinoma of the oral cavity will have occult nodal metastases. Contralateral metastases are more common in tumors that approach or cross the midline. In general, cervical lymph node involvement from oral cavity primary sites is predictable and orderly, spreading first to upper, then middle, and subsequently lower cervical nodes. Hematogenous invasion is generally a rare initial presentation for oral cavity cancer. The lungs are the most common site of distant metastases; skeletal and hepatic metastases occur less often. Mediastinal lymph node metastases are considered distant

Oral Cavity Cancer: Diagnosis and Treatment

141

Fig. 9 Lip carcinoma. (A, B) Photographs show extensive ulceration and infiltration with erythroleukoplakia of the right lower lip and labial mucosa. Extension across the midline is seen intraorally (white arrow in A) and ulceration with breakdown of the skin is seen extraorally (black arrow in B). (C, D, E) Axial postcontrast T1-weighted images from superior to inferior show an enhancing mass extending from the skin surface to the buccal surface of the mandible, without evidence of bone invasion. Note the area of ulceration superiorly (arrow in C). From Hagiwara, M., Nusbaum, A., Schmidt, B. L. (2012). MR assessment of oral cavity carcinomas. Magnetic Resonance Imaging Clinics of North America 20(3), 473–494.

metastases, except level VII or anterior superior mediastinal lymph nodes cephalad of the innominate artery. An increased risk of distant metastasis is associated with a higher burden of lymph node involvement, advanced primary tumor stage, and recurrent disease.

Presentation and Diagnosis Clinical Presentation Although the oral cavity is an anatomic region readily accessible to visual inspection and palpation, many patients present with advanced-stage disease because of vague and initially painless symptoms. About 30%–40% of patients with cancer of the oral cavity harbor cervical lymph node metastases at the time of diagnosis. The most common presenting symptom for cancer of the oral cavity is a nonhealing lesion. Similarly, persistent pain in the mouth that does not resolve is also very common. Other presenting signs and symptoms include persistent leukoplakia or

142

Oral Cavity Cancer: Diagnosis and Treatment

Fig. 10 Left hard palate mucoepidermoid carcinoma. (A) Axial contrast-enhanced CT shows 2-cm enhancing mass with fairly well-circumscribed borders (arrow) in the left hard palate. Well-circumscribed borders are common in minor salivary gland malignancies, making it difficult to differentiate them from benign minor salivary gland tumors. (B, C) Bone window from axial CT images (at the level of the hard palate (B) and just above) show asymmetric widening of the left greater palatine foramen (arrow), compatible with perineural tumor extension. (D) Bone window from axial CT images shows extension of abnormal widening to the left pterygopalatine fossa (arrow). An MR image is needed to assess for proximal intracranial extension to determine if this tumor was resectable. From Aiken, A. H. (2013). Pitfalls in the staging of cancer of oral cavity cancer. Neuroimaging Clinics of North America 23(1), 27–45.

erythema, difficulty chewing or swallowing, limited tongue mobility, bleeding, odynophagia, trismus, constant halitosis, globus sensation, otalgia, loose teeth, ill-fitting or uncomfortable dentures, voice changes, weight loss, numbness, and persistent or enlarging masses in the mouth and/or neck despite antibiotics. Otalgia suggests involvement of cranial nerves V, VII, IX, or Xdthough the auriculotemporal branch of the mandibular division of the trigeminal nerve is the most commonly involved in oral cavity primary tumors (Scarbrough et al., 2003). Facial numbness may suggest involvement of the trigeminal nerve. Hypoesthesia of the lower lip and teeth is often from malignant penetration of the mandible and perineural spread along the inferior alveolar nerve. The presence of trismus may indicate invasion into the pterygoid musculature. The differential diagnosis of lip cancer includes actinic cheilitis, leukoplakia, herpes simplex, syphilitic chancre, keratoacanthoma, hemangioma, and fibroma. The differential diagnosis for an oral tongue cancer includes granular cell myoblastoma, pyogenic granuloma, and hypertrophied papillae. The differential diagnosis for other intraoral cancers includes leukoplakia, candidiasis, and lichen planus.

Diagnostic Evaluation Patients who present with cancer of the oral cavity should undergo a comprehensive history and physical examination. A detailed history of present illness should specifically evaluate for any of the aforementioned symptoms and identify carcinogenic risk factors, such as tobacco, alcohol, or betel nut use. A complete past medical history should be performed to take into account comorbid illnesses when creating an ideal treatment plan. A family medical history should be obtained to evaluate for inheritable genetic mutations. A full review of systems should be performed to evaluate for concurrent illnesses or systemic manifestations of the disease. A meticulous physical examination of the head and neck should be performed, with particular focus on the oral cavity. This typically begins with a full inspection of the oral cavity in good lighting. Bimanual palpation of the oral cavity can help assess extent of induration, fixed tissues, and bony involvement. A thorough palpation of the neck is important to assess regional nodal disease.

Oral Cavity Cancer: Diagnosis and Treatment

143

Evaluation of other areas of the aerodigestive tract for synchronous primaries and functional deficits can be performed by indirect pharyngolaryngoscopy using small mirrors and/or direct pharyngolaryngoscopy using a flexible fiberoptic scope. The next important step in diagnostic evaluation is biopsy to confirm malignancy. Histopathologic sampling can be performed by fine needle aspiration with or without ultrasound guidance, core needle biopsy, incisional biopsy, or exfoliative cytology. Immunohistochemistry staining can be performed on core needle and incisional biopsy specimens. Tumor HPV status can be determined using various testing platforms, including HPV DNA in situ hybridization, HPV RNA in situ hybridization, HPV polymerase chain reaction, and p16 immunochemistry (Duncan et al., 2013). The immunohistochemical staining for p16 is inexpensive, has near universal availability, and is relatively straightforward to interpret (Singhi & Westra, 2010; Lydiatt et al., 2017; Fig. 11). Imaging modalities that may be used for staging, treatment planning, and posttreatment monitoring include X-ray, ultrasound, computed tomography (CT), magnetic resonance imaging (MRI), and positron emission tomography (PET). A chest CT is helpful to exclude lung metastases or identify a synchronous primary lung cancer, especially in patients with a significant smoking history. A head and neck CT, ideally with IV contrast, is the most commonly used modality to determine the extent of soft-tissue and bony involvement as well as to identify occult disease in the neck. The “puffed-cheek” maneuver during CT scanning may be helpful to better visualize and assess extent of invasion for buccal and gingival mucosa primary tumors (Weissman & Carrau, 2001). If CT scanning is not available, then panoramic radiographs can be used to demonstrate mandibular invasion. MRI may be used in cases of contrast allergy, poorly visualized lesions on CT such as from significant dental artifact, to establish the extent of primary tumor, and evaluating for perineural spread. Ultrasound may be used to screen for enlarged lymph nodes that are not clinically detectable, and in experienced hands the accuracy of ultrasound when combined with fine needle aspiration may be superior to CT or MRI for staging the neck (van den Brekel et al., 1993). PET-CT is highly sensitive in staging of patients with squamous cell carcinoma of the head and neck, but it is insensitive to evaluation of patients with glandular tumors, low volume tumors, and cystic metastases (Escott, 2013; Johnson & Branstetter, 2014). A prospective cohort study demonstrated that PET-CT based staging had a significantly higher detection rate of distant metastasis or synchronous cancer compared to other combined modalities (Rohde et al., 2017). The presence of ENE, which is associated with worse outcome, is most commonly diagnosed pathologically, but it may also be identified on physical exam and imaging. Clinically, ENE may be diagnosed by the presence of a matted mass of nodes, involvement of overlying skin or adjacent soft tissues, or clinical signs of nerve invasion. Radiographically, ENE may be diagnosed by an indistinct nodal margin, irregular nodal capsular enhancement, or infiltration into adjacent fat or muscle. Basic blood tests such as complete blood count, comprehensive metabolic panel, and liver function tests are useful to determine overall health and underlying organ function. These can diagnose anemia, malnutrition, kidney disease, or liver disease. Occasionally basic labs may also suggest metastatic disease to the liver or bone. These labs in combination with complete past medical history can also guide appropriate treatment, such as surgery or preferred systemic agent based on predicted treatment-related toxicity. Other diagnostic tests to consider in certain patients include electrocardiogram and pulmonary function tests prior to surgery. Candidates for radiation therapy should ideally undergo dental evaluation prior to initiation of treatment. Additionally, many patients benefit from multidisciplinary referrals to nutritionists and speech therapists.

Staging Historically, cancers of the oral cavity were predominately staged based on the size of the primary tumor; invasion into surrounding structures; the size, laterality, and number of nodal metastases; and the presence or absence of distant metastatic disease. The American Joint Committee on Cancer (AJCC) 8th Edition introduced two new parameters in oral cavity staging: depth of invasion and

Fig. 11 Usefulness of reangled images to visualize squamous cell carcinoma of the right oral tongue. (A) Axial CT through the oral cavity is nondiagnostic because of dental amalgam and extensive streak artifact. (B) Delayed and reangled axial contrast-enhanced CT image shows a large enhancing right oral tongue mass (arrow), effacing the right sublingual space. From Aiken, A. H. (2013). Pitfalls in the staging of cancer of oral cavity cancer. Neuroimaging Clinics of North America 23 (1), 27–45.

144

Oral Cavity Cancer: Diagnosis and Treatment

ENE. Effectively, depth of invasion will increase the T category by 1 for each 5 mm of tumor depth until
10 mm. Pathological presence of ENE will increase the nodal category by 1 to either N2a or N3b depending on the size, laterality, and quantity of nodes involved.

Management General Management Treatment goals for cancer of the oral cavity include eradication of malignancy, preservation or restoration of form and function, avoidance or minimization of treatment sequelae, and prevention of second cancers. The ideal treatment to achieve these goals, whether as single modality or in combination, depends on various disease-, treatment-, provider-, and patient-related factors. Examples of these respective factors include tumor stage and location; potential treatment morbidity and convenience; multidisciplinary team expertise and available technology; and patient performance status and comorbid disease. A multidisciplinary approach is paramount in the optimal management of patients with cancer of the oral cavity. Multidisciplinary evaluation prior to treatment disposition helps to ensure consensus recommendations are made and to facilitate interdisciplinary coordination of care. Early stage oral cavity tumors (e.g., T1-T2, N0-N1) are generally managed with single modality locoregional treatment consisting of surgery or radiation therapy. Surgery is generally preferred, but radiation therapy is appropriate when overall health or functional status precludes an operation, when there is an anticipated poor functional or cosmetic outcome, or simply when a patient declines surgery. For patients with locally advanced oral cavity tumors (e.g., T3-T4, N2-N3), upfront surgery followed by adjuvant radiation therapy with or without concurrent chemotherapy is generally preferred. However, in patients unable to undergo surgical resection, definitive chemoradiotherapy has been shown to be a reasonable option.

Prognostic and Predictive Factors A series of nomograms and retrospective analyses have identified factors associated with recurrence and survival. One such nomogram predicts preoperative prognosis in patients with oral cavity squamous cell carcinoma and was designed after reviewing demographic, host, and tumor characteristics of 1617 patients treated primarily with surgery at a single-institution tertiary care cancer center between 1985 and 2009 (Montero et al., 2014). The most influential predictors of both recurrence and cancer-specific mortality probability were tumor size, nodal status, subsite, and bone invasion. A retrospective analysis of a single United States institution’s 35-year experience treating 230 primary invasive squamous cell carcinomas of the oral cavity with surgery and postoperative radiotherapy (PORT) demonstrated that the following variables were significant on multivariate analysis for locoregional control: positive margins, vascular invasion, perineural invasion, extracapsular extension, and T stage (Hinerman et al., 2004). One of the consistently reported prognostic factors for oral cavity cancers is depth of primary tumor invasion, which is associated with occult cervical metastases, especially in oral tongue cancer. Although the oral cavity subsites, study designs, and techniques for measuring depth of invasion varied significantly, a review of > 50 studies identified tumor thickness/depth of invasion as predictive of nodal involvement and survival (Pentenero et al., 2005). A multicenter international study of patients with T1-T2, N0 oral tongue squamous cell carcinoma identified depth of invasion and worst pattern of invasion as strong predictors for locoregional recurrence and cancer-related mortality (Almangush et al., 2015). Consequently, the authors proposed that multimodality treatment should be utilized for early-stage oral tongue carcinomas with depth of invasion
4 mm or with a growth pattern characterized by small cell islands or satellites. This was supported by a meta-analysis, which analyzed sixteen studies consisting of 1136 patients with oral cavity squamous cell carcinoma to assess the predictive value of tumor thickness for cervical lymph node involvement (Huang et al., 2009). The optimal tumor thickness cutoff point was < 4 mm, yielding a negative predictive value of 95.5% for cervical lymph node involvement. Additionally, the pattern of pathologically involved cervical lymph nodes is associated with patient outcomes. An analysis of prognostic factors in 255 patients with oral cavity squamous cell carcinoma with ENE identified level IV and V nodal metastases and tumor depth
12 mm as independent predictors of 5-year survival (Liao et al., 2011; Fig. 12). These factors divided patients into three prognostic groups. In the low-risk group (no level IV/V metastases and tumor depth < 12 mm), intermediate-risk group (no level IV/V metastases and tumor depth
12 mm), and high-risk group (level IV/V metastases present), the 5-year overall survival rates were 50%, 28%, and 10%, respectively.

Surgery The overall health and functional status of the patient are important determinants in choosing between surgical and nonsurgical approaches. Surgery is generally preferred as definitive treatment of the primary tumor because it is expeditious, convenient, and offers pathologic information to guide treatment of the neck and adjuvant therapy. General principles of surgery include adequate exposure, wide negative margins (e.g., 1–2 cm mucosal margins and 1 cm bone margins), proximal transection of clinicallyinvolved nerves, evaluation of the neck (elective vs. therapeutic), and tracheostomy if postoperative edema and airway obstruction are anticipated (Fig. 13).

Oral Cavity Cancer: Diagnosis and Treatment

I IA B

IIA III

VI

IV

145

IIB

VA VB

Fig. 12 Cervical lymph node levels. From Montero, P. H. and Patel, S. G. (2015). Cancer of the oral cavity. Surgical Oncology Clinics of North America 24(3), 491–508.

(A)

(B)

(C)

(D)

(E)

Fig. 13 Various surgical approaches: (A) peroral, (B) mandibulotomy, (C) lower cheek flap, (D) visor flap, and (E) upper cheek flap. From Montero, P. H., Patel, S. G. (2015). Cancer of the oral cavity. Surgical Oncology Clinics of North America 24(3), 491–508.

Surgical management of the primary tumor For early-stage lesions of the oral cavity, a transoral method of approach to resection generally provides adequate exposure to obtain negative margins. For advanced or poorly visualized tumors, a mandibulotomy or pull-through technique is often utilized for floor of mouth, tongue, and mandibular pathology while a Weber-Ferguson incision or mid-face degloving may be necessary for

146

Oral Cavity Cancer: Diagnosis and Treatment

maxillary or palatal lesions. When compared to open approaches, transoral approaches lead to shorter hospitalizations, and decreased tracheotomy and PEG dependence (Moore et al., 2012; White et al., 2010). Of note, most of the functional outcomes research on transoral versus traditional approaches pertains to oropharyngeal carcinoma but can be extrapolated to the oral cavity. For lip cancers, surgical excision with intraoperative margin analysis with immediate reconstruction is the preferred method of treatment. For lesions involving up to one-third of the lip, excision with primary closure is generally performed. If the proposed excision involves up to two-thirds of the lip, local flaps are required for reconstruction. In general, regardless of the reconstruction needed (e.g., Abbe flap, Estlander flap if the oral commissure is involved, Karapandzic flap, radial forearm free tissue transfer with or without palmaris tendon suspension, etc.), wide local excision is indicated. For cancers that involve the alveolar ridge, resection of a portion of the mandible is sometimes indicated (Genden et al., 2005). Given that the mucosa is densely adherent to the underlying periosteum, it is not uncommon to see bony erosion. While mandibular invasion is associated with decreased overall survival, medullary space invasion (rather than cortical invasion) is associated with decreased disease-specific survival and overall survival (Li et al., 2017). A marginal mandibulectomy, or rim resection of only the alveolar process or the lingual cortex of the mandible, is oncologically safe as long as the tumor abuts or only superficially invades the mandible (Chen et al., 2011; Patel et al., 2008). Physical exam findings (such as fixation and paresthesias) and imaging (CT or MRI) should be used to determine extent of surgical resection preoperatively (Van Cann et al., 2008; Nomura et al., 2005). In dentulous patients, in additional to erosion and infiltration of the bone, the tumor can migrate through the dental socket into the mandible, especially in the setting of dental extractions (Genden et al., 2005). When there is clinical or radiographic evidence of medullary space involvement, a segmental mandibulectomy is indicated. Of particular importance when determining extent of surgery is the status of the mental/inferior alveolar nerve. Clinical or radiographic findings that are concerning for involvement will require a more aggressive resection to obtain microscopically uninvolved bone, marrow, and nerve margins (Fig. 14). Wide local excision (i.e., partial glossectomy) is the treatment of choice for small tumors confined to the tongue without significant posterior extension. In more advanced cases (e.g., deep extension to the extrinsic tongue muscles or involvement of the floor of mouth), a more invasive approach to the tumor is necessary to provide negative margins. The extent of resection (i.e., amount of oral tongue and floor of mouth mucosa) and the approach needed play a significant role in the reconstruction. Small transoral resections of the oral tongue can be closed primarily or left open to heal through secondary intention. Larger defects can be reconstructed with a split-thickness autograft if there is limited floor of mouth involvement. If significant portions of the floor of mouth are involved, one must be concerned about tethering and the associated functional consequences such as dysarthria and dysphagia. If a significant amount of floor of mouth mucosa is resected, local, regional, or free flap reconstruction with thin pliable tissue is necessary to preserve tongue mobility. Regional flaps like the submental island flap can be used for reconstruction without effecting oncologic safety (Kramer et al., 2015; Rigby & Hayden, 2014; Howard et al., 2014). Cancer of the buccal mucosa can generally be excised via transoral approach. The buccinator muscle can be resected along with the primary lesion to provide an appropriate deep margin. Larger and more advanced tumors may require a more extensive ablation (e.g., resection of the pterygoids, subcutaneous fat and overlying skin with a resultant through-and-through defect, etc.) and reconstruction. Small, early-stage lesions of the hard palate can also be removed through a wide local excision with the periosteum as an appropriate deep margin. If the periosteum or the underlying bone is involved, similar to an alveolar ridge lesion, resection of the bone via a maxillectomy is necessary. An inferior or subtotal maxillectomy may be necessary to obtain negative margins. If the nasal cavity or maxillary sinus is entered as a consequence of tumor ablation, the amount of connection will dictate reconstruction. Obturators, local flaps, or free flap reconstruction (bony vs. soft tissue) are all viable options with differing risks and benefits (Fig. 15). Advances in reconstructive surgery, together with improvements in our understanding of the physiology of swallowing and in speech rehabilitation, have helped to reduce the morbidity associated with surgery in the oral cavity. Surgical reconstruction may be

Fig. 14 (A) cT1 squamous cell carcinoma of the right lateral tongue, marked circumferentially with a 1-cm margin. (B) Same specimen after resection; the greatest degree of margin shrinkage is noted near the floor of the mouth. From Shapiro, M. and Salama, A. (2017). Margin analysis: Squamous cell carcinoma of the oral cavity. Oral and Maxillofacial Surgery Clinics of North America 29(3), 259–267.

Oral Cavity Cancer: Diagnosis and Treatment

147

Fig. 15 Fibular (left) and radial forearm (right) free flaps are two common flaps used in oral cavity reconstruction after major resections. From Montero, P. H. and Patel, S. G. (2015). Cancer of the oral cavity. Surgical Oncology Clinics of North America 24(3), 491–508.

performed after confirmation of complete resection with negative surgical margins. The reconstructive ladder ranging from primary closure or secondary intention to free tissue transfer must be employed with special consideration to long term swallowing, speech, and breathing. Small surgical defects of the oral tongue may not require reconstruction and therefore are often allowed to heal by secondary intention while a similarly sized defect of the floor of mouth could lead to debilitating tethering, dysphagia, and dysarthria. Larger or complex defects may be reconstructed by a combination of primary closure, split-thickness skin graft, vascularized cutaneous free flap, regional myocutaneous flap, or microvascular free flap (Urken et al., 1994). Reconstruction of a segmental defect of the mandible is a complex topic that is beyond the scope of this article (Fig. 16).

Surgical management of the neck More than 90% of occult metastatic disease from oral cavity cancer involves nodal groups I–III (submental, submandibular, upper, and middle jugulodigastric nodes), thus a selective neck dissection removing these levels (formerly known as a supraomohyoid neck dissection) is oncologically sound, especially in the clinically N0 neck. Elective supraomohyoid neck dissection has been shown to improve survival rates compared to primary tumor resection alone in early stages of oral squamous cell carcinoma (Brazilian Head and Neck Cancer Study Group, 1998; Kligerman et al., 1994; D’Cruz et al., 2015). In particular, patients with T1-T2, N0 oral tongue cancer with tumor thickness > 4 mm had significant improvement in disease-specific survival with elective neck dissection compared to observation. Because of the survival advantage and the poor salvage rate, elective neck dissection has been the standard of care (Abu-Ghanem et al., 2016). In general, elective neck dissection in the clinically N0 neck is indicated when there is at least a 20% chance of lymph node involvement. The lower lip has a low risk of occult metastasis and therefore does not require elective neck dissection (Ozturk et al., 2015; Gary et al., 1995). A systematic review and meta-analysis demonstrated that the overall rate of cervical metastasis from a maxillary squamous cell carcinoma was 31%, with increasing risk as T-stage advances (Zhang and Peng, 2016). Patients with a primary tumor that involves the oral tongue, floor of mouth, mandibular alveolar ridge, or buccal mucosa present with an increased risk of occult cervical metastasis and should therefore be considered for an elective neck dissection (Capote et al., 2007; Fasunla et al., 2011). Based on lymphatic drainage pathways, midline tumors should be treated with bilateral neck dissections. Recently, results were published on the first 500 patients enrolled on a prospective, randomized controlled trial evaluating the effect of elective node dissection (i.e., ipsilateral neck dissection with the primary surgery) versus therapeutic node dissection (i.e., neck dissection following nodal relapse) on survival in patients with lateralized T1-T2 squamous cell carcinomas of the oral cavity

Fig. 16 After partial glossectomy, a posterior-based buccinator myomucosal flap (based on buccal artery) was harvested (A) and sutured to the residual tongue (B). Photograph taken 6 months postoperatively reveals healthy condition of the transplanted flap without shrinkage (C). From Ahn, D., Lee, G. J., and Sohn, J. H. (2017). Reconstruction of oral cavity defect using versatile buccinator myomucosal flaps in the treatment of cT2–3, N0 oral cavity squamous cell carcinoma: Feasibility, morbidity, and functional/oncological outcomes. Oral Oncology 75, 95–99.

148

Oral Cavity Cancer: Diagnosis and Treatment

(D’Cruz et al., 2015). After a median follow up of 39 months, elective neck dissection resulted in higher rates of 3-year overall and disease-free survival than therapeutic neck dissection with similar rates of adverse events in the elective-surgery and therapeuticsurgery groups (6.6% and 3.6%, respectively). A systematic review and meta-analysis comparing elective neck dissection with observation in patients with clinically nodenegative, early-stage T1-T2 oral tongue squamous cell carcinoma showed that elective neck dissection can significantly reduce the rate of regional nodal recurrence and improve disease-specific survival in this population (Abu-Ghanem et al., 2016). A meta-analysis on selective versus comprehensive neck dissection in oral squamous cell carcinoma patients with clinically nodepositive necks showed that selective neck dissection achieves similar regional control, disease-specific survival, and overall survival compared with comprehensive neck dissection (Liang et al., 2015). Overall, there is increasing evidence that elective neck dissection in oral cavity carcinoma, even early stage, reduces regional recurrence rates, disease specific survival, and improves overall survival compared to observation alone (D’Cruz et al., 2015; Fasunla et al., 2011) (Table 2). Sentinel lymph node biopsy is a less invasive procedure than neck dissection which is routinely used for other nonoral cavity sites of primary malignancy. A German prospective consecutive cohort analysis first correlated sentinel lymph node positivity with survival in early oral cavity and oropharyngeal cancer (Broglie et al., 2013). The authors found that the 3-year overall survival in patients with a negative sentinel lymph node was 98% compared to 71% in sentinel lymph node positive patients. Similar differences were seen in disease-specific survival (95% vs. 76%) and disease-free survival (98% vs. 73%). In addition, they identified correlation between the presence of isolated tumor cells or micrometastases and survival. Other authors have reported the sensitivity and negative predictive value of sentinel lymph node biopsy for oral cancer staging as 71% and 94%, respectively (Rigual et al., 2013). Further research to validate these findings may lead to changes in patterns of care based on the utility of the sentinel lymph node biopsy.

Adjuvant Therapy Following Definitive Surgical Resection Following surgical resection, patients with pathologic features such as advanced primary T stage (T3-T4), lymphovascular space invasion, perineural invasion, positive margins, multiple pathologically involved cervical lymph nodes, and ENE have been shown to be at high-risk for locoregional recurrence (Cooper et al., 1998). PORT, frequently in conjunction with chemotherapy, results in a decreased risk of locoregional relapse for these patients (Lavaf et al., 2008; Maccomb, 1957). A nomogram constructed using data from 590 patients treated at a United States cancer center and validated using data from 417 patients treated at a Brazilian hospital can be used to help decide adjuvant treatment; however, limitations of the nomogram include the lack of important pathologic variables including ENE, perineural invasion, and lymphovascular space invasion, as well as the nonuniform use of radiation therapy between the construction and validation cohorts (39.5% vs. 60.4%) (Gross et al., 2008). More recently this group has created and validated an online nomogram using data from > 1400 patients to estimate locoregional failure-free survival with and without PORT (Wang et al., 2013). Factors predictive of a locoregional control benefit from PORT include high N stage, low T stage, positive margins, young age, and female sex. Adjuvant radiotherapy is preferred over preoperative radiotherapy based on Radiation Therapy Oncology Group (RTOG) 7303, which demonstrated that PORT provided superior locoregional control (70% vs. 58%) when compared to preoperative radiotherapy in patients with advanced head and neck cancer. Overall survival was unaffected by the timing (Tupchong et al., 1991). PORT provides the advantage of avoiding delay in implementation of surgical resection and obtaining complete pathologic staging of the tumor. However, postoperative wound complications may delay the initiation of PORT, and the regional hypoxia that can accompany the postoperative state may diminish the effectiveness of PORT compared to that achievable under conditions of full oxygenation.

Table 2

Types of neck dissections Lymph nodes excised

Radical neck dissection

Levels I–V

Modified radical neck dissection type I

Levels I–V

Modified radical neck dissection type II

Levels I–V

Modified radical neck dissection type III Levels I–V Supraomohyoid neck dissection

Levels I–III

Other structures excised

Structures preserved

Sternocleidomastoid muscle, internal jugular vein, spinal accessory nerve, submandibular gland Sternocleidomastoid muscle, internal Spinal accessory nerve jugular vein, submandibular gland Internal jugular vein, submandibular gland Sternocleidomastoid muscle, spinal accessory nerve Submandibular gland Sternocleidomastoid muscle, internal jugular vein, spinal accessory nerve Submandibular gland Sternocleidomastoid muscle, internal jugular vein, spinal accessory nerve

From Montero, P. H., Patel, S. G. (2015). Cancer of the oral cavity. Surgical Oncology Clinics of North America 24(3), 491–508.

Oral Cavity Cancer: Diagnosis and Treatment

149

Overall treatment time has been shown to be a critical factor in tumor control for cancers of the head and neck, including those of the oral cavity. Improved outcomes have been demonstrated when all radiation therapy is completed within 40 days for oral tongue cancer (Mendenhall et al., 1989). For patients with advanced head and neck cancer treated with combined modality therapy, the overall time from surgery to the completion of radiation therapy has been shown to be inversely related to prognosis. Multiple phase III prospective trials have demonstrated that the cumulative duration of combined modality therapy drives locoregional control and survival (Ang et al., 2001; Rosenthal et al., 2017). Significantly improved outcomes have been reported when the treatment package time was < 85 days, and the treatment package time had a greater impact on locoregional control and survival than the dose of radiation administered (Rosenthal et al., 2017). High-risk patients who had a prolonged delay between surgery and PORT had an improved outcome with an accelerated fractionation schedule compared with those irradiated once daily (Ang et al., 2001).

Adjuvant chemoradiotherapy Cisplatin-based chemotherapy combined with PORT has been compared to PORT alone for medically fit head and neck cancer patients in multiple randomized studies. These studies demonstrated statistically significant (Bernier et al., 2004; Bachaud et al., 1996; Fietkau et al., 2006) or strong statistical trends (Cooper et al., 2004) for improved locoregional control and disease-free survival with the addition of chemotherapy to PORT. Two of these studies demonstrated a statistically significant improvement in overall survival (Bernier et al., 2004; Noronha et al., 2018), while the others have shown numerically improved but not statistically significant survival improvements (Fietkau et al., 2006; Cooper et al., 2004). Of note, at least one quarter of patients in both the European Organization for Research and Treatment of Cancer (EORTC) 22931 and RTOG 9501 studies had oral cavity cancer, 26% and 27% respectively. A pooled analysis of the latter two trials demonstrated improved overall survival with the addition of concurrent chemotherapy to PORT for patients with positive margins and/or ENE (Bernier et al., 2005). More recently, the final results of the Oral Cavity Adjuvant Therapy (OCAT) trial comparing multiple adjuvant treatment strategies (conventionally fractionated radiation with five fractions per week vs. altered fractionation with six fractions per week vs. conventional chemoradiation with weekly cisplatin 30 mg/m2) in patients with resectable high-risk, locally advanced oral cavity cancer (stage III/IV and at least ENE, involvement of 2 or more regional lymph nodes, microscopically positive margins, extensive soft tissue or skin involvement requiring major reconstruction, perineural invasion, or lymphovascular space invasion) were presented (Laskar et al., 2016). No significant difference in locoregional control, disease-free survival, or overall survival was noted between the treatment arms. Planned subgroup analysis showed that chemoradiotherapy with weekly cisplatin resulted in improved locoregional control, disease-free survival, and overall survival compared to conventional radiation for patients with T3-T4, N2-N3 disease with ENE. There was no apparent difference in toxicity or radiotherapy compliance among the three regimens. The results of this trial have to be interpreted carefully, as most of the patients developed oral cavity cancer from betel nut chewing. Therefore, it is not clear if these results can be extrapolated to other areas where oral cavity cancer typically results from other risk factors. Also, a majority of the patients on the trial had gingiva/buccal tumors, so it is not clear if these results can be extrapolated to patients with oral tongue or floor of mouth cancers (Fig. 17). Additionally, the chemotherapy regimen utilized (30 mg/m2 of weekly cisplatin) is considered to be suboptimal (see below), which may account for the lack of difference in outcomes between treatment arms.

Concurrent chemotherapy regimens for adjuvant chemoradiotherapy Many consider bolus cisplatin (100 mg/m2) given every 3 weeks with radiation therapy to be the standard of care for cancers of the head and neck; however, the optimal chemotherapy schedule is unknown. Bolus cisplatin (100 mg/m2) schedules were tested in the previously mentioned RTOG 9501 and EORTC 22931 studies (Bernier et al., 2004; Cooper et al., 2004). A French randomized trial (Bachaud et al., 1996) utilized 50 mg weekly cisplatin, while the ARO 96-3 trial (Fietkau et al., 2006) employed cisplatin 20 mg/m2 and 5-fluorouracil 600 mg/m2 on days 1–5 and 29–33. A phase III randomized trial comparing the use of bolus cisplatin (100 mg/m2) to weekly cisplatin (40 mg/m2) with concurrent adjuvant radiation therapy in patients with high-risk (ENE, positive surgical margins, or pN2-N3) squamous cell carcinoma of the oral cavity demonstrated no difference in overall or locoregional recurrence free survival at 1 year (Tsan et al., 2012). A recently reported study in patients with head and neck cancer (> 90% with oral cavity cancer) receiving adjuvant or definitive concurrent chemoradiotherapy compared cisplatin delivered 30 mg/m2 weekly versus 100 mg/m2 every 3 weeks (Noronha et al., 2018). Bolus cisplatin (100 mg/m2) was found to improve locoregional control, which was the primary endpoint; however, many consider the weekly cisplatin regimen of 30 mg/m2 utilized in this trial to be suboptimal. The role of carboplatin-based chemotherapy regimens given concurrently with PORT versus PORT alone remains poorly defined. Randomized studies have been attempted, but unfortunately these studies were closed early before their accrual goals were met. Preliminary results revealed no significant benefit in locoregional control, disease-free survival, or overall survival from combining carboplatin with PORT (Expert Panel on Radiation Oncology-Head and Neck et al., 2011). RTOG 0234 is a phase II trial of high-risk postoperative patients (positive margin, ENE, and/or
2 pathologically involved cervical nodes) randomized to PORT in combination with cetuximab (400 mg/m2 loading dose followed by 250 mg/m2 weekly) and weekly docetaxel (15 mg/m2) or to PORT with cetuximab (400 mg/m2 loading dose followed by 250 mg/m2 weekly) and weekly cisplatin (30 mg/m2) (Harari et al., 2014). Published results from this study compared favourably to outcomes seen in similar patients treated with PORT and cisplatin (100 mg/m2 every 3 weeks) on RTOG 9501. It should be noted that almost half of the patients enrolled on RTOG 0234 (46.8%) had cancers of the oral cavity. Based on these findings, RTOG 1216,

150

Oral Cavity Cancer: Diagnosis and Treatment

Fig. 17 Contrast-enhanced CT axial image in a 55-year-old male smoker with squamous cell carcinoma of the tongue demonstrates an enhancing soft tissue mass involving the left oral tongue (white arrows) with extension to the midline. There is also a large necrotic-appearing left level 2 lymph node (black arrow) consistent with nodal metastasis. From Meesa, I. R., Srinivasan, A. (2015). Imaging of the oral cavity. Radiologic Clinics of North America 53(1), 99–114.

a randomized phase II/III trial, was designed to determine the effectiveness of docetaxel with or without cetuximab relative to cisplatin in the setting of adjuvant PORT. Currently, no randomized phase III data exist regarding the use of other chemotherapy agents in the postoperative setting for head and neck cancer. As a result, many consider cisplatin 100 mg/m2 every 3 weeks to be the standard.

Adjuvant radiotherapy dose, volume, and modality The optimal adjuvant radiotherapy dose is not well established. Most of the randomized studies which demonstrated a benefit of concurrent chemoradiotherapy in the adjuvant setting utilized doses of 60–66 Gy in 2 Gy daily fractions to high-risk areas (primary tumor bed with a positive margin, or nodal regions with ENE) and doses of 50–54 Gy in 2 Gy fractions to areas at risk for microscopic involvement. In three of the four randomized studies investigating adjuvant chemoradiotherapy, doses > 65 Gy were prescribed to high-risk areas (Bernier et al., 2004; Bachaud et al., 1996; Fietkau et al., 2006), while the other study, RTOG 9501, utilized a dose of 60 Gy with or without an optional 6 Gy boost (Cooper et al., 2004). Despite the fact that little evidence exists to support the higher PORT doses used in these randomized trials over the doses established from PORTalone dose-finding studies (63 Gy for ENE and 57.6 Gy for all other areas), we recommend similar dosing schedules as the randomized trials given that they were associated with significant overall survival benefits in patients with ENE and positive margins. Frequently, the radiotherapy treatment volume used in the adjuvant setting for head and neck cancer encompasses the primary tumor site and the bilateral cervical lymph node regions, yet it is unclear whether both always need to be included. For patients with completely resected primary tumors and negative margins whose only indication for PORT is pathologic cervical lymphadenopathy, some clinicians would recommend therapy only to the neck. Similarly, for patients who have undergone a comprehensive neck dissection with pathologically uninvolved cervical lymph nodes and have only a positive margin in the primary tumor bed as their sole indication for treatment, clinicians may recommend PORT to the tumor bed only. Another setting where a less comprehensive radiation volume may be recommended is for patients with well-lateralized primary tumors where the reported patterns of progression suggest that this may be appropriate (Expert Panel on Radiation Oncology-Head and Neck et al., 2011). Intensity modulated radiation therapy (IMRT) is the recommended treatment modality for head and neck cancers whenever high-dose, curative-intent irradiation is administered, whether definitively or adjuvantly (Mell et al., 2005). The benefits of IMRT are further discussed later in this article in the IMRT subsection of Definitive Radiation Therapy.

Oral Cavity Cancer: Diagnosis and Treatment

151

Definitive Radiation Therapy Historically, the oral cavity has been divided into subsites, as the location of the primary tumor would determine the radiation modalities that could be utilized to deliver dose. As external beam radiotherapy delivery techniques have improved, these distinctions have become less clinically relevant. For early stage oral cavity cancer, radiotherapy alone can result in excellent local control and survival rates; however, local control decreases with more advanced tumor stage (Fig. 18). The optimal dose fractionation schedule for patients receiving definitive radiotherapy alone for oral cavity cancer has not yet been established. Randomized data (Beitler et al., 2014; Overgaard et al., 2003; Overgaard et al., 2010) and meta-analyses (Budach et al., 2006; Bourhis et al., 2006) support an overall survival benefit with the use of accelerated fractionation or hyperfractionated radiotherapy, such that some form of altered fractionation should be utilized with treating oral cavity cancer patients with radiotherapy alone. One multi-institutional, international, randomized study comparing an accelerated regimen of 66–70 Gy delivered in 2 Gy daily fractions 6 days per week to the same dose delivered 5 days per week found improved locoregional control (42% vs. 30%, p ¼ .004), disease-free survival (50% vs. 40%, p ¼ .03), and a trend toward improved overall survival (35% vs. 28%, p ¼ .07) with accelerated fractionation (Overgaard et al., 2010). When analysed with respect to HPV status, accelerated fractionated radiotherapy resulted in improved local control for both p16-positive (hazard ratio [HR] 0.56 [95% confidence interval (CI) 0.33–0.96]) and p16-negative tumors (HR 0.77 [95% CI 0.60–0.99]) (Lassen et al., 2011). However, when a more intensive accelerated regimen of 1.8 Gy twice daily to 59.4 Gy was compared to 70 Gy in 2 Gy fractions in stage III/IV head and neck cancer patients, no statistically significant benefits were seen in terms of locoregional control or overall survival (Poulsen et al., 2001). It is unknown if the lack of benefit seen was due to the regimen used, or due to inclusion criteria as the benefit for acceleration in some randomized studies was less significant for those with stage IV disease as well as those with a larger nodal disease burden (Overgaard et al., 2010). The meta-analysis of radiotherapy in carcinoma of the head and neck (MARCH) Collaborative Group pooled 15 randomized studies (including 6515 patients, 13% with oral cavity cancer) and compared standard fractionation to either accelerated or hyperfractionated radiotherapy. Altered fractionation schedules were associated with an 8.5% absolute reduction in local failure and a 3.4% absolute improvement in 5-year overall survival. Further analysis demonstrated no significant interaction of tumor site on overall survival or tumor control. Heterogeneity in the patients included on the accelerated and hyperfractionated trials did not allow for direct comparisons, but patients treated with a hyperfractionated radiotherapy regimen were noted to have an 8.2% absolute improvement in overall survival at 5 years compared to only a 2% absolute improvement in those treated with an accelerated fractionation schedule (Bourhis et al., 2006). This meta-analysis was updated recently with increased patient numbers and follow up, and it confirmed the earlier finding that altered fractionation was associated with an improvement in overall survival, with an absolute benefit of 1.2% at 10 years (Lacas et al., 2017). This benefit also appeared to be limited to patients receiving hyperfractionated therapy, with an absolute benefit in overall survival of 3.9% at 10 years. RTOG 9003 compared standard fractionation (70 Gy in 2 Gy/day fractions), hyperfractionation (81.6 Gy in 1.2 Gy/fraction, twice-daily), split-course accelerated fractionation (67.2 Gy in 1.6 Gy/fraction twice-daily with a 2 week rest after 38.4 Gy), and accelerated fractionation with a concomitant boost (72 Gy in 1.8 Gy/fraction for 14 fractions followed by 1.8 Gy morning and 1.5 Gy afternoon boost to gross disease) for patients with cancers of the head and neck, excluding only nasopharynx; however, only about 10% of patients had oral cavity cancer (Fu et al., 2000). In the initial report, locoregional control was improved with hyperfractionation and accelerated concomitant boost as compared to the standard fractionation. A trend toward improved disease-free survival for patients treated with hyperfractionation (37.6% vs. 31.7%, p ¼ .067) or accelerated concomitant boost (39.3% vs. 31.7%, p ¼ .054) was also noted. With longer follow up, patients receiving a hyperfractionated regimen demonstrated

Fig. 18 Right lateral tongue squamous cell carcinoma crossing midline, with involvement of the genioglossus muscles. (A) Photograph shows an extensive area of mucosal irregularity and leukoplakia involving the right lateral tongue. The patient was unable to move the tongue due to tumor infiltration and pain. Coronal T1-weighted postcontrast fat-saturated (B) and axial STIR (C) images show a T2-hyperintense enhancing mass in the right tongue. The coronal image shows inferior extension into the floor of mouth/sublingual space (solid white arrow in B) and medial involvement of the genioglossus muscle. The axial image shows extension across the midline (dashed arrow in C). From Hagiwara, M., Nusbaum, A., and Schmidt, B. L. (2012). MR assessment of oral cavity carcinomas. Magnetic Resonance Imaging Clinics of North America 20(3), 473–494.

152

Oral Cavity Cancer: Diagnosis and Treatment

improved overall survival (HR 0.81, p ¼ .05) when patients were censored at 5 years (Beitler et al., 2014). Hyperfractionated patients also exhibited less late toxicity when compared to accelerated fractionated patients. Most notably, in patients who were disease-free at 5 years, only 4.8% of hyperfractionated patients had feeding tubes versus 13.0% of patients receiving accelerated fractionation with concomitant boost. It is difficult to determine the clinical significance of these toxicity data today since all of patients in the RTOG study were treated with parallel-opposed fields and the blocking techniques were rudimentary with only minimal sparing of the parotid glands and pharyngeal constrictor muscles. Currently, contemporary IMRT and image-guided radiotherapy (IGRT) techniques allow for increased sparing of these structures, which likely significantly reduces late treatment-related morbidity.

Intensity modulated radiotherapy Intensity modulated radiotherapy (IMRT) has been widely adopted for the treatment of head and neck cancers (Mell et al., 2005). Retrospective single institution (Studer et al., 2007; Studer et al., 2012) and SEER-Medicare (Beadle et al., 2014) analyses have shown significant improvements in local control and cause-specific survival, respectively, with the use of IMRT in oral cavity cancer patients. Helical tomotherapy and volumetric modulated arc therapy (VMAT) are advanced forms of IMRT. While theoretical consideration of second cancers exists with IMRT (Hall, 2006), this technology has the ability to minimize normal organ exposure to radiation. This is particularly important for oral cavity cancer patients as pharyngeal constrictor dose is associated with dysphagia (Eisbruch et al., 2004; Caudell et al., 2010) and dose to the parotid, submandibular, and minor salivary glands are associated with xerostomia (Nutting et al., 2011). In addition, IMRT has the potential to decrease acute dermatitis. A prospective, randomized trial demonstrated the beneficial impact of IMRT on the reduction of xerostomia in patients with cancer of the oropharynx or hypopharynx, using the late effects of normal tissues scale, unstimulated and sodium citrate stimulated parotid saliva from each parotid orifice and the floor of mouth, and patient-reported quality of life (Nutting et al., 2011; Fig. 19).

Brachytherapy Brachytherapy is the placement of radioactive materials in close proximity to or inside of tumors and was developed prior to IMRT or the use of concurrent chemotherapy. It allowed practitioners to deliver a tumoricidal dose of radiation to tumors in the head and neck while minimizing mandibular dose. Improved external beam radiotherapy planning and delivery techniques as well as a recognition that the incidence of osteoradionecrosis resulting from brachytherapy is significantly underreported, has resulted in a substantial decrease in the utilization brachytherapy for oral cavity tumors. Historically, brachytherapy’s role in oral cavity cancer has been to boost the dose to gross disease following external beam radiotherapy. Typically, catheters are implanted under general anesthesia in the operating room, and as such oral cavity brachytherapy is a highly operator-dependent procedure (Fig. 20). Given the historically poor locoregional control rates for oral cavity tumors treated with external beam radiotherapy alone, intensifying the treatment with brachytherapy makes logical sense. Higher rates of locoregional control have been achieved using an integrated treatment approach of external beam radiotherapy directed at the primary and bilateral neck, followed by a brachytherapy boost (Harrison, 1997). Complications of brachytherapy for oral cavity tumors can include mucosal ulceration, soft tissue necrosis, and osteoradionecrosis of the mandible (YamazakI et al., 2013). The risk of complications appears to be related to the technique of implantation, but may reach as high as 30%. Quality of life analyses comparing a combined regimen of brachytherapy plus external beam radiotherapy to surgery followed by PORT favored the combination of brachytherapy and external beam radiotherapy, suggesting that in experienced hands this is a reasonable treatment method for head and neck carcinoma (Levendag et al., 2004). There is limited information regarding the use of brachytherapy and chemotherapy concurrently, and as such this approach should be avoided outside of a formal trial setting. Brachytherapy guidelines The American Brachytherapy Society (ABS) has published guidelines for the use of high dose rate (HDR) brachytherapy for head and neck cancer, including specific guidelines for oral cavity tumors (Nag et al., 2001). The ABS panel members advise that whenever possible, a spacer be constructed and inserted at the time of the HDR treatment to reduce doses to normal tissues such as the mandible and mucosa of the palate. The relative risk of normal tissue injury for lesions near the mandible, such as the floor of mouth, is greater than for oral tongue, where the bone is not contiguous with the target volume. The interval between external beam radiotherapy and brachytherapy should be minimized (within 1–2 weeks, depending on the degree of recovery from mucositis), and the total duration of combined radiation therapy should be kept as short as possible (ideally < 8 weeks) to minimize tumor cell repopulation. Expert panel evidence, as well as single institution series, recommend external beam radiotherapy doses of 40–50 Gy to the primary tumor (or 45–60 Gy when the neck is treated for microscopic disease) followed by a HDR brachytherapy boost of 2.7–3.0 Gy per fraction for 6–7 doses with locoregional control of implanted tumors reaching 80%. Similar to the ABS, the European Brachytherapy Group (GEC) and European Society for Therapeutic Radiology and Oncology (ESTRO) developed joint guidelines for the use of brachytherapy in the treatment of head and neck carcinoma based on consensus recommendations given the lack of randomized trial data (Mazeron et al., 2009; Kovács et al., 2017). For oral cavity tumors, these guidelines recommend 46–50 Gy of external beam radiotherapy followed by a 15–30 Gy low dose rate brachytherapy boost or equivalent HDR dose. In order to minimize overall treatment time, HDR brachytherapy is often administered as two fractions per day separated by at least six hours.

Oral Cavity Cancer: Diagnosis and Treatment

153

Fig. 19 (Top) CT simulation for radiation treatment planning with patient in supine position, immobilized with five point customized thermoplastic face mask covering the shoulders and a bite block in place to depress the tongue away from minor salivary glands. (Middle) Axial CT image showing the isodose lines for the intensity-modulated primary radiation fields encompassing the high-risk elective nodal regions and the primary tumor. (Bottom) Sagittal CT image showing the isodose lines for the intensity-modulated boost radiation fields to treat the primary tumor.

Concurrent chemoradiotherapy for locoregionally advanced oral cavity cancer Although surgical resection followed by adjuvant radiation with or without chemotherapy is preferred for patients with large infiltrative primary tumors of the oral cavity, definitive concurrent chemoradiotherapy has been shown to be a reasonable option for patients who are unable to undergo surgical resection. In addition, surgical resection may be possible but not recommended for certain patients given the associated surgical morbidity and likely necessity of adjuvant chemoradiotherapy, which has a similar toxicity profile to definitive intent chemoradiotherapy (Fig. 21).

154

Oral Cavity Cancer: Diagnosis and Treatment

Fig. 20 (A) Technique of afterloading catheters insertion. (B) Implantation of afterloading catheters. (C) CT-based planning of dose distribution. From Tucek L, Petera J, Sirák I, Vosmik M, Dolezalová H, Brokesová S, Hodek M, Kasaová L, Paluska P. (2011) Hyperfractionated high-dose rate brachytherapy in the treatment of oral tongue cancer. Reports of Practical Oncology and Radiotherapy 16(6), 243–247.

Multiple studies have demonstrated a benefit with concurrent chemoradiotherapy in the definitive treatment of head and neck cancer, including oral cavity cancer (Adelstein et al., 2003; Brizel et al., 1998; Wendt et al., 1998). However, the use of concurrent chemoradiotherapy for patients with locoregionally advanced oral cavity cancer without distant metastasis is based primarily on the results of the meta-analysis of chemotherapy in head and neck cancer (MACH-NC), which demonstrated a 6.5% absolute improvement in overall survival at 5 years from the use of concurrent chemoradiotherapy compared to radiotherapy alone (Pignon et al., 2009). However, the beneficial effects of chemotherapy appeared to decrease with age. Patients with age > 70 years did not derive a significant benefit, likely because more advanced age is associated with increasing frailty and more competing risks of mortality. Subgroup analysis demonstrated the benefit of adding chemotherapy to locoregional treatment alone for patients with oral cavity cancer, with a HR of 0.87 (95% CI 0.80–0.93) and an absolute improvement in 5-year overall survival of 5.1% (Blanchard et al., 2011). While this meta-analysis did not show a significant interaction between chemotherapy timing (adjuvant, neoadjuvant, vs. concomitant) and survival for oral cavity primaries, this may only be a consequence of a lack of power. There are no randomized trials of chemoradiotherapy versus radiotherapy alone in the definitive treatment of patients with oral cavity cancer. A multiinstitution retrospective analysis investigated the outcomes of patients with locally advanced oral cavity cancer treated with definitive chemoradiotherapy on protocols between 1994 and 2008 and demonstrated 5-year overall and progressionfree survival rates of 66% and 67%, respectively (Stenson et al., 2010). This cohort was updated recently to include patients treated up to 2014 and demonstrated 5-year overall and progression-free survival rates of 63% and 59%, respectively (Melotek et al., 2016). Five-year locoregional and distant control were reported as 79% and 87%, respectively. Long-term toxicities of this regimen were acceptable, with crude rates of osteoradionecrosis and sustained need for a feeding tube of 19% and 14%, respectively. A separate study from the same institution reported the outcomes of patients with oral cavity cancer treated on protocol with chemoradiotherapy using IMRT and demonstrated estimated 5-year locoregional progression-free survival, disease-free survival, and overall survival of 71%, 76%, and 76%, respectively (Pederson et al., 2011). By contrast, another single institution series of oral cavity cancer patients treated with definitive chemoradiation between 1990 and 2011 reported worse outcomes with a 3-year disease-specific survival of 38% and actuarial 5-year overall survival of 15% (Scher et al., 2015). A recent National Cancer Database (NCDB) analysis reported a 3-year overall survival of 38% for patients with locally advanced oral cavity squamous cell carcinoma treated with definitive concurrent chemoradiotherapy (Spiotto et al., 2017). Notably, patients treated with chemoradiotherapy were more likely to be older than 60 years old, have more comorbidities, and have more advanced primary tumor and/or nodal stage compared to patients treated surgically with PORT.

Concurrent chemoradiotherapy regimens The associated toxicities of concurrent chemoradiotherapy have stimulated the search for the optimal chemotherapy regimen. Cisplatin given every 3 weeks at 100 mg/m2 is often cited as the standard regimen. For oropharyngeal cancer, the Groupe d’Oncologie Radiothérapie Tête et Cou (GORTEC) randomized study demonstrated an overall survival advantage using a carboplatin/5fluorouracil regimen specifically chosen to avoid cisplatin-related tinnitus, renal dysfunction, and emesis (Calais et al., 1999). Randomized studies comparing alternative cisplatin dosing schedules (such as 30–40 mg/m2 weekly or 20 mg/m2/day on days 1–5 and 22–26) to bolus cisplatin or to other chemoradiotherapy platforms have been conducted in nasopharyngeal cancer patients (Lee et al., 2016) or in the postoperative oral cavity setting (Tsan et al., 2012). However, multiple randomized studies that compared radiotherapy alone to concurrent chemoradiotherapy using nonbolus cisplatin schedules including daily cisplatin (6 mg/m2/day) (Jeremic et al., 2000), weekly cisplatin (40 mg/m2) (Sharma et al., 2010), or cisplatin (20 mg/m2/day) given on days 1–5 repeated every 3 weeks (Huguenin et al., 2004) had comparable outcomes to bolus cisplatin. As previously mentioned, a recently reported study in patients with head and neck cancer (> 90% with oral cavity cancer) receiving adjuvant or definitive concurrent chemoradiotherapy compared cisplatin delivered 30 mg/m2 weekly versus 100 mg/m2 every 3 weeks (Noronha et al.,

Oral Cavity Cancer: Diagnosis and Treatment

155

Fig. 21 T4a floor of mouth squamous cell carcinoma with bone invasion and necrotic right level II nodal metastasis. (A) Photograph shows a large exophytic and ulcerated mass in the right floor of mouth extending across the midline. (B, C) Axial postcontrast fat-saturated T1-weighted images from superior to inferior show a large heterogeneously enhancing mass with areas of necrosis in the right floor of mouth (solid white arrows) extending across the midline anteriorly. Gross tumor invasion of the mandible with erosion through the outer cortex is seen (solid black arrows in C). No evidence was seen on imaging of perineural invasion in the inferior alveolar canal. A large necrotic right level II node is seen posteriorly (double-lined white arrows). (D, E) T2-weighted fat-saturated images from superior to inferior similarly show a mass in the floor of mouth (solid arrows) extending across the midline, with anterior extension through the mandible into the submental region (dashed white arrow in E). Note the low T2 signal intensity of most of the mass, indicating hypercellularity. The necrotic right level II node (double lined white arrow in D) is also seen. From Hagiwara, M., Nusbaum, A., Schmidt, B. L. (2012). MR assessment of oral cavity carcinomas. Magnetic Resonance Imaging Clinics of North America 20(3), 473–494.

2018). Bolus cisplatin (100 mg/m2) was found to improve the primary endpoint of locoregional control; however, many consider the weekly cisplatin regimen of 30 mg/m2 to be suboptimal. National Comprehensive Cancer Network (NCCN) guidelines recommend cisplatin 40 mg/m2 if a weekly regimen is utilized, but high-dose cisplatin (100 mg/m2 every 3 weeks for three cycles) remains the preferred regimen.

156

Oral Cavity Cancer: Diagnosis and Treatment

Concurrent chemotherapy and accelerated radiotherapy Two recently reported randomized studies showed that there was no benefit to accelerated radiotherapy over conventionally fractionated radiotherapy when delivered with concurrent platinum-based chemotherapy. GORTEC 99-02 randomized locoregionally advanced head and neck cancer patients to either very accelerated radiotherapy 64.8 Gy (1.8 Gy twice daily); 70 Gy (2 Gy daily over 7 weeks) with concurrent carboplatin 70 mg/m2 and 5-fluorouracil 600 mg/m2/day on days 1–4, 22–25, and 43–46; or 70 Gy (2 Gy daily to 40 Gy, then 1.5 Gy twice daily) with carboplatin 70 mg/m2 and 5-fluorouracil 600 mg/m2/day on days 1–4 and 29–33. Outcomes in both chemoradiotherapy arms were similar, and both were superior to very accelerated radiotherapy with lower rates of acute toxicity and percutaneous gastrostomy tube placement during therapy and at 5 years (Bourhis et al., 2012). Similarly, RTOG 0129 randomized patients to 72 Gy accelerated concomitant boost radiotherapy with two cycles of cisplatin 100 mg/m2 or to 70 Gy daily radiotherapy with three cycles of cisplatin 100 mg/m2. Again, there was no benefit to accelerated chemoradiotherapy seen in 8-year overall survival (accelerated 48% vs. conventional 48%), locoregional progression (37% vs. 39%) or worst grade 3–5 toxicity (38% vs. 37%) (Nguyen-Tan et al., 2014). Interestingly, on an unplanned post-hoc analysis, it appeared that standard fractionation patients who received one cycle of chemotherapy had worse outcomes than those who received the two or three cycles of chemotherapy. The value of acceleration via the use of a simultaneous integrated boost, a commonly used approach for IMRT delivery, must therefore be questioned when concurrent chemotherapy is being used as part of the treatment regimen.

Acute toxicities of chemoradiotherapy The use of concurrent chemoradiotherapy in patients with oral cavity cancer is a life-changing event. Toxicities may include fatigue, nausea, emesis, thickened secretions, xerostomia, mucositis, dysphagia, odynophagia, alopecia, dermatitis, anemia, neutropenia, hoarseness, Lhermitte’s syndrome, and infection. Dysphagia is perhaps the most difficult acute and late complication. Prior to any treatment, oral cavity cancer patients are less likely to be affected by tumor-induced dysphagia than those with other cancers of the head and neck (Stenson et al., 2000). A retrospective review of prospectively collected data demonstrated that head and neck cancer patients with more advanced tumor stage (T3-T4) treated with chemoradiotherapy manifested an improvement in swallowing function that outweighed treatment-related toxicity, likely due to reduction in tumor volume (Salama et al., 2008). The study also noted a trend for worse postchemoradiotherapy swallowing function in patients with increasing age, whereas patients with good performance status were less likely to have worse swallowing function. Since dysphagia can significantly impact nutritional status, early therapeutic interventions with swallowing exercises to strengthen the pharyngeal musculature are recommended. Patients should be instructed to swallow as large a volume as possible during and after treatment, and perform exercises shown to improve swallowing ability (Rosenthal et al., 2006). Prospectively, dysphagia has been associated with exceeding specific dosimetric thresholds to the pharyngeal constrictors or laryngeal apparatus when treating other disease sites, which should be taken into consideration during radiotherapy planning, as discussed further below (Eisbruch et al., 2011).

Late toxicities of chemoradiotherapy Late toxicities of chemoradiotherapy for patients with oral cavity cancer can include dysphagia, fibrosis, osteoradionecrosis, trismus, xerostomia, dental caries, feeding tube dependence, hearing loss, kidney damage, and neuropathy. The RTOG reported late toxicities following chemoradiotherapy based on a pooled analysis of 230 locally advanced head and neck cancer patients treated on three prospective chemoradiotherapy trials (Machtay et al., 2008). Patients with older age and increased tumor stage (T3-T4) were more likely to experience late toxicity, as well as those who underwent a posttreatment neck dissection. When late toxicity was analyzed by primary tumor site, patients with oropharyngeal and oral cavity primary tumors were statistically less likely to experience late toxicity; however, it should be noted that only 5% of patients had an oral cavity primary. The incidence of late chemoradiotherapy toxicities are illustrated in single and multiinstitution experiences. For example, one multiinstitution retrospective analysis reported that only 7.8% of oral cavity cancer patients were dependent on gastric tube feedings for all caloric intake after a median follow up of 3.25 years post-chemoradiotherapy, while the remainder of patients were able to maintain their weight and nutrition by oral intake alone (Stenson et al., 2010). The investigators also reported that 18.4% of patients developed osteoradionecrosis requiring surgical intervention, with over half requiring only one procedure to achieve satisfactory results. These results were found to be similar in an updated analysis including patients treated up to 2014 with a median follow up of 5.7 years, with no difference in the rate of osteoradionecrosis across treatment decades (Melotek et al., 2016). In a separate analysis of patients with oral cavity cancer treated with chemoradiotherapy utilizing IMRT, it was noted that although a majority of patients required feeding tube placement at some point during the course of therapy, < 15% had a feeding tube at the time of last follow up (Pederson et al., 2011). This study also showed an acceptable rate of osteoradionecrosis, 14%, after a median follow up of 60 months (Table 3).

Deintensification of therapy for HPV-positive oral cavity cancer patients Given better outcomes of HPV-related oropharyngeal cancer patients, in combination with the known acute and late effects of surgery, chemotherapy, and radiotherapy, many are investigating strategies to maintain high rates of tumor control and survival while deintensifying curative intent treatment in locoregionally advanced patients. However, because of the weaker association between HPV and cancer of the oral cavity as discussed previously in the Epidemiology and Etiology section, deintensification

Table 3

Selected modern series reporting rates of osteoradionecrosis No.

Primary

RT technique

Median follow-up (months)

ORN incidence (%)

Risk factors/comments

Tsai et al. (2013)

402

T1-T2 oropharynx

31

7.5

Higher V50 and V60 associated with ORN

Gevorgyan et al. (2013)

1575

Head and neck

26

0.84

3D-CRT (vs. IMRT) predictive of severity of ORN

Duarte et al. (2014)

158

Head and neck

NR

6.3

3D-CRT (10.1%) vs. IMRT (0%) ORN rate

Chen et al. (2016)

1692

Oral

3D-CRT (17%) IMRT (83%) 3D-CRT IMRT 3D-CRT (63%) IMRT (37%) IMRT

36.8 (mean)

6.2

Maesschalck et al. (2016)

234

Oropharynx

653

Head and neck

3D-CRT: 4.9 years IMRT: 3.2 years NR

11

De Felice et al. (2016)

5.5

Raguse et al. (2016)

149

Head and neck

3D-CRT (62%) IMRT (38%) 3D-CRT (11%) IMRT (89%) 3D-CRT (70%) IMRT (30%)

Primary site (floor of mouth, buccal mucosa, retromolar trigone, or gum), segmental mandibulectomy, total dose >75 Gy associated with ORN No difference in ORN rate between 3D-CRT and IMRT

41

25.5

Studer et al. (2016) Owosho et al. (2017)

531 1023

Oral cavity, salivary gland, mesopharynx Oral cavity, oropharynx

IMRT IMRT

38 52.5

7 4.3

Smoking associated with ORN persistence, no dosimetric factors associated with ORN Any comorbidity, pre-RT mandibular surgery, poor oral hygiene, and insufficient dentoalveolar surgery associated with ORN Marginal or periosteal bone resection associated with ORN Poor periodontal status, history of alcohol use and radiation dose associated with ORN

Abbreviations: ORN, osteoradionecrosis; 3D-CRT, 3-dimensional conformal radiotherapy; IMRT, intensity-modulated radiotherapy; NR, not reported. From Moon, D. H., Moon, S. H., Wang, K., Weissler, M. C., and Chera, B. S. (2017). Incidence of, and risk factors for, mandibular osteoradionecrosis in patients with oral cavity and oropharynx cancers. Oral Oncology 72, 98–103.

Oral Cavity Cancer: Diagnosis and Treatment

Study

157

158

Oral Cavity Cancer: Diagnosis and Treatment

of therapy for oral cavity tumors is not an active area of research. Furthermore, in contrast to findings for oropharyngeal cancer, some studies suggest worse prognosis with HPV-positive oral cavity cancer compared to HPV-negative disease (Lee et al., 2012; Duray et al., 2012).

Systemic Therapy The use of chemotherapy to treat head and neck cancer has evolved from its initial role only in the recurrent or metastatic setting to standard use in the definitive treatment setting. After nearly six decades of clinical research, a wealth of data demonstrate the benefit of concurrent chemotherapy administration in the definitive treatment of head and neck cancer with radiotherapy as summarized previously. For locoregionally advanced disease, the utility of neoadjuvant cytotoxic chemotherapy has been evaluated, as well as the safety and efficacy of targeted systemic agents and concomitant radiotherapy with or without chemotherapy.

Induction chemotherapy prior to definitive local therapy The use of neoadjuvant chemotherapy prior to surgical resection or radiotherapy for oral cavity cancer patients has been tested in multiple randomized studies. In particular, one trial investigated the use of induction chemotherapy (cisplatin and 5fluorouracil) followed by surgery with or without adjuvant radiation therapy (for high-risk patients only) in locally advanced resectable oral cavity squamous cell carcinoma and demonstrated no benefit in overall survival between the treatment arms at 5 years (Licitra et al., 2003). More recently, a randomized phase III trial evaluating the use of induction chemotherapy (docetaxel, cisplatin, and 5-fluorouracil) in locally advanced resectable oral cavity squamous cell carcinoma demonstrated no significant benefit in overall or disease-free survival from the addition of preoperative chemotherapy to surgery and adjuvant radiotherapy (Zhong et al., 2013). However, post-hoc analysis indicated that patients in the experimental arm who had a clinical response or  10% viable tumor cells on pathology demonstrated improved locoregional and distant control, as well as increased overall survival. Whether induction chemotherapy prior to concurrent chemoradiotherapy improves survival when compared to chemoradiotherapy alone was an area of active study advocated by some given that distant metastases are frequently a site of first failure for patients with locoregionally advanced head and neck cancer (Brockstein et al., 2004). However, mature results from multiple randomized studies did not demonstrate improved survival with induction chemotherapy followed by concurrent chemoradiotherapy (Cohen et al., 2014; Haddad et al., 2013). It is unknown if the lack of demonstrated survival benefit was due to the therapy itself or the fact that these studies were initiated prior to robust knowledge of the behavior of HPV-associated oropharyngeal cancers. Therefore, inclusion of these patients with more favorable outcomes were not accounted for in the study design and may complicate interpretation of these studies.

Targeted agents and radiotherapy Epidermal growth factor receptor (EGFR) is over-expressed in nearly 80%–90% of cases of head and neck squamous cell carcinoma and correlates with poor prognosis and radiation resistance. Preclinical evidence showed that blocking EGFR restores radiation sensitivity and enhances cytotoxicity, which led to clinical trials evaluating this class of agents. Cetuximab was approved in combination with radiation for the treatment of locally advanced head and neck squamous cell carcinoma based on a single randomized study comparing radiotherapy 70–76.8 Gy with or without weekly cetuximab (loading dose of 400 mg/m2 followed by 250 mg/ m2). Combination therapy improved locoregional control, disease-free survival, and overall survival; however, this study did not include oral cavity cancer patients (Bonner et al., 2006; Bonner et al., 2010). Based upon lower-level evidence, there is expert consensus that cetuximab is appropriate in oral cavity cancer (category 2B recommendation). Updated analyses show that the addition of cetuximab for head and neck cancer remains beneficial regardless of p16 status (Rosenthal et al., 2016), and the presence of a prominent rash is a prognostic factor (Bonner et al., 2010). Consideration should be given to the use of altered radiation fractionation with cetuximab since improved overall survival was seen in patients who were treated with accelerated concomitant boost and hyperfractionated radiotherapy. Cetuximab should be used with caution in specific geographic areas (particularly the southeastern United States) where severe anaphylactic reactions can occur. These are mediated by an immunoglobulin E response to the galactose-alpha-1,3-galactose oligosaccharide found on the Fab portion of the cetuximab heavy chain (Chung et al., 2008). The fully humanized monoclonal antibody to EGFR, panitumumab, has a much lower rate of severe allergic reactions, but no level-1 evidence exists to support equivalent efficacy to cetuximab for head and neck cancer. Other than geographic limitations, there is no consensus as to which patient populations are better served by concurrent chemotherapy versus concurrent cetuximab with radiation therapy. Certain patient factors, such as renal dysfunction or overall poor functional status, prompt physicians to utilize cetuximab preferentially over cisplatin with radiotherapy, but no level-1 evidence supports these as indications. Of note, those same medical conditions were exclusion criteria for the randomized trial that demonstrated cetuximab’s benefit over radiation therapy alone (Fig. 22).

Targeted agents in combination with cytotoxins and radiotherapy A meta-analysis of several phase II trials and retrospective studies comparing radiation therapy with concurrent cetuximab versus concurrent cisplatin found better survival and locoregional control in patients treated with radiation and cisplatin (Petrelli et al., 2014). RTOG 0522, which notably did not include oral cavity cancer patients, failed to show benefit of the combination of cetuximab with chemoradiotherapy over standard chemoradiotherapy, and the addition of cetuximab was associated with more grade

Oral Cavity Cancer: Diagnosis and Treatment

159

Fig. 22 Squamous cell carcinoma of the left buccal mucosa, T4a. (A) Axial contrast-enhanced CT shows large left buccal mass (arrows) extending to maxillary RMT. (B) Axial contrast-enhanced CT (inferior to A) shows inferior and lateral extension to involve the skin (arrow). Involvement of the skin is critical to note, because the tumor is upstaged to T4a. From Aiken, A. H. (2013). Pitfalls in the staging of cancer of oral cavity cancer. Neuroimaging Clinics of North America 23(1), 27–45.

3–4 toxicity (Ang et al., 2014). More recently, the published results of the GORTEC 2007-01 phase III randomized trial which included oral cavity cancer patients demonstrated that three cycles of concomitant carboplatin 70 mg/m2/day and 5-fluoruracil 600 mg/m2/day on days 1–4 with cetuximab-radiotherapy markedly improved progression-free survival and locoregional control with a nonsignificant gain in overall survival compared to cetuximab-radiotherapy without chemotherapy (Tao et al., 2018). To date the triple drug combination of cetuximab, carboplatin, and 5-fluouracil has not been compared to cisplatin alone. Therefore, single agent platinum-based chemoradiotherapy remains the standard of care for locally advanced squamous cell carcinoma of the head and neck. Oral cavity cancer patients have also been included in studies evaluating the addition of bevacizumab to chemoradiotherapy. The addition of bevacizumab to 5-fluorouracil, hydroxyurea, and twice daily radiotherapy did not confer any benefit in a study of locoregionally advanced head and neck cancer patients, of which oral cavity was the most common primary site (31%) (Salama et al., 2011). Bevacizumab has also been integrated with erlotinib (synchronous dual inhibition of VEGF and EGFR) and cisplatin (33 mg/m2 cisplatin on days 1–3 of weeks 1 and 5) together with 1.25 Gy twice-daily radiotherapy to 70 Gy. Fifteen percent of patients had oral cavity primary tumors. At a median follow-up of 46 months, 3-year locoregional control and overall survival were promising compared to historical series at 86% and 85%, respectively (Yoo et al., 2012). Soft tissue and osteoradionecrosis occurred in both series, and careful attention should be paid to the results of ongoing studies integrating bevacizumab to multiple chemoradiotherapy platforms for locoregionally advanced head and neck cancer patients including those with oral cavity tumors.

Posttreatment Management and Surveillance Following definitive therapy for oral cavity cancer, patients should be seen regularly for clinical evaluation. Initial posttreatment follow up varies based on the duration and severity of acute treatment-related toxicities in order to provide appropriate supportive care and ensure adequate nutrition. Typically patients are seen every 1–3 months for the first year, then every 2–6 months for the second year, then every 4–8 months for years 3 through 5, and then every 12 months thereafter. At each of these visits, a detailed history, thorough physical examination of the head and neck region, and direct fiberoptic laryngoscopy (if possible) should be performed. Periodic blood tests may be ordered to assess for adequate posttreatment organ function, such as the thyroid following neck irradiation or kidneys following cisplatin. Carotid ultrasounds may also be ordered to screen for accelerated carotid atherosclerosis causing hemodynamically significant stenosis following neck irradiation, although there is no consensus on the optimal timing. An annual low-dose CT scan should be performed for patients that meet lung cancer screening criteria based on smoking history. Additional imaging studies may be obtained if the clinical history or physical exam is concerning for any changes that may indicate tumor recurrence. Patients should also be encouraged to undergo regular dental evaluation, including dental cleaning and topical fluoride treatment. If dental extraction or other surgical intervention is planned within the irradiated field, then discussion with the radiation oncologist should be initiated for consideration of referral for hyperbaric oxygen treatment. After definitive radiotherapy or chemoradiotherapy, follow-up imaging should be performed within the first three months of treatment completion in patients with node-positive presentations. In the neck, a radiographic complete response on CT is defined as nonenhancing, nonnecrotic nodal tissue < 1.5 cm in size and is associated with 100% long-term disease control in the neck (Liauw et al., 2006). Surveillance CT imaging of the primary site does not add additional information to physical or fiberoptic examination and therefore should not be performed unless clinically indicated (Sullivan et al., 2011).

160

Oral Cavity Cancer: Diagnosis and Treatment

PET-CT is more often used as the sole imaging modality following the completion of radiotherapy to assess nodal response. A prospective analysis of over 100 node-positive head and neck cancer patients followed with PET-CT at around 12 weeks posttherapy and again 4 weeks later (if equivocal or greater residual activity was present on the 12 week scan) treated with neck dissection for metabolic persistence helped to clarify the role of PET-CT scan following chemoradiotherapy. Irrespective of residual CT abnormalities in the neck, a negative neck on the PET scan at 12 weeks, defined as the absence of metabolic activity, was associated with no isolated nodal progression (Porceddu et al., 2011). The negative predictive value of PET was 98.1% (95% CI 93.2–99.8) compared to 96.8% with CT (95% CI 88.8–99.6), and false positive readings were seen in only 1.8% of PET-CT scans compared to 38% of CT scans, yielding a positive predictive value of 77.8% for PET-CT and 14% for CT. Long-term follow-up of this patient cohort found that PET-CT-directed posttherapy neck management adequately spared > 90% of patients from a neck dissection, with only one patient experiencing a neck recurrence after initial metabolic resolution (Sjovall et al., 2015). Of note, these patients also had a contrast enhanced CT scan performed 6 weeks after completion of treatment, and all patients with a negative contrastenhanced CT scan were subsequently found to have a negative PET. Therefore, 12-week PET imaging may not be necessary for patients who have had a prior negative posttreatment contrast-enhanced CT scan. More recently, a randomized study of 564 patients compared the utility of post-chemoradiotherapy PET-CT-based neck surveillance versus a planned neck dissection in patients with clinical AJCC 7th Edition N2 or N3 disease. Two-year overall survival was 84.9% (95% CI 80.7–89.1) for patients undergoing PET-CT-based neck surveillance compared to 81.5% (95% CI 76.9–86.3) in the group receiving a planned neck dissection. No differences were noted in the 2-year locoregional control rates between groups, with 91.9% (95% CI 88.5–95.3) control in the PET-CT surveillance group and 91.4% (95% CI 87.8–95.0) control in the planned neck dissection group. Additionally, PET-CT-based surveillance was more cost effective, saving an average of $2190 per patient. For these reasons, PET-CT-based surveillance of head and neck cancer patients should be considered standard post chemoradiotherapy (Mehanna et al., 2016). Approximately 90% of tumor recurrences will occur within the first 18 months following treatment. Patients without evidence of disease two or more years after treatment are at greater risk of developing a secondary cancer than recurrence of their original disease. Patients who continue to use tobacco and alcohol are at the highest risk. Should a second primary head and neck cancer occur, identification at an early stage is paramount in order to achieve the best cure rate with the least invasive means of treatment. Therefore, close follow-up is essential in patients with cancers of the oral cavity (Fig. 23).

Treatment of Recurrent and Metastatic Oral Cavity Cancer The appropriate management of metastatic or recurrent oral cavity cancer depends largely on the extent of disease, the prior therapy administered, and whether recurrence is local, regional, or both. If there is distant disease recurrence, systemic therapy approaches will likely assume primary importance. In the case of small recurrences at the primary site for patients treated with primary excision only, further excision with or without PORT is often recommended. For larger recurrences in patients who received radiation as part of their initial management, the rate of surgical salvage is quite low; however, in some cases, further resection may be considered for palliation or as an attempt at curative treatment, particularly in the setting of a clinical trial. Reirradiation is possible, but with the potential for increased toxicity. Systemic therapy and palliative care are also treatment options for this group of patients. The risks and benefits of each treatment should be discussed with the individual patient.

Fig. 23 Squamous cell carcinoma of the left retromolar trigone (RMT), T4b N2b. (A) Axial contrast-enhanced CT demonstrates large tumor centered in the left RMT (arrows). (B) Axial contrast-enhanced CT (cranial to A) shows obvious invasion of the posterior maxilla involving the maxillary sinus (long arrow) from tumor growing superiorly along the pterygomandibular raphe. Tumor is therefore upstaged to T4a. Also note tumor extension to medial pterygoid plate and masticator space (short arrows), resulting in overall T4b stage. (C) Axial contrast-enhanced CT shows metastatic level I and II nodes (arrows). Some have central low density, a very specific sign for metastasis. From Aiken, A. H. (2013). Pitfalls in the staging of cancer of oral cavity cancer. Neuroimaging Clinics of North America 23(1), 27–45.

Oral Cavity Cancer: Diagnosis and Treatment

161

Systemic therapy for recurrent and metastatic oral cavity cancer The standard of care for patients with recurrent or metastatic oral cavity or oropharyngeal cancer is systemic therapy with platinumbased chemotherapy. Phase III trials have investigated cisplatin (with or without 5-fluorouracil or methotrexate), 5-fluorouracil, single-agent methotrexate, and carboplatin with 5-flurouracil (Forastiere et al., 1992; Anon, 1990; Jacobs et al., 1992). Many drugs, in addition to platinum agents, 5-fluorouracil, and methotrexate, have been shown in phase II trials to exhibit single agent activity such as paclitaxel (Forastiere et al., 1998), docetaxel (Dreyfuss et al., 1996), gemcitabine (Catimel et al., 1994), ifosfamide (Sandler et al., 1998), vinorelbine (Saxman et al., 1998), pemetrexed (Pivot et al., 2001), capecitabine (Martinez-Trufero et al., 2010), and irinotecan (Murphy, 2005). Single agent cisplatin (100 mg/m2) has been shown to improve overall survival compared to bestsupportive care (Morton et al., 1985) and superior to single agent methotrexate in a phase III randomized trial (Anon, 1990). Many randomized trials have tried to improve survival with combination cisplatin-based regimens. Cisplatin (100 mg/m2) with 5-fluorouracil (1000 mg/m2/day) demonstrated improved response rates (32% vs. 17%, p ¼ .035), but did not improve median survival (5.7 months) over cisplatin alone (Jacobs et al., 1992). Similar results were seen when cisplatin/5-fluorouracil, carboplatin/5-fluorouracil, and methotrexate alone were compared in a randomized phase III trial. Cisplatin/5-fluorouracil and carboplatin/5-fluorouracil demonstrated increased response rates (32% and 21%, respectively) compared to methotrexate alone (10%), but neither regimen improved median survival (6.6 and 5.6 months, respectively) compared to methotrexate alone (5.0 months) (Forastiere et al., 1992). Combination treatments were associated with worse toxicity. Similar response rates and survival have been seen with cisplatin/paclitaxel versus cisplatin/5-fluorouracil, providing a more easily administered treatment option for patients (Gibson et al., 2005). Adding agents targeting EGFR to platinum-based systemic therapy has been shown to improve overall survival compared to platinum-based systemic therapy alone in a phase III study. The EXTREME study randomized recurrent and metastatic head and neck cancer patients to cisplatin 100 mg/m2 on day 1 or carboplatin AUC ¼ 5 on day 1 along with 5-fluorouracil 1000 mg/m2/ day on days 1–4 every 3 weeks, with or without cetuximab 250 mg/m2 (following a loading dose of 400 mg/m2) that was continued until disease progression or patient intolerance (Vermorken et al., 2008). The addition of cetuximab to chemotherapy resulted in an improvement in median overall survival from 7.4 to 10.1 months and an improvement in median progression-free survival from 3.3 to 5.6 months. Currently there are no phase III trials to suggest that addition of tyrosine kinase inhibitors (including gefitinib or erlotinib) to platinum-based therapy provides a survival benefit. Ongoing studies are investigating the role of bevacizumab and other targeted agents in combination with chemotherapy. For patients with platinum-refractory recurrent and metastatic head and neck cancer, immune checkpoint inhibitors targeting the programmed cell death protein-1 (PD-1) pathway have significantly changed second-line therapy. The Checkmate 141 study randomized 361 patients (48.5% oral cavity cancers) to the anti-PD-1 antibody nivolumab 3 mg/kg every 2 weeks or investigator choice of 40–60 mg/m2 methotrexate, 30–40 mg/m2 docetaxel, or 250 mg/m2 cetuximab (after a loading dose of 400 mg/m2). Patients receiving nivoluma