The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes: Neuropsychological Endophenotypes and Biomarkers [Volume I, 1 ed.] 1402094639, 9781402094637, 9781402094644

Table of contents :
Contents to Volume 1......Page 8
Foreword......Page 6
Contributors to Volume 1......Page 11
Part I: Methodological and Technological Advances......Page 15
1. Where Do We Stand in the Quest for Neuropsychiatric Biomarkers and Endophenotypes and What Next?......Page 16
2. Methodological and Statistical Issues in the Use of Biomarkers in Clinical and Research Studies......Page 35
3. Challenging the Genetic Complexity of Schizophrenia by Use of Intermediate Phenotypes......Page 52
4. Translational Medicine: Functional Biomarkers for Drug Development of "Cognitive Enhancers" in Schizophrenia......Page 68
5. Leveraging High-Dimensional Neuroimaging Data in Genetic Studies of Neuropsychiatric Disease......Page 98
6. Proteomics as a New Tool for Biomarker-Discovery in Neuropsychiatric Disorders......Page 114
7. Schizophrenia Endophenotypes as Treatment Targets......Page 123
Part II: Neuropsychological, Neurocognitive and Neurophysiological Domains......Page 133
8. Neuropsychological Endophenotypes in Schizophrenia and Bipolar I Disorder: Yields from the Finnish Family and Twin Studies......Page 134
9. Is More Cognitive Experimental Psychopathology of Schizophrenia Really Necessary? Challenges and Opportunities......Page 150
10. Intellectual Functioning as an Endophenotype for Schizophrenia......Page 164
11. Emotion Recognition Deficits as a Neurocognitive Marker of Schizophrenia Liability......Page 172
12. The Use of Neurocognitive Endophenotypes in Large-Scale Family Genetic Studies of Schizophrenia......Page 186
13. Neurocognitive Endophenotypes for Bipolar Disorder: Evidence from Case-Control, Family and Twin Studies......Page 203
14. Trait and State Markers of Schizophrenia in Visual Processing......Page 219
15. Visual Scanning Abnormalities as Biomarker for Schizophrenia......Page 229
16. Biomarkers and Endophenotypes in Eating Disorders......Page 235
17. Movement Abnormalities: A Putative Biomarker of Risk for Psychosis......Page 246
Contents to Volumes 2, 3, and 4......Page 266
Contributors to Volumes 2, 3, and 4......Page 270
D......Page 277
N......Page 278
W......Page 279

Citation preview

The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes

THE HANDBOOK OF NEUROPSYCHIATRIC BIOMARKERS, ENDOPHENOTYPES AND GENES

Volume 1: Neuropsychological Endophenotypes and Biomarkers Volume 2: Neuroanatomical and Neuroimaging Endophenotypes and Biomarkers Volume 3: Metabolic and Peripheral Biomarkers Volume 4: Molecular Genetic and Genomic Markers

Michael S. Ritsner Editor

The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes Volume 1

Neuropsychological Endophenotypes and Biomarkers

Editor Michael S. Ritsner, M.D., Ph.D. Associate Professor of Psychiatry, the Rappaport Faculty of Medicine Technion - Israel Institute of Technology, Haifa and Sha’ar Menashe Mental Health Center, Hadera, Israel

ISBN 978-1-4020-9463-7

e-ISBN 978-1-4020-9464-4

Library of Congress Control Number: 2008942052 © Springer Science + Business Media B.V. 2009 No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording or otherwise, without written permission from the Publisher, with the exception of any material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work. Printed on acid-free paper springer.com

Foreword

Common genetically influenced neuropsychiatric disorders such as schizophrenia spectrum disorders, major depression, bipolar and anxiety disorders, epilepsy, neurodegenerative and demyelinating disorders, Parkinson and Alzheimer’s diseases, alcoholism, substance abuse, and drug dependence are the most debilitating illnesses worldwide. They are characterized by their complexity of causes and by their lack of pathognomonic laboratory diagnostic tests. During the past decade many researchers around the world have explored the neuropsychiatric biomarkers and endophenotypes implicated, not only in order to understand the genetic basis of these disorders but also from diagnostic, prognostic, and pharmacological perspectives. These fields have therefore, witnessed enormous expansion in new findings obtained by neuropsychological, neurophysiological, neuroimaging, neuroanatomical, neurochemical, molecular genetic, genomic and proteomic analyses, which have generated a necessity for syntheses across the main neuropsychiatric disorders. The challenge now is to translate these findings into meaningful etiologic, diagnostic and therapeutic advances. This four volume collection of Handbooks offers a broad synthesis of current knowledge about biomarker and endophenotype approaches in neuropsychiatry. Since many of the contributors are internationally known experts, they not only provide up-to-date state of the art overviews, but also clarify some of the ongoing controversies, future challenges and proposing new insights for future researches. The contents of the volumes have been carefully planned, organized, and edited in close collaboration with the chapter authors. Of course, despite all the assistance provided by contributors and others, I alone remain responsible for the content of these Handbooks including any errors or omissions, which may remain. The Handbook is organized into four interconnected volumes covering five major sections. Volume 1 “Neuropsychological Endophenotypes and Biomarkers” contains 17 chapters composed of two parts emphasizing schizophrenia as a prototype. The first section serves as an introduction and overview of methodological issues of the biomarker and endophenotype approaches in neuropsychiatry and some technological advances. Chapters review definitions, perspectives, and issues that provide a conceptual base for the rest of the collection. The second section comprises chapters in v

vi

Foreword

which the authors present and discuss the neuropsychological, neurocognitive and neurophysiological candidate biomarkers and endophenotypes. Volume 2 “Neuroanatomical and Neuroimaging Endophenotypes and Biomarkers”, focuses on neuroanatomical and neuroimaging findings obtained for wide spectra of neuropsychiatric disorders. Volume 3 “Metabolic and Peripheral Biomarkers”, explores several specific metabolic and peripheral biomarkers, such as neuroactive steroid biomarkers, cortisol to DHEA molar ratio, mitochondrial complex, biomarkers of excitotoxicity, melatonin, retinoic acid, abnormalities of inositol metabolism in lymphocytes, and others. Volume 4 “Molecular Genetic and Genomic Markers” contains chapters devoted to searching for novel molecular genetic and genomic markers in less explored areas. This volume includes an Afterword written by Professor Robert H. Belmaker. Similarly to other publications contributed to by diverse scholars from diverse orientations and academic backgrounds, differences in approaches and opinions, as well as some overlap, are unavoidable. I believe that this collection is probably the first of its kind to go beyond the neuropsychiatric disorders and delve into the neurobiological basis for diagnosis, treatment, and prevention. The take-home message is that principles of the biomarker-endophenotype approach may be applied no matter what kind of neuropsychiatric disorder afflicts our patients. The Handbook is designed for use by a broad spectrum of readers including neuroscientists, psychiatrists, neurologists, endocrinologists, pharmacologists, psychologists, general practitioners, geriatricians, graduate students, health care providers in the fields of neurology and mental health, and others interested in trends that have crystallized in the last decade, and trends that can be expected to evolve in the coming years. It is hoped that this collection will also be a useful resource for the teaching of psychiatry, neurology, psychology and mental health. With much gratitude, I would like to acknowledge the contributors from 16 countries for their excellent cooperation. In particular, I am most grateful to Professor Irving Gottesman for his support of this project. His unending drive and dedication to the field of psychiatric genetics never ceases to amaze me. I wish to acknowledge Professor Robert H. Belmaker, distinguished biological psychiatrist, who was very willing to write the afterword for these volumes. I also wish to take this opportunity to thank my close co-workers and colleagues Drs. Anatoly Gibel, Yael Ratner, Ehud Susser, Stella Lulinski, Rachel Mayan, Professor Vladimir Lerner and Professor Abraham Weizman for their support and cooperation. Finally, I am forever indebted to my wife Galina Ritsner, sons Edward and Yisrael for their understanding, endless patience and encouragement when it was most required. I sincerely hope that these four interconnected volumes of the Handbook will further knowledge in the complex field of neuropsychiatric disorders. February, 2009

Michael S. Ritsner Editor

Contents to Volume 1

Foreword ........................................................................................................... Michael S. Ritsner

v

Contributors to Volume 1 ................................................................................

xi

Part I 1

2

3

4

5

6

7

Methodological and Technological Advances

Where Do We Stand in the Quest for Neuropsychiatric Biomarkers and Endophenotypes and What Next? ............................. Michael S. Ritsner and Irving I. Gottesman

3

Methodological and Statistical Issues in the Use of Biomarkers in Clinical and Research Studies ................................... Ryan J. Van Lieshout and Peter Szatmari

23

Challenging the Genetic Complexity of Schizophrenia by Use of Intermediate Phenotypes ........................................................ Assen Jablensky

41

Translational Medicine: Functional Biomarkers for Drug Development of “Cognitive Enhancers” in Schizophrenia ....................................................................................... Georg Winterer Leveraging High-Dimensional Neuroimaging Data in Genetic Studies of Neuropsychiatric Disease .................................... Cinnamon S. Bloss, Trygve E. Bakken, Alexander H. Joyner, and Nicholas J. Schork Proteomics as a New Tool for Biomarker-Discovery in Neuropsychiatric Disorders ................................................................ Thomas J. Raedler, Harald Mischak, Holger Jahn, and Klaus Wiedemann Schizophrenia Endophenotypes as Treatment Targets......................... Stephen I. Deutsch, Barbara L. Schwartz, Richard B. Rosse, John Mastropaolo, Ayman H. Fanous, Abraham Weizman, Jessica A. Burket, and Brooke L. Gaskins

57

87

103

113

vii

viii

Contents to Volume 1

Part II

8

9

Neuropsychological, Neurocognitive and Neurophysiological Domains

Neuropsychological Endophenotypes in Schizophrenia and Bipolar I Disorder: Yields from the Finnish Family and Twin Studies ...................................................................................... Annamari Tuulio-Henriksson, Jonna Perälä, Irving I. Gottesman, and Jaana Suvisaari Is More Cognitive Experimental Psychopathology of Schizophrenia Really Necessary? Challenges and Opportunities .................................................................................... Angus W. MacDonald, III

10

Intellectual Functioning as an Endophenotype for Schizophrenia ..... Odette de Wilde

11

Emotion Recognition Deficits as a Neurocognitive Marker of Schizophrenia Liability ......................................................... Renata Schoeman, Dana J.H. Niehaus, Liezl Koen, and Jukka M. Leppänen

12

13

14

15

The Use of Neurocognitive Endophenotypes in Large-Scale Family Genetic Studies of Schizophrenia .............................................. William P. Horan, Tiffany A. Greenwood, David L. Braff, Raquel E. Gur, and Michael F. Green

125

141

155

163

177

Neurocognitive Endophenotypes for Bipolar Disorder: Evidence from Case-Control, Family and Twin Studies ...................... Eugenia Kravariti, Fergus Kane, and Robin M. Murray

195

Trait and State Markers of Schizophrenia in Visual Processing ................................................................................. Yue Chen, Daniel Norton, and Ryan McBain

211

Visual Scanning Abnormalities as Biomarker for Schizophrenia ..................................................................................... Patricia E.G. Bestelmeyer

221

16

Biomarkers and Endophenotypes in Eating Disorders ........................ Carolina Lopez, Marion Roberts, and Janet Treasure

17

Movement Abnormalities: A Putative Biomarker of Risk for Psychosis ................................................................................ Vijay A. Mittal and Elaine F. Walker

227

239

Contents to Volume 1

ix

Contents to Volumes 2, 3, and 4 ......................................................................

259

Contributors to Volumes 2, 3, and 4 ...............................................................

263

Index ..................................................................................................................

271

Contributors to Volume 1

Trygve E. Bakken M.Sc., Scripps Genomic Medicine and Scripps Translational Science Institute; Medical Scientist Training Program and Graduate Program in Neurosciences, University of California, San Diego, CA, USA E-mail: [email protected] Patricia E.G. Bestelmeyer, Ph.D., Post-Doc Centre for Cognitive Neuroimaging Department of Psychology, Glasgow, UK E-mail: [email protected] Cinnamon S. Bloss, Ph.D., Research Scientist, Scripps Genomic Medicine and Scripps Translational Science Institute, Scripps Health and The Scripps Research Institute, La Jolla, CA, USA E-mail: [email protected] David L. Braff, M.D., Professor, Department of Psychiatry, University of California, San Diego, CA, USA E-mail: [email protected] Jessica A. Burket, B.S., Mental Health Service Line, Department of Veterans Affairs Medical Center, Washington, USA Yue Chen, Ph.D., Director, Visual Psychophysiology Laboratory, McLean Hospital, Department of Psychiatry, Harvard Medical School, Belmont, MA, USA E-mail: [email protected] Stephen I. Deutsch, M.D., Ph.D., Mental Health Service Line, Department of Veterans Affairs Medical Center, Washington, Department of Psychiatry, Georgetown University School of Medicine, Washington, USA E-mail: [email protected] Ayman H. Fanous, M.D., Mental Health Service Line, Department of Veterans Affairs Medical Center, Washington, Department of Psychiatry, Georgetown University School of Medicine, Washington, USA Brooke L. Gaskins, B.A., Mental Health Service Line, Department of Veterans Affairs Medical Center, Washington, USA Irving I. Gottesman Professor, Departments of Psychiatry and Psychology, University of Minnesota, Minneapolis, MN, USA E-mail: [email protected]

xi

xii

Michael F. Green, Ph.D., Professor, Semel Institute, University of California, Los Angeles, CA, USA E-mail: [email protected] Tiffany A. Greenwood, Ph.D., Assistant Adjunct Professor of Psychiatry, Department of Psychiatry, University of California, San Diego, CA, USA E-mail: [email protected] Raquel E. Gur, M.D., Ph.D., The Karl and Linda Rickels Professor and Vice Chair for Research Development, Departments of Psychiatry, Neurology and Radiology, Director, Neuropsychiatry Section, University of Pennsylvania Medical Center, Philadelphia, PA, USA E-mail: [email protected] William P. Horan, Ph.D., VA Greater Los Angeles Healthcare system & University of California, Los Angeles, CA, USA E-mail: [email protected] Assen Jablensky, M.D., D. Med.Sci., Professor of Psychiatry, School of Psychiatry and Clinical Neurosciences, The University of Western Australia, Director, Centre for Clinical Research in Neuropsychiatry, Australia E-mail: [email protected] Holger Jahn University of Hamburg, Department of Psychiatry, Hamburg, Germany Alexander H. Joyner M.Eng., Scripps Genomic Medicine and Scripps Translational Science Institute; Graduate Program in Biomedical Sciences, University of California, San Diego, CA, USA E-mail: [email protected] Fergus Kane, Ph.D. student at the section of General Psychiatry, Department of Psychiatry, Institute of Psychiatry, London, UK Liezl Koen, M.B. Ch.B., M.Med. (Psych), Department of Psychiatry, University of Stellenbosch, South Africa E-mail: [email protected] Eugenia Kravariti, M.A., M.Sc., Ph.D., Lecturer, NIHR Biomedical Research Centre for Mental Health, South London and Maudsley NHS Foundation Trust and Institute of Psychiatry, King’s College London, UK E-mail: [email protected] Jukka M. Leppänen, Ph.D., Assistant Professor, Department of Psychology, University of Tempere, Tempere, Finland E-mail: [email protected] Ryan J. Van Lieshout The Offord Centre for Child Studies, McMaster Children’s Hospital and Department of Psychiatry and Behavioural Neurosciences, McMaster University, Hamilton, Ontario, Canada E-mail: [email protected] Carolina Lopez Eating Disorders Research Unit, Department of Academic Psychiatry, King’s College London, UK E-mail: [email protected]

Contributors to Volume 1

Contributors to Volume 1

xiii

Angus W. MacDonald, III, Ph.D., Associate Professor, Departments of Psychology and Psychiatry, University of Minnesota, Minneapolis, Minnesota, USA E-mail: [email protected] John Mastropaolo, Ph.D. Mental Health Service Line, Department of Veterans Affairs Medical Center, Washington, USA Ryan McBain McLean Hospital, Belmont, MA, USA Harald Mischak Mosaiques Diagnostics and Therapeutics AG, Hannover, Germany Vijay A. Mittal, Ph.D., Postdoctoral Scholar, Department of Psychology, University of California Los Angeles, USA E-mail: [email protected] Robin M. Murray, Professor of Psychiatry, Institute of Psychiatry, King’s College London, UK E-mail: [email protected] Dana J.H. Niehaus, M.B. Ch.B., M.Med. (Psych.), D.Med. (Psych.), FC Psych., Department of Psychiatry, University of Stellenbosch, South Africa E-mail: [email protected] Daniel Norton McLean Hospital, Belmont, MA, USA Jonna Perälä, M.D., Researcher, National Public Health Institute, Department of Mental Health and Alcohol Research, Helsinki, Finland E-mail: [email protected] Thomas J. Raedler, M.D., Associate Professor, Department of Psychiatry, Faculty of Medicine, University of Calgary, Calgary, Alberta, Canada E-mail: [email protected]; [email protected] Michael S. Ritsner, M.D., Ph.D., Associate Professor of Psychiatry and Head of Cognitive and Psychobiology Research Laboratory, The Rappaport Faculty of Medicine, Technion – Israel Institute of Technology, Haifa and Chair, Acute Department, Sha’ar Menashe Mental Health Center, Hadera, Israel E-mail: [email protected] Marion Roberts, Eating Disorders Research Unit, Department of Academic Psychiatry, Institute of Phychiatry, King’s College London, 5th Floor Bermondsey Wing, Guy’s Hospital, London, SE1 9RT ddi. 0207 188 0181 E-mail: [email protected]; www.eatingresearch.com Richard B. Rosse, M.D., Mental Health Service Line, Department of Veterans Affairs Medical Center, Washington, Department of Psychiatry, Georgetown University School of Medicine, Washington, USA Renata Schoeman, M.B. Ch.B., M.Soc. Sc., M.Med. (Psych.), FC Psych., Department of Psychiatry, University of Stellenbosch, South Africa E-mail: [email protected]

xiv

Nicholas J. Schork, Ph.D., Director of Research, Scripps Genomic Medicine; Director of Biostatistics and Bioinformatics, The Scripps Translational Science Institute; Professor, Molecular and Experimental Medicine, Scripps Health and The Scripps Research Institute, CA, USA E-mail: [email protected] Barbara L. Schwartz, Ph.D., Mental Health Service Line, Department of Veterans Affairs Medical Center, Washington, Department of Psychiatry, Georgetown University School of Medicine, Washington, USA Jaana Suvisaari, M.D., Ph.D., Academy research fellow, National Public Health Institute Department of Mental Health and Alcohol Research, Helsinki, Finland E-mail: [email protected] Peter Szatmari Department of Psychiatry and Behavioural Neurosciences, McMaster University, Hamilton, Ontario, Canada E-mail: [email protected] Janet Treasure Psychological Medicine Department, King’s College London, Institute of Psychiatry, London, UK E-mail: [email protected] Annamari Tuulio-Henriksson, Ph.D., Senior Researcher, National Public Health Institute Department of Mental Health and Alcohol Research, Helsinki, Finland E-mail: [email protected] Elaine F. Walker, Ph.D., Samuel Candler Dobbs Professor of Psychology and Neuroscience, Department of Psychology, Emory University, USA E-mail: [email protected] Abraham Weizman, M.D., Professor of Psychiatry, Research Unit, Geha Mental Health Center and the Laboratory of Biological Psychiatry at Felsenstein Medical Research Center, Sackler Faculty of Medicine, Tel-Aviv University, Ramat-Aviv, Tel-Aviv E-mail: Israel. [email protected] Klaus Wiedemann University of Hamburg, Department of Psychiatry, Hamburg, Germany Odette de Wilde, Ph.D., Academic Medical Center, University of Amsterdam, Department of Psychiatry, The Netherlands E-mail: [email protected] Georg Winterer, M.D., Ph.D., Associate Professor, Department of Psychiatry, Heinrich-Heine University, Duesseldorf, and Institute of Neurosciences and Biophysics, Juelich Research Centre, Juelich, Germany E-mail: [email protected]

Contributors to Volume 1

Part I

Methodological and Technological Advances

Chapter 1

Where Do We Stand in the Quest for Neuropsychiatric Biomarkers and Endophenotypes and What Next? Michael S. Ritsner and Irving I. Gottesman

Abstract Although biomarker science is a field that is advancing rapidly in medicine as a whole, neuropsychiatric disorders are still characterized by an absence of the biomarkers and laboratory tests that will promote new diagnostic and prognostic procedures. Recent advances in genomic, genetic, epigenetic, neuroscience, proteomic and metabolomic knowledge and technologies have opened the way to searching for biomarkers, however, it is still a relatively new field for neuropsychiatry. In addition, candidate endophenotypes, important trait markers widely used for genetic studies, are useful for the development of heritable diagnostic and prognostic biomarkers (endophenotype strategy). This chapter provides definitions of biomarkers and endophenotypes, elucidating their types and properties that will make them useful in neuropsychiatric research and practice. Recent results in the schizophrenia and mood disorders literature that illustrate the usefulness of biomarkers and endophenotypes are also reviewed. We predict that both biomarker and endophenotypic approaches will open new avenues for practically important applications of genetics, neuroscience and “omics” advantages in neuropsychiatry. Keywords Neuropsychiatric disorders • biomarker • neuroscience • genomics • proteomics • metabolomics • endophenotypes • behavioral markers • schizophrenia • mood disorders

Michael S. Ritsner Department of Psychiatry, Rappaport Faculty of Medicine, Technion – Israel Institute of Technology, Haifa, and Sha’ar Menashe Mental Health Center, Israel Irving I. Gottesman Departments of Psychiatry and Psychology, University of Minnesota, Minneapolis, MN, USA

Abbreviations CSF: Cerebrospinal fluid; DLPFC: Dorsolateral prefrontal cortex; EEG: Electroencephalography; GWAS: Genome-wide association studies; 1H NMR: Nuclear magnetic resonance spectroscopy; LCECA: Liquid chromatography with electrochemical coulometric array detection; MALDI-TOF/TOF-MS: Tandem mass spectrometric fragmentation and de novo protein sequencing; PD: Parkinson’s disease; SELDITOF-MS: Fingerprinting profiling techniques

Introduction The diagnosis of neuropsychiatric disorders such as schizophrenia, major depression, bipolar and schizoaffective disorders, Parkinson and Alzheimer diseases, multiple sclerosis, epilepsy and others is still based largely on clinical examinations eliciting reports of symptoms by patients. This time-tested approach, deficient in specificity and sensitivity, has significant limitations for predicting diagnosis, onset, course of illness, and response to treatment. The field of biomarkers has made tremendous progress during the past several years. In particular, biomarkers have become accepted tools in practice. For example, in three major clinical areas – diagnosis of myocardial infarction, diagnosis and management of heart failure, and diagnosis and management of inflammatory conditions in general and systemic infections1 – promises have turned into reality. According to the NIMH Strategic Plan2 (http://www. nimh.nih.gov/about/strategic-planning-reports/index. shtml) to accelerate the identification of biomarkers and

M.S. Ritsner (ed.), The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes, © Springer Science + Business Media B.V. 2009

3

M.S. Ritsner and I.I. Gottesman

4

behavioral indicators for mental disorders, it will be important to: 1. Support the development of integrated profiles/panels of clinically relevant and validated biomarkers and behavioral indicators (e.g., genes, proteins, brain images, behaviors, or a combination), creating “biosignatures” of disorders. A single biomarker is not likely to be sufficient to indicate the presence of a disorder, but a configuration or combination of biomarkers and behavioral indicators of small effect might do so. 2. Support studies to identify biomarkers and behavioral indicators for different stages of illness and recovery (e.g., biomarkers for onset vs. relapse, biomarkers indicating risk vs. resilience). 3. Support research that examines biomarkers that may be common to mental disorders and other medical disorders (e.g., inflammatory markers for heart disease) in order to identify shared molecular pathways that contribute to development of mental disorders (Strategy 1.3: Identify and integrate biological markers [biomarkers] and behavioral indicators associated with mental disorders, p. 7). Despite the hope for biomarkers and the progress made so far, many challenges lie ahead. Consider that the total number of biomarkers of interest can be estimated

to be ~1,133,000, of which our genome accounts for ~20,500 genes (http://www.genome.gov/12011238; May 9, 2008), transcriptome, 100,000,3 proteome, 1,000,000,4 and metabonome, ~2,500–3,000.5 In recent years, a new translational approach, which aimed to connect basic research to patient care more directly, has been applied, wherein researchers attempt to examine genetic, neurobiological, and behavioral responses concomitantly.6,7 One challenge in applying the translational approach is to identify and characterize candidate biomarkers that may later be used for the identification of at-risk individuals; another, is early detection of illness (including presymptomatic detection of brain dysfunction), so as to facilitate identifying biomarkers reflecting pathophysiologic processes underlying complex neuropsychiatric disorders.8–12 In order to examine the evidence concerning research activity in the biomarker field, we carried out a MEDLINE database search for the last 8 years (from 01.01.2000 up to 31.12.2007), using the terms “biomarker” plus “schizophrenia” or “mood disorders” or “Alzheimer’s disorder” or “epilepsy” or “multiple sclerosis” or “heart failure” or “inflammatory conditions” (reviews were excluded). Figure 1.1 shows the growing the number of publications from 2000 to 2007 for these conditions. Of 1,646 publications found for 2007, 30%

600 Heart failure Alzheimer's disorder

500

Multiple sclerosis Inflammatory conditions Schizophrenia

Publications

400

Mood disorders Epilepsy

300

200

100

0 2000

2001

2002

2003

2004

2005

2006

2007

Fig. 1.1 Publications related to biomarkers have been added to MEDLINE between the years 2000 and 2007

1

Where Do We Stand in the Quest for Neuropsychiatric Biomarkers and Endophenotypes and What Next?

were related to heart failure, 16.7%, to Alzheimer’s disorder, 16.1%, to multiple sclerosis, 14.4%, to inflammatory conditions, 8.9%, to schizophrenia, 8.2%, to mood disorders, and 5.8%, to epilepsy. In this introductory chapter, we will review the state of the art and future challenges in the use of the biomarker and endophenotype strategies to study neuropsychiatric disorders. In particular, this chapter describes and compares current definitions and main properties of biomarkers and endophenotypes, progress in applications of these strategies to schizophrenia and mood disorders. Finally, some of the recent technological advantages in biomarker discovery are discussed. Many more data and specific details regarding these strategic concepts are presented in the chapters to follow.

5

Definitions and Properties Biological Marker Strategy Table 1.1 presents definitions of some main terms used in this book. Although there is no widely accepted definition of what constitutes a biomarker, an NIH study group committed to the following definition: “a characteristic that is objectively measured and evaluated as an indicator of normal biologic processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention”13 (http://www.genomicglossaries.com/CONTENT/Biomarkers.asp). Overall, a biomarker is an indicator of the presence or extent of a

Table 1.1 Overview of some definitions used in the field of marker’s research Terms Definition Biological marker

Measurable and quantifiable biological parameters which serve as indices for health – and physiology related assessments, such as disease risk, psychiatric disorders, environmental exposure and its effects, disease diagnosis, metabolic processes, substance abuse, pregnancy, cell line development, epidemiologic studies, etc. Clinical endpoint A characteristic or variable that reflects how a patient feels, functions or survives Surrogate biomarker A “surrogate marker” can be defined as “… a laboratory measurement or physical sign that is used in therapeutic trials as a substitute for a clinically meaningful endpoint that is a direct measure of how a patient feels, functions, or survives and is expected to predict the effect of the therapy.” Surrogate endpoints are often physiological or biochemical markers that can be relatively quickly and easily measured, and that are taken as being predictive of important clinical outcomes Difference between a The primary difference is that a biomarker is a “candidate” surrogate marker, whereas a surrogate biomarker and marker is a test used, and taken, as a measure of the effects of a specific treatment. A biomarker that a surrogate marker is intended to substitute for a clinical endpoint, for example in a treatment trial of a disease-modifying therapy. A surrogate endpoint is expected to predict clinical benefit (or harm, or lack of benefit or harm) based on epidemiological, therapeutic, pathophysiological or other scientific evidence Disease-modifying therapy A treatment that affects the underlying pathophysiology of the disease rather than purely its symptoms (although there may be symptom improvement as a result of these treatments) Validation It is the process of assessing the assay or measurement performance characteristics and qualification is evidentiary process of linking a biomarker with biology and clinical endpoints DNA marker A cloned chromosomal locus with allelic variation that can be followed directly by a DNA- based assay Genetic markers A single gene (DNA) for which a mutation, deletion, SNP or some other feature provides predictive value. A phenotypically recognizable genetic trait, which can be used to identify a genetic locus, a linkage group, or a recombination event Genomic markers A measurable DNA or RNA characteristic that is an indicator of normal biologic processes, pathogenic processes, and/or response to therapeutic or other inventions Metabolite biomarkers A pattern of metabolites that is able to discriminate or predict Metabolomic biomarker A metabolomic biomarker that predicts disease, measures progression, or monitors therapy potentially could be a single molecule, as well as a pattern of several molecules Proteomic biomarkers A protein expression pattern that is able to discriminate or predict Physiological biomarkers Various physiological responses have been measured and utilized as biomarkers. These include studies of such basic physiological functions as respiration, changes in growth rate, feeding, excretion, etc. Physiological responses are used to provide integrated measures of an organism’s well-being, based on a range of different functional attributes Source: Based on references14–17

6

biological process that is directly linked to the clinical manifestations and outcome of a particular disease. Mueller et al.1 (2008) consider a biomarker to be a protein or other macromolecule that is associated with a biological process or regulatory mechanism. Hence, measurement of this biomarker in blood, for example, might provide quantitative information that could be clinically helpful regarding such a mechanism. Depending on the application, there are different types of biomarkers: • Antecedent markers for evaluation of risk to a disorder. • Screening markers for the early detection of diseases. • Diagnostic markers for identification of disorder. • Biomarker signatures are indicators of a disease state that are usually linked to an ongoing pathophysiology and thus may also provide information and insights into the underlying molecular mechanisms of a given disease.18 • Prognostic markers to predict the likely course of the disease. In general, whenever a biomarker has been determined to be a valid predictor of disease, it should be used in risk assessment and public health policy. • Stratification markers that predict the likelihood of a drug response. Biomarkers may play a prominent role in use of pharmacogenetics, pharmacogenomics and pharmacoproteomics for development of personalized medicine. In order for a diagnostic biomarker to be useful, certain criteria need to be met. These criteria include the following (adapted from8,19): 1. The biomarker should reflect some basic pathophysiological process, and detect a fundamental feature of the disease. 2. The biomarker should be specific for the disease compared with related disorders. 3. The biomarker should not reflect clinical symptomatology and consequences of the disease. 4. The biomarker can be measured repeatedly over time and should be reproducible. 5. The biomarker should be measured in noninvasive and easy-to-perform tests that can be done at the bedside or in the outpatient setting. 6. The biomarker should not cause harm to the individual being assessed. 7. The biomarker should be reliable in many testing

M.S. Ritsner and I.I. Gottesman

environments/labs. 8. The biomarker should be cost effective. The concepts of state and trait have been enjoyed wide usage in personality psychology and in other areas of psychology.20–22 The distinction between them, as well as, the distinction of trait vs. state markers, has often assumed considerable importance in theory and research.23 A trait marker represents the properties of the behavioral, neuropsychological and biological processes that play an antecedent role in the pathophysiology of the neuropsychiatric disorder, whereas a state marker reflects the status of clinical manifestations in patients. An important kind of trait marker termed ‘endophenotype’, which relates only to genetically influenced phenotypical characteristics of patients, may be more closely related to the genotype than are the clinical manifestations of the disorder and, therefore, is widely used for genetically informed studies (Gottesman, Gould,24 and Chapter 8 by Annamari Tuulio-Henriksson et al. in this book). In Fig. 1.2, a relationship is shown between biomarkers and endophenotypes (it is the well-known triangle within triangle figure borrowed from Gould and Gottesman25). Terms closely related to trait marker include elementary phenotype, intermediate phenotype, risk indicator, risk marker, and vulnerability marker.26–29 Peripheral tissues suitable for exploring pathophysiological hypotheses and possibly for providing useful peripheral biomarkers for the diagnosis of neuropsychiatric disorders include skin fibroblasts, platelets, lymphocytes, as well as body fluids such as plasma or cerebrospinal fluid (CSF). Numerous studies have been performed to identify the potential trait markers in depression,30,31 bipolar disorder,32,33 schizophrenia,34–37 obsessive-compulsive disorder,38 Alzheimer disorder,39,40 alcoholism,41,42 and other neuropsychiatric disorders. Overall, markers may be biological (endogenous and exogenous in origin) and non- biological such as environmental (e.g., markers for environmental tobacco smoke43,44), psychological, and others. Genetic biomarkers are useful for diagnosis, patient stratification and prognostic or therapeutic categorization. During the past few years, a novel genetic approach – genomewide association studies (GWAS) – has demonstrated its potential to identify common genetic variants associated with complex diseases such as diabetes, as well as, to discovery of the genetic biomarkers.45–47 Genetic studies have identified genetic biomarkers or causal

1

Where Do We Stand in the Quest for Neuropsychiatric Biomarkers and Endophenotypes and What Next?

As po soc pu iat lat ed ion wi s th

Bi olo gic

so me

al m illn ar ess k in er s

Fig. 1.2 Possible relationships between markers and endophenotypes Current criteria for an endophenotype, to be distinguished from biomarkers, are designed to direct clinical research in psychiatry toward genetically and biologically meaningful conclusions. This schema (genes to endophenotypes to symptoms to disease), allowing for epigenetic, environmental, and stochastic influences, can be applied to other complex diseases as well. None of the components in the figure can be definitive; many more elements exist or await discovery (©2007 I.I. Gottesman and T.D.Gould, used with permission)

7

Endophenotypes 1) 2) 3) 4) 5)

Gottesman 2005

gene variants in more than 1,300 Mendelian diseases during the past 25 years.48 Genetic biomarkers are present at birth, theoretically enabling the institution of timely preventative or ameliorative measures. Optimum biomarker development and application will require a team approach because of the multifaceted nature of biomarker selection, validation, and application, using such techniques as pharmacoepidemiology, pharmacogenetics, pharmacogenomics, and functional proteomics; bioanalytical method development and validation; disease process and therapeutic intervention assessments; pharmacokinetic modeling and simulation to improve and refine drug development.49 A pressing challenge in biomarker research is determining the validity of the biomarker. A three-phased approach to validation of biomarkers involves dealing with a range of characteristics that include the intrinsic qualities of the biomarker, its determinants, and the analytic procedure.50

Endophenotype Strategy One important advance, in this context, has been conceptualization of the endophenotype strategy. This strategy was originally proposed for neuropsychiatric

Associated with illness Heritable State-independent (but may require challenge/provocation) Familial co-segregation with illness Found in some unaffected relatives (probabilism vs. determinism) Gould & Gottesman, 2005

genetics by Gottesman and Shields51,52 (who adapted it from insect genetics, John and Lewis53) in order to reduce the heterogeneity and complexity of research into common, multifactorial, genetically-influenced disorders. It would also boost the statistical power in that unaffected relatives with the candidate endophenotypes would now be included in designs. The term endophenotype refers to a set of quantitative, heritable, trait-related deficits typically assessed by biochemical, endocrine, neuroanatomical, neurophysiological, neuropsychological, and other methods24,52,54–56 (see Fig. 1.3 for a summary of domains used for searching neuropsychiatric biomarkers and endophenotypes and Fig. 1.1 in Chapter 8). Criteria useful for the identification of endophenotypes have been proposed, adapted, and refined over time (see24,51,52,54–60 for a comprehensive review and Table 1.2 for a summary of recent findings). Ideally, an endophenotype would have a simpler pattern of inheritance and a smaller number of genes explaining its variance than the associated symptoms of psychopathology. Additionally, it would be a product of one of the etiologic paths leading to disorder expression.61 Table 1.2 presents the main properties of endophenotypes compared to biomarkers. As can be seen, current criteria for an endophenotype, to be distinguished from other biological markers, are designed to direct clinical

M.S. Ritsner and I.I. Gottesman

8

ENDOPHENOTYPES

Cellular systems

Metabolic alterations DNA & GENES: Mendelian Major genes Polygenes Penetrance Expressivity

DNA markers

PHENOTYPES:

The geno-phenotype gradient Genomics Proteomics

Trait markers

Neuroimaging patterns Neuroanatomical features Neurophysiological features Environment

Neuropsychological functions

Response to drugs

Behavioral Disorder Subtypes Syndromes Symptoms

Metabolomics, metabonomics

BIOLOGICAL MARKERS

State markers

Fig. 1.3 Some domains used for searching biomarkers and endophenotypes for neuropsychiatric disorders. The biomarker concept has involved physiological tests, neuroimaging, biochemical, genetic, genomic panels, and proteomic technologies.

In addition to biological measures, biomarkers and endophenotypes may include neuropsychological and cognitive tasks thought to reflect brain mechanisms that underlie the disorder

research in neuropsychiatry toward genetically and biologically meaningful constructs that precede the appearance of symptoms and the diseases themselves. There is a growing consensus that an endophenotype approach may be utilized to overcome the difficulties regarding the ambiguities inherent in phenotypic description and to facilitate the identification of the susceptibility or protective genes of a wide spectrum of neuropsychiatric disorders (reviewed63,64). Indeed, this approach is being applied to schizophrenia,65–71 major depression and bipolar disorder,58,72–74 attention-deficit hyperactivity disorder,75 autism,76 alcohol dependence,77 and other complex conditions (see also other chapters in this volume). Lastly, although, the definition and use of endophenotypes in animal models of psychiatric illness is a developing area,25,78–80 there is a shift away from traditional animal models to more focused research dealing with an endophenotype-style approach, genetic models and incorporation of new findings from human neuroimaging and genetic studies.81,82

Technological Advantages Biomarker discovery has grown (see Fig. 1.1) during the past decade, driven by the availability of powerful new ‘omics’ technologies, like proteomic, and metabolomics. These technological advances facilitate the profiling of proteins and metabolites of large samples, as well as, providing interesting insights into disease mechanisms by capturing the dynamic nature of disease-related alterations.12 These new approaches may be able to detect unique patterns associated with neuropsychiatric disorders by measuring all the available proteins or metabolic pathways.

Proteomics Proteomics is the large-scale study of proteins, their structures and functions, of the expressed protein complement of cells, tissues, organs, or biological fluids.

1

Characteristics Definition

Endophenotype

Biological marker (biomarker)

An endophenotype is an internal phenotype discoverable by a “biochemical test or microscopic examination”

An endophenotype is ‘intermediate An endophenotype is characteristic, phenotype’ that forms the causal usually assessed in a laboratory, that links between genes and overt reflect the actions of genes predisexpression of disorder posing an individual to a specific disorder, even in the absence of any diagnosable pathology Heritability Heritable Heritable Highly heritable, it’s attributed to shared genetic, rather than environmental, factors Relation to given An endophenotype is a measurIt should be associated with causes It reflects a discrete and well-understood complex able component unseen by the of the disorder. Numerous neurobiological mechanism that is disorder unaided eye along the endophenotypes should affect a both informative for the pathophysipathway between disease and given complex disorder ology of a disorder and indicative of distal genotype the action of a limited number of genes State dependence State-independent State-independent State-independent General population Family

Relatives

Measurement

Reference

The endophenotype is associated with illness in the population

Endophenotype should vary continuously in the general population Endophenotype and illness The level of the endophenotype co-segregate should correlate with an individual’s level of genetic risk independent of his or her symptomatology or diagnosis The correlation on the endophenoAn endophenotype identified in typic trait between relatives probands is found in their discordant for the disorder unaffected relatives at a should be as high as the higher rate than in the general correlation between two population62 concordant relatives An endophenotype may be Endophenotype should optimally be neurophysiological, biochemimeasured across several levels of cal, endocrinological, analysis neuroanatomical, cognitive, or neuropsychological in nature [24] [54]

–

A biomarker is objectively measured substance used for evaluation disorder’s risk, or disorder, or change in the illness course, or response to a therapeutic intervention A biomarker could be environmental, genetic, or multifactorial in origin The biomarker should be associated to diagnose disease risk, presence of disease in an individual, or to tailor treatments for the disease. It can be thought of as a form of risk factor that is not causal The biomarker may be either trait or state-related The biomarker should be associated with illness in the general population

Endophenotype co-segregates with illness

The biomarker does not should be co-segregate with illness within families

An endophenotype is evident in unaffected members

Biomarker is characteristic seen in an individual with the disorder, in those at high risk for the disorder, and in some unaffected relatives

The endophenotype’s measurement should be rapid and easy, so that it can be readily acquired in large numbers of patients. High test-retest reliability [56]

The biomarker may be the external substance itself, a substance that is introduced into an organism

Where Do We Stand in the Quest for Neuropsychiatric Biomarkers and Endophenotypes and What Next?

Table 1.2 Comparison of main characteristics of the endophenotypes with biological markers

[14–17] 9

M.S. Ritsner and I.I. Gottesman

10

It includes the identification of proteins and changes in those proteins that occur in different biological and disease states.83 Proteomics is much more complicated than genomics, mostly because, while an organism’s genome is rather constant, a proteome differs from cell to cell and constantly changes through its biochemical interactions with the genome and the environment.84 Analytical proteomics involves identifying, quantifying, and characterizing the posttranslational modifications of the proteins in different proteomes; that is, changes in protein structure, that occur after the process whereby genetic information is translated into a working protein.85 Functional proteomics aims to obtain the same type of information on proteins that form functional complexes, and also to determine the sites and dynamics of interactions within the complexes. Pienaar and colleagues86 based on the major advances in the fast-growing field of neuroproteomics in Parkinson’s disease, defined neuroproteomics as an emerging tool to establish disease-associated protein profiles, while also generating a greater understanding as to how these proteins interact and undergo posttranslational modifications. Posttranslational modifications of proteins are an important part of the proteome that may play an essential role in determining function.87 Technological advances (polyacrylamide gel electrophoresis, electrospray ionization, matrix-assisted laser desorption/ionization [MALDI], analysis of MALDI-derived peptides in Time-of-Flight analyzers, and multidimensional protein identification technology and bioinformatics for data handling and interpretation) allow a large-scale identification of peptide sequence and post-translational modifications. Proteomic analysis can reveal protein expression levels, posttranslational modifications and protein-protein interactions. These proteomic tools have the power to identify quantitative and qualitative protein patterns in postmortem brain tissue, cerebrospinal fluid (CSF) or serum, thus increasing the knowledge about etiology and pathogenesis of brain diseases.88 Proteomic technologies allow identifying not only proteins but also the nature of their posttranslational modifications thus enabling the elucidation of signal transduction pathways and their deregulation under pathological conditions. The linkage of information about proteome changes with functional consequences lead to the development of functional proteomic studies. Comparing protein profiles in healthy and

disease states provides an opportunity to establish specific diagnostic and prognostic biomarkers. In addition, proteomic studies of the effects of medication – in vitro and in vivo – might help to design specific pharmaceutical agents with fewer side effects. It is now possible within a single study to identify numerous proteins and perform relative quantifications on them in two different samples. Moreover, combination of proteomic biomarkers with clinical phenotype, metabolite changes, and genetic haplotype information is promising for the physician assessment of individual risk profile.89 Comparisons can be made between groups of animals in different physiological states or in response to experimental treatment. Differences between normal individuals and those in disease states can form the foundation for elucidation of causative factors of disease and the identification of biomarkers for the diseased state.90

Metabolomics Metabolomics is defined as the quantitative measurement of the dynamic multiparametric metabolic response of living systems to pathophysiological stimuli or genetic modification.91 In other words, metabolomics is the systematic study of the unique chemical fingerprints that specific cellular processes leave behind – specifically, the study of their small-molecule metabolite profiles.92 At the center of metabolomics is the concept that an individual’s metabolic state is a close representation of the individual’s overall health status. This metabolic state reflects what has been encoded by the genome and modified by environmental factors.93 Metabolomics or metabolomic profiling, aims to profile all the small molecule metabolites found within a cell, tissue, organ, or organism and use this information to understand a biological manipulation such as a drug intervention or a gene knockout. While neither mass spectrometry or NMR spectroscopy, the two most commonly used analytical tools in metabolomics, can provide a complete coverage of the metabolome, compared with other functional genomic tools for profiling biological moieties the approach is cheap and high throughput.94,95 The metabolome has come to be defined as the complete set of metabolites in a given cell, tissue, biological sample

1

Where Do We Stand in the Quest for Neuropsychiatric Biomarkers and Endophenotypes and What Next?

or sub-fraction of a biological sample. The size of the metabolome is a matter of (possibly irrelevant) debate and depends strongly on the definition of what to include and on the analytical platforms and methods used to assess it.96 Metabolomics is a powerful tool for identifying any disturbances in normal homeostasis of metabolic processes, including those involving carbohydrates, fatty acids, amino acids, and nucleic acids. These normal metabolic processes are highly regulated and are vital for cell structure and function. These functions are catalyzed by various enzymes, which are in turn regulated by specific genes. Defects in genes/enzymes could lead to chronic diseases in humans and in some cases mortality. Metabonomics offers several advantages over genomics and proteomics97: • It has a relatively small number of biomarkers (~2,500–3,000). • It can be applied noninvasively in biofluids (plasma, urine, feces, etc.), which can be considered “expanded clinical chemistry.” • Combined toxicogenomic and metabonomic data in animal models can help focus on evaluating metabonomic biomarkers of clinical relevance. • Several diseases, particularly those from inborn errors of metabolism, have been characterized with specific clinical biomarkers in relation to clinical pathology. Several analytical metabonomics platforms have been applied to study potential biomarkers associated with myocardial ischemia,98 liver and epithelial ovarian cancer,99 and diabetes.95,100

Clinical Implications Both the biomarker and endophenotypic strategies have already been successful in investigation of many complex neuropsychiatric disorders. Today the identification of schizophrenia and mood disorders biomarkers and endophenotypes are a crucial step towards improving current diagnosis, developing new presymptomatic treatments, identifying high-risk individuals and disease subgroups, and assessing the efficacy of preventative interventions at a rate that is not currently possible. Overall, markers and endophenotypes can be categorized into the following groups:

11

neuroanatomical and neuroimaging, neurodevelopmental, physiological or electrophysiological, biochemical or metabolic alterations, psychological and neurocognitive deficits. Tables 1.3 and 1.4 list commonly investigated traits, which have been discussed as candidates for markers or endophenotypes for schizophrenia or mood disorders, or both of them, revealing a number of features (for more details see other chapters in this volume). Potential markers and candidate endophenotypes have been suggested for schizophrenia, including a variety of structural brain pathology, minor physical anomalies, neurocognitive deficits, various eventrelated potentials measured by electroencephalography (EEG), and olfactory identification deficits, neurocognitive deficits, including impairments in executive functioning, attention and memory domains, several electrophysiologic findings, such as sensorymotor gating deficits, and smooth pursuit eye-tracking abnormalities (reduced gain during smooth pursuit and increased saccade frequency), and biochemical alterations (Table 1.3). Presently, none of these measures has satisfactory performance characteristics in terms of predictive validity, noninvasiveness, ease of testing and low cost that would enable their widespread use.101 In addition, several genetic or DNA markers under current investigation that can increase a person’s risk for schizophrenia, albeit with very modest increments in odds ratios: COMT, DAT1, AKT1, 5-HTTLPR, GRIN1, GMR3, DAOA, NRG1, DTNBP1, MTHFR, NPAS3, DISC1, RGS4, HOPA, RTN4R, and PPP3CC102,103 (http:// www.schizophreniaforum.org/res/sczgene/default. asp; see also chapter by Ritsner and Susser in this volume). Biological markers and endophenotypes for major depression and bipolar disorders under current investigation also include neuroanatomical (e.g. subcortical gray matter volume abnormalities), neurophysiological (e.g. fronto-temporal alterations within the first 200 ms during an attentional task), biochemical (e.g. impaired response to corticotropin-releasing hormone), psychological (e.g. cyclothymic personality traits), and others (reviewed,10,58 see Table 1.4). It should be noted that biomarkers and endophenotypes may not be specific to the disorder and are shared across different conditions from mood disorders to schizoaffective disorders and schizophrenia: e.g. suppression of P50 auditory evoked responses, P50 sensory

M.S. Ritsner and I.I. Gottesman

12

Table 1.3 Selected candidates for biological marker and endophenotype measures in schizophrenia Kind of endophenotype or biomarker

Measure

Reference

Neuroanatomical or neuroimaging traits

Inward deformation of the anterior and posterior regions of the thalamus (thalamic shape abnormalities) Lateral ventricular enlargement Decreased volume of left and total anterior insular lobule Cerebellar abnormalities Olfactory bulb abnormalities Minor physical anomalies and neurologic soft signs P300 wave anomalies Smooth pursuit eye movement abnormalities Sensory-motor gating deficits (e.g. deficits in pre-pulse inhibition, PPI) Deficits in inhibitory functions of sensory gating measured via suppression of the P50 evoked potential An auditory event-related potential that provides an index of auditory sensory memory (mismatch negativity) Oculomotor measures regarding saccades and antisaccades Increased reflexive errors on the antisaccade task A weighted combination of electrophysiological features: mismatch negativity, P50, P300, and antisaccades Aggregability of red blood cells Phenylthiocarbamide perception Early visual processing deficits, as measured by clear amplitude reductions in the occipital P1 component of the visual event-related potential Increased alpha defensins in T cell lysates Decreased expression of reelin receptor VLDLR in peripheral lymphocytes Ceruloplasmin, C3 and C4 blood levels Elevation of serum levels of C-reactive protein Serum soluble L-selectin Niacin (vitamin B3) skin test Plasma homovanillic acid Peripheral-type benzodiazepine receptors Platelet 5-HT2A binding Brain-derived neurotrophic factor and nerve growth factor Neural growth factors in the peripheral blood of patients Face recognition deficits Visual form perception deficits Visual sustained attention deficits Deficits in prepulse inhibition Executive cognitive impairment Impairments in attention and memory domains Visuospatial working memory, verbal memory, language, oculomotor scanning/psychomotor speed, and general intelligence Intellectual asymmetry with a relative superiority of verbal skills to spatial skills Temperament types, emotional distress, emotion-oriented coping and self-constructs A source monitoring deficit

[104]

Neurodevelopmental signs Physiological or electrophysiological anomalies

Biochemical or immunological alterations

Psychological or neurocognitive deficits

gating ratio, inhibition of leading saccades during smooth pursuit eye movements, and cancellation of reflexive saccades in the antisaccade eye movement

[105, 106] [107] [108] [109] [110] [111] [112–118] [119–121] [122] [123, 124] [125] [126] [127] [128] [129] [130]

[131] [132] [133] [134] [135] [136–138] [139] [35] [140] [141] [37] [142–144] [145] [146,147] [56] [148] [149,150] [151–154] [155] [156, 157] [158]

task, and serum cortisol to dehydroepiandrosterone (DHEA) molar ratio (Table 1.4). Thus, these conditions may have some common biological basis, includ-

1

Where Do We Stand in the Quest for Neuropsychiatric Biomarkers and Endophenotypes and What Next?

13

Table 1.4 Selected candidates for biological marker and endophenotype in mood disorders Type of endophenotype or biomarker Bipolar disorder Neuroanatomical or neuroimaging traits Biochemical alterations Psychological features Major depression disorder (MDD) Physiological or electrophysiological anomalies Biochemical alterations

Psychological features

Measure

Reference

Subcortical gray matter volume abnormalities Hippocampal glutamate concentrations The fluorescence intensity of the membrane potential Melatonin Cyclothymic personality traits

[159] [160] [161] [162] [163]

Fronto-temporal alterations within the first 200 ms during an attentional task N-terminal pro-B-type natriuretic peptide Response to corticotropin-releasing hormone Waking salivary cortisol levels Hypothalamic-pituitary-adrenal activity 42 amino-acid form of beta-amyloid (Abeta42), cerebrospinal fluid Platelet imipramine binding, 5-HT1A receptor expression, soluble interleukin-2 receptor and interleukin-6 in serum, brain-derived neurotrophic factor in serum, blood folate levels, suppression of the dexamethasone suppression test Platelet adenylyl cyclase activity Dexamethasone/corticotropin releasing hormone test The melancholic type of personality

[164]

Bipolar and Major depression disorders Physiological or electrophysiological Increased REM density anomalies Pupillary reactivity Psychological Impairment of executive function (attentional set shifting) Schizophrenia, schizoaffective and bipolar disorders Physiological or electrophysiological Suppression of P50 auditory evoked responses, P50 sensory gating anomalies ratio, inhibition of leading saccades during smooth pursuit eye movements, and cancellation of reflexive saccades in the antisaccade eye movement task Biochemical alterations Serum cortisol to dehydroepiandrosterone molar ratio

ing SNPs or genes, that is reflected in common markers and/or endophenotypes. While proteomics has excelled in several disciplines in medicine (cancer, injury and aging), only a few studies on basic proteomics have been conducted for neuropsychiatric disorders. For instance, a PubMed search shows that of the 7,987 articles on the topic of proteomics, only 19 mention schizophrenia, while 1,339 were on cancer proteomics.188 Proteomics technologies have been used in the detection of differences between healthy individuals and patients suffering from motor neuron disease,189 Huntington disease,190 Parkinson disease,191 and schizophrenia192 and other neuropsychiatric disorders (see193–197 for a comprehensive review). Recently, about 330 unique proteins with deranged levels and modifications have been detected

[165] [166] [167] [168] [169] [170]

[171] [172] [173] [174] [175] [176] [177–183]

[184–187]

by proteomics approaches to be related to neuropsychiatric disorders. They are mainly involved in metabolism pathways, cytoskeleton formation, signal transduction, guidance, detoxification, transport, and conformational changes.196 However, the exact cellular and molecular mechanisms underlying these conditions have not been fully investigated. Two-dimensional (2D) gel electrophoresis of the cerebrospinal fluid (CSF) in unmedicated suicide attempters and non-attempters with major depressive disorder revealed that suicide attempters differed from non-attempters in one protein with an approximate molecular weight of 33 kiloDalton (kD) and an isoelectric point of 5.2.198 Proteomic analysis of the anterior cingulate cortex in the major psychiatric disorders provides support for the view that cytoskeletal

14

and mitochondrial dysfunction are important components of the neuropathology of the major psychiatric disorders.199 Findings from the high-throughput proteomic analysis of the dorsolateral prefrontal cortex (DLPFC) from schizophrenia, bipolar, and normal control cohorts demonstrate the usefulness of ProteinChipbased SELDI-TOF protein profiling in gaining insight into the molecular pathology of schizophrenia and bipolar disorder as it points to changes in protein levels characterizing these conditions.200 Holmes et al.192 used nuclear magnetic resonance 1 ( H NMR) spectroscopy in conjunction with computerized pattern recognition analysis in order to investigate metabolic profiles of CSF samples from a total of 152 drug-naïve or minimally treated patients with firstonset paranoid schizophrenia and healthy controls. Discriminant analysis showed a highly significant separation of these patients away from healthy controls. The alterations identified in drug-naïve patients could be validated in a test sample set achieving a sensitivity and specificity of 82% and 85%, respectively. Surface-enhanced laser desorption ionization mass spectrometry was employed to profile proteins and peptides in a total of 179 CSF samples (58 schizophrenia patients, 16 patients with depression, 5 patients with obsessive-compulsive disorder, 10 patients with Alzheimer disease, and 90 controls18). Authors found a highly significant differential distribution of samples from healthy subjects away from drug-naïve patients with first-onset paranoid schizophrenia. The key alterations were the up-regulation of a 40-amino acid VGFderived peptide, the down-regulation of transthyretin at approximately 4 kDa, and a peptide cluster at approximately 6,800–7,300 Da. These schizophreniaspecific protein peptide changes were replicated in an independent sample set. Both experiments achieved a specificity of 95% and a sensitivity of 80% or 88% in the initial study and in a subsequent validation study, respectively. Levin et al.201 used a set of 55 clinical serum samples from schizophrenia patients and healthy volunteers to show that the label-free proteomic approach yields reproducible results across a large number of samples and can be used to accurately measure the relative protein abundance. Authors identified 1,709 serum proteins covering a dynamic range of over three orders of magnitude. Prabakaran et al.202 investigated liver tissue and red blood cells (RBC) from schizophrenia patients and

M.S. Ritsner and I.I. Gottesman

controls using 2-D DIGE proteomic analysis. Analysis of the schizophrenia RBC proteome revealed eight proteins significantly altered in samples from 20 schizophrenia patients compared to 20 controls. Six of the altered proteins in the liver and four of the altered RBC proteins are related to oxidative stress. These results suggest that at least some of the pathological processes associated with the schizophrenia disease process can be traced in peripheral tissue. If peripheral cells can be used as a disease surrogate, promising new investigative avenues could be explored. Hirano et al.203 applied two global ‘omics’ approaches to develop an inventory of differentially expressed proteins and genes in Wig rat, a promising animal model of attention-deficit hyperactivity disorder (ADHD). Interestingly, ten proteins that were indentified in this study were also previously reported in studies involving neurodegenerative diseases and psychiatric disorders, such as Alzheimer’s disease (AD), Parkinson’s disease, and schizophrenia. Moreover, some of the proteins identified in the midbrain were involved in synaptic vesicular transport, suggesting abnormality in neurotransmitter release in this region. Bogdanov et al.191 utilized metabolomic profiling using high performance liquid chromatography coupled with electrochemical coulometric array detection (LCECA) to look for biomarkers in plasma useful for the diagnosis of 66 patients with Parkinson’s disease (PD) versus 25 controls. They also measured 8-hydroxy2-deoxyguanosine (8-OHdG) levels as a marker of oxidative damage to DNA. Authors found a complete separation of the two groups, and then determined the variables, which played the greatest role in separating the two groups and applied them to PD subjects taking dopaminergic medications. 8-OHdG levels were significantly increased in PD patients, but overlapped controls. Two other markers of oxidative damage were measured in our LCECA profiles. These findings show that metabolomic profiling with LCECA coulometric array has great promise for developing biomarkers for both the diagnosis, as well as monitoring disease progression in PD. Overall, this brief survey suggests that new molecular techniques can yield fresh insights into neuropsychiatric disorders. Additional studies are needed to confirm findings from these studies and to address many open questions through discovering useful biomarkers. We also need to remind ourselves of Seymour Kety’s204 cautions about the many ways that false positive

1

Where Do We Stand in the Quest for Neuropsychiatric Biomarkers and Endophenotypes and What Next?

findings can be generated by the special circumstances surrounding the lives of our patients.

Conclusions and Future Directions Current definitions of criteria, main types and properties of biomarkers and endophenotypes, some ‘omics’ technological advantages, and applications of biomarkerendophenotype strategy were briefly discussed in this chapter. There is growing interest in the concept of biomarkers and multivariate endophenotypes, which provide new opportunities and new challenges to understanding and treating neuropsychiatric disorders. In recent years, a translational approach, which examines genetic, neurobiological, and behavioral responses concomitantly, has been applied, and has yielded significant progress.6,7 According to this approach, neuropsychiatric research attempts to directly connect basic research to patient care. One challenge to applying this approach in studying complex disorders is to identify and characterize particular altered responses that can serve as trait markers. Studying trait markers, as opposed to state markers, make the complex disorder more tractable at genetic, neurobiological and psychological levels.205 Indeed, technological advantages and neuroscience tools that will lead to diagnoses based on biomarkers and treatments targeted to pathophysiology rather than symptoms should be more effective and eventually more accessible.206 The biologically distinct abnormalities identified by neuroscience should make treatments more effective and safer. While several biomarkers and endophenotypes have been reported, the heterogeneity and the complexity of neuropsychiatric disorders have greatly slowed down biomarker research. Therefore, the impact of biomarker strategy and technologies in neuropsychiatry is still in its infancy. No diagnostic test or other application of wide clinical use has yet emerged from these researches, although some enthusiasts of pharmacogenomics, less cautious than others, believe in marketing initial positive findings.207 In particular, selecting, evaluating and applying biomarkers in drug discovery and exploratory drug development do substantially shorten the time to reach a critical decision point. Furthermore, the use of biomarkers in early drug development helps

15

to streamline clinical development by determining whether the drug is reaching and affecting the molecular target in humans, delivering findings that are comparable to preclinical data, and by providing a measurable endpoint that predicts desired or undesired clinical effects. Thus, appropriateness of biomarkers depends on the stage of development, development strategy and the medical indications.208 The challenge now for neuropsychiatry is to identify biomarkers and putative endophenotypes that will allow us to diagnose and treat patients based on the biological factors. Although endophenotypes should be important tools for searching the candidate biomarkers for neuropsychiatric disorders, several impediments remain60,209 with the use of endophenotypes as biomarkers since many of them do not fit suggested criteria such as, for instance, the biomarker should be: (i) specific for the disease compared with related disorders, (ii) reproducible, (iii) measured in noninvasive and easy-to-perform tests that can be done at the bedside or in the outpatient setting, (iv) reliable across many testing environments/ labs, and (vi) inexpensive. Thus, it should be apparent that although great progress has been made in advancing neuropsychiatric biomarkers and endophenotypes, much remains to be done.

References 1. Mueller C, Müller B, Perruchoud AP. Biomarkers: past, present, and future. Swiss Med Wkly. 2008;138:225–229. 2. The National Institute of Mental Health Strategic Plan. NIMH, 2007, p. 37. http://www.nimh.nih.gov/about/strategicplanning-reports/index.shtml 3. Harrigan GG. Metabolomics: a ‘systems’ contribution to pharmaceutical discovery and drug development. Drug Discov World. 2006; Available at: http://www.ddwonline.com/data/pdfs/metabolomics.pdf. Accessed June 27, 2007. 4. Harrison PM, Kumar A, Lang N, et al. A question of size: the eukaryotic proteome and the problems in defining it. Nucleic Acids Res. 2002; 30:1083–1090. 5. Dettmer K, Hammock BD. Metabolomics – a new exciting field within “omics” sciences. Environ Health Perspect. 2004; 112:A396–A397. 6. Rosenberg RN. Translational research on the way to effective therapy for Alzheimer disease. Arch Gen Psychiatry. 2005; 62:1186–1192. 7. Hillered L, Vespa PM, Hovda DA. Translational neurochemical research in acute human brain injury: the current status and potential future for cerebral microdialysis. J Neurotrauma. 2005; 22:3–41.

16 8. Henley SM, Bates GP, Tabrizi SJ. Biomarkers for neurodegenerative diseases. Curr Opin Neurol. 2005; 18:698–705. 9. Shaw LM, Korecka M, Clark CM, et al. Biomarkers of neurodegeneration for diagnosis and monitoring therapeutics. Nat Rev Drug Discov. 2007; 6:295–303. 10. Phillips ML, Vieta E. Identifying functional neuroimaging biomarkers of bipolar disorder: toward DSM-V. Schizophr Bull. 2007; 33:893–904. 11. Bailey P. Biological markers in Alzheimer’s disease. Can J Neurol Sci. 2007; 34 Suppl 1:S72–S76. 12. Schwarz E, Bahn S. The utility of biomarker discovery approaches for the detection of disease mechanisms in psychiatric disorders. Br J Pharmacol. 2008; 153(Suppl 1): S133–S136. 13. Atkinson AEA. Biomarkers and surrogate endpoints: preferred definitions and conceptual framework. Clin Pharmacol Ther. 2001; 69:89–95. 14. Biomarkers Definition Working Group, 1998, Gregory Downing, NIH Initiatives in Surrogate Endpoints and Endpoint Analysis, Non-clinical Studies Subcommittee, Advisory Committee for Pharmaceutical Science presentation, 2000. 15. Biomarkers Definitions Working Group: Biomarkers and surrogate endpoints: preferred definitions and conceptual framework. Clin Pharmacol Ther. 2001; 69:89–95. 16. Russell K. Biomarkers and surrogate markers: an FDA perspective. NeuroRx. 2004; 1:189–195. 17. Wagner J, Merck A. Conference on Biomarkers Discovery and Validation, Oct. 14–18, 2004 http://bigdaddy.scripps. edu/darlene/Asilomar/pages/abstracts/jwagner.htm 18. Huang JT, Leweke FM, Oxley D, et al. Disease biomarkers in cerebrospinal fluid of patients with first-onset psychosis. PLoS Med. 2006; 3(11):e428. 19. Sunderland T, Gur RE, Arnold SE. The use of biomarkers in the elderly: current and future challenges. Biol Psychiatry. 2005; 58:272–276. 20. Alston WP. Traits, consistency, and conceptual alternatives for personality theory. J Theor Soc Behav. 1975; 5:17–48. 21. Zuckerman M. General and situation-specific traits and states: new approaches to assessment of anxiety and other constructs. In: Zuckerman M, Spielberger CD (eds), Emotion and anxiety: new concepts, methods, and applications. Erlbaum, Hillsdale, NJ; 1976, pp. 133–174. 22. Zuckerman M. The distinction between trait and state scales is not arbitrary: comment on Allen and Potkay’s “on the arbitrary distinction between traits and states.” J Pers Soc Psychol. 1983; 44:1083–1086. 23. Fridhandler BM. Conceptual note on state, trait, and the state-trait distinction. J Pers Soc Psychol. 1986; 50:169–174. 24. Gottesman II, Gould TD. The endophenotype concept in psychiatry: etymology and strategic intentions. Am J Psychiatry. 2003; 1(60):636–645. 25. Gould TD, Gottesman II. Psychiatric endophenotypes and the development of valid animal models. Genes Brain Behav. 2006; 5:113–119. 26. Adler LE, Freedman R, Ross RG, et al. Elementary phenotypes in the neurobiological and genetic study of schizophrenia. Biol Psychiatry. 1999; 46:8–18. 27. Jones PB, Tarrant CJ. Developmental precursors and biological markers for schizophrenia and affective disorders:

M.S. Ritsner and I.I. Gottesman specificity and public health implications. Eur Arch Psychiatry Clin Neurosci. 2000; 250:286–291. 28. Gould TD, Manji HK. The molecular medicine revolution and psychiatry: bridging the gap between basic neuroscience research and clinical psychiatry. J Clin Psychiatry. 2004; 65:598–604. 29. Agarwal DP. The genetics of alcohol metabolism and alcoholism. Indian J Hum Genet. 2001; 1:25–32. 30. Brittlebank AD, Scott J, Williams JM, Ferrier IN. Autobiographical memory in depression: state or trait marker? Br J Psychiatry. 1993; 162:118–121. 31. Sharma P, Rao K, Subbakrishna DK. Vulnerability to depression: a study of trait and state factors. Indian J Clin Psychol. 2001; 28:194–203. 32. Clark L, Goodwin GM. State- and trait-related deficits in sustained attention in bipolar disorder. Eur Arch Psychiatry Clin Neurosci. 2004; 254:61–68. 33. Zalla T, Joyce C, Szöke A, et al. Executive dysfunctions as potential markers of familial vulnerability to bipolar disorder and schizophrenia. Psychiatry Res. 2004; 121:207–217. 34. Ilani T, Ben-Shachar D, Strous RD, et al. A peripheral marker for schizophrenia: increased levels of D3 dopamine receptor mRNA in blood lymphocytes. Proc Natl Acad Sci USA. 2001; 98:625–628. 35. Ritsner M, Modai I, Gibel A, et al. Decreased platelet peripheral-type benzodiazepine receptors in persistently violent schizophrenia patients. J Psychiatr Res. 2003; 37:549–556. 36. Chen Y, Bidwell LC, Norton D. Trait vs. state markers for schizophrenia: identification and characterization through visual processes. Curr Psychiatry Rev. 2006; 2:431–438. 37. van Beveren NJ, van der Spelt JJ, de Haan L, Fekkes D. Schizophrenia-associated neural growth factors in peripheral blood. Eur Neuropsychopharmacol. 2006; 16:469–480. 38. Rao NP, Reddy YC, Kumar KJ, et al. Are neuropsychological deficits trait markers in OCD? Prog Neuropsychopharmacol Biol Psychiatry. 2008 June 8. 39. Gasparini L, Racchi M, Binetti G, et al. Peripheral markers in testing pathophysiological hypotheses and diagnosing Alzheimer’s disease. FASEB J. 1998; 12:17–34. 40. Ward M. Biomarkers for Alzheimer’s disease. Expert Rev Mol Diagn. 2007; 7:635–646. 41. Farren CK, Tipton KF. Trait markers for alcoholism: clinical utility. Alcohol 1999; 34:649–665. 42. Helander A. Biological markers in alcoholism. J Neural Transm Suppl. 2003; 66:15–32. 43. Kuusimäki L, Peltonen K, Vainiotalo S. Assessment of environmental tobacco smoke exposure of Finnish restaurant workers, using 3-ethenylpyridine as marker. Indoor Air. 2007; 17:394–403. 44. Okoli CT, Hall LA, Rayens MK, Hahn EJ. Measuring tobacco smoke exposure among smoking and nonsmoking bar and restaurant workers. Biol Res Nurs. 2007; 9:81–89. 45. Kingsmore SF, Kennedy N, Halliday HL, et al. Identification of diagnostic biomarkers for infection in premature neonates. Mol Cell Proteomics. 2008;7:1863 –1875. 46. Burmeister M, McInnis MG, Zöllner S. Psychiatric genetics: progress amid controversy. Nat Rev Genet. 2008; 9:527–540.

1

Where Do We Stand in the Quest for Neuropsychiatric Biomarkers and Endophenotypes and What Next?

47. Craddock N, O’Donovan MC, Owen MJ. Genome-wide association studies in psychiatry: lessons from early studies of non-psychiatric and psychiatric phenotypes. Mol Psychiatry. 2008; 13:649–653. 48. Botstein D, Risch N. Discovering genotypes underlying human phenotypes: past successes for mendelian disease, future approaches for complex disease. Nat Genet. 2003; 33 Suppl:228–237. 49. Colburn WA. Biomarkers in drug discovery and development: from target identification through drug marketing. J Clin Pharmacol. 2003; 43:329–341. 50. Bonassi S, Neri M, Puntoni R. Validation of biomarkers as early predictors of disease. Mutat Res./Fund Mol Mech Mutagen. 51. 2001; 480–481:349–358. 51. Gottesman II, Shields J. Schizophrenia and genetics; a twin study vantage point. Academic, New York. 1972. 52. Gottesman II, Shields J. Genetic theorizing and schizophrenia. Br J Psychiatry. 1973; 122:15–30. 53. John B, Lewis KR. Chromosome variability and geographical distribution in insects: chromosome rather than gene variation provide the key to differences among populations. Science. 1966; 152:711–721. 54. Cannon TD, Keller MC. Endophenotypes in the genetic analyses of mental disorders. Annu Rev Clin Psychol. 2006; 2:267–290. 55. Hasler G, Drevets WC, Manji HK, Charney DS. Discovering endophenotypes for major depression. Neuropsychopharmacology. 2004; 29:1765–1781. 56. Turetsky BI, Calkins ME, Light GA, et al. Neurophysiological endophenotypes of schizophrenia: the viability of selected candidate measures. Schizophr Bull. 2007; 33:69–94. 57. Braff DL, Freedman R, Schork NJ, Gottesman II. Deconstructing schizophrenia: an overview of the use of endophenotypes in order to understand a complex disorder. Schizophr Bull. 2007; 33:21–32. 58. Hasler G, Drevets WC, Gould TD, Gottesman II, Manji HK. Toward constructing an endophenotype strategy for bipolar disorders. Biol Psychiatry. 2006; 60:93–105. 59. Chan RC, Gottesman II. Neurological soft signs as candidate endophenotypes for schizophrenia: a shooting star or a Northern star? Neurosci Biobehav Rev. 2008; 32:957–971. 60. Walters JT, Owen MJ. Endophenotypes in psychiatric genetics. Mol Psychiatry. 2007; 12:886–890. 61. Frederick JA, Iacono WG. Beyond the DSM: defining endophenotypes for genetic studies of substance abuse current psychiatry reports. 2006; 8:144–150. 62. Leboyer M, Bellivier F, Nosten-Bertrand M, et al. Psychiatric genetics: search for phenotypes. Trends Neurosci. 1998; 21:102–105. 63. Braff DL, Greenwood TA, Swerdlow NR, et al. Advances in endophenotyping schizophrenia. World Psychiatry. 2008; 7:11–18. 64. Pearlson GD, Folley BS. Endophenotypes, dimensions, risks: is psychosis analogous to common inherited medical illnesses? Clin EEG Neurosci. 2008; 39:73–77. 65. Freedman R, Adler LE, Leonard S. Alternative phenotypes for the complex genetics of schizophrenia. Biol Psychiatry. 1999; 45:551–558. 66. Skuse DH. Endophenotypes and child psychiatry. Br J Psychiatry. 2001; 178:395–396.

17

67. Sporn AL, Greenstein DK, Gogtay N, et al. Progressive brain volume loss during adolescence in childhood-onset schizophrenia. Am J Psychiatry. 2003; 160:2181–2189. 68. Braff DL, Freedman R. Endophenotypes in studies of the genetics of schizophrenia. In: Davis KL, Charney DS, Coyle JT, Nemeroff C (eds) Neuropsychopharmacology: the fifth generation of progress. Lippincott Williams & Wilkens, Philadelphia, PA; 2002, pp. 703–716. 69. Egan MF, Goldberg TE. Intermediate cognitive phenotypes associated with schizophrenia. Methods Mol Med. 2003; 77:163–197. 70. Lenzenweger MF. Schizophrenia: refining the phenotype, resolving endophenotypes. Behav Res Ther. 1999; 37:281–295. 71. Gur RE, Calkins ME, Gur RC, et al. The consortium on the genetics of schizophrenia: neurocognitive endophenotypes. Schizophr Bull. 2007; 33:49–68. 72. Glahn DC, Bearden CE, Niendam TA, Escamilla MA. The feasibility of neuropsychological endophenotypes in the search for genes associated with bipolar affective disorder. Bipolar Disord. 2004; 6:171–182. 73. Lenox RH, Gould TD, Manji HK. Endophenotypes in bipolar disorder. Am J Med Genet. 2002; 114:391–406. 74. Benes FM. Searching for unique endophenotypes for schizophrenia and bipolar disorder within neural circuits and their molecular regulatory mechanisms. Schizophr Bull. 2007; 33:932–936. 75. Waldman ID. Statistical approaches to complex phenotypes: evaluating neuropsychological endophenotypes for attention-deficit/hyperactivity disorder. Biol Psychiatry. 2005; 57:1347–1356. 76. Belmonte MK, Cook EH Jr, Anderson GM, et al. Autism as a disorder of neural information processing: directions for research and targets for therapy. Mol Psychiatry. 2004; 9:646–663. 77. Dick DM, Jones K, Saccone N, et al. Endophenotypes successfully lead to gene identification: results from the collaborative study on the genetics of alcoholism. Behav Genet. 2005; 10:1–15. 78. Gould TD, Einat H. Animal models of bipolar disorder and mood stabilizer efficacy: a critical need for improvement. Neurosci Biobehav Rev. 2007; 31:825–831. 79. Harrison-Read PE. Models of mania and antimanic drug actions: progressing the endophenotype approach. J Psychopharmacol. 2008 [epub ahead of print]. 80. Flint J, Shifman S. Animal models of psychiatric disease. Curr Opin Genet Dev. 2008 Aug 11 [epub ahead of print]. 81. Cryan JF, Slattery DA. Animal models of mood disorders: recent developments. Curr Opin Psychiatry. 2007; 20:1–7. 82. Powell SB, Geyer MA. Overview of animal models of schizophrenia. Curr Protoc Neurosci. 2007; Chapter 9:Unit 9.24. 83. Scobioala S, Klocke R, Michel G, et al. Proteomics: state of the art and its application in cardiovascular research. Curr Med Chem. 2004; 11:3203–3218. 84. Anderson NL, Anderson NG. Proteome and proteomics: new technologies, new concepts, and new words. Electrophoresis. 1998; 19:1853–1861. 85. Liebler DC. Analytical proteomics approaches to analysis of protein modifications: tools for studying proteome-environment interactions. In Toxicogenomics. Wiley. 2004, pp. 283–297.

18 86. Pienaar IS, Daniels WM, Götz J. Neuroproteomics as a promising tool in Parkinson’s disease research. J Neural Transm. 2008;115:1413 – 30. 87. Sigal A, Milo R, Cohen A, et al. Dynamic proteomics in individual human cells uncovers widespread cell-cycle dependence of nuclear proteins. Nat Methods. 2006; 3:525–531. 88. Hünnerkopf R, Grassl J, Thome J. Proteomics: biomarker research in psychiatry. Fortschr Neurol Psychiatr. 2007; 75:579–586. 89. Balestrieri ML, Giovane A, Mancini FP, Napoli C. Proteomics and cardiovascular disease: an update. Curr Med Chem. 2008; 15:555–572. 90. Blake CA. Physiological proteomics: cells, organs, biological fluids, and biomarkers. Exp Biol Med. 2005; 230:785–786. 91. Harrigan GG, Goodacre R. (eds), Metabolic profiling: its role in biomarker discovery and gene function analysis. Kluwer, Boston, MA; 2003. 92. Daviss B. Growing pains for metabolomics. Scientist. 2005; 19:25–28. 93. Kaddurah-Daouk R. Metabolic profiling of patients with schizophrenia. PLoS Med. 2006; 3(8):e363 doi:10.1371/ journal.pmed.0030363. 94. Goodacre R, Vaidyanathan S, Dunn WB, et al. Metabolomics by numbers: acquiring and understanding global metabolite data. Trends Biotechnol. 2004; 22:245–252. 95. Griffin JL, Vidal-Puig A. Current challenges in metabolomics for diabetes research: a vital functional genomic tool or just a ploy for gaining funding? Physiol Genomics. 2008; 34:1–5. 96. Kell DB. Metabolomic biomarkers: search, discovery and validation. Expert Rev Mol Diagn. 2007; 7:329–333. 97. Vangala S, Tonelli A. Biomarkers, metabonomics, and drug development: can inborn errors of metabolism help in understanding drug toxicity?. AAPS J. 2007; 9:E284–E297. 98. Sabatine MS, Liu E, Morrow DA, et al. Metabolomic identification of novel biomarkers of myocardial ischemia. Circulation. 2005; 112:3868–3875. 99. Odunsi K, Wollman RM, Ambrosone CM, et al. Detection of epithelial ovarian cancer using 1H-NMR-based metabonomics. Int J Cancer. 2005; 113:782–788. 100. Yang J, Xu G, Hong Q, et al. Discrimination of type 2 diabetic patients from healthy controls by using metabonomics method based on their serum fatty acid profiles. J Chromatogr B Analyt Technol Biomed Life Sci. 2004; 813:53–58. 101. Copolov D, Crook J. Biological markers and schizophrenia. Aust N Z J Psychiatry. 2000; 34 Suppl:S108–S112. 102. Bertram L. Genetic research in schizophrenia: new tools and future perspectives. Schizophr Bull. 2008; 34:806–812. 103. Allen NC, Bagade S, McQueen MB, et al. Systematic meta-analyses and field synopsis of genetic association studies in schizophrenia: the SzGene database. Nat Genet. 2008; 40:827–834. 104. Harms MP, Wang L, Mamah D, et al.Thalamic shape abnormalities in individuals with schizophrenia and their nonpsychotic siblings. J Neurosci. 2007; 27:13835–13842. 105. Shenton ME, Dickey CC, Frumin M, McCarley RW. A review of MRI findings in schizophrenia. Schizophr Res. 2001; 49:1–52.

M.S. Ritsner and I.I. Gottesman 106. McDonald C, Marshall N, Sham PC, et al. Regional brain morphometry in patients with schizophrenia or bipolar disorder and their unaffected relatives. Am J Psychiatry. 2006; 163:478–487. 107. Makris N, Goldstein JM, Kennedy D, et al. Decreased volume of left and total anterior insular lobule in schizophrenia. Schizophr Res. 2006; 83:155–171. 108. Marcelis M, Suckling J, Woodruff P, et al. Searching for a structural endophenotype in psychosis using computational morphometry. Psychiatry Res. 2003; 122:153–167. 109. Turetsky BI, Moberg PJ, Arnold SE, et al. Low olfactory bulb volume in first-degree relatives of patients with schizophrenia. Am J Psychiatry. 2003; 160:703–708. 110. John JP, Arunachalam V, Ratnam B, Isaac MK. Expanding the schizophrenia phenotype: a composite evaluation of neurodevelopmental markers. Compr Psychiatry. 2008; 49:78–86. 111. Bramon E, McDonald C, Croft RJ, et al. Is the P300 wave an endophenotype for schizophrenia? A meta-analysis and a family study. Neuroimage. 2005; 27:960–968. 112. Levy DL, Holzman PS, Matthysse S, et al. Eye tracking and schizophrenia: a selective review. Schizophr Bull. 1994; 20:47–62. 113. Flechtner KM, Steinacher B, Mackert A. Subthreshold symptoms and vulnerability indicators (e.g., eye tracking dysfunction) in schizophrenia. Compr Psychiatry. 2000; 41(2 Suppl 1):86–89. 114. Tsuang M. Schizophrenia: genes and environment. Biol Psychiatry. 2000; 47:210–220. 115. McDowell JE, Brown GG, Paulus M, et al. Neural correlates of refixation saccades and antisaccades in normal and schizophrenia subjects. Biol Psychiatry. 2002; 51:216–223. 116. Iacono WG, Moreau M, Beiser M, et al. Smooth-pursuit eye tracking in first-episode psychotic patients and their relatives. J Abnorm Psychol. 1992; 101:104–116. 117. Amador XF, Malaspina D, Sackeim HA, et al. Visual fixation and smooth pursuit eye movement abnormalities in patients with schizophrenia and their relatives. J Neuropsychiatry Clin Neurosci. 1995; 7:197–206. 118. Ross RG, Olincy A, Mikulich SK, et al. Admixture analysis of smooth pursuit eye movements in probands with schizophrenia and their relatives suggests gain and leading saccades are potential endophenotypes. Psychophysiology. 2002; 39:809–819. 119. Blackwood D. P300, a state and a trait marker in schizophrenia. Lancet. 2000; 355:771–772. 120. Cadenhead KS, Swerdlow NR, Shafer KM, et al. Modulation of the startle response and startle laterality in relatives of schizophrenic patients and in subjects with schizotypal personality disorder: evidence of inhibitory deficits. Am J Psychiatry. 2000; 157:1660–1668. 121. Light GA, Braff DL. Measuring P50 suppression and prepulse inhibition in a single recording session. Am J Psychiatry. 2001; 158:2066–2068. 122. Freedman R, Adams CE, Adler LE, et al. Inhibitory neurophysiological deficit as a phenotype for genetic investigation of schizophrenia. Am J Med Gen. 2000; 97:58–64. 123. Umbricht D, Koller R, Schmid L, et al. How specific are deficits in mismatch negativity generation to schizophrenia? Biol Psychiatry. 2003; 53:1120–1131.

1

Where Do We Stand in the Quest for Neuropsychiatric Biomarkers and Endophenotypes and What Next?

124. Umbricht D, Krljes S. Mismatch negativity in schizophrenia: a meta-analysis. Schizophr Res. 2005; 76:1–23. 125. Ross RG, Meinlein S, Zerbe GO, et al. Saccadic eye movement task identifies cognitive deficits in children with schizophrenia, but not in unaffected child relatives. J Child Psychol Psychiatry. 2005; 46:1354–1362. 126. Calkins ME, Curtis CE, Iacono WG, Grove WM. Antisaccade performance is impaired in medically and psychiatrically healthy biological relatives of schizophrenia patients. Schizophr Res. 2004; 71:167–178. 127. Price GW, Michie PT, Johnston J, et al. A multivariate electrophysiological endophenotype, from a unitary cohort, shows greater research utility than any single feature in the Western Australian family study of schizophrenia. Biol Psychiatry. 2006; 60:1–10. 128. Barshtein G, Ponizovsky AM, Nechamkin Y, Ritsner M, et al. Aggregability of red blood cells of schizophrenia patients with negative syndrome is selectively enhanced. Schizophr Bull. 2004; 30:913–922. 129. Moberg PJ, McGue C, Kanes SJ, et al. Phenylthiocarbamide (PTC) perception in patients with schizophrenia and firstdegree family members: relationship to clinical symptomatology and psychophysical olfactory performance. Schizophr Res. 2007; 90:221–228. 130. Yeap S, Kelly SP, Sehatpour P, et al. Early visual sensory deficits as endophenotypes for schizophrenia: high-density electrical mapping in clinically unaffected first-degree relatives. Arch Gen Psychiatry. 2006; 63:1180–1188. 131. Craddock RM, Huang JT, Jackson E, et al. Increased alpha defensins as a blood marker for schizophrenia susceptibility. Mol Cell Proteomics. 2008;7:1204 – 1213. 132. Suzuki K, Nakamura K, Iwata Y, et al. Decreased expression of reelin receptor VLDLR in peripheral lymphocytes of drug-naive schizophrenic patients. Schizophr Res. 2008; 98:148–156. 133. Morera AL, Henry M, García-Hernández A, FernándezLópez L. Acute phase proteins as biological markers of negative psychopathology in paranoid schizophrenia. Actas Esp Psiquiatr. 2007; 35:249–252. 134. Dickerson F, Stallings C, Origoni A, et al. C-reactive protein is associated with the severity of cognitive impairment but not of psychiatric symptoms in individuals with schizophrenia. Schizophr Res. 2007; 93:261–265. 135. Iwata Y, Suzuki K, Nakamura K, et al. Increased levels of serum soluble L-selectin in unmedicated patients with schizophrenia. Schizophr Res. 2007; 89:154–160. 136. Messamore E. Relationship between the niacin skin flush response and essential fatty acids in schizophrenia. Prostaglandins Leukot Essent Fatty Acids. 2003; 69:413–419. 137. Smesny S, Rosburg T, Riemann S, et al. Impaired niacin sensitivity in acute first-episode but not in multi-episode schizophrenia. Prostaglandins Leukot Essent Fatty Acids. 2005; 72:393–402. 138. Smesny S, Klemm S, Stockebrand M, et al. Endophenotype properties of niacin sensitivity as marker of impaired prostaglandin signalling in schizophrenia. Prostaglandins Leukot Essent Fatty Acids. 2007; 77:79–85. 139. Sumiyoshi T, Kurachi M, Kurokawa K, et al. Plasma homovanillic acid in the prodromal phase of schizophrenia. Biol Psychiatry. 2000; 47:428–433. 140. Arranz B, Rosel P, San L, et al. Low baseline serotonin-2A receptors predict clinical response to olanzapine in first-

19

episode schizophrenia patients. Psychiatry Res. 2007; 153:103–109. 141. Buckley PF, Pillai A, Evans D, et al. Brain derived neurotropic factor in first-episode psychosis. Schizophr Res. 2007; 91:1–5. 142. Sachs G, Steger-Wuchse D, Kryspin-Exner I, et al. Facial recognition deficits and cognition in schizophrenia. Schizophr Res. 2004; 68:27–35. 143. Calkins ME, Gur RC, Ragland JD, Gur RE. Face recognition memory deficits and visual object memory performance in patients with schizophrenia and their relatives. Am J Psychiatry. 2005; 162:1963–1966. 144. Leppänen JM, Niehaus DJ, Koen L, et al. Deficits in facial affect recognition in unaffected siblings of Xhosa schizophrenia patients: evidence for a neurocognitive endophenotype. Schizophr Res. 2008; 99:270–273. 145. Kimhy D, Corcoran C, Harkavy-Friedman JM, et al. Visual form perception: a comparison of individuals at high risk for psychosis, recent onset schizophrenia and chronic schizophrenia. Schizophr Res. 2007 Dec; 97(1–3):25–34. 146. Chen WJ, Faraone SV. Sustained attention deficits as markers of genetic susceptibility to schizophrenia. Am J Med Genet. 2000; 97:52–57. 147. Wang Q, Chan R, Sun J, et al. Reaction time of the Continuous Performance Test is an endophenotypic marker for schizophrenia: a study of first-episode neuroleptic-naive schizophrenia, their non-psychotic first-degree relatives and healthy population controls. Schizophr Res. 2007; 89:293–298. 148. Birkett P, Sigmundsson T, Sharma T, et al. Executive function and genetic predisposition to schizophrenia-the Maudsley family study. Am J Med Genet B Neuropsychiatr Genet. 2007; 147B:285–293. 149. Egan MF, Goldberg TE, Gscheidle T, et al. Relative risk of attention deficits in siblings of patients with schizophrenia. Am J Psychiatry. 2000; 157:1309–1316. 150. Sitskoorn MM, Aleman A, Ebisch SJ, et al. Cognitive deficits in relatives of patients with schizophrenia: a metaanalysis. Schizophr Res. 2004; 71:285–295. 151. Goldberg TE, Torrey EF, Gold JM, et al. Genetic risk of neuropsychological impairment in schizophrenia: a study of monozygotic twins discordant and concordant for the disorder. Schizophr Res. 1995; 17:77–84. 152. Cannon TD, Huttunen MO, Lonnqvist J, et al. The inheritance of neuropsychological dysfunction in twins discordant for schizophrenia. Am J Hum Genet. 2000; 67:369–382. 153. Glahn DC, Therman S, Manninen M, et al. Spatial working memory as an endophenotype for schizophrenia. Biol Psychiatry. 2003; 53:624–626. 154. Saperstein AM, Fuller RL, Avila MT, et al. Spatial working memory as a cognitive endophenotype of schizophrenia: assessing risk for pathophysiological dysfunction. Schizophr Bull. 2006; 32:498–506. 155. Kravariti E, Toulopoulou T, Mapua-Filbey F, et al. Intellectual asymmetry and genetic liability in first-degree relatives of probands with schizophrenia. Br J Psychiatry. 2006;188:186–187. 156. Ritsner M, Susser E. Temperament types are associated with weak self-construct, elevated distress and emotionoriented coping in schizophrenia: evidence for a complex vulnerability marker? Psychiatry Res. 2004; 128:219–228. 157. Ritsner MS, Ratner Y, Gibel A, Weizman R. Positive family history is associated with persistent elevated emotional

20 distress in schizophrenia: evidence from a 16-month follow-up study. Psychiatry Res. 2007; 153:217–223. 158. Brunelin J, d’Amato T, Brun P, et al. Impaired verbal source monitoring in schizophrenia: an intermediate trait vulnerability marker? Schizophr Res. 2007; 89:287–292. 159. Ladouceur CD, Almeida JR, Birmaher B, et al. Subcortical gray matter volume abnormalities in healthy bipolar offspring: potential neuroanatomical risk marker for bipolar disorder? J Am Acad Child Adolesc Psychiatry. 2008; 47:532 –539. 160. Colla M, Schubert F, Bubner M, et al. Glutamate as a spectroscopic marker of hippocampal structural plasticity is elevated in long-term euthymic bipolar patients on chronic lithium therapy and correlates inversely with diurnal cortisol. Mol Psychiatry. 2008 [epub ahead of print]. 161. Thiruvengadam AP, Chandrasekaran K. Evaluating the validity of blood-based membrane potential changes for the identification of bipolar disorder I. J Affect Disord. 2007; 100:75–82. 162. Srinivasan V, Smits M, Spence W, et al. Melatonin in mood disorders. World J Biol Psychiatry. 2006; 7:138–151. 163. Hantouche EG, Akiskal HS. Toward a definition of a cyclothymic behavioral endophenotype: which traits tap the familial diathesis for bipolar II disorder? J Affect Disord. 2006; 96:233–237. 164. Kemp AH, Hopkinson PJ, Hermens DF, et al. Frontotemporal alterations within the first 200 ms during an attentional task distinguish major depression, non-clinical participants with depressed mood and healthy controls: a potential biomarker? Hum Brain Mapp. 2008 Jan 7 [epub ahead of print]. 165. Politi P, Minoretti P, Piaggi N, et al. Elevated plasma N-terminal ProBNP levels in unmedicated patients with major depressive disorder. Neurosci Lett. 2007; 417:322–325. 166. Steimer T, Python A, Schulz PE, Aubry JM. Plasma corticosterone, dexamethasone (DEX) suppression and DEX/ CRH tests in a rat model of genetic vulnerability to depression. Psychoneuroendocrinology. 2007; 32:575–579. 167. Mannie ZN, Harmer CJ, Cowen PJ. Increased waking salivary cortisol levels in young people at familial risk of depression. Am J Psychiatry. 2007; 164:617–621. 168. Friess E, Schmid D, Modell S, et al. Dex/CRH-test response and sleep in depressed patients and healthy controls with and without vulnerability for affective disorders. J Psychiatr Res. 2008;42:1154 –1162. 169. Gudmundsson P, Skoog I, Waern M, et al. The relationship between cerebrospinal fluid biomarkers and depression in elderly women. Am J Geriatr Psychiatry. 2007; 15:832–838. 170. Mössner R, Mikova O, Koutsilieri E, et al. Consensus paper of the WFSBP Task Force on Biological Markers: biological markers in depression. World J Biol Psychiatry. 2007; 8:141–174. 171. Hines LM, Tabakoff B. WHO/ISBRA study on state and trait markers of alcohol use and dependence investigators. Platelet adenylyl cyclase activity: a biological marker for major depression and recent drug use. Biol Psychiatry. 2005; 58:955–962. 172. Ising M, Horstmann S, Kloiber S, et al. Combined dexamethasone/corticotropin releasing hormone test predicts

M.S. Ritsner and I.I. Gottesman treatment response in major depression - a potential biomarker? Biol Psychiatry. 2007; 62:47–54. 173. Stanghellini G, Bertelli M, Raballo A. Typus melancholicus: personality structure and the characteristics of major unipolar depressive episode. J Affect Disord. 2006; 93: 159–167. 174. Modell S, Ising M, Holsboer F, Lauer CJ. The Munich vulnerability study on affective disorders: premorbid polysomnographic profile of affected high-risk probands. Biol Psychiatry. 2005; 58:694–699. 175. Silk JS, Dahl RE, Ryan ND, et al. Pupillary reactivity to emotional information in child and adolescent depression: links to clinical and ecological measures. Am J Psychiatry. 2007; 164:1873–1880. 176. Clark L, Sarna A, Goodwin GM. Impairment of executive function but not memory in first-degree relatives of patients with bipolar I disorder and in euthymic patients with unipolar depression. Am J Psychiatry. 2005; 162:1980–1982. 177. Kumra S, Sporn A, Hommer DW, et al. Smooth pursuit eye-tracking impairment in childhood-onset psychotic disorders. Am J Psychiatry. 2001; 158:1291–1298. 178. Kathmann N, Hochrein A, Uwer R, Bondy B. Deficits in gain of smooth pursuit eye movements in schizophrenia and affective disorder patients and their unaffected relatives. Am J Psychiatry. 2003; 160:696–702. 179. Louchart-de la Chapelle S, Nkam I, Houy E, et al. A concordance study of three electrophysiological measures in schizophrenia. Am J Psychiatry. 2005; 162:466–474. 180. Martin LF, Hall MH, Ross RG, et al. Physiology of schizophrenia, bipolar disorder, and schizoaffective disorder. Am J Psychiatry. 2007; 164:1900–1906. 181. de Wilde OM, Bour LJ, Dingemans PM, Koelman JH, Linszen DH. A meta-analysis of P50 studies in patients with schizophrenia and relatives: differences in methodology between research groups. Schizophr Res. 2007; 97: 137–151. 182. Patterson JV, Hetrick WP, Boutros NN, et al. P50 sensory gating ratios in schizophrenics and controls: a review and data analysis. Psychiatry Res. 2008; 158:226–247. 183. Hong LE, Turano KA, O’Neill H, et al. Refining the predictive pursuit endophenotype in schizophrenia. Biol Psychiatry. 2008; 63:458–464. 184. Young AH, Gallagher P, Porter RJ. Elevation of the cortisol-dehydroepiandrosterone ratio in drug-free depressed patients. Am J Psychiatry. 2002; 159:1237–1239. 185. Ritsner M, Maayan R, Gibel A, et al. Elevation of the cortisol/dehydroepiandrosterone ratio in schizophrenia patients. Eur Neuropsychopharmacol. 2004; 14:267–273. 186. Ritsner M, Gibel A, Maayan R, et al. A. State and trait related predictors of serum cortisol to DHEA(S) molar ratios and hormone concentrations in schizophrenia patients. Eur Neuropsychopharmacol. 2007; 17:257–264. 187. Gallagher P, Watson S, Smith MS, et al. Plasma cortisoldehydroepiandrosterone (DHEA) ratios in schizophrenia and bipolar disorder. Schizophr Res. 2007; 90:258–265. 188. Lakhan SE. Schizophrenia proteomics: biomarkers on the path to laboratory medicine? Diagn Pathol. 2006; 1:11 doi:10.1186/1746-1596-1-11. 189. Rozen S, Cudkowicz ME, Bogdanov M, et al. Metabolomic analysis and signature in motor neuron disease. Metabolomic. 2005; 2:101–108.

1

Where Do We Stand in the Quest for Neuropsychiatric Biomarkers and Endophenotypes and What Next?

190. Underwood BR, Broadhurst D, Dunn WB, et al. Huntington disease patients and transgenic mice have similar pro-catabolic serum metabolite profiles. Brain. 2006; 129:877–886. 191. Bogdanov M, Matson WR, Wang L, et al. Metabolomic profiling to develop blood biomarkers for Parkinson’s disease. Brain. 2008; 131:389–396. 192. Holmes E, Tsang TM, Huang JT, et al. Metabolomic profiling of CSF: evidence that early intervention may impact on disease progression and outcome in schizophrenia. PLoS Med. 2006; 3:e327. 193. Marcotte ER, Srivastava LK, Quirion R. cDNA microarray and proteomic approaches in the study of brain diseases: focus on schizophrenia and Alzheimer’s disease. Pharmacol Ther. 2003; 100:63–74. 194. Vercauteren FG, Bergeron JJ, Vandesande F, et al. Proteomic approaches in brain research and neuropharmacology. Eur J Pharmacol. 2004; 500:385–398. 195. Schmidt O, Schulenborg T, Meyer HE, et al. How proteomics reveals potential biomarkers in brain diseases. Expert Rev Proteomics. 2005; 2:901–913. 196. Fountoulakis M, Kossida S. Proteomics-driven progress in neurodegeneration research. Electrophoresis 2006; 27: 1556–1573. 197. Kobeissy FH, Sadasivan S, Liu J, et al. Psychiatric research: psychoproteomics, degradomics and systems biology. Expert Rev Proteomics. 2008; 5:293–314. 198. Brunner J, Bronisch T, Uhr M, et al. Proteomic analysis of the CSF in unmedicated patients with major depressive disorder reveals alterations in suicide attempters. Eur Arch Psychiatry Clin Neurosci. 2005; 255:438–440. 199. Beasley CL, Pennington K, Behan A, et al. Proteomic analysis of the anterior cingulate cortex in the major psychiatric disorders: evidence for disease-associated changes. Proteomics. 2006; 6:3414–3425.

21

200. Novikova SI, He F, Cutrufello NJ, Lidow MS. Identification of protein biomarkers for schizophrenia and bipolar disorder in the postmortem prefrontal cortex using SELDITOF-MS ProteinChip profiling combined with MALDITOF-PSD-MS analysis. Neurobiol Dis. 2006; 23:61–76. 201. Levin Y, Schwarz E, Wang L, et al. Label-free LC-MS/ MS quantitative proteomics for large-scale biomarker discovery in complex samples. J Sep Sci. 2007; 30: 2198–2203. 202. Prabakaran S, Wengenroth M, Lockstone HE, et al. 2-D DIGE analysis of liver and red blood cells provides further evidence for oxidative stress in schizophrenia. J Proteome Res. 2007; 6:141–149. 203. Hirano M, Rakwal R, Shibato J, et al. Proteomics- and transcriptomics-based screening of differentially expressed proteins and genes in brain of Wig rat: a model for attention deficit hyperactivity disorder (ADHD) research. J Proteome Res. 2008; 7:2471–2489. 204. Kety SS. Biochemical theories of schizophrenia. II. Science. 1959; 129:1590–1596. 205. Saccuzzo DP, Braff DL. Information-processing abnormalities: trait- and state-dependent components. Schizophr Bull. 1986; 12:447–459. 206. Insel TR, Quirion R. Psychiatry as a clinical neuroscience discipline. JAMA. 2005; 294:2221–2224. 207. Colburn WA, Lee JW. Biomarkers, validation and pharmacokinetic-pharmacodynamic modelling. Clin Pharmacokinet. 2003; 42:997–1022. 208. Kuhlmann J, Wensing G. The applications of biomarkers in early clinical drug development to improve decision-making processes. Curr Clin Pharmacol. 2006; 1:185–191. 209. Flint J, Munafò MR. The endophenotype concept in psychiatric genetics. Psychol Med. 2007; 37:163–180.

Chapter 2

Methodological and Statistical Issues in the Use of Biomarkers in Clinical and Research Studies Ryan J. Van Lieshout and Peter Szatmari

Abstract The study of biomarkers has great promise for advancing knowledge in clinical practice and research in mental health. Biomarkers have the potential to improve our ability to diagnose individuals with a psychiatric disorder, to screen populations at risk of developing disorders, to provide prognostic information to those who manifest prodromal signs or to those who already have a disorder, and to help individuals make decisions about the most appropriate treatments (treatment response). However, there are many methodological and statistical issues that must be kept in mind as biomarkers are validated in research studies and before they can be considered useful for clinical practice. Keywords Methods • mental disorders

•

statistics

•

biological markers

Abbreviations DST Dexamethasone suppression test; SSRI Selective serotonin reuptake inhibitor; RNA Ribonucleic acid, APOE Apolipoprotein E; RCT Randomized controlled trial; COMT Catechol-Omethyltransferase; CI Confidence interval; SD Standard deviation; n Sample size; Zα/2 The standard normal deviate for α/2; α Probability of committing a type I error; level of statistical significance

R. J. Van Lieshout and P. Szatmari () The Offord Centre for Child Studies, McMaster Children’s Hospital and Department of Psychiatry and Behavioural Neurosciences, McMaster University, Hamilton, Ontario, Canada.

Background Knowledge of the biological underpinnings and correlates of psychiatric disorders is expanding rapidly and can be used to guide the diagnosis and treatment of individuals suffering from these illnesses. Biomarkers have great potential on their own or in combination with traditional risk factors to reduce our reliance on subjective assessment methods, improve the management of individuals with mental illness, and to inform research into the pathological mechanisms underlying these disorders. A biomarker is any biological measurement taken on an individual and can refer to a genetic or genomic variant or the resulting RNA or proteins associated with those variants. A biomarker can also be a step in a physiological pathway, a metric of neuroimaging or any other biological outcome of interest. Biomarkers are, in essence, measurement tools, and their potential lies in their ability to more closely reflect underlying causal mechanisms with less measurement error than do subjective reports of symptoms or observations of behavior.1 Biomarkers have the potential to improve our ability to diagnose individuals with a psychiatric disorder, to screen populations at risk of developing disorders, and to provide prognostic information to those who manifest prodromal signs or to those who already have a disorder. Biomarkers can also be used to help individuals make decisions about the most appropriate treatments, that is, to predict treatment response. Specific methodological issues within this area of research are too numerous to cover exhaustively, and so we will limit our discussion to more general, clinical uses of biomarkers, leaving the discussion of biomarkers as indicators of etiology to the remaining chapters of this volume. Given the enormous number of potential

M.S. Ritsner (ed.), The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes, © Springer Science + Business Media B.V. 2009

23

24

R.J. Van Lieshout and P. Szatmari

biomarkers and assays that are possible, it is also not feasible to discuss the specifics of each test or technology. Instead we discuss more general methodological and statistical issues likely to be relevant to a wide range of biomarker studies and argue for the need for close attention to these as a means of moving the field from promise to fulfilled potential When biomarkers are assessed for their usefulness in aiding diagnosis; they are validated by comparison with a ‘gold standard’ reference measure. Typically this takes the form of a DSM2 or ICD3 defined diagnostic category. The same reference is often used in studies examining the usefulness of a biomarker for prediction and prognosis; the only difference being the time of application of this reference standard. In the case of treatment response, the comparator could be a prespecified level of improvement on a symptom measure. When both the biomarker and the outcome are dichotomous (that is, they each have only two possible levels), their relationship can be represented by a 2 × 2 table (Table 2.1). In this case, the term ‘outcome’ can refer to any of a number of variables, among them diagnosis, prognosis or treatment response. The same principles of good research methodology apply to studies from the laboratory to the clinical arena. While research of lesser methodological quality is often completed more quickly than studies with superior designs, tests that have not been properly validated can actually slow the introduction of biomarkers into clinical use by providing incorrect information to physicians and patients, and undermining the public’s faith in their usage. Flaws in study design and analysis typically lead to biased results, often taking the form of overly optimistic estimates of the usefulness of biomarkers. These frequently take the form of insufficient power and multiple testing. There are several examples of biomarkers in psychiatry that initially showed great promise; the dexamethasone suppression test as a

Table 2.1 2 × 2 Table illustrating the comparison of a biomarker to an outcome when both are dichotomous Outcome

Positive Biomarker Negative

Positive

Negative

TP (True positive) FN (False negative)

FP (False positive) TN (True negative)

diagnostic tool for major depressive disorder, cytochrome p450 isoenzymes as predictors of response to selective serotonin reuptake inhibitors (SSRI) and APOE genotypes for prediction and the tropicamide eye drop test for diagnosis in Alzheimer’s disease. However, these have not been widely adopted, largely because later studies utilizing increased methodological rigor raised serious questions about their validity and usefulness.

The Challenge of Biomarker Discovery in Psychiatry The usefulness of a marker is limited by the reference standards it is validated against. The use of imperfect comparator measures imposes limits on the reliability and validity of putative markers and hinders their development in clinical practice and research. In the past, attempts at validating biomarkers in psychiatry have largely been limited to comparisons with various DSM or ICD defined diagnostic categories, though this has been expanded more recently to include more objective measures. The process of making a psychiatric diagnosis is inherently subjective, based on the assessors’ observations and interpretations of the reports of patients and others; along with their associated error. When these ‘gold standards’ are imperfect, they can over- or underestimate the performance and usefulness of the biomarker test being evaluated. When possible, clinical interviews should be supplemented with more objective measures, such as pathology findings (e.g. in Alzheimer’s disease). Unfortunately, such confirmatory tests do not exist for the majority of psychiatric disorders. In addition to problems with measurement, issues surrounding what currently defined mental disorders actually represent limit the discovery and use of biomarkers in psychiatry. Extant nosological systems are based mainly on phenomenological features and usually view disorder as dichotomous, ignoring the continuous nature and high rates of comorbidity seen in those with mental illness (see ref.4). Some disorders in psychiatry are progressive and so issues of disease stage may introduce further error. It is also not clear that each mental disorder represented in the DSM or ICD represents a single disease or pathologic process

2

Methodological and Statistical Issues in the Use of Biomarkers in Clinical and Research Studies

and there may be significant overlap in many of the disorders that these classification systems define as distinct. Moreover, certain syndromes may be comprised of a number of etiologically and pathogenetically heterogeneous conditions bound together by a final common pathway. Genes and other biomarkers may not correspond to these prescribed syndromes but instead to other, more proximal characteristics of the disorder such as individual traits or biochemical processes. If one is unable to accept that the psychiatric diagnosis or endophenotype used to define an outcome accurately represents the concept under study, then one will not be able to believe that the biomarkers used to detect these are valid. However, if one considers these standards acceptable, it is possible to approach and interpret such studies as meaningful. Psychiatric disorders are complex genetic syndromes with multifactorial etiologies, making the determination of which symptoms, features or processes to measure, complicated. The genetic complexity of many psychiatric disorders along with the role of epigenetic influences has already limited our ability to apply pharmacogenomics and pharmacogenetics to the treatment of psychiatric problems. Complexity of disorder and a lack of understanding of the mechanisms by which psychotropic medications exert their effects may help to explain the relative lack of usefulness of cytochrome p450 genotypes in predicting SSRI response in depressed patients. While variation in these genes is known to explain variability in SSRI plasma levels, it is not clear that ‘therapeutic’ serum drug levels are sufficient for treatment response. However, side effects, which are known to more closely correlate with serum drug levels can be predicted well by specific genetic variants both for antidepressants and antipsychotics.5 Prior to conducting studies of biomarkers in psychiatry, pragmatic issues such as acceptability, feasibility and cost should be considered. First and foremost, collection of the biomarker must be acceptable to the population from which it is taken. Reliance on biopsy samples may negate the use of biomarker studies for certain disorders. Costs must also be considered early on as these need to be acceptable to patients, researchers and funding agencies. The use of biomarkers in psychiatry is further complicated by the fact that the human brain comprises only 2% of our body mass and the blood brain barrier may impede the

25

release of many of these markers to the systemic circulation, leading to difficulties in collection and measurement.6

Planning Studies of Psychiatric Biomarkers One should have a clear idea of the stage of development and proposed use of biomarkers prior to their utilization in clinical research studies. A systematic approach to the generation of an appropriate research question, sampling procedure and measurement strategy will help to ensure that a study does not fail prior to its analysis phase. Appropriate, answerable questions with specific objectives and testable hypotheses should be laid out in advance. Study questions should consider the proposed relationship of the marker to the outcome as well as the confounders, mediating and moderating variables that might affect the association. The nature and effects of these types of variables need to be anticipated and taken into account in order for researchers to make valid conclusions. Next, study design, sampling, measurement and analytical issues and likely sources of error including chance and bias need to be considered. These will be discussed in greater depth in subsequent sections of this chapter. The process of biomarker discovery can be conceptualized as proceeding through several stages. In the developmental stage, little may be known about the marker and studies at this point are often designed to assess whether or not a biomarker has promise for future application. The process of biomarker discovery can be hypothesis driven or exploratory, as in the screening of markers using proteomic or metabolomic techniques. These approaches can be useful for paring down the number of candidate markers in a cost-effective way. If a measure appears to be linked to an outcome of interest, technical details and feasibility are commonly assessed at this stage. These studies are not usually population based and instead usually rely on small sample sizes.7 Research at this stage can be useful as long as it is understood that it does not accurately reflect the usefulness of the marker in clinical practice.8 If the putative marker demonstrates potential, the hypotheses generated in the development phase should be evaluated in a series of subsequent validation studies, each with increasing methodological rigor.

26

R.J. Van Lieshout and P. Szatmari

These studies generally involve comparing the test to a gold standard comparator and assessing its ability to discriminate in diagnostic studies or to predict in prognostic or treatment studies. Early in validation this phase, one attempts to determine the extent to which the biomarker test provides different results in patients with the disorder compared to those without it. Further studies will then assess if those with specific test results have the desired outcome in a more heterogeneous sample containing those more closely resembling individuals to whom the test will eventually be applied. A hierarchy of evidence for diagnostic tests that are applicable to the validation stage of biomarker evaluation has been proposed.9 It takes into account sampling considerations, as well as the importance of using masked comparisons of the biomarker test to the gold standard. This classification parallels in some ways (from Level 5 to Level 1 below) the phases through which biomarkers should pass prior to their clinical application.9 Level

Criteria

1 An independent, masked comparison with reference standard among an appropriate population of consecutive patients 2 An independent, masked comparison with reference standard among non-consecutive patients or confined to a narrow population of study patients 3 An independent, masked comparison with an appropriate population of patients, but reference standard not applied to all study patients 4 Reference standard not applied independently or masked 5 Expert opinion with no explicit critical appraisal, based on physiology, bench research of first principles

practice. The benefits of proper evaluation include improved diagnosis and care, reduced costs and the elimination of useless tests. Studies of lower methodological quality are known to overestimate the effect size and usefulness of diagnostic tests10,11 and treatment interventions.9 Factors known to lead to such overestimates include lack of randomization of subjects, the use of small sample sizes, inappropriate blinding of those administering, interpreting or receiving a test, and spectrum bias, the use of cases with more severe disease than is present in the population of patients one wishes to generalize results to.10,11 For example, while the use of a more homogeneous group of severely ill patients will help to detect an effect or relationship (by shifting the distribution of ‘scores’ on the biomarker test to the right; see Fig. 2.1), a diagnostic test validated in such a way will be useless when the time comes to apply it clinically because it will be unable to discriminate between less severely diseased and non-diseased patients, the population to which the test will most likely be applied to. Many tests can distinguish between very ill and very healthy patients, but only those that can discriminate between finer gradations of illness are likely to be of clinical value. Verification bias or the differential application of confirmatory or gold standard tests in those with positive or negative results on the biomarker test under study, also overestimate their usefulness. Studies that fail to provide a description of their population or the test examined are also known to lead to excessive estimates of accuracy.10

If a biomarker demonstrates adequate sensitivity, specificity and predictive value in the validation stage, its usefulness as an intervention can then be assessed. This would ideally take the form of a randomized controlled trial (RCT) where a group of patients are randomized to receive the biomarker test or not, and its effect on diagnostic or prognostic outcomes and patient and societal implications are assessed. To date few if any psychiatric biomarkers have provided evidence of usefulness at this stage.

The Need for Proper Methodological Evaluation Proper methodological evaluation of biomarkers is vital to their appropriate usage and uptake into clinical

Fig. 2.1 Biomarker test thresholds and their effects on diagnostic test accuracy (Adapted from ref.49)

2

Methodological and Statistical Issues in the Use of Biomarkers in Clinical and Research Studies

Study Design Selection of an appropriate study design for the assessment of biomarkers depends on its stage of development as well as feasibility and ethical considerations. If a putative marker has demonstrated promise in the development phase, its evaluation in the validation stage usually occurs in series of studies with increasing methodological rigor. The hierarchy of study design, in terms of increasing methodological rigor, progresses from cross-sectional and case-control studies to cohort to prospective randomized controlled trials.12 Prior to widespread clinical usage, the relevance and robustness of a biomarker would ideally be evaluated using large, prospective, randomized, controlled, double-blind clinical trials; or cohort studies with adjustments made for putative confounding factors and interactions between the biomarker and outcome. However, it has been suggested that in some circumstances, cross-sectional studies can also provide reasonable evidence for the accuracy of a test.13 Regardless of the stage of the biomarker of interest, vital elements of study design including whether the study was prospective or retrospective, whether samples were stratified or matched and how long they were followed should be reported. Of all study designs, a properly conducted randomized, double-blind controlled trial is the most methodologically rigorous, least susceptible to bias and not only demonstrates causality, but balances known and unknown confounders in all groups under study. Randomized controlled trials also minimize the likelihood that biased ascertainment or adjudication of outcomes will affect results because both groups are treated equally. In these types of studies, researchers randomize subjects to the biomarker (or biomarker plus the remainder of the diagnostic or prognostic package if applicable) or control group (in which the biomarker is not measured) and subsequently assess outcomes. Unfortunately, such studies are costly, may be difficult to justify ethically and are limited to addressing narrow issues over shorter periods of follow-up. Thus, they are often reserved for specific, mature research questions. Randomized controlled trials have not been widely used to date in the examination of biomarkers in psychiatry, which likely reflects the early stage of development of many psychiatric biomarkers. However, even biomarker tests that have

27

become commercially available, such as the Amplichip® p450, a test used to assess cytochrome p450 isoenzyme 2D6 genotypes, have not undergone prospective RCT evaluation.5 The remaining possible study designs are described below in order of descending methodological rigor. When RCTs are not feasible, observational study designs such as cohort or case-control studies are employed. These types of studies cannot establish causality and are susceptible to selection and information biases, as well as the influence of confounders. Selection bias exists when the groups studied differ in important respects other than the variables of interest. Information bias is present when data about the biomarker, outcome or covariates are collected in different ways between study groups (as occurs in studies that fail to employ blinding). Both can lead to incorrect estimates of the usefulness of a biomarker. Observational studies are also susceptible to confounding. Confounders are variables that are independently associated with the biomarker and with the disorder or process it marks, but are not directly involved in the pathway between the two. When present, confounders can produce unpredictable alterations in the estimate of the strength of the relationship between predictor and outcome. Failure to consider and assess relevant covariates may have contributed to the difficulties encountered in attempts to replicate cytochrome p450 isoenzyme findings in depression treatment response with SSRIs. For example, even if one were to assume that clinical response correlates linearly with increasing serum SSRI or serotonin levels, numerous other factors are known to contribute to this outcome including environmental, dietary, demographic and additional medical conditions. Failure to consider these factors may combine to overwhelm the variation in serum levels accounted for by p450 enzyme differences and may account for the lack of usefulness of this test in the clinical setting.14 Prospective cohort studies, where individuals begin free of the outcome at baseline, are the strongest observational study design as they reduce selection and information biases. They are useful in situations where exposures are rare but outcomes common and are of particular use in studies of prognostic markers. Barnett et al.15 utilized this methodology in their study of the effects of catechol-O-methyltransferase (COMT) polymorphisms on cognitive function in children. They assessed COMT polymorphisms at birth and then

28

cognitive outcomes later in life in the context of the Avon Longitudinal Study of Parents and Children. Prospective cohort studies also minimize recall bias and the systematic error associated with memory. However, this type of study can be prohibitively expensive, especially when the outcome is uncommon and when the time-lag elapsed between predictor and outcome is lengthy. Confounders cannot be protected against in cohort studies, but can be adjusted in the analysis phase if anticipated and measured. The analysis and conduct of cohort studies is particularly challenging when following individuals with disorders or symptoms that may change over time or in groups where attrition may be high. Loss to follow up is a concern in prospective cohort studies and increases with time to outcome. Its effects on study results are unpredictable, especially if loss is differential between exposed and non-exposed groups. These types of studies can also be limited by over-diagnosis or the identification of cases that may not be representative of those detected in normal clinical settings. In other words, cases in this context might not have come to clinical attention if they had not been enrolled in the study. Retrospective cohort studies select patients based on previously recorded exposures (or measurements) and assess outcomes in the present Such studies are cheaper and can be conducted more quickly than prospective studies and so have been used more widely, but have higher susceptibility to bias and overestimation of effect.11 The case-control study is a retrospective, observational study design where the recruitment starting point for cases is the current presence of a desired endpoint/outcome, not an earlier exposure, as it is in cohort studies. A control group is formed that is intended to be similar to the case group with the exception of the outcome. Ideally, controls are drawn from the same population base and at the same time as cases, and in validation studies, they should have symptoms similar to the disorder of interest but not have the disorder itself. These types of studies are frequently used to assess biomarkers to be used as diagnostic tests. In this design, the risk of finding false associations due to unknown factors in the population that vary according to case and control status, is always present. Case-control studies are also useful when outcomes are rare (e.g. schizophrenia) and there is a long latency between exposure and outcome. They can

R.J. Van Lieshout and P. Szatmari

also provide rapid answers to research questions. The validity of the conclusions drawn from these studies varies according to the sampling method employed however. If only severe cases are chosen, spectrum bias is present and estimates of effect may be inappropriately inflated. The limitations associated with case-control studies also do not permit them to estimate predictive values or likelihood ratios (because the researcher manipulates the prevalence of the disorder in the study) and should only be used when other methodologies are not feasible.16 Finally, many studies of biomarkers in psychiatry utilize cross-sectional study methodology because of their ease of conduct and relatively low cost. These types of studies provide results more quickly than the above study designs, but provide little useful information about causality and are subject to a larger number of validity limiting biases.

Sampling Sampling in the context of biomarker studies refers to the selection of individuals from a population of interest as well as to the acquisition of a biomarker from a study participant. For the sake of this section, it is taken to refer to the former, whereas the section on measurement will address the latter. Sampling provides the researcher with a group of subjects that are intended to be representative of the population from which the biomarker will be drawn, but provides a feasible number of individuals for study.17 Its conduct is often dictated by the stage of development of the biomarker and the study purpose. Development phase studies designed to assess if a biomarker is present in a disorder tend to utilize highly selected, homogeneous samples so as to increase the likelihood of finding a relationship between marker and outcome. However, later phase validation studies benefit from the use of more heterogeneous samples, especially if the eventual goal of the biomarker is the differentiation of one disorder from another. The constitution of a sample can greatly affect the results of a study. For example, estimates of the heritability of a biomarker or trait in major depressive disorder vary significantly depending on whether less severely affected community samples or more severe clinical samples are utilized.18 The characteristics of the sample,

2

Methodological and Statistical Issues in the Use of Biomarkers in Clinical and Research Studies

their source, as well as applied inclusion and exclusion criteria should always be reported in biomarker studies. Two types of sampling strategies can be employed. First, convenience or non-probability samples are those selected arbitrarily and are readily available to the researcher. They are simpler to collect but are prone to bias. This method has been employed in most studies of psychiatric biomarkers to date. Conversely, in probability sampling, every unit of the population of interest has the same chance of being selected. Most statistical tests are based on the assumption that a sample has been drawn randomly from the population of interest. This sampling strategy is less prone to error and is required for the generalizability of results from the study sample to the population at large. Researchers considering sampling strategies should consider issues of feasibility as well as bias, and so inclusion and exclusion criteria should permit as representative a sample as possible whilst facilitating study completion. As a result, a compromise is often reached such that convenience samples that attempt to utilize a consecutive sample of patients are used. The use of consecutive patients helps to minimize some selection biases associated with convenience samples.11,19 Prior to widespread use however, a biomarker should be tested in a population that is representative of the one to which the test will be applied in clinical or research studies. Biases in sampling are due to sampling error. Sampling error refers to the differences in estimates associated with the use of samples rather than entire populations. It can only be measured in the context of probability sampling and so this is the only setting in which this error can be adjusted in the analysis phase. Missing values or loss to follow-up can also introduce bias and threaten the validity of results though it does not definitively indicate that bias is present. In some cases, imputation, the process by which missing values are estimated and then included in the dataset to ensure ‘completeness’, is used to account for missing data. Studies of biomarkers should provide a description of missing values and how these were dealt with. An ideal sampling strategy attempts to minimize error associated with the sampling process. Proper sampling begins with the definition of the target population, the group to which the test will eventually be applied. In defining the target population, it is often useful to consider the patient’s diagnosis, stage of illness, age and gender among other characteristics

29

thought to be relevant to the research question. It is vital that reliable and valid measures are used to define the characteristics of one’s sample. Error in diagnosis or other means of classifying subjects is a form of measurement error and can invalidate or lead to inconsistent results across studies. After the target population is determined, the sampling frame, a list of sampling units (distinct, identifiable sampling elements; usually single patients), is specified. This permits access to the individuals who are eligible for the study. Sample selection methods are then applied using probability or non-probability sampling and the sample size determined.20 The hypothetical example of determining the usefulness of a serum marker in predicting antipsychotic response in first-episode schizophrenia (FES) patients can be used to illustrate the challenges associated with sampling in studies of psychiatric biomarkers. In this case, the target population would be all individuals experiencing their first episode of schizophrenia. However it is impossible to sample every case of new onset schizophrenia and so one must select a sample from the population of FES patients. One might choose to draw a sample from a clinic that specializes in providing care to individuals with first episode psychotic disorders. However, the eligibility criteria for enrolment in such a clinic may lead to the exclusion of cases above a specified age or those who present more covertly, such as individuals with prominent negative symptoms. A more representative or generalizable sample might be drawn from the practices of family physicians though this strategy may also miss cases. While reducing sample heterogeneity will increase the likelihood of finding an effect, it reduces generalizability and our ability to accurately determine if a marker can discriminate between groups that might be tested in a community sample of individuals with that psychiatric disorder. One must keep in mind however that there is no such thing as a perfect study and so attempts to minimize or control biases may be the best one can do. The properties of biomarker and other tests including their validity are not necessarily inherent features of the test itself, but are dependent on the situation and individuals in which the test is studied. Thus, it is vital that the study samples are described in a manner that permits study replication, as failure to do so is associated with an inappropriate elevation of effect sizes.10 Unfortunately this is not always done and has limited

30

our ability to compare studies in a number of areas including studies of biomarkers such as magnetic resonance spectroscopy in schizophrenia.21 Detailed reports of populations, methods and results also permit a determination of the generalizability of the study and permits meta-analytic synthesis of results across studies. An adequate sample size is vital to ensuring control of random error and providing the power required to detect differences between groups. Acquiring and maintaining adequate sample sizes is especially challenging in psychiatry because the processes required to diagnose patients is labor intensive and rates of dropout can be high. Before conducting a study, one must determine the size of the sample required to ensure that estimates drawn from it are precise enough to permit the detection of any differences of importance. One must ensure that attrition rates are factored into these calculations. Sample sizes can be increased in some cases by the use of endophenotypes (or more precisely, intermediate phenotypes22) as these may reduce the complexity of phenotypes and lie closer to the basic mechanisms underlying a disorder than the clinical diagnosis itself. Sample size calculations are based on parameters that are frequently estimated based on pilot data or hypotheses, as well as anticipated variation in both the population and biomarker assay, and putative effect sizes or differences. The probability of committing a type I error (alpha) or concluding that a difference exists between two groups when one does not (usually set at 0.05) must also be specified as well as an acceptable level of beta (usually 0.20, giving a power of 0.80), the probability of stating that there is no difference between two groups when there actually is, or type II error. While a thorough discussion of the technical details of sample size calculations is beyond the scope of the current chapter, the reader is directed to Kelsey et al.16 or Chow et al.23 for further details. Sample size calculators are available on the Internet and links to several can be found at www. statpages.org. Small sample sizes will severely limit the ability of the researcher to assess multivariate data and are a major limitation in studies of psychiatric biomarkers. Indeed, studies of the relationship of SSRI treatment response and cytochrome p450 enzymes have generally utilized small sample sizes and have not provided the opportunity to differentiate between different groups of individuals.14 Given the large sample sizes

R.J. Van Lieshout and P. Szatmari

needed to detect differences in many studies of biomarkers, multi-center collaborations and tissue banking may offer solutions. In studies designed to assess the accuracy of diagnostic biomarkers, a commonly encountered pitfall is that of workup or verification bias. This occurs when patients with a particular result on the biomarker test are preferentially referred for application of the gold standard assessment. Differential selection of those with positive results artificially inflate the true and false positive cell values in a 2 × 2 table (see Table 2.1). Such a sample will therefore not properly reflect those individuals the test will eventually be applied to. Unfortunately, unless a record of consecutive referrals from which cases are selected is available, the extent of this type of bias cannot be minimized.24 Nierenberg and Feinstein25 have proposed a set of study phases for evaluating diagnostic biomarkers that illustrates the need for increasingly hetero geneous samples as one progresses from developmental to validation stages of test evaluation. Studies these authors refer to as phase I are designed to assess the strength of the relationship between marker and outcome, and utilize only those with obvious or severe disease. If the marker demonstrates some utility, they suggest that these patients should then be compared to healthy controls to assess if the test is able to coarsely discriminate between disease and non-disease. Should such studies fail, the marker is unlikely to be of use and investigation should be abandoned. According to these authors, subsequent stages should use more heterogeneous samples of cases and controls; not just those with severe disease or those without comorbidity. They suggest that the usefulness of a test should be examined in a population of patients that will resemble those to whom the biomarker is intended to be used with. This has proven especially important in the search for diagnostic biomarkers for Alzheimer’s disease, which shares not only clinical but pathological features with a number of other dementias.26 In order to be useful diagnostically, a diagnostic test for Alzheimer’s should differentiate it from other dementias that are similar symptomatically as well as pathologically. Only at these later stages of evaluation can the diagnostic accuracy and clinical utility of the test be fully determined. Failure to follow this approach to evaluating a biomarker can lead to premature uptake of a test and

2

Methodological and Statistical Issues in the Use of Biomarkers in Clinical and Research Studies

its subsequent failure. One notable example of this is the use of the dexamethasone suppression test (DST) for diagnosing major depressive disorder. Initially, claims of moderate sensitivity and high specificity were made for the test, however these were not supported when proper sampling was conducted.25 Studies carried out prior to its widespread use are best described as developmental and few met criteria for biomarker validation. Studies used to support the clinical use of the DST included neither a wide spectrum of patients with the disorder nor those with comorbid conditions or symptoms that might produce false positive results (e.g. Alzheimer’s disease, obsessive compulsive disorder and schizophrenia). Had such individuals been included in studies prior to its clinical use, the high false positive rate of the test would have been detected and it would not have been widely used.

Measurement Useful biomarkers should be both reliable and valid. Reliability or precision is the freedom of measurement from random error. Internal validity or accuracy is the extent to which a measurement represents the true value of interest and is free of systematic error. The type of validity assessed for biomarkers in psychiatry is most often ‘criterion’ validity, an assessment of how well a biomarker value correlates with the ‘gold standard’ measure. Reliability and validity are attributes of all tests and are especially important for gold standard tests used in validation studies. Failure to use reliable and valid reference measures can invalidate a study and so attention to detail at this stage is vital. Measurement error (or the departure of values generated by the tool or scale from the true value) can be comprised of both random and systematic error. When the measurement of biomarkers is unreliable, extreme results gleaned from them on one occasion tend to get closer to the mean on subsequent testing, a concept known as regression to the mean,1 and attenuate effect sizes. If the degree of measurement error of an outcome differs between those with and without the biomarker, its effects can be even less predictable and the observed effect size can either increase or decrease relative to the true effect size. Inconsistency in measurement,27 endpoints28 and inadequate symptom assessment29 have all been proposed to account for the failure

31

to replicate studies relating cytochrome p450 isoenzyme mutations and SSRI response in depression. Standardization of methods not only reduces measurement error, but allows results across studies to be integrated and combined statistically, increasing the power to detect true effects.30 A certain amount of error is inherent in every measurement and can rarely be eliminated, but understanding its sources, and how to minimize these can lead to increased confidence in the conclusions drawn from biomarker studies. The major argument for the use of biomarkers is to a large extent the hope that these will have less measurement error than the assessment of subjective signs and symptoms. However, biomarkers are not entirely immune to this type of error. The best defense against measurement error is careful study design and an understanding of the potential sources of random and systematic error. Measurement error can be reduced by standardizing measurements, establishing a study protocol or operations manual, training and assessing observers, automating and refining the instruments used, taking repeated measurements and by blinding observers, subjects and those interpreting the tests.17 The type of biological material acquired, how it was stored and measured, as well as the remainder of the assay method protocol should be reported in detail. If error is anticipated, measured and quantified, it can be adjusted in the analysis phase of the study (though unmeasured error may still be present).31 Prospective data collection allows better data control and permits assessment of data integrity and consistency and reduces bias relative to retrospective collection. Therefore, it should be conducted when possible.32 The assessment of both the biomarker and its gold standard comparator should also be sufficiently described in a study so that the impact of the potential biases associated with them can be assessed by readers.33 Sources of measurement error include the observer, instrument and subject in the assessment of all study variables. Observer error can occur in the interpretation of, for example, MRI scans or diagnostic interview results and subject error can be present in the reporting of symptom experience or cortisol level given the time of day. Error can be minimized by blinding these individuals to the study purpose and standardizing sampling, measurement, training and interpretation processes. Blinding those interpreting tests reduces review bias, as knowledge of the result

32

of the gold standard or biomarker test can lead to incorrectly elevated indices of accuracy. The use of component phenotypes, a component of a mental disorder rather than the whole disorder itself, may decrease measurement error and increase statistical power by reducing the phenotypic complexity of the concept under investigation.34 This has been used in the study of autism where use of age of phrase speech onset among sibling pairs with autism increased the linkage signal on chromosome 7q relative to the diagnosis of autism itself.35 Participant factors such as one’s genes, sex, ethnicity, adiposity, diet, medication use, smoking, alcohol intake, exercise, comorbid disorders, and timing or the physiological or psychological state at the time of testing can also introduce error into measurements of biomarkers. Failure to measure and account for these sources of variability has likely contributed to the failure of a number of tests, among them the initially promising tropicamide eye drop test for Alzheimer’s disease diagnosis.36 If it is not possible to conduct a randomized trial where inter-individual differences would be minimized, careful measurement of these potential confounding factors and adjustment using multivariate regression techniques should be conducted if the sample size permits and if these are true confounding variables. The episodic nature of many mental disorders introduces another element of complexity to psychiatric biomarkers in that they may be state-dependent or trait-dependent, and vary according to the stage of illness. This variation may have complicated the interpretation of certain magnetic resonance spectroscopy findings in schizophrenia.21 Care should be taken to assess and consider whether such differences are likely to affect biomarkers or their assays (e.g. using longitudinal studies). The assays used to measure the biomarkers themselves sometimes lack precision. This was the case in early studies of the use of Western blots to assess amyloid protein precursor levels in CSF for their association with Alzheimer’s disease.37 Means of sample transfer, handling and storage can also induce unwanted analytical variability in samples.38 Assays selected should be able to analyze large numbers of samples and provide consistent results across laboratories. Those doing the collection and/or the testing should be blind to the group membership. Assays from all study groups should be represented in each run of samples so as to

R.J. Van Lieshout and P. Szatmari

control for inter-assay variability. Reliability of comparators requires uniform staff training procedures and the assessment of the inter-rater reliability of assessors to ensure that consistency is optimized.

Validation and Analysis The statistical analysis of studies of biomarkers should focus not only on statistical significance but also the strength of the relationship between the marker and outcome, the validity of the test, and its effects on clinical decision making and therapeutic outcomes. All statistical methods including means of variable selection, model building and how model assumptions were verified should be described in all studies. Before a biomarker is considered to be valid; chance, bias and confounding factors must be taken into account. The likelihood of a result being due to chance (as assessed by tests of statistical significance) can be decreased by increasing sample size or by using methods to improve reliability. Tests of statistical significance are necessary but not sufficient features of a valid biomarker test because increasing sample size can reduce p-values to significant levels, regardless of a test’s usefulness. Bias or systematic error can be detected by examining the sampling and measurement phases of the study and determining if and by how much they depart from the ideal study plan. If a relationship is found between a biomarker and an outcome, and it is deemed not to be spurious (due to chance or bias), the relationship can also be due to confounders. Confounding factors are frequently present and may account for the relationship noted between antioxidant enzyme abnormalities and schizophrenia. It is well-known that cigarette smoking can lead to alterations in antioxidant compounds independent of schizophrenia. Given that rates of smoking are increased in those with schizophrenia, smoking (and not schizophrenia itself) may account for the observed abnormalities.39 Smoking, gender and one’s ability to cope with stress might also confound the relationship noted between increased levels of the pro-inflammatory cytokine interleukin-6 and major depressive disorder.40 Thus, independent studies indicating that changes in biomarkers are not due to confounders should be carried out or, in the least, these variables need to be mea-

2

Methodological and Statistical Issues in the Use of Biomarkers in Clinical and Research Studies

sured and adjusted for statistically in the analysis phase of studies. In cases where one is attempting to develop a diagnostic test or predict a result from a combination of tests or features, concerns about the relationship between biomarker and outcome being present secondary to a confounder are not a problem. However, if one is attempting to assess the relevance of the marker to outcome or treatment response, confounding should be taken into account. Confounders are ideally controlled for in the design phase of observational studies though they can be adjusted for in its analysis as well. Excluding subjects with a particular value of a confounder, a process known as specification is one way of controlling for confounders in the design phase of observational studies. Matching on these types of variables is another possibility and doing so with respect to demographic characteristics is recommended by some.7 However, this will require knowledge of the confounder in advance and can limit sample size and generalizability. Matching also has a number of pitfalls as doing so based on demographic factors can introduce additional error if any of these factors are related to diagnosis. In some situations, as in the use of psychotropic medications, it may not be possible to match on relevant factors. Here, confounders can be controlled for by stratification, grouping according to the levels of the confounder or statistical adjustment using regression methods, though these too have limitations. The appropriate statistical tests for the analysis of biomarkers depends on the purpose of the study, as well as the type of variable examined. When continuous variables, those with a numerically continuous range of values, are categorized into dichotomous or categorical ones, statistical power is reduced. This loss of power may be as great as the amount lost when one

33

third of the data is simply discarded!41 Such categorization also increases the risk of false positive results and conceals any non-linearity in the relationship between dependent and independent variables.41 Thus, this practice should be avoided if possible. This is one situation where quantitative, genetically informative phenotypes may offer advantages over DSM or ICD diagnoses. The use of continuous data may require that special techniques be used to ensure that they are amenable to parametric statistical analysis. This may include the need for mathematical transformations of variables so that their distributions more closely resemble a normal distribution or so that more constant variance is present. A number of measures of agreement between the results of a biomarker in comparison to a gold standard reflect the ability of a test to discriminate between health and disease. When both biomarker and outcome are dichotomous, as in the assessment of risk for Alzheimer’s disease risk by homozygosity (vs. heterozygosity) of the APOE ε4 gene, study results can be expressed in a 2 × 2 table (Table 2.2). In this situation, these measures include sensitivity, specificity and positive and negative predictive values. The meaning of the terms sensitivity and specificity varies across disciplines however. To most laboratory scientists, sensitivity is the minimal level of a substance an assay can detect and specificity refers to the ability of a test to detect a particular analyte among a group of substances that are structurally related. To most clinicians and epidemiologists however, sensitivity reflects the ability of a test to detect a condition of interest when the disease is actually present. In this context, it is also defined as the proportion of subjects with the disease in whom the test gives the right answer. Specificity reflects the recognition of absence of a disorder by a test when the disorder is truly absent.

Table 2.2 2 × 2 Table for comparison of biomarker test to gold standard comparator with dichotomous outcomes Gold standard (outcome)

Positive Biomarker Test Negative

Positive

Negative

TP (True positive)

FP (False positive)

FN (False negative) TP TP + FN Sensitivity =

TN (True negative) TN Specificity = TN + FP

Positive predictive Value =

Negative predictive Value =

TP

TP + FP

TN

TN + FN

34

It has also been referred to as the proportion of subjects without the disease in whom the test gives the correct answer (in this case, a negative result). The remainder of our discussion refers to these terms in their clinical or epidemiological sense. Sensitivity and specificity do not vary with disease prevalence, though they can change as the spectrum of patients enrolled does.42 For example, if more severely disordered patients are included in a study, they are more likely to have a distribution of values further away from controls (see Fig. 2.1), and, if the diagnostic threshold remains the same, the test would appear to have higher sensitivity than it would in a sample where less ill cases were included. The use of differing control groups can lead to variation in estimates of sensitivity and specificity as has been the case in studies of CSF total tau protein in those with Alzheimer’s disease.37 This again highlights the importance of validating tests on the populations in which they are intended to be used on. As we discover more and more promising genetic loci in psychiatric disorders, an increasing body of evidence suggests that many control particular neural processes or systems rather than a particular disorder. For example, in schizophrenia, while smooth pursuit eye movement abnormalities, neurological soft signs, and prepulse inhibition abnormalities were once thought to be specific to the disorder, subsequent studies have shown this not to be true.43–45 Brain derived neurotrophic factor levels, which are known to be low in those with depressed mood, do not appear to be specific to major depressive disorder, but are also shared by those with depressive personality traits.46 APOE genotyping also has little utility as a diagnostic test in Alzheimer’s disease, as it is neither sufficiently sensitive nor specific to differentiate Alzheimer’s from other dementias or normal controls.47 Serum amyloid β1–42 is another biomarker originally thought to have potential for Alzheimer’s disease diagnosis but it also lacks the requisite sensitivity and specificity though it may help to predict risk in combination with other factors.48 The limitations of our current syndromic classification of mental disorders suggest that further dissection of disorders into informative sub-phenotypes is required. These findings also highlight the importance of assessing the specificity of results in studies of biomarkers with heterogeneous samples, especially if one wishes to use it diagnostically or as a marker of disease specific processes.

R.J. Van Lieshout and P. Szatmari

A test’s positive predictive value reflects the proportion of individuals with positive tests that give the right answer (have the disease) whereas negative predictive values represent the fraction of subjects with negative tests who actually do not have the disorder. A high positive predictive value is of particular use in biomarkers designed to aid in the process of diagnosis. A positive test result in this context suggests that the probability of having a disorder is high. However, predictive values change depending on the prevalence of the disorder in the sampled population.49 A graphical representation of these concepts can be used to further aid in their illustration. Figure 2.1 is a plot of biomarker measures or test scores against the number of individuals or units with each score. Nondiseased individuals are represented by the distribution of scores on the left and diseased on the right. The central limit theorem dictates that these will approximate normal distributions as long as sample sizes are above 30 (below these, binomial techniques should be used). The cutoff point or diagnostic or decision threshold is represented by its labeled hatched vertical line. As with most tests in medicine, the distribution of scores on biomarker tests in both groups overlaps. The various rates, true positive, false positive and the concepts based on these, sensitivity, specificity and so forth, will vary as the characteristics of the population vary. For example, a more heterogeneous group of cases (diseased persons) will have a wider distribution of scores and for the same cut-point, a higher false negative rate (and lower sensitivity and power). For variables or tests with more than two levels or that are ordinal or continuous in nature, likelihood ratios and ‘area under the receiver operating characteristic curve’ (AUC) can be used to assess the validity of biomarkers. In these cases, many values of sensitivity and specificity are possible depending on the cutoff point or threshold chosen to define positive tests. The receiver operating curve (ROC) plot is a graphical representation of the sensitivities and specificities of a given test at various levels or scores on the test of interest. It plots the true positive rate (sensitivity) against the false positive rate (1-specificity) for a number of cutoff points selected by the researcher (see Fig. 2.2). The ROC curve can also be used to help investigators select cutoffs based on the characteristics of the disease and the test that they desire. The desired properties of a test will depend on its purpose. For example, tests that are deemed useful and accurate for diagnosis

2

Methodological and Statistical Issues in the Use of Biomarkers in Clinical and Research Studies

Fig. 2.2 Three hypothetical ROC curves. Curve A represents the perfect gold standard (AUC = 1), Curve B a more typical result (AUC = 0.85) and Curve C represents random chance (AUC = 0.5). The diagnostic accuracy of a test improves as one moves from Curve C to Curve A

may produce unacceptably high rates of false negatives when the prevalence of a disorder is high, thus making them potentially harmful for population screening purposes. Markers designed to assess disease progression do not necessarily have to possess high specificity but should be sensitive to change. Shifting diagnostic thresholds, either to increase sensitivity or specificity, always comes at a cost. As can be seen in Fig. 2.1, shifting the diagnostic threshold or cutoff to the right increases specificity, but at the price of reducing sensitivity. If the purpose of a test is to confirm the presence of a disorder, specificity should be high, whereas if one wishes to rule out a condition, high sensitivity would be desired. It has been suggested that when screening for Alzheimer’s disease, a highly sensitive test would be desired (at the cost of specificity), whereas a test used in clinical settings, high specificity may be more desirable.26 When deciding on a cutoff based on ROC curves, one must also balance the importance of detection with the impact on the patient of receiving a false positive diagnosis of a life-threatening condition. Thus, when applying a test to an individual, all four outcomes, true and false positives and negatives need to be considered because each type of results has different implications for individuals.

35

The area under the ROC curve incorporates both sensitivity and specificity as well as the degree of overlap between healthy and disordered populations. Values for different tests can be compared using its values. A perfect test reaches the upper left corner of the plot and earns an AUC value of one (perfect accuracy and no population overlap; see curve A in Fig. 2.2), whereas an AUC of 0.5 suggests that there are no differences in the groups based on the test and that it is of no diagnostic value. Rarely can a test completely distinguish between groups, particularly among complex disease with multiple comorbidities, such is the case in psychiatric disorders, and so careful selection of cutoff points is vital.49 Likelihood ratios also provide useful clinical information. They allow the researcher to use all of the information in a test as the results can be applied to all possible test values. For each result, the LR is the likelihood of that result in someone with the disease compared to the likelihood of the same result in a person without disease. Likelihood ratios do not make a diagnosis but are designed to modify one’s pretest probability of disorder (the clinical suspicion one has based on the available clinical evidence prior to the administration of the test). To use a LR, one must convert the pretest probability to a pretest odds using the formula: Odds = Probability/1 − Probability, apply the LR to get a posttest odds and then convert back to probabilities: Probability = Odds/1 + Odds. Likelihood ratios > 10 or < 0.1 provide meaningful changes and are likely to usefully inform clinical decisions; whereas those 0.5 are not.50 For dichotomous tests, likelihood ratios for positive and negative test results can be calculated using sensitivities and specificities. LR for a Positive Test =

Sensitivity 1 − Specificity

LR for a Negative Test =

1 − Sensitivity Specificity

All results, regardless of whether they are measures of diagnostic accuracy or the strength of a relationship between a biomarker and an outcome should be reported with accompanying confidence intervals. Most frequently the 95% confidence interval is given. It represents the range of values within which the true value for the whole population from which the study sample was drawn lies 95% of the time. The calculation

36

of confidence intervals is appropriately conducted using the binomial distribution for sample sizes less than 30 and using the normal distribution for groups that exceed this size.51 Confidence intervals help to inform the reader as to the numerical stability or reliability of estimates of calculated population parameters. Because confidence intervals vary according to the standard error of the value (CI = Mean ± Zα/2(SD/n) ), they become narrower as the sample size increases.51 Developmental phase studies have some unique analytical considerations. In this type of research, the number of markers assessed often exceeds the number of samples. As a result, the risk of overfitting of the data (the finding of a chance pattern that increases with an increased variable to sample size ratio), is high. The chance of reporting false positive results or making a type I error (concluding that a finding is significant when it is actually due to chance) increases as the ratio of variables to sample size increases. This can be addressed by a variety of methods including using split samples, where the data are randomly divided into two samples; one for fitting the mathematical model used to determine the strength of relationship for the biomarker and another for assessing its validity or replicability. The explosion of technology related to the study of genes and their variants has enabled researchers to conduct thousands of comparisons per study. While powerful, analyses involving multiple comparisons of genes and gene-environment interactions also increase the likelihood of type I error. The optimal method of dealing with multiple comparisons is still actively debated though many statisticians state that the Bonferonni correction is most suitable. In fact, type I error due secondary to multiple comparisons has been postulated to be at least partly responsible for the failures to replicate the initial promising findings of cytochrome p450 gene variants to predict psychotropic drug response.29

Combining Biomarkers When deciding on whether or not to use a biomarker test, one must consider whether the patient will be better off having had the test than they would be if they had not undergone it. This requires consideration of whether the sample and setting in which the biomarker has been tested is similar enough to the current situation of interest in the clinic to justify applying it. If the

R.J. Van Lieshout and P. Szatmari

test is unlikely to apply to your patient, to give a valid result, or alter outcomes, it is not worth performing. Outcomes such as quality of life and peace of mind of the patient should also be considered. Unfortunately, the quality and type of data required to make decisions based on biomarker tests is often not available for most, if not all, psychiatric biomarkers. The Standards for Reporting of Diagnostic Accuracy (STARD) document provides a series of evidencebased recommendations applicable to the planning, conduct, analysis and reporting of studies of diagnostic tests that can also be applied to the assessment of the validity of biomarker based assays in psychiatry. They are listed below (Table 2.3).32 Given that psychiatric disorders are complex biopsychosocial illnesses, it is unlikely that any single biomarker will accurately predict outcomes for individual patients. Moreover, simple measurement of a marker and acquisition of a result does not make it a clinically useful or successful test as there are a number of steps that must be undertaken prior to it proving its clinical mettle. If a marker is used as part of a package of clinical factors or characteristics, or its use requires adherence to protocols regarding its conduct and interpretation, it may be wise to evaluate the package rather than the individual marker itself.13 Thus, when studying such combinations of biomarkers, it is vital that the entire protocol be standardized, applied and studied. This is highlighted by the lack of usefulness of validated biomarkers when, if not collected interpreted and the results acted on properly, are of little use. Therapeutic monitoring of tricyclic antidepressant drug levels can provide benefits when the information is used properly52 but are not useful if is not.53 This has also been demonstrated in studies of mood stabilizers.54 The error associated with each individual biomarker is compounded when it is combined with others. For example, diagnostic prediction of schizophrenia using four relatively well-established endophenotypes for schizophrenia, while superior to any single measure, is still imperfect and may reflect error in the measurement of each component.55 The most useful techniques that have been used to evaluate multiple diagnostic markers are regression models. These techniques can also be used to identify mediator variables; those present in the causal pathway from predictor to outcome and through which the independent variable acts to produce changes in the dependent variable. They are also used to assess for the presence of effect moderation (statistical interactions),

2

Methodological and Statistical Issues in the Use of Biomarkers in Clinical and Research Studies

37

Table 2.3 Standards for reporting of diagnostic accuracy Section of paper Title/Abstract/ Keywords Introduction Methods Participants

Test methods

Results Participants

Test results

Estimates

Discussion

Identify the article as a study of diagnostic accuracy (recommend MeSH heading ‘sensitivity and specificity’) State the research questions or study aims, such as estimating diagnostic accuracy or comparing accuracy between tests or across participant groups The study population: The inclusion and exclusion criteria, setting and locations where data were collected Participant recruitment: Was recruitment based on presenting symptoms, results from previous tests, or the fact that the participants had received the index tests or the reference standard? Participant sampling: Was the study population a consecutive series of participants defined by the selection criteria in item 3 and 4? If not, specify how participants were further selected Data collection: Was data collection planned before the index test and reference standard were performed (prospective study) or after (retrospective study)? The reference standard and its rationale Technical specifications of material and methods involved including how and when measurements were taken, and/or cite references for index tests and reference standard Definition of and rationale for the units, cut-offs and/or categories of the results of the index tests and the reference standard The number, training and expertise of the persons executing and reading the index tests and the reference standard Whether or not the readers of the index tests and reference standard were blind (masked) to the results of the other test and describe any other clinical information available to the readers Methods for calculating or comparing measures of diagnostic accuracy, and the statistical methods used to quantify uncertainty (e.g. 95% confidence intervals) Methods for calculating test reproducibility, if done When study was performed, including beginning and end dates of recruitment Clinical and demographic characteristics of the study population (at least information on age, gender, spectrum of presenting symptoms) The number of participants satisfying the criteria for inclusion who did or did not undergo the index tests and/or the reference standard; describe why participants failed to undergo either test (a flow diagram is strongly recommended) Time-interval between the index tests and the reference standard, and any treatment administered in between Distribution of severity of disease (define criteria) in those with the target condition; other diagnoses in participants without the target condition A cross tabulation of the results of the index tests (including indeterminate and missing results) by the results of the reference standard; for continuous results, the distribution of the test results by the results of the reference standard Any adverse events from performing the index tests or the reference standard Estimates of diagnostic accuracy and measures of statistical uncertainty (e.g. 95% confidence intervals) How indeterminate results, missing data and outliers of the index tests were handled Estimates of variability of diagnostic accuracy between subgroups of participants, readers or centers, if done Estimates of test reproducibility, if done Discuss the clinical applicability of the study findings

the situation where the association between a predictor variable and an outcome depends on the levels of a third distinct variable.56 Regression analysis was used for this purpose by Caspi et al.57 to provide support for the theory that a polymorphism in the COMT gene moderates the association between cannabis use in adolescence and the risk of developing psychosis in adult-

hood. Thus, variants in the COMT gene may be of value in identifying populations at risk of developing psychosis. While chi square tests can be used to assess for such statistical interactions, they do not permit adjustment for other variables in this context. Linear and logistic regression models attempt to predict the cumulative probability of either a continuous or binary

38

event over a particular period. Proportional hazards regression can also be used and can deal with patients lost to follow-up. More recently, artificial neural networks have also been proposed as useful means of predicting outcomes and seem to overcome the limitations of regression analyses, namely dealing with increasing numbers of non-linear and interacting factors, the type that are often seen in proteomic and genetic studies.58

Conclusions and Future Directions The use of biomarkers offers great promise for the development of more effective means of diagnosing and treating patients with mental disorders. However, studies involving these technologies have the potential to mislead or even harm if they fail to employ appropriate methodological and statistical techniques in their planning and conduct. Exciting discoveries in the use of biomarkers in psychiatry will be facilitated by further refinements in the description and classification of mental disorders, cross-disciplinary and multi-site collaboration, and the use of multivariate statistical techniques in both the analysis and combination of putative biomarkers. Researchers wishing to undertake or critically appraise studies of biomarkers utilizing a variety of methodologies may wish to refer to the STARD,33 REMARK,59 CONSORT60 or STROBE61 statements for further guidance regarding methodological and reporting issues. Adherence to the principles discussed in this chapter will not only provide the means to properly evaluate the utility of biomarkers but will help to catalyze their regulatory approval and use.

References 1. Szatmari P, Maziade M, Zwaigenbaum L, et al. Informative phenotypes for genetic studies of psychiatric disorders. Am J Med Genet B Neuropsychiatr Genet 2007;144:581–588 2. American Psychiatric Association. Diagnostic and statistical manual of mental disorders. 4th Ed. Text Revision. Arlington, VA: American Psychiatric Press; 2000 3. World Health Organization. International statistical classification of diseases and related health problems: tenth revision. 2nd Ed. Geneva, Switzerland: World Health Organization; 2004 4. Helzer JE, Kraemer HC, Krueger RF. The feasibility and need for dimensional psychiatric diagnosis. Psychol Med 2006;36:1671–1680

R.J. Van Lieshout and P. Szatmari 5. Maier W, Zobel A. Contribution of allelic variations to the phenotype of response to antidepressants and antipsychotics. Eur Arch Psychiatry Clin Neurosci 2008;258(1, Suppl): 12–20 6. Land JM. Evidence-based medicine: evaluation of biomarkers. In: Trull AK, Demers LM, Holt DW, Johnston A, Price CP, Tredger MJ (eds.) Biomarkers of disease: an evidence-based approach. Cambridge: Cambridge University Press; 2002, pp. 390–397 7. Hulka BS, Margolin BH. Methodological issues in epidemiologic studies using biologic markers. Am J Epidemiol 1992;135:200–209 8. Simon R. Validation of pharmacogenomic biomarker classifiers for treatment selection. Cancer Biomark 2006;2:89–96 9. Moore RA. Evidence-based medicine: evaluation of biomarkers. In: Trull AK, Demers LM, Holt DW, Johnston A, Price CP, Tredger MJ (eds.) Biomarkers of disease: an evidence-based approach. Cambridge: Cambridge University Press; 2002, pp. 3–15 10. Lijmer JG, Willem Mol B, Heisterkamp S, et al. Empirical evidence of design-related bias in studies of diagnostic tests. JAMA 1999;282:1061–1066 11. Rutjes AWS, Reitsma JB, DiNisio M, et al. Evidence of bias and variation in diagnostic accuracy studies. CMAJ [serial online] 2006;174:DOI:10.1503/cmaj.050090 12. Baker SG, Kramer BS, McIntosh M, et al. Evaluating markers for the early detection of cancer: overview of study designs and methods. Clin Trials 2006;3:43–56 13. Schunemann HJ, Oxman AD, Brozek J, et al. Grading quality of evidence and strength of recommendations for diagnostic tests and strategies. BMJ 2008;336:1106–1110 14. Rasumussen-Torvik LJ, McAlpine DD. Genetic screening for SSRI drug response among those with major depression: great promise and unseen perils. Depress Anxiety 2007;24:350–357 15. Barnett JH, Heron J, Ring SM, et al. Gender-specific effects of the catechol-O-methyltransferase Val108/158Met polymorphism on cognitive function in children. Am J Psychiatry 2007;164:142–149 16. Hulley SB, Cummings SR, Browner WS, Grady DG, Newman TB. Designing Clinical Research. 3rd Ed. Philadelphia, PA: Lippincott, Williams & Wilkins; 2007 17. Kelsey JL, Whittemore AS, Evans AS, et al. Methods in observational epidemiology. 2nd Ed. New York: Oxford University Press; 1996, pp. 311–340 18. McGuffin P, Cohen S, Knight J. Homing in on depression genes. Am J Psychiatry 2007;164:195–197 19. Pai M, Flores LL, Pai N, et al. Diagnostic accuracy of nucleic acid amplification tests for tuberculous meningitis: a systematic review and meta-analysis. Lancet Infect Dis 2003;3:633–643 20. Boyle MH. Sampling in epidemiologic studies. In: Verhulst FC, Koot HM (eds.) The epidemiology of child and adolescent psychopathology. Oxford: Oxford University Press; 1995, pp. 66–84 21. Keshavan MS, Stanley JA, Pettegrew JW. Magnetic resonance spectroscopy in schizophrenia: methodological issues and findings – part II. Biol Psychiatry 2000;48:369–380 22. Carlson CS, Eberle MA, Kruglyak L, Nickerson DA. Mapping complex disease loci in whole-genome association studies. Nature 2004;429:446–452

2

Methodological and Statistical Issues in the Use of Biomarkers in Clinical and Research Studies

23. Chow S-C, Shao J, Wang H. Sample size calculations in clinical research. Basel, Switzerland; Marcel Dekker Inc; 2003 24. Reid MC, Lachs MS, Feinstein AR. Use of methodological standards in diagnostic test research. JAMA 1995;274: 645–651 25. Nierenberg AA, Feinstein AR. How to evaluate a diagnostic marker test. JAMA 1988;259:1699–1702 26. Litvan I. Methodological and research issues in the evaluation of biological diagnostic markers for Alzheimer’s disease. Neurobiol Aging 1998;19:121–123 27. Bolanna AA, Arranz MJ, Mancama D, et al. Pharmacogenomics – can genetics help in the care of psychiatric patients? Int Rev Psychiatry 2004;16:311–319 28. Jones DS, Perlis RH. Pharmacogenetics, race, and psychiatry: prospects and challenges. Harvard Rev Psychiatry 2006;14:92–106 29. Bondy S, Spellmann I. Pharmacogenetics of antipsychotics: useful for the clinician? Curr Opin Psychiatry 2007;20: 126–130 30. Gordon E, Liddell BJ, Brown KJ, et al. Integrating objective gene-brain-behavior markers of psychiatric disorders. J Integr Neurosci 2007 Mar;6(1):1–34. 31. Kelsey JL, Whittemore AS, Evans AS, et al. Methods in observational epidemiology. 2nd Ed. New York: Oxford University Press; 1996:355 32. Bossuyt PM, Reitsma JB, Bruns DE, et al. The STARD statement for reporting studies of diagnostic accuracy: explanation and elaboration. Clin Chem 2003;49:7–18 33. Streiner DL. Diagnosing tests: using and misusing diagnostic and screening tests. J Pers Assess 2003;81:209–219 34. Almasy L. Quantitative risk factors as indices of alcoholism susceptibility. Ann Med 2003;35:337–343 35. Alarcon M, Cantor R, Lieu J, et al. Evidence for a language quantitative trait locus on chromosome 7q in multiple autism families. Am J Hum Genet 2002;70:60–71 36. Kardon RH. Drop the Alzheimer’s drop test. Neurology 1998;50:588–591 37. Frank RA, Galasko D, Hampel H, et al. Biological markers for therapeutic trials in Alzheimer’s disease. Proceedings of the biological markers working group; NIA initiative on neuroimaging in Alzheimer’s disease. Neurobiol Aging 2003;24:521–536 38. Michels Blanck H, Bowman BA, Cooper GR, et al. Laboratory issues: use of nutritional biomarkers. J Nutr 2003;133:888S–894S 39. Yao JK, Reddy RD, van Kammen DP. Oxidative damage and schizophrenia. An overview of the evidence and its therapeutic implications. CNS Drugs 2001;15:287–310 40. Haack M, Hinze-Selch D, Fenzel T, et al. Plasma levels of cytokines and soluble cytokine receptors in psychiatric patients upon hospital admission: effects of confounding factors and diagnosis. J Psychiatr Res 1999;33:407–418 41. Altman DG, Royston P. The cost of dichotomizing continuous variables. BMJ 2006;332:1080 42. Sackett D, Straus S. On some clinically useful measures of the accuracy of diagnostic tests. ACP J Club 1998; 129:A17–A19 43. Kathmann N, Hochrein A, Uwer R, et al. Deficits in gain of smooth pursuit eye movements in schizophrenia and affective disorder patients and their unaffected relatives. Am J Psychiatry 2003;160:696–702

39

44. Bender S, Weisbrod M, Resch F. Which perspectives can endophenotypes and biological markers offer in the early recognition of schizophrenia? J Neural Transm 2007;114:1199–1215 45. Gottesman II, Gould TD. The endophenotype concept in psychiatry: etymology and strategic intentions. Am J Psychiatry 2003;160:636–645 46. Mossner R, Mikova O, Koutsilieri E, et al. Consensus paper on the WFSBP task force on biological markers: biological markers of depression. World J Biol Psychiatry 2007;8:141–174 47. Shaw LM, Korecka M, Clark CM, et al. Biomarkers of neurodegeneration for diagnosis and monitoring of therapeutics. Nat Rev Drug Discov 2007;6:295–303 48. Bailey P. Biological markers in Alzheimer’s disease. Can J Neurol Sci 2007;34(1, Suppl):S72–S76 49. Vitzhum F, Behrens F, Anderson NL, et al. Proteomics: from basic research to diagnostic application. A review of requirements and needs. J Proteome Res 2005;4:1086–1097 50. Jaeschke R, Guyatt G, Sackett DL. Users’ guides to the medical literature. III. How to use an article about a diagnostic test. A. Are the results of the study valid? JAMA 1994;271:389–391 51. Fleiss JL. Statistical methods for rates and proportions. 2nd Ed. New York: Wiley-Interscience; 1981 52. Burke MJ, Preskorn SH. Therapetuci drug monitoring of antidepressants. Cost implications and relevance to clinical practice. Clin Pharmacokinet 1999;37:147–165 53. Vuille F, Amey M, Baumann P. Use of serum level monitoring of antidepressants in clinical practice. Pharmacopsychiatry 1991;24:190–195 54. Mann K, Hiemke C, Lotz J, Schmidt LG, Lackner KJ, Bates DW. Appropriateness of plasma level determinations for lithium and valproate in routine care of psychiatric inpatients with affective disorders. J Clin Psychopharmacol 2006;26:671–673 55. Price GW, Michie PT, Johnston J, et al. A multivariate electrophysiological endophenotype, from a unitary cohort, shows greater research utility than any single feature in the Western Australian Family study of schizophrenia. Biol Psychiatry 2006;60:1–10 56. Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol 1986;51:1173–1182 57. Caspi A, Moffitt TE, Cannon M, et al. Moderation of the effect of adolescent-onset cannabis use on adult psychosis by a functional polymorphism in the catechol-O-methyltransferase gene: longitudinal evidence of a gene X environment interaction. Biol Psychiatry 2005;57:1117–1127 58. Bostwick DG, Adolfsson J, Burke HB, et al. Epidemiology and statistical methods in prediction of patient outcome. Scand J Urol Nephrol 2005;39(3);94–110. 59. McShane LM, Altman DG, Sauerbrei W, et al. Reporting recommendations for tumour marker prognostic studies (REMARK). Br J Cancer 2005;93:387–391 60. Moher D, Schulz KF, Altman D for the CONSORT Group. The CONSORT statement: revised recommendations for improving the quality of reports of parallel-group randomized trials. JAMA 2001;285:1987–1991 61. von Elm E, Altman DG, Egger M, et al. Strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ 2007;335:806–808

Chapter 3

Challenging the Genetic Complexity of Schizophrenia by Use of Intermediate Phenotypes Assen Jablensky

Abstract Schizophrenia is a complex disorder, whose phenotypic variation and likely extensive genetic heterogeneity is not adequately captured by current clinical classifications. Despite a century of research, the field abounds in inconsistent empirical findings and conceptual controversies. How far can genetics take us in understanding its causes and what is the role of the environment? Is the disorder rooted in neurodevelopment or in neurodegeneration? Can we bridge the gap between the objective measurement of brain function and the subjective phenomenology of schizophrenia? These, and other unresolved fundamental issues, lead to questions about the status of schizophrenia as a putative disease entity and to attempts at its “deconstruction” by using intermediate (endo-) phenotypes. Endophenotypes are objectively measurable, biologically anchored heritable traits, which co-segregate with clinical illness in pedigrees and may also be expressed in clinically unaffected family members. This chapter reviews the phenotypic variation and likely etiological heterogeneity underlying the clinical phenotype of schizophrenia; outlines the conceptual foundation and criteria for the application of endophenotype research strategies; and provides an overview of promising endophenotype-based approaches including measures of cognition, electrophysiological brain responses, and brain imaging techniques. The design and findings of the Western Australian Family Study of Schizophrenia provide an illustrative example of the application of an endophenotype approach to parsing the complexity of the disorder with a view to facilitating its genetic analysis. A. Jablensky () School of Psychiatry and Clinical Neurosciences, Centre for Clinical Research in Neuropsychaitry, The University of Western Australia, Perth, Australia

Keywords Schizophrenia • phenotypic variation • endophenotypes • cognition • neurophysiology • genetics

Introduction The concept of endophenotype refers broadly to “measurable components, unseen by the unaided eye, along the pathway between disease and distal genotype”.1 Related, but not fully synonymous, terms are intermediate, elementary or correlated phenotype, subclinical or lower-level trait. Essentially, endophenotypes are objectively measurable, biologically anchored heritable traits co-segregating with clinical illness in pedigrees and may also be expressed, partially or fully, in clinically unaffected biological relatives. They are hypothesized to be more proximal to the primary biological defect and, hence, more sensitive to the “genetic signal” than the complex and often fuzzy clinical phenotype.2 Though overlapping, the concepts of endophenotype and biomarker refer to different biological properties. While both are objectively measurable, biomarkers index a dysfunction or morphological feature which is more tightly associated with clinical illness but may not be heritable or co-segregating within pedigrees. Some, but not all, biomarkers are endophenotypes, and some, but not all, endophenotypes can also serve as biomarkers. The term “endophenotype” originated in studies of plant and insect genetics,3 and was introduced into psychiatric research in the early 1970s by Gottesman and Shields,4 primarily with reference to schizophrenia, to fill the gap between the search for susceptibility genes and the “elusive disease process”, incompletely revealed by its symptoms. Biological traits exhibiting

M.S. Ritsner (ed.), The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes, © Springer Science + Business Media B.V. 2009

41

42

intermediate phenotype properties have proved to be effective tools for gene discovery in a variety of complex diseases (e.g. measures of serum iron in idiopathic haemochromatosis; long QT interval in ischaemic heart disease; characteristic EEG abnormalities in juvenile myoclonic epilepsy). In psychiatric genetics, the endophenotype concept started gaining ground during the 1990s, with a rapidly increasing number of publications in the last 5 years.5 The growing interest in endophenotypes is to a large extent a reflection of the limited success of genetic linkage and association studies, based solely on clinical diagnosis as the phenotype, to produce robust and replicable evidence of specific susceptibility genes in any of the major psychiatric disorders.6 While a number of different neuropsychiatric disorders have been subject in recent years to endophenotype-based research strategies (e.g. attention-deficit/hyperactivity disorder7; bipolar disorder8; autism9), the majority of studies utilizing this novel approach have been focused on schizophrenia.

Limitations of the Diagnostic Category as a Phenotype for Genetic Research With lifetime risk at ∼1% and nearly ubiquitous presence in diverse populations,10 schizophrenia has remained refractory to numerous efforts to unravel its aetiology and underlying neurobiology. Although its broad heritability is estimated at ∼80%, genetic linkage and association studies have so far made only modest contributions to the key problem of “connecting phenotype with the genotype”.11 No specific gene variant, or a combination of genes, has been definitively shown to contribute to the causation of the disorder.12 While genetic complexity (nonmendelian inheritance, polygenes, genetic heterogeneity, and unknown extent of environmental influences) is indeed a major hindrance, often invoked to call for ever larger study samples,13 the problem of a fallible phenotype has received relatively little consideration. The majority of extant genetic studies have been predicated solely on the diagnostic category of schizophrenia (defined by current DSM or ICD criteria) as the phenotype, based on psychopathological symptoms, their course, and associated behavioural deviance or impairment.

A. Jablensky

The diagnostic criteria of DSM-IV and ICD-10 were conceived with a view to achieving three goals: (i) to identify groups of patients with broadly similar clinical presentation and prognosis; (ii) to facilitate early diagnosis and choice of treatment; and (iii) to define a homogeneous heritable diagnostic category for aetiological research. While the first two goals, related to clinical utility, have been largely achieved, the third has not been attained.14 Symptomatology, elicited from patients’ reports and behavioural observation, though useful for clinical purposes, may not be a reliable indicator of the underlying neurobiology of the disease process. The clinical phenomenology of schizophrenia is characterised by extensive pleomorphism, foreshadowed in E. Bleuler’s15 notion of a “group of the schizophrenias”, including a number of aetiologically distinct disorders sharing a broad, “common final pathway” of clinical phenotype expression. Notwithstanding the acceptable level of inter-rater agreement that can be achieved on the categorical diagnosis, the symptoms of schizophrenia span a wide range of psychopathology and display extensive inter-individual variability and temporal inconstancy. Part of its clinical manifestations, in particular some of the negative symptoms, may actually result from compensatory adaptations and behaviours16 that vary over the course of the illness and its treatments. As no symptom is pathognomonic or necessary, and variable subsets of symptoms can be sufficient for the diagnosis, patients may be allocated to the diagnostic category of schizophrenia without sharing a single symptom, sign or type of impairment. This allows for a wide margin of classification error (phenotype misspecification) in samples selected for biological research exclusively by the current diagnostic criteria, and the biological homogeneity of patients in such samples is open to question. Clinical populations, selected on the diagnostic category are more likely than not, to contain an admixture of aetiologically different disorders. The corollary of this realistic conjecture is that parsing such samples into component subtypes should reduce biological heterogeneity. In this respect, schizophrenia may not be different from other complex diseases (such as retinitis pigmentosa,17 Type 2 diabetes,18 or breast cancer19) where “splitting” and stratification strategies have resulted in critical advances towards the discovery of contributory genes. While in mendelian diseases phenotypes are usually dependable pointers to genotypes (and vice versa), such reciprocity is certainly ruled out in the case of

3

Challenging the Genetic Complexity of Schizophrenia

diseases such as schizophrenia where the clinical phenotype is the end product of intractably complex interactions between genotypes, environments and individual neurodevelopmental histories that may include epigenetic regulation of gene expression. As pointed out by Gottesman and Gould,1 such complexity calls for use of “more optimally reduced measures of neuropsychiatric functioning…than behavioural “macros”. Judiciously selected and carefully evaluated endophenotypes should provide such narrowly constrained measures.

Early Explorations of Endophenotypes In one of the earliest studies explicitly embracing an endophenotype concept, Matthyse et al.20 and Holzman21 examined smooth-pursuit eye movements (SPEM) in twins discordant for schizophrenia and found a nearly perfect concordance for SPEM dysfunction in the monozygotic twins, whereas only ∼50% of the dizygotic twins were concordant. Based on these findings, Holzman22 proposed a model postulating a genetically transmitted latent trait which may cause either schizophrenia or poor eye tracking, or both. The trait was thought to be associated with a disease process that could independently invade one brain region or another, giving rise to the diverse symptoms. Analysis of apparently divergent manifestations, such as SPEM and clinical schizophrenia, may facilitate discovery of a common aetiology, as previously demonstrated in neurology, where the phenotypes of Friedreich’s ataxia and hereditary cerebellar ataxia proved to be two manifestations of a single pathological process affecting anatomically distinct regions.23 Analogous reasoning should apply to psychiatric disorders, by parsing the phenotype into simpler components that could be studied biologically. Similarly, Cromwell24 argued that cognitive, biobehavioural and other markers might account for more of the genetic variance than does the clinical diagnosis. He proposed a list of “things to do before the geneticist arrive”, which included identification of “schizophrenia-related variants” (SRV) by: (a) variables known to be associated with schizophrenia, but not necessarily part of its diagnosis; (b) vulnerability markers that emerge earlier than the symptoms of schizophrenia; and (c) similar characteristics found among mentally healthy biological relatives

43

of patients. Examples of SRV indicators were lexical priming (resistance, in patients with schizophrenia and in their biological relatives, to the normally occurring interference from an incongruent priming word); and the reaction time (RT) crossover effect (the amount of slowing in RT induced by alternating preparatory intervals). A theoretical framework, in which the endophenotype idea was implicitly embedded, was provided by Paul Meehl’s25 concept of schizotaxia. Meehl regarded schizophrenia as a “loose syndrome”, a cluster of seemingly unrelated phenomena, which has at its basis a genetically determined, generalized CNS integrative defect, a “slippage at the synapse”, whose phenotypic expression he termed schizotaxia, a subclinical neurobehavioural syndrome predisposing to schizophrenia. As objective indicators of schizotaxia, he proposed neurophysiological tests, such as SPEM and the P50 potential; attentional measures of signal-to-noise ratio; syntactical and semantic aberrations; and soft neurological signs. Meehl expected this model to predict functional impairments in domains involving complex integration of inputs, e.g. from multiple sensory modalities, and domains demanding finely tuned control by their subsystems: “whatever is wrong with the schizotaxic CNS is ubiquitous, a functional aberration present throughout, operating everywhere from the sacral cord to the frontal lobes … a functional parametric aberration of the synaptic control system … an integrative defect analogous to dyslexia”.25

Endophenotype Criteria Endophenotypes offer a novel approach to reducing the complexity and heterogeneity of schizophrenia that could be either an alternative or a complement to symptom-based phenotypes. The translation of the endophenotype concept into research strategies calls for explicit definitions and criteria. The following checklist of criteria or properties of endophenotypes is based on the original criteria proposed by Gottesman and Shields,1 their subsequently revision, and further versions and refinements proposed by Almasy and Blangero2; Bearden and Freimer26; Leboyer et al.27; Cannon and Keller8; Snitz et al.28; Gur et al.,16 and Turetsky et al.,29 on behalf of the Consortium on the Genetics of Schizophrenia (COGS).

44

Viable endophenotypes must exhibit any number of the following properties: 1. A stable, state-independent trait (observable regardless of the patient’s symptoms, clinical status or disease stage); the trait may be detectable prior to the onset of clinical symptoms. 2. The trait should co-segregate with clinical illness within families and the variation in its distribution within pedigrees should exhibit at least a moderate level of heritability. 3. A modest to large odds ratio or effect size of the trait prevalence in affected individuals as compared to healthy controls should be demonstrable. 4. The trait should be present at a higher rate in nonaffected family members than in the general population (high relatives-risk ratio, λr). 5. At the population level, the trait should vary continuously as a normally distributed trait and be associated with the incidence and severity of illness at specifiable thresholds. 6. The trait should possess good biometric or psychometric properties, analyzable on a quantitative scale (statistically more powerful than dichotomization), and show acceptable test-retest reliability and reproducibility. 7. There should be a plausible association with the genetic causes of the disorder, rather than with its concomitants or consequences, such as effects of medication or cerebral degeneration due to disease progression. 8. The trait should be related to known or suspected neurobiological pathways and well-characterized neural system models relevant to the disease. 9. The genetic analysis of endophenotypes should prove to be more powerful than analysis of the disorder itself (the effect sizes of genetic associations will be greater in relation to the endophenotype than to the clinical syndrome). 10. Endophenotypes should be translatable into animal models (involving homologies of expression across species). 11. The administration of the relevant test measures in the context of research should be cost-effective and logistically practicable. While hardly any of the currently proposed or actually utilized endophenotype variables is likely to meet all of the above conditions, adherence to the criteria should ensure greater stringency in the selection of

A. Jablensky

potentially useful endophenotypes. Notably, the currently emerging consensus on endophenotype criteria leaves out entirely some of the earlier desiderata, such as that endophenotypes should have a simpler genetic architecture than the illness phenotype, or that they should be specific to a given diagnostic category. There is little doubt that the genetic basis of some endophenotypes is likely to be as complex as the genetics of the disease itself, and involve multiple genes and polymorphisms. While endophenotypes can provide useful trait markers, they may not be any easier to dissect at a genetic level than the disorders to which they are related.30 Furthermore, the requirement that an endophenotype should be specifically associated with schizophrenia, and not with other disorders, is unrealistic, as it is becoming increasingly clear that overlapping sets of genes may influence susceptibility to disease across the traditional nosological boundaries.31,32

Cognitive Endophenotypes Cognitive deficit, as a salient feature of schizophrenia, was part of Kraepelin’s original definition of dementia praecox, essentially characterised by “weakening of the mainsprings of volition, lowered mental efficiency, unsteadiness of attention, inability to sift, arrange and correct ideas, and to accomplish mental grouping of ideas”.33 Meta-analytic or systematic reviews of the evidence16,34 suggest that on the criteria of consistency and effect sizes of reported findings, cognitive dysfunction in schizophrenia ranks as a prime candidate for the endophenotype domain. Compromised higher cognitive function (low IQ) prior to the onset of disease has been shown to be a significant risk factor for schizophrenia in two large, population-based studies.35,36 There is remarkable agreement in the literature that deficits in multiple cognitive domains predate the onset of clinical symptoms; are not attributable to antipsychotic medications; persist over the course of the illness; are unrelated to its duration; occur in non-psychotic relatives; are qualitatively characteristic of schizophrenia as compared to other psychotic disorders; and exhibit the properties of a stable trait. Cognitive deficits are thus likely to be a core feature of schizophrenia (not an epiphenomenon of the illness), and meet most of the endophenotype criteria. This conclusion is under-

3

Challenging the Genetic Complexity of Schizophrenia

scored by the meta-analysis by Heinrichs and Zakzanis37 of 204 studies published between 1980 and 1994 (comprising a total of 7,420 schizophrenia patients and 5,865 controls), in which effect sizes (Cohen’s d) and the U statistic (degree of non-overlap) were calculated for 22 neurocognitive test variables ranging from IQ, verbal memory and attention to executive function and language. Neurocognitive deficit was found to be a reliable and well replicated finding in schizophrenia, though no single test or cognitive construct was capable of separating perfectly schizophrenia patients from normal controls. Seven widely used measures achieved effect sizes greater than 1.0 (i.e. 60–70% non-overlap between the cases and controls): global verbal memory (1.41), bilateral motor skills (1.30), performance IQ (1.26), the continuous performance task (1.16), word fluency (1.15), the Stroop task (1.11) and WAIS-R IQ (1.10). Although a subset of ∼50% of patients had nearly normal performance, significant cognitive impairment was common in schizophrenia and exceeded the deficits found in some neurological disorders, justifying the view that “schizophrenia is a neurological disorder that manifests itself in behaviour”.37 However, cognitive deficits in schizophrenia are variable in scope and severity.38 They range from pervasive, generalized dysfunction, through patchy focal disorders, to mild focal deficits or nearly normal performance. Yet converging evidence points to specific deficits in verbal declarative memory (mainly in the early encoding stage) and in working memory as major sources of variance.39 This observation has prompted attempts at delineating particular profiles or subtypes. While conventional cluster analyses tend to distribute patients into subgroups of severely compromised, intermediate and mildly affected performance, “classical” fine-grain neuropsychological analyses (case studies of individual profiles rather than group means; delineation of generalised/differential deficits; search for “double dissociations”) have identified subtypes and patterns of dysfunction that parallel the amnestic syndromes in coarse brain disease, such as Huntington’s (HD), Parkinson’s (PD) or Alzheimer’s (AD). Paulsen et al.40 elicited from ∼50% of schizophrenia patients a ‘subcortical’ (striatal, HD/PD-type) memory profile combining prominent retrieval deficit with absence of significant storage deficits (rapid forgetting). Another 15% had a ‘cortical’ (hippocampal-thalamic,

45

AD-type) profile (primary encoding and storage impairment, with an excess of irrelevant word intrusions on free and cued recall), while the profiles of the remaining 35% did not deviate significantly from those of the controls. These findings were replicated by Turetsky et al.41 and supported by neuroimaging data suggesting ventricular enlargement with preserved temporal lobe gray matter in the ‘subcortical’ group, and left-hemisphere temporal and frontal volume reduction in the ‘cortical’ group. Dickinson et al.42 estimated that over 30% of the variance in cognitive test performance by schizophrenia patients could be explained by a general largeeffect (“g”) factor, affecting the integration of multiple intermodal brain functions into “core” cognitive operations such as concept formation and reasoning skills. Further variance, however, could be explained by a number of independent, small-effect variables selectively affecting specific functions, such as processing speed and visual memory. Promising as they are, these approaches to “splitting schizophrenia” were limited by small sample size, as well as by insufficient efforts to integrate multi-domain data (e.g. neuroimaging and neurophysiological measurements) into multivariate composite traits that might increase their capacity to parse the deficits characterising schizophrenia. With a few exceptions,43–46 cognitive deficits have not yet been systematically tested as endophenotypes in molecular genetic studies. The cognitive indicators with moderate to high effect sizes, which meet the criteria of endophenotypes and have recently been reviewed for the COGS Consortium16 include, the domains of attention, verbal declarative memory, working memory, face memory, and emotion processing. The salient characteristics of these measures are summarized in Table 3.1. It is, however, likely that most cognitive measures available today assess interrelated facets of complex interactive neural networks, rather than isolated processing modules.47 Many of the tests reviewed above are influenced by several component processes and their analysis in isolation from one another may lead to a spurious impression of selective impairment where in fact generalized deficit might be present.37 If several endophenotype measures share a common genetic basis, analysing them jointly as patterns (without multiple testing) is likely to increase power. 26,29,46,48

46

Table 3.1 Overviewa of selected cognitive candidate endophenotypes in schizophrenia research Heritability/segregation within families

CPT-IP d = 1.51 CPT-DS d = 1.29 (WAFSSb)

Test-retest: CPT-IP r = 0.56 (SZ) 0.73 (C) CPT-DS r = 0.65 SZ) 0.72 (C) Both stable across clinical states

Linkage to 6p24 (WAFSS) Twin studies: CPT-IP h2 = 0.39–0.49(C) Association with the 22q11 deletion CPT-DS h2 = 0.51–0.57(C) syndrome CPT-DS λr = 9 (sibs) λr = 12 (parents)

Verbal declarative memory The most consistently Found deficit in SZ patients. (VDM) VDM deficits related to WMS-III LM negative symptoms CVLT RAVLT

d = 1.41 – 2.39

Test-retest: r = 0.62–0.64 (SZ) r = 0.74–0.88 (C)

VDM deficits associated Twin studies: with decreased h2 = 0.47–0.63 (C) volume of hippocamh2 = 0.21–0.49 (SZ) pus; association with Mild deficits in unaffected DISC1 reported family members

Working memory (WM) Consistently found deficits in SZ patients Online maintenance of information: Spatial delayed response Digits forward Maintenance + manipulation N-back tasks Digits backward Letter-number sequencing (LNS)

LNS > 1.4 SD (SZ vs. controls) LNS d = 0.66 (unaffected family x vs. controls)

Constancy over time; minimal Twin studies: correlation with positive h2 = 0.43–0.49 (C) symptoms; moderate h2 = 0.36–0.42 (SZ) correlation with negative symptoms

WM deficits associated with DLPFC and posterior parietal dysfunction Likely role of COMT val158met polymorphism and DISC1

Face memory and emotion Frequently found deficits processing in SZ patients Face recognition tasks Emotion recognition tasks

No published data

Possibly stable trait; more data required

Associations (fMRI): Faces: right fusiform gyrus and frontotemporal circuitry Emotion: amygdale and hippocampus. Genetic association not known.

Cognition Attention Continuous Performance Task (CPT) Identical Pairs (CPT-IP): working memory load Degraded Stimulus (CPT-DS): perceptual disambiguation

a

Association with disorder

Effect size

SZ patients consistently worse than controls CPT-IP associated with negative symptoms; CPT-DS associated with cognitive disorganization

Based on references16,37,39–42,45,46,88 Western Australian Family Study of Schizophrenia

b

Twin studies: h2 = 0.33 (faces) h2 = 0.37 (emotion) Mild deficits in unaffected family members

Underlying neurobiology/ genetics

A. Jablensky

Trait stability/state independence

Domain/tests

3

Challenging the Genetic Complexity of Schizophrenia

Neurophysiological Endophenotypes Five potential endophenotypes in the domain of neurophysiology have been extensively characterized in clinical studies, but for the majority, relatively little is at present known regarding their genetic underpinnings. The P50 component of an early (pre-attentive) event-related potential (ERP) has been systematically assessed as a measure of auditory sensory gating.29,48–50 P50 is a positive deflection occurring upon the presentation of a pair of clicks with a 500 ms interstimulus interval. The P50 wave generated to the second (test) click is normally suppressed relative to the P50 response to the first (conditioning) click, likely due to the activation by the first click of cholinergic and GABAergic inhibitory neural circuitry in the hippocampus. Impaired suppression of P50 is present in the majority of patients with schizophrenia, as well as in a proportion of their unaffected first-degree relatives. Estimates of heritability range between 0.44 in twins49 and 0.53 in a recent COGS-sponsored study.51 P50 is largely uninfluenced by typical antipsychotic drugs, but some amelioration of the deficit may occur with administration of atypical antipsychotics. Genetic linkage has been found to a region on chromosome 15q13–14, and subsequently association with a polymorphism in the promoter of the CHRNA7 gene within that region, coding for the α-7 subunit of the nicotinic receptor.50 The neural substrate of the P50 response has been extensively investigated in animal models and involves primarily the hippocampus (the CA3 and CA4 inhibitory interneurons modulating the activity of glutamatergic pyramidal cells), though temporoparietal and prefrontal cortical areas are also likely to play a role.52 The neurobiological significance of the sensory gating mechanism is thought to be in enhancing the processing of strong, likely ‘relevant’, stimuli at the expense of blocking weak, ‘irrelevant’ ones. A gating dysfunction, such as occurring in schizophrenia patients, can result in hippocampal synaptic occlusion with local blockage of neural processing.29 Prepulse inhibition of startle (PPI) is an operational measure of sensorimotor gating, in which weak acoustic prestimuli presented at intervals of 30–300 ms prior to a startle-eliciting stimulus reduce the magnitude of the blink reflex component of the startle response.53 It is a stable neurobiological marker that been investigated extensively in both animal models and in human

47

studies, and can be reproduced reliably in multi-site studies.54 PPI is mediated by circuitry involving the limbic cortex, striatum, pallidum, and pontine tegmentum, interacting with the primary startle circuit at the level of the brain stem (the caudal reticular nucleus of the pons). The PPI response is sensitive to pharmacological manipulation (e.g. it is enhanced by nicotine and degraded by administration of d-amphetamine). In patients with schizophrenia, PPI is deficient, in the sense that inhibition of the startle is reduced or fails, and the brain remains overly responsive to the second stimulus. Atypical antipsychotics, however, tend to normalize the PPI response. In both humans and rodents, PPI is significantly heritable. The inhibitory deficit tends to be present in first-degree relatives of schizophrenia patients.53 Heritability has been estimated at ∼0.50 in twin studies55; and at 0.24 in a recent COGS study.51 Apart from schizophrenia, PPI deficits have been reported in Huntington’s disease,56 seizure disorders,57 Tourette syndrome,58 and in carriers of the chromosome 22q11 deletion.59 Mismatch negativity (MMN) is a negative ERP component, recorded as an automatic response to lowprobability deviant sounds in a sequence of standard acoustic stimuli, while the subject’s attention is directed elsewhere. The stimulus deviance can be designed to be either one of pitch, or of duration (in schizophrenia patients, the effect size of the response is significantly greater for duration as compared to pitch variation).60 The likely generators of the MMN potential include the primary and secondary auditory cortices, with possible contributions from the dorsolateral prefrontal cortex. Neurobiologically, the MMN is regarded as a response of the auditory system to the detection of a deviation from an established pattern of acoustic stimulation in primary sensory memory.61 Reduction in MMN amplitude can be elicited in the majority of schizophrenia patients,62 and in a proportion of their unaffected first-degree relatives,63 though replications of the latter finding have not been consistent. Relatively little is known about the heritability of MMN. Recent twin studies64,65 produced heritability estimates in the 0.63–0.68 range, but this is yet to be replicated. The MMN deficit appears to be relatively specific to schizophrenia.66 Interestingly, deviant MMN responses have been reported to occur in patients who are carriers of the chromosome 22q11 deletion.67 It remains uncertain whether the MMN deficit is present

48

in the early stages of schizophrenic disorders, or it only appears with disease progression, being an index of neurodegeneration. P300 is a complex group of ERP components, related to a family of frontal (P3a) and parietal (P3b) generators comprising the hippocampus, superior temporal gyrus, inferior parietal lobe, frontal cortex, and thalamus.68 The electrogenesis of P300 is thought to depend significantly on the neurodevelopmental maturation and integrity of the laminar and columnar organization of the cortex.69 It has been extensively studied in schizophrenia and a range of other disorders, including alcoholism. P300 can be elicited in practically all sensory modalities, but auditory stimuli produce the largest effect sizes. Experimentally, P300 is generated in an oddball paradigm in response to attended infrequent target stimuli (e.g. involving a pitch difference) which require an overt (attentional effortful) reaction. While the latency of the P300 response is age-dependent (increasing with age), the amplitude is a stable trait, relatively unaffected by aging. P300 reflects several different cognitive processes, including attention, contextual updating of working memory, and attribution of salience to a deviant stimulus. It is still debatable whether P300 deficits arise in a ‘top down’ fashion from a primary dysfunction in the prefrontal heteromodal cortex, or “bottom-up” from an impairment at the level of initial cortical response and stimulus encoding. Reduced amplitude of the oddball P300 response is a robust abnormality, observed in the majority of schizophrenia patients (effect size d = 0.74, meta-analysis in Jeon and Polich70). Some findings suggest that a reduced P300 amplitude over the left hemisphere, with an amplitude maximum over the right hemisphere, may be specific to schizophrenia.71 As a trait abnormality, the P300 amplitude decrement is independent of medication effects, length of illness, or severity of symptoms, and can be elicited in clinically unaffected first-degree relatives of schizophrenia patients.29 Heritability of the P300 amplitude variance has been consistently estimated at 0.48–0.61,72 which ranks it among the most heritable endophenotypes. Genetic studies have reported suggestive linkages of P300 to regions on chromosomes 2, 5, 6 and 17.73 An important finding is the co-segregation of P300 amplitude reduction with the translocation on chromosome 1 in a large Scottish family that led to the identification of DISC1 as a strong candidate gene in schizophrenia.74

A. Jablensky

The antisaccade (AS) task is an oculomotor paradigm, assessing inhibitory capacity, in which a fixation cue appears unpredictably on the screen in an eccentric location. The subject is instructed to inhibit the reflexive pro-saccade response, and instead make a volitional saccade in the opposite, mirror-image direction. This involves the ability to retain, manipulate and transform sensory information about location into a goal-directed, willed action. Errors are recorded if the saccade is in the wrong direction, or is dysmetric. The neurobiology underlying antisaccade performance is relatively well established, and involves the parietal and prefrontal dorsolateral cortices, as well as the supplementary oculomotor areas. Patients with schizophrenia produce high error rates, reduced accuracy of antisaccades, and longer latencies to correct response.75 The AS task is not influenced by the clinical state of the patient, and the effects of antipsychotic medication, if any, tend to be in the direction of mild improvement. The data on deficits in AS performance among unaffected first-degree relatives are not entirely consistent, with some studies reporting increased AS error rates,76 while others fail to find such deficit.77 Heritability of the AS performance is still a contentious issue, with one study based on twin data suggesting >0.50 concordance78; recent COGS findings of a comparable level of heritability (0.49, estimated by component variance analysis in nuclear families with a schizophrenia proband51; and studies reporting inconsistent findings in relatives of schizophrenia patients.79,80 The genetic basis of the AS task is largely unexplored; as in other neurocognitive tasks, individuals with the chromosome 22q11 deletion tend to have a compromised AS performance.81 An overview of the main characteristics of the neurophysiological endophenotypes discussed in this section is provided in Table 3.2.

Neuroimaging Endophenotypes “Imaging genetics” is an emerging strategy for mapping neural structures and brain activity as a function of genotype.82,83 Evolving powerful techniques for three-dimensional cortical surface mapping, as well as novel functional neuroimaging methods,84,85 are certain to have a profound impact on the whole field of endophenotype research into schizophrenia, ultimately

3

Table 3.2 Overviewa of selected neurophysiological candidate endophenotypes in schizophrenia research Trait stability/State independence

Heritability/Segregation within families

Association with disorder

Effect size

SZ patients fail to attenuate response to second (test) stimulus

Test-retest: Amplitude: r = 0.66–0.73 d = 0.78 Suppression ratio: Suppression deficit present in both d = 0.54 acutely psychotic (WAFSSb) and stabilized patients. Clozapine reduces deficit

Cholinergic activation of hippocampal Twin studies: CA3/CA4 interneurons inhibits the h2 = 0.53–0.68 firing of pyramidal neurons. Unaffected relatives and Involvement of temporoparietal and high-risk subjects prefrontal circuits likely. Association show less suppression with the CHRNA7 gene coding for than controls, but data the α-7 subunit of the nicotinic are inconsistent receptor

Failure of automatic inhibition SZ patients are deficient in automatic inhibition of Prepulse inhibition of startle startle. The response is reflex (PPI) influenced by atypical antipsychotics and nicotine. DA agonists reduce, NMDA antagonists increase PPI

Not available

Test-retest: r > 0.90 Longitudinal stability of PPI across clinical states has been little investigated

PPI regulated by a limbic cortico-striatoTwin studies: pallido-pontine circuit interacting h2 > 0.50 with the primary startle circuit at the PPI deficits found in pons reticular nucleus. PPI is unaffected family abnormal in the 22q11 deletion members. syndrome. Possible role of the Males produce greater PPI D2-like receptor G-protein and than females. PPI in hippocampal α5 subunit of the Asians > Caucasians GABAA receptor

Stimulus deviance detection Deficit is present in the majority of SZ patients Mismatch negativity (MMN): (not ameliorated by Formation of an early atypical antipsychotics). (pre-attentive) auditory memory trace; automatic Presence of MMN deficit in first-episode patients comparison process. MMN uncertain tests the integrity of the primary auditory memory network

d ∼ 1.0 (metaanalysis) d = 0.74 (WAFSS)

MMN is reduced in the 22q11 deletion Test-retest: No formal h2 estimates available. Abnormality syndrome. Possible association with r = 0.78 (duration is present in a the COMT val158met polymorphism MMN) proportion of r = 0.53 (frequency unaffected family MMN) members Intraclass r = 0.9 over 1 year

Working memory updating/ stimulus salience evaluation Auditory P300 event-related potential Composite of: P3a (frontal); P3b (parietal)

Meta-analysis: Test-retest: Twin studies: d = 0.89 r = 0.81–0.91 (2 weeks) h2 = 0.60 (amplitude) r = 0.59–0.61(1–2years) Unaffected family members similar to d = 0.59 (latency) probands. Familial WAFSS: deficit most evident d = 0.91 for P3a (amplitude)

Sensory gating P50 event-related potential (suppression ratio and amplitude difference)

Amplitude decrement and increased latency in SZ patients: one of the most consistent findings in the disorder

Underlying neurobiology/genetics

Challenging the Genetic Complexity of Schizophrenia

Neurophysiology/ psychophysiology

Amplitude decrement correlated with smaller left superior temporal gyrus. Genetic linkage to 6p24. Possible association with DISC1 and the DRD2 receptor.

(continued)

49

50

Table 3.2 (continued) Neurophysiology/ psychophysiology Saccadic dysfunction Antisaccade task (AS) Inhibition of reflexive prosaccade; performance of antisaccade to a mirror location

Association with disorder

Effect size

SZ patients: increased error rate; longer latencies to correct AS; reduced spatial accuracy

d = 0.99 (patients vs controls, WAFSS) d = 0.99 (relatives vs. controls)

Trait stability/State independence

Heritability/Segregation within families

Test-retest: Twin studies: r = 0.87 (2 years) h2 = 0.57 COGS inter-site study: Error rate in unaffected family members: r = 0.77–0.96 inconsistent data Deficit stable across clinical states

Underlying neurobiology/genetics Sensorimotor reprogramming; DLPFC, lateral interparietal area, supplementary eye field neurons Linkage to 22q11–12 (COMT effect?)

a

Based on references46,49–55,60–66,69–74,75–81,88 Western Australian Family Study of Schizophrenia Abbreviations in Tables 3.1 and 3.2: C = healthy controls, CVLT = California Verbal Learning Test, d = effect size (Cohen’s d), DLPFC = dorsolateral prefrontal cortex, fMRI = functional magnetic resonance imaging, h2 = heritability coefficient, LNS = letter-number sequencing task, r = correlation coefficient (Pearson), RAVLT = Rey’s Auditory Verbal Learning Test, SD = standard deviation, SZ = schizophrenia, WMS-III LM = Wechsler Memory Scale, 3rd Revision, λr = relative risk to patient’s family members

b

A. Jablensky

3

Challenging the Genetic Complexity of Schizophrenia

leading to future whole genome – full brain association studies. For the time being, however, the applications of imaging endophenotypes to the genetics of complex psychiatric disorders are limited by problems of reliability and reproducibility; risk of spurious findings; and lack of high-dimensional analytical methods that would control for Type I error.86,87

Endophenotypes in the Western Australian Family Study of Schizophrenia The Western Australian Family Study of Schizophrenia (WAFSS) is one of the first research projects (initiated in 1996) to apply systematic endophenotyping to the search for susceptibility genes in schizophrenia.46,88 A core aim was to address explicitly the problem of heterogeneity in schizophrenia. The design was based on the hypothesis that the syndrome of schizophrenia comprises several subtypes that could be teased out by objective endophenotype measurements of brain function and by exploring their genetic underpinnings. By the end of 2007, the study population comprised 895 individuals, including 473 members (157 affected with schizophrenia or schizophrenia spectrum disorder) of 126 nuclear families, 195 additional singleton cases of schizophrenia, and 161 unrelated healthy controls. Clinical assessment included standardized diagnostic interviews; developmental history (detailed maternal interviews and case notes); and treatment documentation from the Western Australian psychiatric case register. Participants completed a comprehensive neurocognitive assessment. The core battery of cognitive tasks focused on the domains of general ability (premorbid and current IQ); verbal learning and memory; sustained attention; speed of information processing; behavioural lateralization; and psychometric measures of schizotypy, temperament and character. The cognitive evaluation of the phenotype was complemented by electrophysiological measures (P50, N1, MMN, P300 and an antisaccade task), as well as structural MRI in a subset of the participants. In the analysis of the endophenotypes, the neurocognitive and personality measures were aggregated into a limited number of quantitative traits, using grade of membership (GoM) analysis,89 a version of latent

51

structure analysis which defines latent groups (“pure types”, PT) and allows individuals to resemble each PT to a quantifiable degree, yielding a set of quantitative traits for use as covariates in genetic analysis. Two PTs represented >90% of the schizophrenia patients and 23% of their unaffected first-degree relatives, while another two comprised normal controls and clinically healthy relatives with age-related cognitive deficits. The two schizophrenia PTs presented contrasting cognitive patterns: one of generalised cognitive deficit (labeled as CD subtype) and one cognitively spared (CS subtype). The CD phenotype displays pervasive deficit across the majority of cognitive domains, with the most affected functions including encoding in verbal memory (98% of CD cases scoring ≥ 1 SD below the performance level of controls); current IQ (91% scoring ≥ 1 SD below controls); sustained attention/working memory (90% scoring ≥ 1 SD below controls); and verbal memory delayed recall (87% scoring ≥ 1 SD below controls). Comparable deficits arise in the visual memory span and visuospatial working memory.90 Memory impairment is present at a very short delay (immediate recall), is stable with increasing delay intervals and points to compromised encoding, rather than retention/forgetting. Notably, 12.7% of the clinically unaffected first-degree relatives of CD cases exhibit similar deficits, though of attenuated severity. Most of the ‘cognitively spared’ (CS) patients perform within 1 SD of the control levels on all functions. The electrophysiological data91 reveal further significant deficits in the CD subtype relative to both CS and controls: reduced P50 amplitude and ratio; abnormal PPI; low amplitudes of N1 and P300; high error rates and low self-correction rates on the antisaccade task.93 Coherence of ERP-evoked gamma band oscillations is within normal range but power is reduced in both CD and CS. On structural MRI, CD cases have volume reductions in the left medial temporal lobes and the hippocampus, while reduction in the frontal lobes is minimal. The phenotypic data clearly point to cognitive deficit, particularly in learning and encoding in verbal memory, as a core feature of a homogeneous group of CD cases who resemble Kraepelin’s dementia praecox.33 The CD subtype is also associated with an increased rate of pregnancy complications. Impairments of early development (learning difficulties, marked introversion and social withdrawal), and a higher incidence of schizophrenia in the pedigrees

52

A. Jablensky

of CD cases, characterize the CD subtype as a likely neurodevelopmental disorder. While positive psychotic symptoms do not differentiate clearly between the CD and CS subtypes, CD patients consistently display more negative symptoms, particularly poverty of speech, poor nonverbal communication and diminished social drive. The cognitive and electrophysiological responses of CD patients suggest impaired synaptic function at the early stages of sensory encoding with a cascade effect along the timeline of cognitive processing. Whole-genome linkage analysis of the WAFSS sample produced evidence for a distinct genetic basis of the CD subtype (a locus on chromosome 6p25-24, from which the CS subtype was definitively excluded (Fig.3.1).46 In sum, the data support the notion of two or more distinct neurocognitive subtypes of schizophrenia which warrant systematic examination of the molecular networks involved.

Phenotypic variability has been confounding the search for the causes of schizophrenia since the inception of the diagnostic category.88 The inconsistent and poorly replicated results of genetic linkage and association

studies using the diagnostic category as the sole schizophrenia phenotype raise critical questions about the validity of the current nosology of schizophrenia. Geneticists are facing a difficult predicament, seeking to discover specific genes contributing to an overinclusive and likely heterogeneous diagnostic category for which no specific biological substrate has yet been demonstrated. The genetic polymorphisms and neurobiological deficits underlying schizophrenia are multiple, varied, and partly shared with predisposition to other disorders. Such polymorphisms and deficits need not be intrinsically pathological and may represent extreme variants of normal structure and function. Above a certain density threshold, their additive or non-linear interaction could give rise to the diagnostic symptoms in affected subjects, but subclinical manifestations as endophenotype traits will be detectable in otherwise healthy people, with a higher relative risk in biological relatives of probands. While reasoning along such lines is increasingly common among researchers, the approaches proposed to deal with the phenotype bottleneck in schizophrenia research differ substantially. On one hand, there are proposals to abandon the “Kraepelinian dichotomy” of schizophrenic and affective disorders in favour of a psychosis continuum.92 On the other hand, there is an emerging research agenda seeking narrowly constrained phenotypes that may tag distinct

a

b

Deconstructing Schizophrenia?

1.0

0.9 0.8

0.5

ch6p linked unlinked

0.4

CS giks

− 1SD MEAN + 1SD

0.6 CD giks

−1SD MEAN + 1SD

0.8

0.7

0.6 ch6p linked unlinked

0.4

− 1SD MEAN + 1SD

0.3 0.2

− 1SD MEAN + 1SD

0.1

0.2

0.0

0.0 CD

CS

Fig. 3.1 Genetic findings (linkage to chromosome 6p25-22) in the Western Australian Family Study of Schizophrenia (WAFSS46) for 93 fully characterized families (388 members) phenotypically classified into a cognitive deficit (CD) and a cognitively spared (CS) subtype, based on grade of

CD

CS

membership (g ik) scores on composite quantitative traits CD and CS, integrating ten neurocognitive endophenotypes. ■ = linked family; O =nonlinked family. (a) Distribution of the families on trait CD; (b) Distribution of the families on trait CS

3

Challenging the Genetic Complexity of Schizophrenia

subtypes of schizophrenia and resolve part of its aetiological heterogeneity. Candidate endophenotypes, or biomarkers of pathogenetic processes affecting cognition, brain morphology and neurophysiology, constitute the mainstay of this approach. Genetic linkage and association studies employing endophenotypes have so far produced preliminary results that call for replication. Subtyping strategies are supported by mounting evidence that sample stratification, particularly using quantitative endophenotype traits as covariates, can reduce heterogeneity and substantially increase power.93,95 This approach has been highly successful in the genetics of other complex diseases. Its application to schizophrenia genetics will bring the disorder into the mainstream of current research into the common genetic diseases. In contrast to otherwise powerful, but hypothesisfree approaches, such as whole-genome association studies, which are not predicated on prior neurobiological information, endophenotype-guided studies can be guided by specific hypotheses based on neuroscience and molecular biology. There are, however, caveats to consider. First, a cognitive or neurophysiological variable meeting most of the endophenotype criteria may not be causally involved in the clinical phenomenology of the disorder. In schizophrenia (as in many other psychiatric disorders), we are still facing an “explanatory gap” between objectively measurable neurocognitive dysfunction and subjectively experienced symptoms, such as primary delusions, passivity experiences, or thirdperson auditory hallucinations. Secondly, genes influencing neurocognitive or neuroanatomical endophenotypes may selectively exert their effects only at early developmental stages (in utero or in early infancy), whereas their expression in later life would be altered by environmental stress, substance use, or medication. Thirdly, we do not have at present any validated simple, relatively low-cost covariate measure of a large effect size, analogous to the QT interval, glucose tolerance index, or serum iron level, that would enable the splitting of the clinical phenotype along genetic fault lines. Instead, endophenotypebased studies of schizophrenia (or any other psychiatric disorder for that matter) will require measurement of multiple variables, thus constraining the effective sample size. Strategies preempting such methodological pitfalls should include efforts to achieve proper

53

standardization of evidence-based endophenotype testing batteries which comprise a “core” of wellestablished, standard tests as a reference point, but also allow for additional experimental, customdesigned applications. As proposed by Gerlai,94 such batteries should be organized hierarchically, starting from broader, less specific tests that cover major domains of cognition and are sensitive to multiple deficits, and then proceeding to increasingly focused tests, e.g. for specific mechanisms of memory or attention. It would be desirable to increase the information density of the tests (i.e. the number of measures that can be obtained from a single test), as well as their flexibility, in the sense of capacity to tap into a broader spectrum of brain functions. What kind of data would constitute supportive evidence for distinct component disorders or subtypes within schizophrenia? Converging evidence from endophenotype-based studies suggests that measures of neurocognitive dysfunction provide the largest effect sizes among a host of “candidate” endophenotypes, being also cost-efficient for phenotyping large samples. In particular, several patterns of short-term and working memory impairment against a background of generalized cognitive deficit have been replicated across studies and are present in a substantial proportion of schizophrenia patients. Since most of the neurocognitive tests tap into several component processes, composite endophenotypes, integrating multiple neurocognitive measures, are more likely to capture variation that is genetically influenced than single-feature endophenotypes. The subtypes generated by such approaches should be capable of classifying individuals, rather than variables, and the resulting classification is likely to be polythetic (based on subsets of correlated features, rather than on the simultaneous presence of all defining attributes). Whether subtypes are discrete taxa, i.e. identifiable by zones of discontinuity with other subtypes; dimensional, representing continua with fuzzy boundaries; or hybrid (class-quantitative, with dimensions superimposed on discrete categories), is testable with taxometric methods common in biological classifications. In the context of genetic research, the most significant criterion of their validity will be the gain in predictive power and process understanding, in the sense of mechanistic explanation of disease phenomena.

54

Conclusion and Future Directions To sum up, the dissection of the syndrome of schizophrenia into modular endophenotypes with specific neurocognitive or neurophysiological underpinnings is beginning to be perceived as a promising approach in schizophrenia genetics. The current evidence is neither final nor static, and needs to be re-examined as new concepts and technologies coming from molecular genetics, neuroscience, cognitive science, or brain imaging bring forth new perspectives on disease causation and brain function. This must be complemented by a refined, reliable and valid phenotyping involving correlated neurobiological features. The study of endophenotypes cutting across the conventional diagnostic boundaries may reveal unexpected patterns of associations with symptoms, personality traits, or behaviour. The mapping of clinical phenomenology on specific brain dysfunction and genetics may in the future substantially recast the present nosology of psychiatric disorders.

References 1. Gottesman II, Gould TD. The endophenotype concept in psychiatry: etymology and strategic intentions. American Journal of Psychiatry 2003;160:636–645 2. Almasy L, Blangero J. (2001) Endophenotypes as quantitative risk factors for psychiatric disease: rationale and study design. American Journal of Medical Genetics (Neuropsychiatric Genetics) 2001;105:42–44 3. Sturtevant AH. A History of Genetics. Cold Spring Harbor Laboratory Press 2001/Electronic Scholarly Publishing Project (www.esp.org/books/sturt/history/readbook.html) 4. Gottesman II, Shields J. Genetic theorizing and schizophrenia. British Journal of Psychiatry 1973;122:15–30 5. Braff DL, Greenwood TA, Swerdlow NR, et al. Advances in endophenotyping schizophrenia. World Psychiatry 2008; 7:11–18 6. Sullivan PF. The genetics of schizophrenia. PLoS Medicine 2005;2:614–618 7. Doyle AE, Faraone SV, Seidman LJ, et al. Are endophenotypes based on measures of executive function useful for molecular genetic studies of ADHD? Journal of Child Psychology and Psychiatry 2005;46:774–803 8. Cannon TD, Keller MC. Endophenotypes in the genetic analysis of mental disorders. Annual Review of Clinical Psychology 2006;2:267–290. 9. Abrahams BS, Geschwind DH. Advances in autism genetics: on the threshold of a new neurobiology. Nature Reviews Genetics 2008;9:341–355

A. Jablensky 10. Jablensky A. Epidemiology of schizophrenia: the global burden of disease and disability. European Archive of Psychiatry and Clinical Neuroscience 2000;250:274–285 11. Botstein D, Risch N. Discovering genotypes underlying human phenotypes: past successes for mendelian disease, future approaches for complex disease. Nature Genetics 2003;33, Suppl:228–237 12. Harrison PJ, Weinberger DR. Schizophrenia genes, gene expression, and neuropathology: on the matter of their convergence. Molecular Psychiatry 2005;10:40–68 13. Allen NC, Bagade S, McQueen MB, et al. Systematic metaanalyses and field synopsis of genetic association studies in schizophrenia: the SzGene database. Nature Genetics 2008;40:827–834 14. Kendell R, Jablensky A. Distinguishing between the validity and utility of psychiatric diagnoses. American Journal of Psychiatry 2003;160:4–12 15. Bleuler E. Lehrbuch der Psychiatrie. Springer: Berlin; 1920 [reprinted English translation: Textbook of psychiatry. Arno: New York; 1976] 16. Gur RE, Calkins MN, Gur RC, et al. The Consortium on the Genetics of Schizophrenia: neurocognitive endophenotypes. Schizophrenia Bulletin 2007;33:49–68 17. Hartong DT, Berson EL, Dryja TP. Retinitis pigmentosa. Lancet 2006;368:1795–1809 18. Gaulton KJ, Willer CJ, Li Y, et al. Comprehensive association study of Type 2 diabetes and related quantitative traits with 222 candidate genes. Diabetes 2008 (PMID: 18678618) 19. Pharoah PD, Antoniou AC, Easton DF, et al. Polygenes, risk prediction, and targeted prevention of breast cancer. New England Journal of Medicine 2008;358:2760–2763 20. Matthysse S, Holzman PS, Gusella JF, et al. Linkage of eye movement dysfunction to chromosome 6p in schizophrenia: additional evidence. American Journal of Medical Genetics (Neuropsychiatric Genetics) 2004;128B:30–36 21. Holzman PS. Genetic latent structure models: implications for research on schizophrenia. Psychological Medicine 1987;17:271–274 22. Holzman PS. The role of psychological probes in genetic studies of schizophrenia. Schizophrenia Research 1994;13:1–9 23. Kaplan S, Herkovitz S, Shapiro E. A universal mechanism ties genotype to phenotype in trinucleotide diseases. PLoS Computational Biology 2007;3:e235 (PMCID: PMC2082501) 24. Cromwell RL, Elkins IJ, McCarthy ME, O’Neil TS. Searching for the phenotypes of schizophrenia. Acta Psychiatrica Scandinavica 1994;90 (Suppl 384):34–39 25. Meehl PE. Toward an integrated theory of schizotaxia, schizotypy, and schizophrenia. Journal of Personality Disorders 1990;4:1–99 26. Bearden CE, Freimer NB. Endophenotypes for psychiatric disorders: ready for primetime? Trends in Genetics 2006;22:306–313 27. Leboyer M, Belliver F, NostenBertrand, et al. Psychiatric genetics: search for phenotypes. Trends in Neurosciences 1998;21:102–105 28. Snitz BE, MacDonald AW, Carter CS. Cognitive deficits in unaffected first-degree relatives of schizophrenia patients: a meta-analysis review of putative endophenotypes. Schizophrenia Bulletin 2006;32:179–194

3

Challenging the Genetic Complexity of Schizophrenia

29. Turetsky BI, Calkins ME, Light G, et al. Neurophysiological endophenotypes in schizophrenia: the viability of selected candidate measures. Schizophrenia Bulletin 2007;33:69–94 30. Flint J, Munafo MR. The endophenotype in psychiatric genetics. Psychological Medicine 2007;37:163–180 31. Skuse DH. Endophenotypes and child psychiatry. British Journal of Psychiatry 2001;178:395–396 32. Owen MJ, Craddock N, Jablensky A. The genetic deconstruction of psychosis. Schizophrenia Bulletin 2007;33: 905–911 33. Kraepelin E. Psychiatrie. 8 Auflage. Barth:Leipzig 1909 [reprinted English translation: Dementia praecox and paraphrenia. Krieger: Huntington, New York; 1971] 34. Heinrichs RW. Meta-analysis and the science of schizophrenia: variant evidence or evidence of variants? Neuroscience and Biobehavioral Reviews 2004;28:379–394 35. Davidson M, Reichenberg A, Rabinowitz J, et al. Behavioral and intellectual markers for schizophrenia in apparently healthy male adolescents. American Journal of Psychiatry 1999;156:1328–1335 36. Zammit S, Allebeck P, David AS, et al. A longitudinal study of premorbid IQ scores and risk of developing schizophrenia, bipolar disorder, severe depression, and other nonaffective psychoses. Archives of General Psychiatry 2004;61:354–360 37. Heinrichs RW, Zakzanis KK (1998) Neurocognitive deficit in schizophrenia: a quantitative review of the evidence. Neuropsychology 1998;12:426–445 38. Joyce EM, Rosier JP (2007) Cognitive heterogeneity in schizophrenia. Current Opinion in Psychiatry 2007;20:268–272 39. Cirillo MA, Seidman LJ. Verbal declarative memory dysfunction in schizophrenia: from clinical assessment to genetics and brain mechanisms. Neuropsychology Review 2003;13:43–77 40. Paulsen JS, Heaton RK, Sadek JR, et al. The nature of learning and memory impairments in schizophrenia. Journal of the International Neuropsychological Society 1995;1:88–99 41. Turetsky BI, Moberg PJ, Mozley LH, et al. Memorydelineated subtypes of schizophrenia: relationship to clinical, neuroanatonical, and neurophysiological measures. Neuropsychology 2002;16:481–490 42. Dickinson D, Iannone VN, Wilk CM, et al. General and specific cognitive deficits in schizophrenia. Biological Psychiatry 2004;55:826–833 43. Egan MF, Goldberg TE, Kolachana BS, et al. Effect of COMT val108/158met genotype on frontal lobe function and risk for schizophrenia. Proceedings of the National Academy of Sciences 2001;98:6917–6922 44. Bilder RM, Volavka J, Csobor P, et al. Neurocognitive correlates of the COMT val158met polymorphism in chronic schizophrenia. Biological Psychiatry 2002;52:701–707 45. Paunio T, Tuulio-Henriksson A, Hiekkalinna T, et al. Search for cognitive trait components of schizophrenia reveals a locus for verbal learning and memory on 4q and for visual working memory on 2q. Human Molecular Genetics 2004;13:1693–1702 46. Hallmayer JF, Kalaydjieva L, Badcock J, et al. Genetic evidence for a distinct subtype of schizophrenia characterized by pervasive cognitive deficit. American Journal of Human Genetics 2005;77:468–476 47. Dickinson D, Gold JM (2008) Less unique variance than meets the eye: overlap among traditional neuropsychological dimensions in schizophrenia. Schizophrenia Bulletin 2008 [PMID:17986678]

55 48. Martin LF, Hall M-H, Ross RG, et al. Physiology and schizophrenia, bipolar disorder, and schizoaffective disorder. American Journal of Psychiatry 2007;164:1900–1906 49. Young DA, Waldo M, Ruttledge JH, et al. Heritability of inhibitory gating of the P50 auditory evoked potential in monozygotic and dizygotic twins. Neuropsychobiology 2001;33:113–117 50. Leonard S, Freedman R. Genetics of chromosome 15q13q14 in schizophrenia. Biological Psychiatry 2006;15:115–122 51. Greenwood TA, Braff DL, Light GA, et al. Initial heritability analyses of endophenotypic measures for schizophrenia. The Consortium on the Genetics of Schizophrenia. Archives of General Psychiatry 2007;64:1242–1250 52. Grunwald T, Butros NN, Pezer N, et al. Neuronal substrates of sensory gating within the human brain. Biological Psychiatry 2003;53:511–519 53. Cadenhead KS, Swerdlow NR, Shafer KM, et al. Modulation of the startle response and startle laterality in relatives of schizophrenic patients and in subjects with schizotypal personality disorder: evidence of inhibitory deficit. American Journal of Psychiatry 2000;157:1660–1668 54. Swerdlow NR, Sprock J, Light GA, et al. Multi-site studies of acoustic startle and prepulse inhibition in humans: initial experience and methodological considerations based on studies by the Consortium on the Genetics of Schizophrenia. Schizophrenia Research 2007;92:237–251 55. Anokhin AP, Heath AC, Myers E, et al. Genetic influences on prepulse inhibition of startle reflex in humans. Neuroscience Letters 2003;353:45–48 56. Swerdlow NR, Paulsen J, Braff DL, et al. Impaired prepulse inhibition of acoustic and tactile startle in patients with Huntington’s disease. Journal of Neurology, Neurosurgery and Psychiatry 1995;58:192–200. 57. Braff DL, Geyer MA, Swerdlow NR. Human studies of prepulse inhibition of startle: normal subjects, patient groups, and pharmacological studies. Psychopharmacology 2001;156:234–258 58. Castellanos FX, Fine EJ, Kaysen D, et al. Sensorimotor gating in boys with Tourette’s syndrome and ADHD: preliminary results. Biological Psychiatry 1996;39:33–41 59. Sobin C, Kiley-Brabeck K, Karayiorgou M. Lower prepulse inhibition in children with the 22q11 deletion syndrome. American Journal of Psychiatry 2005;162:1090–1099 60. Todd J, Michie PT, Schall U, et al. Deviant matters: duration, frequency, and intensity deviants reveal different patterns of mismatch negativity reduction in early and late schizophrenia. Biological Psychiatry 2008;63:58–64 61. Näätanen R, Jacobsen T, Winkler I. Memory-based or afferent processes in mismatch negativity (MMN): a review of the evidence. Psychophysiology 2005;42:25–32 62. Catts SV, Shelley AM, Ward PB, et al. Brain potential evidence for an auditory sensory memory deficit in schizophrenia. American Journal of Psychiatry 1995;152:213–219 63. Michie PT, Innes-Brown H, Todd J, Jablensky AV. Duration mismatch negativity in biological relatives of patients with schizophrenia spectrum disorders. Biological Psychiatry 2002;52:749–758 64. Hall MH, Schulze K, Rijsdijk F, et al. Heritability and reliability of P300, P50 and duration mismatch negativity. Behavioral Genetics 2006;36:845–857 65. Ahveninen J, Jääskeläinen IP, Osipova D. Inhertied auditory-cortical dysfunction in twin pairs discordant for schizophrenia. Biological Psychiatry 2006;60:612–620

56 66. Umbricht D, Krljes S. Mismatch negativity in schizophrenia: a meta- analysis. Schizophrenia Research 2005;76:1–23 67. Baker K, Baldeweg T, Sivagnanasundaram S, et al. COMT val108/158 Met modifies mismatch negativity and cognitive function in 22q11 deletion syndrome. Biological Psychiatry 2005;58:23–31 68. Linden DE. The P300: where in the brain is it produced and what does it tell us? Neuroscientist 2005;11:563–576 69. Hegerl U, Juckel G, Müller-Schubert A, et al. Schizophrenics with small P300: a subgroup with a neurodevelopmental disturbance and a high risk for tardive dyskinesia? Acta Psychiatrica Scandinavica 1995;91:120–125 70. Jeon YW, Pollich J. Meta-analysis of P300 and schizophrenia: patients, paradigms, and practical implications. Psychophysiology 2003;40:684–701 71. McCarley RW, Salisbury DF, Hirayasu Y, et al. Association between smaller left posterior superior temporal gyrus volume on magnetic resonance imaging and smaller P300 amplitude in first-episode schizophrenia. Archives of General Psychiatry 2002;59:321–331 72. Wright MJ, Hansell NK, Geffen GM, et al. Genetic influence on the variance in P3 amplitude and latency. Behavioral Genetics 2001;31:555–565 73. Porjesz B, Begleiter H, Wang K, et al. Linkage and linkage disequilibrium mapping of ERP and EEG phenotypes. Biological Psychology 2002;61:229–248 74. Blackwood DH, Fordyce A, Walker MT, et al. Schizophrenia and affective disorders – cosegregation with a translocation at chromosome 1q42 that directly disrupts brain-expressed genes: clinical and P300 findings in a family. American Journal of Human Genetics 2001;69:428–433 75. Curtis CE, Calkins ME, Iacono WG. Saccadic disinhibition in schizophrenia patients and their first-degree biological relatives A parametric study of the effects of increasing inhibitory load. Experimental Brain Research 2001;137: 228–236 76. Clementz BA, McDowell JE, Zisook S. Saccadic system functioning among schizophrenia patients and their firstdegree biological relatives. Journal of Abnormal Psychology 1994;103:277–287 77. Brownstein J, Krastoshevsky O, McCollum C, et al. Antisaccade performance is abnormal in schizophrenia patients but not in their biological relatives. Schizophrenia Research 2003;63:13–25 78. Malone SM, Iacono WG. Error rate on the antisaccade task: heritability and developmental change in performance among preadolescent and late-adolescent female twin youth. Psychophysiology 2002;39:664–673 79. Levy DL, O’Driscoll G, Matthyse S, et al. Antisaccade performance in biological relatives of schizophrenia patients: a meta-analysis. Schizophrenia Research 2004;71:113–125

A. Jablensky 80. Calkins ME, Curtis CE, Iacono WG, et al. Antisaccade performance is impaired in medically and psychiatrically healthy biological relatives of schizophrenia patients. Schizophrenia Research 2004;71:167–178 inconsistent data on relatives 81. Myles-Worsley M, Coon H, McDowell, et al. Linkage of a composite inhibitory phenotype to a chromosome 22q locus in eight Utah families. American Journal of Medical Genetics 1999;88:544–550 82. Meyer Lindenberg A, Weinberger DR. Intermediate phenotypes and genetic mechanisms of psychiatric disorders. Nature Reviews Neuroscience 2006;7:818–827 83. Gur RE, Keshavan MS, Lawries SM. Deconstructing psychosis with human brain imaging. Schizophrenia Bulletin 2007;33:921–931 84. Davatzikos C, Shen D, Gur RC, et al. Whole-brain morphometric study of schizophrenia revealing a spatially complex set of focal abnormalities. Archives of General Psychiatry 2005;62:1218–1227 85. Davidson LL, Heinrichs RW. Quantification of frontal and temporal lobe brain-imaging findings in schizophrenia: a metaanalysis. Psychiatry Research Neuroimaging 2003;122:69–87 86. Glahn DC, Thompson PM, Blangero J (2007) Neuroimaging endophenotypes: Strategies for finding genes influencing brain structure and function. Human Brain Mapping 2007;28: 488–501 87. Bearden CE, van Erp TGM, Thompson PM, et al. Cortical mapping of genotype-phenotype relationships in schizophrenia. Human Brain Mapping 2007;28:519–532 88. Jablensky A. Subtyping schizophrenia: implications for genetic research. Molecular Psychiatry 2006;11:815–836 89. Manton KG, Woodbury MA, Tolley DH. Statistical Applications Using Fuzzy Sets. Wiley, New York; 1994 90. Badcock JC, Badcock DR, Read C, et al. Examining encoding imprecision in spatial working memory in schizophrenia. Schizophrenia Research 2007;100:144–152 91. Price GW, Michie PT, Johnston J, et al. A multivariate electrophysiological endophenotype, from a unitary cohort, shows greater research utility than any single feature in the Western Australian Family Study of Schizophrenia. Biological Psychiatry 2006;60:1–10 92. Craddock N, Owen MJ. Rethinking psychosis: the disadvantages of a dichotomous classification now outweigh the advantages. World Psychiatry 2007;6:84–91 93. Moldin SO. Indicators of liability to schizophrenia: perspectives from genetic epidemiology. Schizophrenia Bulletin 1994;20:169–184. 94. Gerlai R. Phenomics: fiction or the future? Trends in Neuroscience 2002;10:506–509 95. Tan H-Y, Callicott JH, Weinberger DR. Intermediate phenotypes in schizophrenia genetics redux: is it a no brainer? Molecular Psychiatry 2008;13:233–238

Chapter 4

Translational Medicine: Functional Biomarkers for Drug Development of “Cognitive Enhancers” in Schizophrenia Georg Winterer

Abstract Biomarkers are the bridge between basic and clinical research and enable us to speed up the process of drug development. In the process of drug development of “cognitive enhancers” for the treatment of schizophrenia, the usefulness of biomarkers is increasingly realized. In this chapter, we will introduce the concept of biomarkers and discuss its application with regard to drugs that are developed for the purpose of cognitive enhancement in schizophrenia. While focussing on functional biomarkers, i.e., functional magnetic resonance imaging (fMRI) and electrophysiology (EEG), it will be clearly delineated what is currently feasible, what kind of limitations still exist and what needs to be done in the future to optimise the available functional biomarkers. In addition, those functional biomarkers are listed which are thought to qualify as biomarkers for cognitive deficits in schizophrenia or which are at least close to what one may call a valid biomarker for drug development. Keywords Schizophrenia • translational medicine

•

cognition

•

biomarkers

Abbreviations FDA Food and Drug Administration; EMEA European Medicines Agency; NIH National Institutes of Health; NIMH National Institute of Mental Health; MATRICS Measurement and Treatment Research to Improve Cognition in Schizophrenia; IND Investigational new drug; MRI Magnetic resonance

G. Winterer () Department of Psychiatry, Heinrich-Heine University, Duesseldorf, Germany and Institute of Neurosciences and Biophysics, Juelich Research Centre, Juelich, Germany.

imaging; fMRI Functional magnetic resonance imaging; BOLD Blood oxygenation level dependent; EEG Electrophysiology; ERP Event-related potentials; MEG Magnetoencephalography; PSP Postsynaptic potential; PET Positron emission tomography; SPET Single-photon emission tomography; MRS Magnetic resonance spectroscopy; ICA Independent component analysis; MNI Montreal Neurological Institute; ROI Region of interest; ECG Electrocardiogram; ASL Arterial spin labelling; pharmfMRI Pharmacological fMRI; PFC Prefrontal cortex; DA Dopamine; MMN Mismatch negativity; P300 Event-related potential with positive amplitude around 300 ms after stimulus; COMT Catechol-O-methyltransferase;

Introduction Schizophrenia is a major mental disorder characterized by positive and negative symptoms, as well as persistent neurocognitive deficits. All currently approved medications for schizophrenia function by blocking dopamine D2 receptors, and have proved to be effective primarily against positive and – to some extent – against negative symptoms. By contrast, cognitive symptoms frequently persist, and lead to disability and poor long-term outcome.1 As of today, all available drugs for the treatment of schizophrenia have only little – if any – impact on this symptom complex. In order to test novel compounds for their effects on the clinical endpoint cognitive performance, the translational medicine approach is now increasingly considered to be the way to go. In the case of drug discovery and development, translational medicine – as defined in Wikipedia – typically refers to the

M.S. Ritsner (ed.), The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes, © Springer Science + Business Media B.V. 2009

57

58

“translation” of basic research into real therapies for real patients. The emphasis is on the linkage between the laboratory and the patient’s bedside, without a real disconnect. This is often called the “bench to bedside” definition. However, translational medicine can also have a much broader definition, referring to the development and application of new technologies in a patient driven environment – where the emphasis is on early patient testing and evaluation. When it comes to the development of novel drugs for the treatment of cognitive deficits in schizophrenia, the translational medicine concept requires some explanation. Until recently, using behavioral tasks which involve more or less specific cognitive domains including attention, working memory or verbal memory as outcome measures for drug effects have been the classical approach. More recently, however, functional biomarkers obtained from functional magnetic resonance imaging (fMRI) and electrophysiology (EEG) are also increasingly considered to be implemented in drug trials of “cognitive enhancers”. Among several reasons outlined in more detail below, the perhaps most important rationale for using biomarkers is related to the recognition that the process of traditional outcome-based drug discovery and development can be long, expensive and uncertain.2–4 Knowledge as obtained from biomarkers at an early point in time during this process may help to speedup decision-making about candidate molecules and therapeutic concepts which will facilitate the development process by ensuring that if we are to fail, we fail fast.5 The use of any kind of biomarkers requires an awareness of and a reasonable integration in already available traditional concepts of drug development. In the traditional concept of drug development of cognitive enhancers, cognitive tasks as clinical endpoints need to fulfill particular requirements in order to be accepted by the national and international bodies for the approval of drugs such as the U.S. Food and Drug Administration (FDA) or the European Medicines Agency (EMEA). Therefore, considerable efforts have been recently undertaken to meet these requirements. A prime example in this regard is the NIMH (National Institute of Mental Health) initiative titled “Measurement and Treatment Research to Improve Cognition in Schizophrenia” (MATRICS) which was designed to support the development of pharmacological agents for improv-

G. Winterer

ing the neurocognitive impairments that are a core feature of schizophrenia (http://www.matrics.ucla. edu/matrics-recommendations-frame.htm). It is the explicit goal of MATRICS to advance the development of drugs for treating these impairments in schizophrenia by addressing important obstacles that have interfered with drug development. Among these barriers were: 1. A lack of a consensus on the best measure of neurocognitive deficits for clinical trials 2. Concerns regarding the adequacy of neuropsychological test scores as sole endpoints in clinical trials and the potential need for co-primary measures of functioning 3. The lack of a consensus regarding the appropriate design of trials for registering an agent for treating cognitive deficits 4. A lack of consensus regarding the best approaches for identifying promising compounds 5. Concerns regarding obstacles that interfered with collaborations among government, industry, and academia Acknowledging these obstacles, it was agreed that several essential criteria need to be satisfied for a consensus cognitive battery for clinical trials in schizophrenia: 1. Reliable and valid assessment of cognition at the level of all individual major cognitive domains 2. Inclusion of the following cognitive domains: speed of processing, attention/vigilance, working memory, verbal learning and memory, visual learning and memory, reasoning and problem solving, and social cognition 3. High test-retest reliability 4. High utility as a repeated measure 5. Demonstrated relationship to functional outcome 6. Demonstrated tolerability and practicality Behavioral measures of cognition do have, however, a number of inherent disadvantages when used for drug development that cannot be easily overcome without using co-primary functional measures – even in the ideal case that all of the above requirements are met. First, most of the above named cognitive domains are poorly phylogenetically conserved which makes it difficult to use animal models within a translation research framework for drug development.6,7 According to Sherwood et al.,7 the modern human mind may be

4

Translational Medicine

conceived as a mosaic of traits inherited from a common ancestry with our close relatives, along with the addition of evolutionary specializations within particular domains. These modern human-specific cognitive and linguistic adaptations appear to be correlated with enlargement of the neocortex and related structures. Accompanying this general neocortical expansion, certain higher-order unimodal and multimodal cortical areas have grown disproportionately relative to primary cortical areas. Anatomical and molecular changes have also been identified that might relate to the greater metabolic demand and enhanced synaptic plasticity of modern human brains. Finally, the unique brain growth trajectory of modern humans has made a significant contribution to our species’ cognitive and linguistic abilities. Thus, these developmental changes can pose considerable difficulties with regard to the “translation” of basic research into real therapies for real patients. Therefore, difficulties arise in the decisionmaking process with regard to the investigational new drug (IND) application when trying to extrapolate results from preclinical animal studies to the use in humans. Second, behaviorally assessed cognitive domains are usually highly complex in terms of the number and interaction of involved neural circuits and alternative processing modes of these circuits. This is even true for relatively simple behavioral measures like processing speed in reaction time tasks which involves peripheral nerve conduction velocity plus different processing modes in several brain circuits including regions for the regulation of vigilance and attention, for primary and secondary stimulus processing as well as for premotor and motor regions.8 To make things even more difficult, deficiencies in one circuit may be compensated by a higher engagement of another circuit as exemplified by recent fMRI studies of schizophrenia patients.9 Similar compensation mechanisms have been described, for instance, for Alzheimer dementia10 and for Multiple Sclerosis – the latter being a particularly good example because it has been repeatedly demonstrated that the magnetic resonance imaging (MRI) “lesion load” throughout the brain is mostly poorly correlated with neurological and cognitive performance in this illness because symptoms rather depend on the “reserve capacity” of the brain.11,12 MRI lesion outcomes have therefore not been regarded by either regulatory authorities or expert investigators13 as adequate to serve as primary outcome measures in definitive Phase 3 trials. In an analogous way, quite

59

heterogenous conditions in the brain may be responsible for abnormal performance in a certain cognitive domain of schizophrenia patients which itself is considered being a clinically heterogeneous disorder. Thus, given that numerous brain circuits have been demonstrated to be more or less affected in schizophrenia, it is quite likely that abnormalities in different brain circuits may contribute to the overall performance in a particular cognitive domain. In behavioral studies of cognitive performance, novel drugs that target a particular brain circuit therefore may be easily dismissed as ineffective during Phase 1 and 2 studies with usually small subject samples investigated because of a lack of statistical power. By extension, one could argue that behavioral measures are most useful in large sample Phase 3 studies while clear limitations exist with respect to early proof-of-concept required for successful Phase 1b/2a studies submitted to the regulatory agencies. The limitations of behavioral measures during Phase 1/2 for the development of cognitive enhancers are now increasingly realized. It has been proposed that a possible way to overcome these limitations is to use functional biomarkers to measure brain function directly. For instance, on September 9–10, 2004, MATRICS convened representatives from NIMH, academia, and industry in Potomac, Maryland, for the final of a series of consensus-oriented conferences. During this 2-day meeting, there were intensive discussions thematically guided by the overall theme – “New Approaches to Assessing and Improving Cognition in Schizophrenia” – with several breakout groups for more focused discussions on functional neuroimaging and electrophysiology biomarkers.14 Discussing first what defines and distinguishes surrogate markers and biomarkers, it was agreed upon that the term surrogate marker implies any marker that would be acceptable in the place of a primary index of a disturbance (clinical endpoint), e.g. any electrophysiological or neuroimaging measure of brain function which is clearly related to cognitive performance. Accordingly, this definition of a surrogate measure essentially follows earlier suggestions whereby a surrogate marker is defined statistically as a response variable for which a test of the null hypothesis of no relationship to the treatment groups under comparison is also a valid test of the corresponding null hypothesis based on the true endpoint.15 There was also a general agreement that beyond the denition of a surrogate

60

endpoint measure a good biomarker would not merely show a statistical relationship with the clinical endpoint but that it would also reveal something about the disease mechanisms and that a desirable property would be that it permits a more efficient prediction of an endpoint than the measurement of the endpoint itself. Along this line, a reasonable biomarker could be used for proof-of-principle and proof-of-concept studies. In the field of drug research, a company presented with a project or proposal to develop a IND will often undertake internal research initially, to prove that the core ideas are workable (proof-of-principle) and feasible, before going further. This use of proof of concept helps establish viability, technical issues, and overall direction. Among several participants of the meeting, however, it was felt that none of the present indexes of cognition-related brain function would, in any strict sense, qualify as a biomarker which brings up the question what is the definition of a biomarker. An NIH (National Institute of Health) study group committed to the following definition in 199816: “a characteristic that is objectively measured and evaluated as an indicator of normal biologic processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention.” Accordingly, this definition is relatively broad encompassing both physiological indicators (functional biomarkers) such as blood pressure, heart rate and brain function measures as well as molecular measures (molecular biomarker) like prostate specific antigen, liver function tests and enzyme or gene expression assays. Thus, it is obvious that under this definition cognition-related brain function indices may principally qualify as functional biomarkers. The critical question in this context, however, is whether any given biomarker is a useful biomarker. With regard to drug development, participants from industry on the MATRICS meeting clearly expressed what they consider to be useful, i.e., reliable measures of brain activity associated with cognition that would be sensitive to changes with medications that were predictive of clinical response.

Functional Biomarkers: The Methods Functional neuroimaging provides a direct way of investigating the pathophysiology of schizophrenia and drug effects on cognition-related biomarkers

G. Winterer

in vivo.17 Whereas the effects of neurotransmitters implicated in schizophrenia, such as dopamine and glutamate, can be directly assessed on the molecular level using molecular imaging techniques such as positron emission tomography (PET), single-photon emission tomography (SPET) and magnetic resonance spectroscopy (MRS), brain function, particularly when associated with the cognitive processes and symptoms associated with the disorder, can be studied using neurophysiological measures like EEG or event-related potentials (ERP) and fMRI which are currently the most widely applied functional biomarkers for drug development of cognitive enhancers in schizophrenia.

Functional Magnetic Resonance Imaging (fMRI) fMRI assays brain activity based on the blood oxygenation level in microcirculation, an indirect measure of brain activity (Blood Oxygenation Level Dependent or BOLD fMRI). The BOLD signal arises because a neuronal event triggers an increase in blood supply to the surrounding capillary beds which overcompensates for oxygen extraction in the corresponding neuronal circuit. This results in a relative change of oxy- versus deoxyhemoglobin, i.e., it leads to a higher oxyhemoglobin concentration while the deoxyhemoglobin concentration is decreasing. As compared to deoxyhemo globin, oxyhemoglobin is less paramagnetic and therefore, the MR signal intensity is increasing in this brain region. An increase of MR intensity is generally interpreted as an increase of neuronal activity. However, it is important to acknowledge at this point that fMRI hemodynamic BOLD signals – like scalp-recorded electromagnetic (EEG) activity (see below) – are primarily derived from postsynaptic potential (PSP) activity and not neuronal discharge, since in humans PSP activity is metabolically most demanding on the neuron and also evokes most of the blood flow changes.18 Thus “activation” as is commonly used in functional neuroimaging mainly reflects PSPs (often from input to the region) and not neuronal discharge (spike rate). In an fMRI experiment, subjects are put into the MRI scanner and are typically required to respond by button press to a series of visual stimuli, although more

4

Translational Medicine

recently researchers have started to apply stimuli in different modalities (e.g. pain, auditory) or to conduct resting-state fMRI19 where no stimuli are presented at all. When stimuli are applied, the temporal profile of stimulus presentation is correlated with the temporal changes of the BOLD signal at each single point (voxel) in the brain image (1. level analysis). A high correlation – as expressed by t-scores or Z-scores – is considered equivalent with a strong BOLD-response. The resulting statistical maps across voxels are then color-coded and frequently overlaid on anatomical images of the same subject. For optimized localization and quantification of the BOLD-signal as well as cross-correlation/co-variance (connectivity) of BOLD-responses between brain areas, additional – more sophisticated statistical analyses can also be applied including independent component analysis (ICA) or functional connectivity analyses. To aid comparisons between subjects or subject groups (2. level analysis), the 3D image of each brain is then transformed so that superficial structures line up, a process known as spatial normalization. Such normalization typically involves not only translation and rotation of a subject’s brain image, but also scaling and nonlinear warping of the brain surface to match a standard template; e.g. Talairach-Tournaux brain map or the more commonly used MNI (Montreal Neurological Institute) template. However, since the anatomical variation between individual brains can be considerable, a major difficulty arising from the spatial normalization procedure is that it may strongly distort the magnitude of each individual’s BOLD-response. As a result, it can sometimes be hard to statistically compare two different subject groups (e.g. receiving two kind of drugs). Therefore, population probability maps such as the Harvard-Oxford cortical and subcortical structural atlases or Jülich histological (cyto- and myelo-architectonic) atlas have become increasingly popular which define “probabilistic” regions of interest (ROI) which then can be used as a basis of a more accurate group comparison. Lately, some researchers have also started to “correct” a subject’s BOLD-response directly for volumetric characteristics. The gold standard, however, still remains to delineate individual ROIs based on anatomical information in each subject and to quantify the BOLD-response within that ROI for subsequent statistical analyses.

61

Neurophysiology Neurophysiological measures are mostly recorded in humans using standard EEG techniques in which an array of up to 256 electrodes are affixed to specific scalp locations and the pattern of electrical activity is monitored while subjects participate in specific experimental paradigms. Scalp recorded electrical activity primarily represents spatial and temporal summation of synchronous current flow through postsynaptic dendritic membranes, i.e., postsynaptic potentials (PSP) of cortical pyramidal neurons that, because of their parallel alignment in the cortex and asymmetric morphology, serve as ‘open field’ generators in the brain and give rise to local field potentials. The basic physiological data underlying most neurophysiological biomarkers that are presently available are similar to that used in qualitative or quantitative EEG assessment. A critical difference is the use of signal averaging. Brain responses to individual sensory, cognitive or motor events are small relative to the background EEG, whereas signal averaging permits such events to stand out. Neurophysiological biomarkers are typically time-locked to sensory events (for example, auditory or visual stimuli). Nevertheless, other types of events can also be defined. For example, in response tasks, electrical responses can be back-averaged from motor responses and can be used to analyse preceding activity. However, the absence of electrical activity cannot be taken as evidence of the lack of response of a given region as the majority of electrical activity generated in the brain is invisible to surface recordings. Even so, over recent years, neurophysiological measures that probe brain activity in regions that are of relevance to schizophrenia have been developed.

Time Versus Frequency Domain Analyses Once neurophysiological data have been recorded, two complementary analytic approaches are used for data analysis. In the first approach, brain activations are viewed simply as a series of amplitude deflections that vary in time and space over the surface of the scalp. Analysis of response amplitude over time has traditionally been referred to as a time-domain analysis or, more accurately, as time-amplitude domain analysis. The sequence of deflections triggered by a given event

62

is referred to as an ERP. The second approach views brain activity as a sum of superimposed oscillations maintained within and between brain regions giving rise to event-related evoked or induced oscillations.20 This approach, also known as time-frequency analysis, captures more information on underlying brain activity than the traditional ERP approach, but is computationally more complex and less standardized across subjects. Notably, recently developed computational models of drug effects on clinical symptoms in schizophrenia are based on the time-frequency rather than time-amplitude approach.21 Once matured, these in silico models may offer considerable advantages in the future over animal models since drug effects can be modelled under more human-like conditions.

Source Localization For both time and frequency domain analyses, additional data concerning underlying mechanisms can be obtained using source analysis approaches that evaluate the electrical sources of scalp-derived activity within the brain. Source localization approaches make specific assumptions regarding the propagation of electrical activity through the brain and scalp, which may lead to localization imprecision. In practice, however, good spatial agreement is typically obtained between fMRI and ERP22 or oscillationbased8 measures.

Magnetoencephalography Although most research with neurophysiological biomarkers has been performed using electrically based techniques, magnetoencephalography (MEG) can be used instead to detect the magnetic fields that accompany electrical current flow through the brain. As magnetic fields are less susceptible to distortion by volume conduction of currents than electrical fields, reconstruction of MEG sources is possible with higher spatial accuracy than those of EEG. Furthermore, MEG offers the advantage of avoiding cumbersome electrode fixation to the scalp. As with EEG-based measures, MEG recordings can be analysed in either the time or the frequency domain. As a result of the electromagnetic field distribution, however, MEG is limited by its insensitivity to sources radial to the scalp

G. Winterer

surface (that is, pointed perpendicular, rather than parallel, to the skull surface) and by its cost: state-of-theart MEG equipment is still considerably more expensive than EEG-equipment and thus its use is currently limited to a few research centers.

EEG and fMRI: Conceptual Issues EEG and fMRI mutually complement each other for several reasons. fMRI provides a relatively good spatial resolution in the range of millimeters. However, in current psychiatric imaging the temporal resolution of fMRI are in the second range, although new advances in modeling the BOLD response23 and in multiple channel acquisition24 suggest the as yet unrealized possibility of resolutions in the hundred millisecond range in the future. This current temporal resolution is several thousand-fold less than electrophysiological measures, which provide a time resolution within a few milliseconds (ms). Moreover, more advanced electrophysiological time-frequency analyses allow the measurement of oscillatory brain patterns1 which are not amenable to BOLD fMRI and which provide information on qualitatively distinct processing modes of certain brain circuits. In contrast to its superior temporal resolution and decomposition of frequency-dependent oscillatory patterns, the EEG’s spatial resolution is limited, since the source of the voltage fluctuations at the scalp can not be precisely identified as coming from particular brain regions (so-called “inverse-problem”). However, for many commonly applied electrophysiological paradigms used for instance in schizophrenia research, extensive characterization of intracerebral electromagnetic sources has been conducted25 which greatly facilitates the assignment of scalp-recorded electrophysiological measurements to their origin of generation. Also, over the past decade, increasingly sophisticated source analysis approaches have been developed that evaluate the electromagnetic sources of scalp-derived activity within the brain. Still, prior validation of these methods must be through some other method, such as intracranial recording, association with structural MRI “lesions” or with functional MRI. Validation of an ERP source solution with fMRI also can be achieved by varying a task such that a particular ERP component in a sequence of ERPs is selectively emphasized. The corresponding local

4

Translational Medicine

change(s) of the BOLD response – ideally under identical task conditions – would be the validation of the source. This is although the BOLD-response will occur seconds after the electrical event. In many cases, however, reliance on combined ERP/fMRI – or alternatively MEG/fMRI measurements – will not be sufficient for validation requiring expert knowledge taking into account intracranial recordings and/or lesion studies. Even so, one may use both modalities to “cross-validate” e.g. task and drug effects in particular brain regions. There are further issues that need to be taken into account in the decision-making process what functional measure to use. A practical problem of great concern is the availability of reliability estimates for specific functional measures which is considered to be crucial for drug studies as pointed out by the MATRICS initiative (s.a.). While numerous studies have confirmed moderate to high test-retest reliability for most ERP- and EEG measures including those which typically used in schizophrenia studies (e.g.,26 relatively little work has been conducted so far in the fMRI field27 although most fMRI researchers would agree that reliability is mostly sufficient. Again, however, reliability issues suggest that one may consider the use of both EEG/ERP and fMRI modalities for cross-validation. An analogous problem exists with regard to the question in as much EEG/ERP and BOLD fMRI is phylogenetically conserved under certain task conditions – an issue that is of particular relevance in the context of translational studies when it is the aim to extrapolate drug effects from animal research to human research.28 As far as electrophysiological research is concerned numerous studies have confirmed over the past decades phylogenetical conservation for most ERPs (e.g.1,25). Electrophysiological experiments in animals (e.g. mice, rats, cats and monkeys) are usually conducted with implanted electrodes but can also be obtained from the scalp-surface. These experiments, which encompass electrophysiological experiments that are typically conducted in schizophrenia patients as well, can be performed in many cases in the awake and mobile animal. Therefore, preclinical electrophysiological animal experiments are highly recommended since these investigations – among other reasons – may greatly facilitate a comparison in subsequent Phase 1 and 2 studies when using functional biomarkers in humans; a comparison which may not be possible when only having conducted behavioral animal stud-

63

ies. With regard to fMRI, very little data exist on the question of phylogenetic conservation – in fact most animal research using MRI so far has been based on structural analyses or MR-spectroscopy rather than on functional analyses. Even so, Borsook et al.5 recently suggested that in preclinical phases, fMRI could be used to focus the selection of clinically relevant CNS animal assays, to enhance target validation based on objective assessment of behavior, to increase the speed of lead optimization based on functional CNS responses, and to define circuits of action based on the neuroanatomy of the “functional signature”, i.e., the functional circuit response. The examples given at this point primarily referred to fMRI-studies conducted in awake monkeys. In fact, a number of successful studies have now been conducted in monkeys using for instance visual object and face recognition, learning tasks or observation of object moving.28–32 Furthermore, successful experiments also have been conducted in small animals with fMRI. For instance, Febo et al.33 performed fMRI in awake rats to identify the neuronal circuits affected by repeated cocaine administration. Rats were given an injection of cocaine or its vehicle for 7 days, abstained from injections for 1 week, and challenged with an intracerebroventricular cocaine injection during functional imaging. Acute cocaine produced robust positive BOLD responses across wellknown monoamine-enriched brain regions, such as the prefrontal cortex, nucleus accumbens, dorsal striatum, sensory cortex, hippocampus, thalamus, and midbrain areas. Repeated cocaine administration resulted in lower BOLD responses in the prefrontal cortex, agranular insular cortex, nucleus accumbens, ventral pallidum, and dorsomedial thalamus, among other brain regions. Reductions in BOLD intensity were not associated with variations in cerebrovascular reactivity between drug naive rats and those repeatedly exposed to cocaine. Therefore, the lower metabolic activation in response to cocaine was interpreted to reflect a reduced neuronal and/or synaptic activity upon repeated administration. Despite these first successful experiments, however, a multitude of methodological problems still awaits a solution. For instance, animals – particularly small animals – frequently need to be immobilized or at least lightly anesthetized in the scanner environment. While the former exerts a tremendous stress on the animal and clearly limits the ability of an animal to respond to the stimuli, the latter is a pharmacological intervention which may make it

64

impossible to conduct an experiment with external stimulus application as of the reduced vigilance-related diminished if not abolished responsiveness of the brain. Moreover, any pharmacological intervention (e.g. anesthesia) – in addition to the drug under investigation – gives rise to difficulties with regard to any comparison with human functional data. To some extent, however, these difficulties may be eventually overcome by adapting the animal to the scanner environment and by using resting-state fMRI without cognition-related task conditions (such as in the experiment with rats conducted by Febo et al.33 It is noteworthy in this context, that under certain circumstances, restingstate fMRI has been demonstrated in humans to be directly related to the stimulus- and task evoked BOLD-response.19 Even so, it is hardly conceivable that complex task paradigms which are typically investigated in schizophrenia patients (e.g. n-back working memory or related tasks) can be implemented in the scanner with animals – not only because of the technical difficulties involved but simply because of the task complexity which is demanding too much of the animal. The problem of translational research with fMRI is further complicated by the presumably different physiological origin of the BOLD-response in small animals as compared to humans – even when leaving aside the important question how much any task-involved brain circuits are phylogenetically conserved with respect to the engaged brain regions. For instance, Attwell and Iadecola34 recently offered an explanation for why the BOLD-response in small animals and humans may reflect different physiological events. Thus, the BOLD-response in humans correlates more closely with synaptic activity than with neuronal firing (action potentials) as compared to small animals where the opposite is the case – an explanation, which is based on estimations of the cortical gray matter energy budget. In rodents, 47% of the energy used on signaling is predicted to support action potentials and 34% to support postsynaptic potentials. However, in primates the developmentally acquired greater number of synapses reverses this relationship and postsynaptic currents account for 74% of the total energy used on signaling while action potentials are thought to consume only about 10% of the signaling energy. Occasionally made claims in this context that the neuronal firing rate predicts synaptically generated PSPs and that therefore, BOLD indirectly reflects the spike rate do not really help because they represent an oversimplification.21

G. Winterer

Another critical issue to be considered is that BOLD fMRI does not measure neural activity directly, but relies on a cascade of physiological events linking neural activity to the generation of MRI signal (neurovascular coupling). However, most of the disease and pharmacological studies performed so far have interpreted changes in BOLD fMRI as “brain activation” ignoring the potential confounds that can arise through drug- or disease-induced modulation of events downstream of the neural activity.35 This issue is especially serious in diseases (like multiple sclerosis, brain tumours and stroke) and drugs (like anaesthetics or those with a vascular action) that are known to influence these physiological events. It is therefore important to identify, characterize and possibly correct these additionally influences that potentially confound the results from BOLD fMRI studies. Although it might well be that such potential confounds are negligible under many conditions, changes of drug-induced vascular reagibility/blood pressure and/or drug effects on the biochemical mechanisms which mediate signalling between neurons/glia cells on one hand and blood vessels on the other hand may all make it difficult to correctly quantify and interpret drug effects on neural activity. Therefore, Iannetti and Wise35 recently suggested a series of experimental measures to improve the interpretability of BOLD fMRI studies: 1. The inclusion of one or more control tasks to explore the function of a brain system which is not expected to be modulated by the drug under investigation in order to assess possible global modulations of signalling and/or vascular reactivity induced by the drug. 2. The recording of physiological parameters during scanning (e.g. ECG, blood pressure, respiration rate, and movements) which – to some extend – allows to assess indirect systemic effects on BOLD response and to correct for physiological noise. 3. Separate experiments to measure baseline and stimulus perfusion scans (e.g. with arterial spin labeling, ASL) which again would help to estimate possible global modulations of signalling and/or vascular reactivity induced by the drug. 4. Vascular reactivity could be relatively specifically assessed by combining drug administration with CO2-challenge. 5. Animal experiments are suggested in those cases when signalling between neurons/glia cells vascular response needs to be examined which is difficult to assess in humans.

4

Translational Medicine

Iannetti and Wise,35 however, acknowledge that this research field is constantly evolving and that a perfect list of corroborative experiments does not exist. In fact, there are a number of problems with the approach outlined by Iannetti and Wise.35 While using a control task and assessing physiological parameters are certainly useful recommendations which can be implemented with little effort, the other suggestions clearly have their limitations both in terms of practicality and/or meaningfulness. For instance, measuring baseline and stimulus-induced perfusion is time consuming and does not allow a differentiation between signalling and/or vascular reactivity induced by the drug. In those cases, when a drug effect is found, it needs to be assessed as a covariate in each study participant – ideally in a separate session in order to account for timedependent pharmacokinetic and pharmacodynamic effects. Accordingly, using perfusion scans and/or assessing the BOLD CO2-response or conducting addtional animal experiments are presumably most useful to exclude any drug effects on signalling and vascular reactivity prior to the proper pharmacological fMRI (pharmfMRI) study in humans. In any case, time and feasibility will be critical factors. For instance, a patent on a investigational new drug starts to run immediately after IND approval. Accordingly, time consuming procedures during phase 1/2a clinical trials represent a serious limitation. In addition, any additional task during a drug testing session to control for drug effects on signalling and/or vascular reactivity would diminish the time available for testing the drug on cognition-related brain function – a problem which may even be exaggerated when two imaging sessions are conducted on the same day. Therefore, it appears to be a more elegant and smarter approach to obtain an additonal measure of neuronal activity using EEG- or MEG-recordings for cross-validation of any drug effects on the BOLD-response. If there is good correspondence between drug modulation in both modalities, this would strongly support the hypothesis that the difference disclosed with the BOLD technique has a significant neural component. If not, one still could conduct the above proposed measurements. Iannetti and Wise35 suggested to do EEG/MEG measurements during separate sessions rather than simultaneously. They argue that the complex set-up of simultaneous fMRI/EEG would prohibit this approach to become a matter of routine. However, technical advances and practical experience with this approach during the past years have made it increasingly easy to collect data

65

simultaneously in both modalities.36 In fact, there are now good reasons to prefer simultaneous measurements over separate sessions. For instance, in placebocontrolled cross-over designs, which are typically conducted in phase 1/2a studies, separate sessions are not very practical because of the time required when taking into account pharmacokinetic/pharmacodynamic considerations and/or repeated measure effects on brain response. Moreover, simultaneous fMRI/EEG offers the advantage to avoid between-sessions effects on the subject and environmental side (between-sessions variability). Also, the simultaneous approach better accounts for non-stationary brain responses (including fluctuations of vigilance) during the task, thus gaining additional statistical power37 – a critical issue in phase 1/2a studies with typically small samples of subjects being investigated.

Schizophrenia and Functional Biomarkers On the MATRICS meeting in 2004 (s.a.), it was recommended to apply functional co-primary measures in the place of a primary index of a cognitive disturbance (clinical endpoint), e.g. any electrophysiological or neuroimaging measure of brain function, with a clear relationship to cognitive performance. While this recommendation intuitively appears to be straight forward, one should keep in mind that this recommendation is somewhat biased because it implicitely takes a top-down approach to brain function. In reality, however, there is no 1:1 look-up table for any functional measure with behaviorally assessed measures – only a statistical relationship. In fact, this statistical relationship will become weaker when dissecting the functional components that contribute to a behavioral measure as of the complexitiy of behavioral measures with respect to the multitude of involved functional components as outlined above. Therefore, in many cases – and in particular when small are subject samples are investigated – one cannot expect to cross-validate the effect of a drug on functional measures with the drug effect on behavior because of statistical power issues. Consequently, the relationship of a given functional measure with a certain behavioral measure needs to be established prior to the drug trial. On the other hand, there are differences between imaging

66

G. Winterer

modalities. It is, for instance, a great advantage of fMRI that the cognitive paradigms used in fMRI investigations have mostly been adaptations of paradigms that have already been intensively investigated for decades on the behavioral level in schizophrenia. Therefore, is relatively easy to relate measures obtained from functional fMRI with behavioral outcome. In contrast, electrophysiological schizophrenia research followed its own path – i.e., it was less influenced by the cognitive approach in schizophrenia research. As a result, it is sometimes difficult to reconcile the two levels of investigation with each other because of differing concepts and terminology. Subsequently, several examples (Table 4.1) – not to be considered a complete list of all potential functional biomarkers – will be provided from the fMRI and EEG domains where the relationship between cognitive performance and brain function has been demonstrated and which are thought to fulfill additional requirements as useful functional biomarkers for cognitive deficits in schizophrenia.

Photon Emission Tomography (SPET) studies which measured blood flow directly.41,42 As explained in more detail below, these fMRI-measures – in part – are currently achieving the maturity to be used as functional biomarkers in schizophrenia research. In addition, there are promising approaches which may provide us with such biomarkers in the very near future. Subsequently, we will limit the discussion to the cognitive domains working memory and selective attention because most pharmaco-fMRI-studies have been conducted in these domains. Also, many of the problems that still exist with these measurements when it comes to pharmaco-fMRI are found in very similar ways in other cognitive domains such as verbal memory, social cognition etc. Prefrontal Cortex: Working Memory and Selective Attention The allocation of attentional resources to portions of the available sensory input can be driven by bottom-up processes, such as orientation towards an oncoming, salient stimulus (also known as stimulus-driven attention or sensory attention); and by top-down processes that reflect expectations (about, for example, where an important stimulus will be located) and goals.43 Shortterm or working memory is the ability to hold information on-line during a short time period44,45 and is fundamental to top-down or selective attention in the sense that whatever requires attention, i.e., is selected (e.g. a spatial location) has to be maintained in a shortterm/working memory. The short-term/working memory then biases competition between the multiple bottom-up items in the stimulus input; the result is an advantage in the neuronal competition between the

Functional Magnetic Resonance Imaging (fMRI) The first fMRI investigations in schizophrenia patients were conducted in the early 1990s of the last century.38–40 Since then, a considerable number of fMRI-studies has been published mostly in the behavioral domains: attention, working memory, verbal memory and social cognition. To some extent, these fMRI-studies were following-up and extending previously published Positron Emmission Tomography (PET) or SingleTable 4.1 List of functional biomarkers Biomarker

Cognitive domain

Qualitya

Drugs

Functional Magnetic Resonance maging (fMRI) N-back task Oddball task

Working memory Attention

++ ++

Dopaminergic Dopaminergic Nicotinergic

Electrophysiology (EEG) Mismatch negativity P300 Synchrony

Pre-attention Selective attention Attention

+++ +++ ++

Glutamatergic Cholinergic Dopaminergic

++ = Medium quality, +++ = High quality Estimation of the quality of biomarkers based on data about measurement reliability, face validity (available pharmacological studies), strength of scientific concept and phylogenetic conservation a

4

Translational Medicine

multiple inputs for the item that receives top-down bias from the short-term memory.43,46 It is widely accepted that the prefrontal cortex serves as a store for short-term memory. The first insights into the neuronal basis of working memory came from animal research. Fuster47 recorded the electrical activity of neurons in the prefrontal cortex (PFC) of monkeys while they were doing a delayed matching task. However, over the past decades it has become consensus that most working memory tasks recruit a network of PFC and other cortical areas most notably the parietal areas. Also, Lebedev et al.49 investigated the discharge rates of single prefrontal neurons as monkeys attended to a stimulus marking one location while remembering a different, unmarked location. Although the task made intensive demands on shortterm memory, the largest proportion of prefrontal neurons represented attended locations, not remembered ones. These findings showed that short-term memory functions cannot account for all, or even most, delayperiod activity in the part of the prefrontal cortex explored. The authors suggested that prefrontal activity during the delay-period contributes more to the process of selective attention than to memory storage. Schizophrenia and working memory/selective attention. Working memory and selective attention are likely the cognitive domains that have been most frequently investigated in schizophrenia with fMRI because many schizophrenia patients show obvious signs of impairments in related tasks. Since both cognitive domains are intrinsically linked with each other (s.a.), tasks like the n-back or oddball task that are frequently employed in imaging experiments with schizophrenia patients generally measure both, although different tasks may emphasize different aspects in this regard. In the n-back task, which emphasizes the working memory component, visual stimuli (e.g. numbers) are continuously presented in which every number is both a probe and a target. In a typical experiment,50 the numbers appear randomly every 1.8 s for 500 ms at set locations at the points of a diamond-shaped box. Subjects are to recall the stimulus seen N previously by means of a fiber-optic response box with buttons arrayed in the same configuration as the stimuli presented on the screen. The task is presented as two counterbalanced runs in which 30-s epochs of 0-back alternated with either 1-back or 2-back. In the oddball task, on the other hand, the emphasis lies on selective attention. Subjects are presented two or more runs of

67

visual (or auditory) stimuli consisting of randomly presented standard stimuli (e.g. a 500-Hz tone (80% of trials) and target stimuli (e.g. a 1,000-Hz tone (10% of trials), and novel stimuli of different sound quality (10% of trials).51 In simplified versions, which are frequently used in electrophysiological experiments (see below), no novel stimuli are applied.52 Consistent with earlier PET-studies,41,42,53 reduced activation in dorsolateral and medial prefrontal (as well as parietal and anterior cingulate) cortex during working memory/selective attention tasks have been most frequently reported with fMRI in schizophrenia.54–56 However, several recent investigations of relatively high-functioning patients or family members of patients with no obvious signs of psychosis either failed to find hypoactivation57,58 or reported the opposite observation (i.e., hyperactivation).59,60 Currently, it appears that abnormal activation in schizophrenia in response to attentional (and/or working memory) demands reflects both insufficient recruitment of brain systems in relatively impaired coritcal areas on one hand and overcommitment of resources for processing in relatively spared cortical regions on the other hand which predicts performance.50,61,62 A further complication in this regard is that the degree of activation may be further influenced by the level intrinsic motivation to adhere to a task63 which itself can be pathologically disturbed as part of the negative symptoms complex. It is obvious that patterns of prefrontal hyper- and hypoactivation introduce a whole lot of complexity when using fMRI-experiments with working memory/ selective attention tasks as biomarkers for cognitive deficits in schizophrenia – in particular when one considers the possibility that the level of intrinsic motivation and the location and number of relatively impaired and spared regions may differ from patient to patient. Further problems in this regard are related to the testretest reliability of certain BOLD-response patterns in these tasks which has hardly been investigated so far,64,65 and to the question whether these tasks can be employed in animal experiments. Currently, there are several efforts under way to attack some or all of these problems. For instance, using a simple two-choice reaction task, Winterer et al.66 suggested that increased single trial variability (“noise”) underlies the abnormal averaged BOLD-response pattern of cortical hypo- and hyperactivation in schizophrenia described above. In principle, this measure may allow a more straight forward analysis of cortical dysfunction since it is related

68

to cognitive impairments in a linear relationship. However, additional work is required to characterize more precisely the nature of the variability of singletrial BOLD-responses in schizophrenia and to better dissect the different sources that can contribute to the variability (i.e., technical and physiological – both non-neuronal and neuronal). Among the sources that may contribute to BOLD-response variability during cognitive tasks, resting state brain activity is perhaps of outstanding importance – a notion which is derived from numerous electrophysiological studies suggesting that event-related potential characteristics such as the ERP-amplitude are predicted by EEG background activity. It is therefore interesting that Fox et al.67 recently conducted an fMRI study based on the notion that resting brain is not silent, but exhibits organized fluctuations in neuronal activity even in the absence of tasks or stimuli. They showed that this intrinsic brain activity persists during task performance and contributes to variability in evoked brain BOLD-responses. In the same study, they addressed if this intrinsic activity also contributes to variability in behavior. As predicted they identified a relationship between human brain activity in the left somatomotor cortex and spontaneous trial-to-trial variability in button press force. They also demonstrated that 74% of this brain-behavior relationship is attributable to ongoing fluctuations in intrinsic activity similar to those observed during resting state. In addition to establishing a functional and behavioral significance of intrinsic brain activity, these results lend additional insight into the origins of variability in human behavior. With regard to schizophrenia biomarker research, these results are of potentially great meaning. For instance, Calhoun et al.19 recently identified with fMRI and using independent component analysis (ICA) a cortically widespread temporally coherent network with strong spatial cross-correlations between resting-state and oddball task condition. Of note, corresponding changes (spectral power) in both networks (resting state vs. oddball) were found when comparing schizophrenia patients with healthy control subjects. There are now several reports in literature on resting fMRI in schizophrenia patients and all agree that patients clearly differ from controls.68,69 It is obvious that using resting state fMRI has tremendous advantages when used as a functional biomarker for cognitive performance. Calhoun et al.19 provide a list of arguments: First, ill subjects are often unable (or unmotivated, see above) to perform tasks consistently

G. Winterer

in the scanner or to fully understand complex instructions. However, at rest, there are no such “task” demands. Second, abnormal task performance often occurs in schizophrenia due to the cognitive disability associated with the disorder. This is often inevitably confounded with concomitant abnormal brain activation in a “chicken and egg” manner. At rest, when there is no task, this problem can be resolved. Third, the occurrence of symptoms in the scanner (for example auditory hallucinations) is usually thought of undesirable “noise” during performance of a cognitive task but at rest may actually be contributing useful diagnostic information. There is, however, still another reason why resting-state fMRI might be valuable as functional biomarker. The point is that opposed to resting fMRI, complex working memory paradigms such as n-back tasks can hardly be applied in animal research – in particular when small animals are investigated. Even less difficult tasks like the oddball task can only be conducted in small animals (see below: P300 section) with certain restrictions such as measuring the brain response to deviant tones (targets) without the requirement of a motor response. Thus, resting fMRI may turn out to be among the most promising functional biomarkers for cognition in schizophrenia research.

PharmfMRI and Working Memory/Selective Attention Dopamine and working memory/selective attention. The hypothesis of abnormal dopamine (DA) signaling in schizophrenia has served for over 40 years as the major heuristic framework for understanding the impaired synaptic mechanisms in schizophrenia, and it remains the primary target of pharmacological treatment.70 This is one reason why the effects of DA on the BOLD-response during working memory and selective attention have been relatively well investigated. Several landmark studies in this regard have been conducted by Weinberger and colleagues – work which was based on the pioneering electrophysiological studies in nonhuman primates of the GoldmanRakic group (e.g.71,72) on DA effects in prefrontal cortex and earlier own studies of the effects of dextroamphetamine in humans, an indirect monoaminergic agonist, on PET-based regional blood flow measures.73,74 Over the past decade, Weinberger and colleagues conducted a series of fMRI studies during

4

Translational Medicine

working memory tasks mainly using either dextroamphetamine (“drug challenge”) or variations of the catechol-o-methyltransferase (COMT) gene, which codes for an enzyme crucially involved in synaptic DA degradation, to probe the effects of DA on human brain function. For instance, Mattay et al.75 conducted a study which was performed with BOLD fMRI to examine physiological correlates of the effects of dextroamphetamine on working-memory performance in healthy controls. In a group analysis dextroamphetamine increased BOLD signal in the right prefrontal cortex during a task with increasing working-memory load that approached working-memory capacity. However, the effect of dextroamphetamine on performance and on signal change varied across individuals. Dextroamphetamine improved performance only in those subjects who had relatively low working-memory capacity at baseline, whereas in the subjects who had high working-memory capacity at baseline, it worsened performance. In subjects whose performance deteriorated, signal change was greater than that in subjects who had an improvement in performance, and these variations were correlated (Spearman rho = 0.89, P < 0.02). The authors argued that the behavioral and neurophysiologic effects of dextroamphetamine are not homogeneous. These heterogenic effects of dextroamphetamine may be explained by genetic variations that interact with the effects of dextroamphetamine. In fact, this was exactly what was found in a subsequent study.76 Amphetamine enhanced the efficiency of prefrontal cortex function assayed with functional MRI during a working memory task in subjects with the high enzyme activity val/val genotype, who presumably have relatively less prefrontal synaptic dopamine, at all levels of task difficulty. In contrast, in subjects with the low activity met/met genotype who tend to have superior baseline prefrontal function, the drug had no effect on cortical efficiency at low-to-moderate working memory load and caused deterioration at high working memory load. These data illustrated an application of functional neuroimaging in pharmacogenomics and extended basic evidence of an inverted-“U” functional-response curve to increasing dopamine signaling in the prefrontal cortex. In addition, the study suggested that individuals with the met/met catechol-O-methyltransferase (COMT) genotype appear to be at increased risk for an adverse response to amphetamine.

69

In a more recent study,77 the question was addressed whether working memory performance may be improved or decreased by amphetamine, depending not only on baseline working memory capacity but also on amphetamine dosage. The modulation of cortical processing was examined in a verbal working memory network by dextroamphetamine (D-amph) using BOLD functional magnetic resonance imaging (fMRI) with healthy participants. The goal of the study was to test the hypothesis of an inverted U-shaped relationship between D-amph dose and processing efficiency of a verbal working memory system. D-amph dosage was increased cumulatively every 2 h across four scanning sessions collected in a single day. The primary measure used for analyses in this study was the extent of activation in brain regions empirically defined as a working memory network. An inverted U-shaped relationship was observed between the amount of D-amph administered and working memory processing efficiency. This relationship was specific to brain areas functionally defined as working memory regions and to the encoding/maintenance phase (as opposed to the response phase) of the task. Thus, the results were found to be consistent with the hypothesis that the neurochemical effects of amphetamine modulate the efficiency of a verbal working memory system. Recently, the effects of amphetamine on the fMRI BOLD-response have also been assessed in small animals. Dixon et al.78 used resting state fMRI to measure changes in regional brain activation following amphetamine administration, either alone or after pre-treatment with the dopamine D1 receptor antagonist SCH23390, or the dopamine D2 receptor antagonist, sulpiride, in anaesthetised rat. After obtaining baseline data, rats (n = 8) were given amphetamine (3 g/kg i.v) and volume data sets collected for 90 min. Acute amphetamine challenge caused widespread increases in BOLD signal intensity in many subcortical structures with rich dopaminergic innervation, with decreases in BOLD contrast observed in the superficial layers of the cortex. Pretreatment with SCH23390 (n = 8, 0.5 mg/kg, i.v) substantially attenuated the increases in BOLD activity in response to amphetamine, with lesser effects on the amphetamine-evoked decreases in BOLD signal. In contrast, sulpiride (n = 8, 50 mg/kg, i.v) predominantly blocked the decrease in BOLD signal, having a smaller effect on the increases in

70

BOLD signal. The authors concluded from their findings that their data are supportive of the notion that different dopamine receptor types are responsible for separate components of the full amphetamine response. Furthermore, they were able to demonstrate the utility of BOLD contrast fMRI as a means of characterising the mechanisms of drug action in the whole brain. Such studies may be of particular use for investigation of localised action and interaction of different dopaminergic agents. Nicotine and working memory/selective attention. As compared to the number of fMRI-studies available on the effects of dopaminergic agents on working memory and selective attention, far less is known about the effects of other agents targeting for instance the nicotinergic or glutamatergic system which both are thought to play a prominent role in schizophrenia pathophysiology and associated cognitive deficits. Relatively well described are the effects of nicotine on attentional networks in the brain. Based on numerous earlier experimental behavioral studies in animal models and humans, which have consistently demonstrated positive effects of nicotine on attentional performance, Lawrence et al.79 used functional MRI to determine the neural substrates of nicotine’s effects on a sustained attention (rapid visual information-processing) task. Performance was associated with activation in a frontoparietal-thalamic network in both smokers and nonsmokers. Along with subtle behavioral deficits, mildly abstinent smokers showed less task-induced brain activation in the parietal cortex and caudate than did nonsmokers. Transdermal nicotine replacement improved task performance in smokers and increased taskinduced brain activation in the parietal cortex, thalamus, and caudate, while nicotine induced a generalized increase in occipital cortex activity. These data suggest that nicotine improves attention in smokers by enhancing activation in areas traditionally associated with visual attention, arousal, and motor activation. Jacoben et al.80 have investigated schizophrenic patients based on the notion that high rates of smoking among schizophrenic patients may reflect an effort to remediate cognitive dysfunction. They compared 13 smokers with schizophrenia and 13 smokers with no mental illness who were withdrawn from tobacco and underwent fMRI scanning twice, once after placement of a placebo patch and once after

G. Winterer

placement of a nicotine patch. During scanning, subjects performed an n-back task with two levels of working memory load and of selective attention load. During the most difficult (dichotic 2-back) task condition, they found that nicotine improved performance of schizophrenic subjects. Nicotine enhanced activation of a network of regions, including anterior cingulate cortex and bilateral thalamus, and modulated thalamocortical functional connectivity to a greater degree in schizophrenic than in control subjects during dichotic 2-back task performance. With respect to the investigation of nicotinergic drug effects on cognition, there is currently a tremendous need for valid biomarkers – simply because a number of drug companies currently have nicotinergic drugs in the “pipeline”. In fact, since the fMRI landmark study by Lawrence et al.79 on the effect of nicotine on attentional network activity in the brain, several studies of healthy subjects have either confirmed or extended the initial finding. Many of these studies have lately been conducted in conjunction with a concerted effort in the Juelich Research Centre in Germany.81–83 Of note, analogous to the effects of DA on attention/working memory, there is now evidence that either an increase or decrease of the BOLDresponse can be found in response to nicotine challenge suggesting that the nicotine effect on BOLD-response may dependent on specific task conditions, clinical proband or patient sample characteristics or genotype among other aspects. For instance, an association of the alpha4 subunit nicotinergic acetylcholine receptor gene and the parietal and prefrontal BOLD-response during oddball task conditions has also been recently reported.84 As part of our effort in the Juelich Research Center to establish a functional biomarker to test nicotine-related drug effects on the attentional network, we have lately implemented a platform which is based on placebo-controlled nicotine-application (nasal spray) during simultanenous fMRI/EEG (before and after controlled nicotine withdrawal on a clinical research unit) (www. nicotine-research.com). These measurements are supplemented by monitoring of plasma cotinin, stress hormones and several other metabolic parameters. In addition, lymphoblast expression studies as well as genomic investigations are conducted, whereby subjects (plus schizophrenia patients) are recruited from a

4

Translational Medicine

large population-based pool of subjects from the general population. The perhaps most important part of this platform, however, is that all procedures are strictly following EMEA and GCP (good clinical practice) rules in order to make sure that the conducted pharmaco-imaging experiments can be used for the approval process (FDA, EMEA) of a novel compound. Since functional imaging laboratories so far rarely adhere to the formal rules imposed by FDA and EMEA, it can be expected that in the near future further functional imaging laboratories, will be required to implement comparable platforms in case they want to participate in clinical phase 1–3 studies.

Electrophysiology/Event-Related Potentials Because of the widespread availability of neurophysiological equipment and expertise, neurophysiological measures have been widely applied in schizophrenia research for decades dating back to resting state EEG studies conducted in the 1940s of the last century followed by ERP-studies in the 1960s. Therefore, several electrophysiological biomarkers in schizophrenia research have eventually achieved a degree of maturity hardly met by any other functional measure (for review also see Javitt et al.1). Below, we will briefly present two traditional schizophrenia functional biomarkers, i.e., the event-related potentials: Mismatch negativity (MMN) and P300 Evoked Potential. In addition, more advanced neural synchrony and oscillation measures will be presented because of the exciting developments that have recently occurred linking human electrophysiological work with basic neuroscience discoveries about the role of neural synchrony or oscillations in neuronal communication. We will first discuss gamma frequency band oscillations (31–100 Hz), a topic of current very high interest and research. Subsequently, we will highlight low-frequency or “theta” synchrony and its inverse called “noise” which captures synchrony of brain oscillations in the frequency range between 0.5–10 Hz. The concept of “noise” has recently also attracted considerable interest in schizophrenia research because it was demonstrated to be a strong endophenotype for schizophrenia.

71

Mismatch Negativity (MMN) Mismatch negativity (MMN) is a relatively short-latency brain event occurring after the presentation of a deviant stimulus in a train of preceding repetitive “standard” stimuli: for example, a 1.2 kHz tone (‘pitch’) is presented 5% of the time among 1 kHz tones or alternatively a 1.0 kHz tone of 100 ms duration is presented 10% of the time among 1 kHz tones of 50 ms duration. The MMN occurs in the interval ~ 100–220 ms following the deviant stimulus, and is calculated by subtracting the ERP to deviant stimuli from that to standard stimuli. It is important to emphasize that ERP components (e.g. P300) following MMN depend on attentive, conscious processing of the deviant stimuli (often by counting) whereas attentive processing is not needed or used for the MMN, where often a distractor task is used to minimize conscious attention; the MMN is thus often referred to as “pre-attentive”. Näätänen et al.85 summarize evidence of a pre-attentive, supratemporal component to be distinguished from a subsequent weaker frontal process associated with an attention shift caused by auditory stimulus change. The supratemporal component is primarily evoked automatically, pre-attentively, and pre-consciously and is thus thought to reflect the operations of sensory (“echoic”) memory, a kind of memory of past stimuli used by the auditory cortex in analysis of temporal patterns. The critical point here – with regard to the use of MMN as functional biomarker for schizophrenia cognitive deficits – is that principally both the pre-attentive (temporal) and/or the attentive (frontal) component may display a functional deficiency independently or in conjunction. Thus, a disturbance in the pre-attentive circuits can secondarily result in abnormal attention-related processes even so the underlying circuits are intact. Vice versa, it is also conceivable that the pre-attentive component is not abnormally changed as opposed to the attentive component. If only a behavioral measure of attention were used, one would be unable to distinguish drug effects on the two components. MMN in schizophrenia. MMN has attracted great interest in schizophrenia because of its abnormality in this disorder and evidence that it may track progression of the disorder. Todd et al.86 have systematically studied the similarities and differences to MMN elicited

72

by duration, pitch and intensity deviants to patients tested within 5 years of illness onset (“short” illness, mean duration 2.6 years) and long duration illness (>5 years, mean 18.9 years). In short duration illness, a clear reduction was evident in MMN to duration and intensity but not frequency deviants, while longer duration illness was associated with a reduction in frequency and in duration to a lesser degree but not intensity MMN. That frequency deviants were not reduced early in the illness but were reduced later is consistent with the findings of Javitt et al.87 and Salisbury et al.88 However, since these three studies were cross-sectional, Salisbury et al.89 examined frequency deviants in the same group of patients studied longitudinally at first hospitalization for psychosis and a mean 1.5 years later. Schizophrenia patients showed progressive reductions of MMN compared with two matched groups of bipolar patients and healthy controls, although all groups were similar in MMN at protocol entrance. Umbricht et al.90 reported specificity of duration MMN reduction in schizophrenia, compared with affective disorder and healthy controls. Salisbury et al.89 found specificity of progression of pitch abnormalities in schizophrenia compared with bipolar disorder and healthy controls. Of note, the progression in Heschl gyrus volume reduction, which correlates with the decline of MMN amplitude, has similarly been reported in schizophrenia but not in controls or bipolar disorder,91 suggesting that diseases with loss of Heschl gyrus volume will similarly show reduction of MMN. Reliability, heritability and MMN. Turetsky et al.92 summarize reliability data for healthy controls as moderate for the ‘pitch’ deviant (0.5) but high for the duration deviant (>0.8). In chronic schizophrenia, Light and Braff93 found duration MMN reliability to be 0.87 while Salisbury et al.89 found pitch MMN reliability 0.74 in first episode schizophrenia but less in healthy controls (0.52) and still lower in first episode bipolar patients (0.22). Of note, the reduction of MMN in chronically-ill schizophrenia patients appears to be trait-like and not ameliorated by either typical (Haldol) or atypical (clozapine) medication.94 Heritability of the duration MMN is substantial (h2 = 60–70) and comparable to the heritability of the P300 amplitude.95 Of the three family studies that have been conducted with MMN to date, two study found a reduction in MMN generation in unaffected family members of patients with schizophrenia,96,97 while the third did not.98

G. Winterer

Glutamate, NMDA-receptor and MMN. The glutamate hypothesis of schizophrenia is currently the basis for a number of efforts to develop novel pharmacological treatments for schizophrenia including associated cognitive deficits.99 MMN has been most extensively investigated with regard to glutamatergic mechanisms, particularly involving NMDA receptors. In the temporal lobe, depth recordings of humans, monkeys, and cats have found sources in the primary auditory cortex (e.g.100). Laminar analysis in auditory cortex indicates that MMN reflects the activity of glutamatergic NMDA-channel mediated influx of current flow in supragranular cortical layers. Javitt and colleagues100 showed in monkeys that deviants elicited MMN-like activity in the supragranular (surface) layers of primary auditory cortex, and, further that this activity was obliterated by application of NMDA-specific antagonists.1,101,102 In accordance with theses findings, reduced MMN was shown in normal subjects following ketamine administration.103 Furthermore, in normal volunteers, MMN amplitude predicts the degree of psychosis induced by ketamine across subjects.90 As opposed to effects of ketamine, MMN is not affected by the psilocybin, a psychotomimetic agent that functions via stimulation of 5-HT2A receptors90 and deficits in MMN generation persist despite treatment with available antipsychotic agents, including 5-HT2A-selective compounds such as clozapine, olanzapine and risperidone. This suggests that dopaminergic and serotonergic systems are not closely related to MMN generation. Taken together, the test-retest reliability and diagnostic specificity of the MMN-amplitude deficit, its relationship to the progressive deterioration at the onset of the illness as well as the relatively selective effects of glutamate antagonists on MMN-amplitude suggest that MMN may serve as a useful functional biomarker to assess the effects of glutamate agonists on pre-attentive processes.

P300 Event-Related Potential The P300 is an ERP that occurs to a stimulus that a subject actively detects and processes. Similar to MMN, P300 is elicited most commonly in the context of the auditory oddball paradigm. However, when P300 is to be elicited, subjects are instructed to selectively attend to the sequence of tones, and either count the deviants, or press a button in response to the deviants.

4

Translational Medicine

Under such conditions, P300 occurs approximately 300 ms following deviant tone presentation. Thus, as opposed to MMN with its pre-attentive component generated in the primary auditory cortex, P300 is clearly dependent on selective attention and it is therefore not unexpected that particular cognitive deficits including memory and attentional deficits104 have been associated with P300-amplitude reduction. Notably, rodent models of the P300 ERP have been recently developed,105,106 which qualifies P300-amplitude as a particularly useful functional biomarker for translational studies. Typically, P300 is divided into two subcomponents, an early P3-like potential (“P3a”) which is topographically located relatively frontally, and a later, more parietal component “P3b”. While the P3b amplitude is larger when the stimulus is rarer, the amplitude of the P3a is increasing when a third novel stimulus is accompanied by an orienting response of the subject. At closer look, it became eventually obvious that P300 reflects the activity of several, mostly simultaneously active bilateral generators. There is clear evidence from multi-site intracortical recordings,107–109 neuroimaging experiments110,111 and electromagnetic source localization studies112,113 for a network of distributed activity in at least half a dozen cortical locations – most notably in the inferior parietal lobule and posterior superior temporal gyrus – corresponding to the scalp recorded P3b. These studies also suggest a contribution from dorsolateral prefrontal cortex and from anterior cingulate cortex to the scalp recorded P3a. It is also likely that many of the deep sources that generate P300-like activity do not propagate to the scalp (e.g., hippocampus). P300 in schizophrenia. Roth and Cannon114 were the first to report a reduction in schizophrenia of P300 (P3b) amplitude in recording sites over the sagittal midline of the head. Since then, nearly all studies have reported a reduction of central P300 amplitude in subjects with chronic schizophrenia. Delays in the onset of P300 are less commonly reported. A meta-analysis by Bramon et al.98 of studies with Cz or Pz recordings found an effect size of 0.85 for amplitude and 0.57 for latency (p’s < 0.001) differences in schizophrenia with no significant influence of antipsychotic medication or duration of illness. Also, the amplitude reduction of P300 is not related to a lack of motivation to adhere to the task in the patients, as it remains reduced relative to controls even when the patients’ performance in detecting the tones is

73

as good as that of controls.115,116 Furthermore, the latency of P300 in schizophrenia patients does vary with task difficulty as in controls, but the amplitude remains reduced. Thus, P300 amplitude is robustly reduced in chronically-ill schizophrenia patients. In addition to the broad reductions of P300 over both left and right hemispheres, first hospitalized ill schizophrenic subjects display an asymmetry in P300 with smaller voltage over the left temporal lobe versus right. This finding was first observed in chronic schizophrenia where this left temporal P300 amplitude abnormality was found to be correlated negatively with the extent of psychopathology as reflected in thought disorder, delusions and auditory hallucinations.117–120 There is also evidence that this left temporal amplitude reduction is diagnostically relatively specific for schizophrenia, at least as contrasted with affective (manic) psychosis,121–124 although more studies, and in different disease entities are needed. It may, however, turn out that diagnostic specificity can be further improved when taking into account not only topographic information but also single evoked potential characteristics of the P300 wave. For instance, there is some evidence – although still controversial – that increased latency variability may contribute to the overall amplitude reduction of the averaged P300 ERP rather than an amplitude reduction of single evoked potential amplitudes per se.125–130 In order to obtain a clearer picture in this regard, it will probably be necessary to further improve P300 single trial analysis procedures.131 In particular, it needs to be taken into account that any single trial analysis of ERPs depends on its signal to noise ratio which is diminished in schizophrenia (see below). The frontally generated P3a is also abnormally altered in schizophrenia. Of note, both increased amplitudes132 and decreased amplitudes133 have been described. However, under standard auditory oddball conditions, the P3a peak amplitude is difficult to detect reliably and, as has been pointed out,134 an abnormally altered P3a amplitude – particularly when found to be increased – may reflect an increased amplitude of “noise power” (see below) rather than a true abnormality of this P300 subcomponent. Reliability and heritability. Measurement stability and heritability of the P3b amplitude is about as high as for the duration deviant MMN,92 i.e., test-retest reliability is in the range of 0.8–0.9 with heritability (h2) around 0.6. As of the trait-like characteristics, because

74

of the relatively high heritability and since P300amplitude reduction is also found in family members of schizophrenia patients, P300-amplitude is generally accepted as a marker of the genetic risk for schizophrenia.92,95,132 A meta-analysis by Bramon et al.98 on P300 in relatives of schizophrenics showed a significant reduction in P300 amplitude (effect size = 0.61). On the other hand, Frodl et al.120 described a negative correlation of P300-amplitude and illness duration. Turetsky et al.118 found a trend toward normalization of the left temporal P300 subcomponent that correlated with change in Brief Psychiatric Rating Scale which was observed in a similar way by Mathalon et al.133 While the latter group found a correlation with changes of positive and negative symptoms, Gallinat et al.119 related this P300-amplitude change to positive but not negative symptoms. Ford and colleagues128 demonstrated that although P300 showed moderate amplitude increases with symptom resolution, however, it did not approach normal values during these periods of remission. Currently, it appears95 that while MMN- and P300-amplitude reductions have both state- and traitlike components, the trait aspect, i.e., the genetic component, may be more pronounced for P300 than for MMN whereas the opposite may be true for the state aspect related to the course of the illness. P300 and pharmacology. As of theoretical considerations on the generation of ERPs, one would expect P300 to be modulated by glutamatergic mechanisms in an analogous way as MMN. However, there is amazingly little data available, so far.135 Nevertheless, over the past few years, our understanding of the molecular, genetic and pharmacological basis of P300 has improved considerably. Preliminary evidence associates the late temporoparietal P3b component with specific genetic variations in the dopamine D3 and D2 receptor genes,136–139 whereas variations in the gene that codes for the dopamine catabolizing enzyme catechol-O-methyltransferase (COMT) have been associated with frontocentral and centroparietal P300 amplitude variations.140–142 However, negative association findings also have been described for P300amplitude.143–145 To some extent conflicting results have also been reported on pharmacological effects on P300-amplitude. Oranje et al.146 did not find an effect of L-dopa or bromocriptine treatment on P300amplitude. In contrast, acute challenge with d-amphetamine was reported to increase P300-amplitude.147 In this context, one may also mention a study conducted

G. Winterer

by Niznikiewicz et al.148 who found a significant increase in the P300 amplitude in patients treated with clozapine relative to baseline, off-medication status. The P300 amplitude improvement was not found in chlorpromazine-treated patients. Involvement of cholinergic mechanisms in auditory evoked P300 has also been observed. Acetylcholine and cholinergic drugs appear to strongly increase P300-amplitude. The opposite observation, i.e., a decrease of P300-amplitude is made for anticholinergic substances. Both effects are likely to be mediated by muscarinic and perhaps also nicotinergic receptors.149–155 In line with these observations in humans, septal cholinergic lesions abolish P300 in cats.156 In summary, decades of P300 research now suggest that P300 has reached a high degree of maturity as a functional biomarker for schizophrenia illness and related attentional deficits. P300 may be considered a useful translational biomarker to track pharmacological interventions with cholinergic drugs – not only in schizophrenia but in any neuropsychiatric disorder involving the cholinergic system including Alzheimer dementia, addiction and attention deficit disorder.1

Gamma Oscillations Gamma-band synchronization appears to be one of the fundamental modes of neuronal activity. Activated neuronal groups engage in rhythmic synchronization in the gamma-frequency band (31–100 Hz) has been documented in the cortex of many brain regions, including the visual, auditory, somatosensory, motor, parietal, and entrorhinal as well as in hippocampus (reviewed by Fries et al.157). It is present in many species, from insects to mammals, including humans and during conditions ranging from simple sensory stimulation to attentional selection, working memory maintenance and perceptual integration.157 At the cognitive level, work in humans suggests that gamma activity reflects the convergence of multiple processing streams in cortex, this “feature binding” giving rise to a unified percept. Gamma oscillations and schizophrenia. Kwon and colleagues first investigated gamma in schizophrenia using an exogenous input of 40 Hz auditory clicks, leading to a steady-state gamma response. The magnitude of the brain response was measured by power, the

4

Translational Medicine

amount of EEG energy at a specific frequency, with the degree of capability of gamma driving being reflected in the power at and near 40 Hz. Schizophrenia patients had, compared with healthy controls, a markedly reduced power at 40 Hz input. Moreover, the phase response curve, a description of the relationship between the timing of each stimulus and the EEG response, suggested that it was an intrinsic oscillator that was being driven. This was because the time duration between the stimulus and the EEG response, as the 40 Hz stimulation continued, was reduced to a duration too short to be explained by simple conduction to the auditory cortex. This phase response curve is very common when a external signal drives a “tuned oscillator”, much as auditory stimuli can set in motion a tuning fork, which oscillates (resonates) at a particular frequency. The abnormal amplitude and phase response in schizophrenic patients raised the possibility that there was an intrinsic deficit in brain circuitry supporting 40 Hz oscillations. Since gamma-band synchrony has been classically linked to perceptual feature-binding, Spencer and colleagues investigated gamma oscillations in schizophrenia using a task designed to invoke visual feature-binding mechanisms.158 In this experiment, subjects discriminated between squares formed by illusory contours (Illusory Square condition) and a control condition (No-Square). The stimuli in each condition are physically identical, but the rotation of the “pac-men” determines whether or not observers perceive a coherent object. In healthy control subjects, Illusory Squares but not the “No-Squares” elicited a gamma oscillation phase-locked to the stimuli at occipital electrodes. (A key methodological point is that, while evoked power and phase-locking factor both reflect synchrony of activity, phase-locking factor is independent of the amplitude of the signal, while evoked power depends on it. Hence gamma band phase coherence is a more sensitive measure of oscillation than gamma band power. Consequently, healthy controls often show differences in phase locking with schizophrenia when amplitude is not significant.158,159) For schizophrenia patients, however, neither stimulus elicited an oscillation even though the stimuli were correctly identified. A follow-up study examined response-locked gamma oscillations in the same paradigm.160 It was reasoned that the neural mechanisms underlying conscious perception might be more correlated with reaction time (measured by button press) than stimulus onset, as has

75

been found in animal single-unit recording studies. In healthy subjects, Illusory Squares elicited a responselocked gamma oscillation at 250 ms before the reaction time button press, also at occipital electrodes. No such oscillation was elicited by No-Square stimuli. Thus, the response-locked oscillation may be a correlate of visual feature-binding processes involved in conscious perception. For the schizophrenia group a response-locked oscillation at occipital electrodes was also elicited by Illusory Squares. However, this response-locked oscillation occurred in a lower frequency range for schizophrenia patients (22–24 Hz) than normal controls (34–40 Hz). This difference in synchronization frequency suggests that synchrony was necessary for the coherent object to be perceived, but the cell assemblies coding the object were unable to synchronize in the normal gamma range in schizophrenia patients. One possible cause of this effect is reduced cortical connectivity. Selemon and GoldmanRakic161 have suggested that reduced connectivity is an important neural substrate of schizophrenia, and a study modeling gamma oscillations found that increased conduction delays lowered the synchronization frequency of a cell assembly.162 Evidence for a close relationship between the response-locked oscillation and core cognitive and neural abnormalities in schizophrenia was found in the strong correlations between positive symptoms (visual hallucinations, thought disorder, and disorganization) and the degree of phase coherence. These data are consistent with studies by Uhlhaas and colleagues,163 who have found correlations between thought disorder and disorganization symptoms and psychophysical measures of visual perception. Notably, gamma band abnormalities may not reflect drug effects since they have also been observed in unmedicated patients164 and new generation antipsychotics may even reverse preexisting gamma abnormalities.165 In addition, gamma deficits may already be present at the onset of illness. Spencer and colleagues158 have recently shown in first episode patients that gamma abnormalities in schizophrenia have a left-sided bias, compatible with structural MRI findings, whereas those in affective (bipolar) psychosis do not have a hemispheric bias. Thus gamma band abnormalities per se may not be specific to schizophrenia but the source of the abnormalities may differ according to the pathophysiology of the disorder, which include structural alterations in schizophrenia not present in bipolar disorder. Gamma abnormalities

76

may occur in other conditions affecting the neural circuit involved in their generation, including attentiondeficit hyperactivity disorder, autism, epilepsy, and Alzheimer’s disease.166,167 Results have been mixed as to whether the gamma deficits observed in schizophrenia patients are also present in persons with schizotypal personality traits or with increased genetic risk for schizophrenia.134,165,168 It also remains to be determined if there is evidence for progression of gamma abnormalities in schizophrenia, as has been observed for the MMN abnormalities. Further investigations in this currently very active topic of study are needed to specify further how gamma oscillation abnormalities in schizophrenia are relevant to its clinical features and whether certain features may be diagnostically specific. Molecular mechanisms of gamma oscillations. As of today, little work has been conducted on the pharmacology of gamma oscillations. However, the molecular mechanisms involved in the generation of gamma activity are increasingly understood which may guide future investigations studying drug effects on gamma. Basically, gamma oscillation generation involves an interaction between pyramidal projection neurons (glutamatergic, excitatory and regular spiking) and, in the cortex, a particular kind of GABAergic inhibitory neuron, one that is fast spiking (high frequency spikes following current injection) containing the calcium binding protein parvalbumin (PARV) and comprised of basket and chandelier cells, and found to be abnormal in post-mortem work in schizophrenia.169 In response to excitatory input from the pyramidal neurons, the group of GABAergic neurons tends to discharge synchronously, as a result of interconnections mediated by both neurotransmitter (GABAergic) and electrical (gap junction) activity. This GABAergic synchronous activity, in turn, synchronizes the population of pyramidal cells to which they project. The timing of the discharges and the time course of resultant alterations in membrane polarization results in the gamma frequency activity. There is now considerable evidence supporting this model of the substrate for gamma oscillations in the cortex is interaction of GABAergic interneurons with glutamatergic neurons, and that this interaction occurs in upper cortical layers, layers II and II (see, for example170). This abnormality of glutamatergic-GABAergic interaction in the production of gamma band oscillations is congruent with accumulating post-mortem evi-

G. Winterer

dence of reduced inhibition of pyramidal neurons in schizophrenia169 and with basic in vitro studies indicating an increased sensitivity of the NMDA synapse on GABAergic neurons to inhibition by psychotomimetics and, consequently to dis-inhibition of the pyramidal targets of the GABA neurons.171,172 Taken together: gamma oscillations are clearly abnormal in schizophrenia and appear to underlie related cognitive deficits. Tremendous progress has been made with regard to the basic molecular and physiological mechanisms of gamma oscillation generation indicating a prominent role of the GABAergic system. This will greatly facilitate the application of gamma oscillations as functional biomarker for cognitive deficits in schizophrenia.

Low Frequency Oscillations and Noise Over the past decades, the neurophysiological and pharmacological basis of synchronous low-frequency oscillations have been investigated in animal models and more recently in humans during various task conditions. By low frequency oscillations we mean oscillations in the delta (1–4 Hz) and theta (4–8 Hz) band, according to our clinical definitions of frequency bands. Theta activity in animals, primarily rodents, has been extensively investigated and reviewed (e.g.,173–175). In studies of rodents (hippocampal) oscillations across a broad range of frequencies between 5–12 Hz are termed theta. Theta is prominent during motor behavior variously described as “voluntary,” “preparatory,” “orienting,” or “exploratory” and during rapid eye movement (REM) sleep. An acceleration of frequency within the theta band is seen, for instance, with stronger orienting responses or stronger exploratory behavior, i.e., slow and fast theta (reviewed in176). In larger animals including cats, dogs and primates the frequency range of theta is usually lower than in rodents ranging from 3–8 Hz. In addition, theta activity can be quite different from the rodent in terms of state- and site-dependency. For instance, in intracranial recordings in human hippocampus, there is no constant theta oscillation during REM sleep, but rather short bursts of 4–7 Hz waves appear in this state and during transitions to wakefulness. Whether neocortical theta oscillations exist that are completely or partly independent from the hippocampal formation is a matter under investigation. So far,

4

Translational Medicine

evidence for independent theta oscillations has been obtained in rabbit cingulate cortex177–179 and more recently in primates. For instance, Tsujimoto et al.180 identified with intracortical recordings a slow wave oscillation source (5–6 Hz) in primate anterior cingulate cortex (ACC) during an executive attentional task whereby synchrony (coherence between electrodes) increased with task performance. In another investigation using subdural and depth recordings which was conducted by Cantero et al.,181 theta waves (4–7 Hz) were observed in the basal temporal lobe and frontal cortex during transitions from sleep to wake, but not in REM. Importantly, Cantero et al. found that theta activity in human hippocampus and in neocortex was not correlated (coherent) supporting the notion of cortical theta generators being independent of hippocampal generators. In this context, an investigation of Guderian and Düzel182 in human subjects is of considerable interest. Using whole-head magnetoencephalography, they found that recollection of personal events is associated with an induced activity increase in a distributed synchronous theta (4.5–7.0 Hz) network, including prefrontal, mediotemporal, and visual areas. This would suggest that independent theta generators may cooperate under certain task conditions. Scalp EEG-recordings in humans have suggested for some years the existence of a frontal midline slow wave oscillator during tasks that require selective attention (0.5–5 Hz),183 problem solving or working memory tasks (5–7 Hz),184,185 i.e., in cognitive conditions that are typically impaired in schizophrenia. Gevins and Smith186 reported that subjects with stronger frontal midline slow wave oscillations (5–7 Hz) have higher cognitive ability. Winterer et al.134 found that lower prefrontal slow wave synchrony, i.e., higher slow wave variability (“noise”) during an oddball task condition predicts attentional performance (for details see below). It is, however, not yet resolved how synchronized slow wave oscillations contribute to task performance. Jensen and colleagues187,188 suggested on the basis of electrophysiological recordings in animals and humans that the power of fast gamma oscillations is modulated by the phase of slow wave oscillations whereby coupled slow wave oscillations might mediate a dynamic link between hippocampal and neocortical including prefrontal areas, thereby allowing to recruit and bind distributed cortical representations. A similar suggestion was made by Jones and Wilson189 who proposed that the coordination of slow wave oscil-

77

lations between prefrontal and hippocampal sites may constitute a general mechanism through which the relative timing of disparate neural activities during cognitive operations can be controlled. In agreement with this notion is a recent report of Jacobs et al.,190 who investigated the temporal relationship between brain oscillations and single-neuron activity in humans during a virtual navigation task with recordings from 1,924 neurons in various brain regions. They found that neuronal activity in various brain regions increases at specific phases of brain oscillations. Neurons in widespread brain regions were phase-locked to oscillations in the theta- (4–8 Hz) and gamma- (30–90 Hz) frequency bands. In hippocampus, phase locking was prevalent in the delta- (1–4 Hz) and gamma-frequency bands. Individual neurons were phase locked to various phases of theta and delta oscillations, but they only were active at the trough of gamma oscillations. The authors proposed that slow wave oscillations facilitate phase coding and that gamma oscillations help to decode combinations of simultaneously active neurons. In this context, it is of interest that Hyman et al.191 recently showed that the majority of the medial prefrontal cells with a significant correlation of firing rate changes with behavior were entrained to slow wave oscillations in hippocampus, which leaves, however, open the question whether entrainment is also present with prefrontal slow wave oscillations. Slow wave oscillations and schizophrenia. In 2000, Winterer et al.192 first reported on the basis of electrophysiological investigations that information processing in schizophrenia might be characterized by a diminished cortical signal to noise ratio resulting from a poor low frequency (0.5–5.5 Hz) phase-synchrony (i.e., increased “noise”) of event-related frontocentral brain oscillations. In this study, electrophysiological recordings from scalp were obtained during a twochoice reaction time task with randomized checkerboard presentation in the left and right hemifield requiring subjects to respond accordingly by button press. “Noise” was calculated as the power µV2 of the averaged event-related EEG subtracted from the mean power (µV2) of the single evoked EEG-responses. Noise power formula: Noise Power =

1 N 2 ⎛1 N ⎞ ui ⎟ ∑ (ui )- ⎜⎝ N ∑ ⎠ N i=1 i=1

2

whereby N is the number of trials and ui is the EEG signal of trial i.

78

Accordingly, this “noise” measure is related to the latency variability measure of ERPs (see above160). However, the two measures are not identical: “noise” – as an oscillatory-based measure – does not exclusively consider the peak latency variability but also the amplitude variability of an evoked EEG-response. In addition, it is not limited to a narrow time window around the peak of an ERP but takes into account the entire evoked response. Several noteworthy findings were obtained in this study: test-retest reliability of the noise measure turned out to be highly stable (Cronbach’s alpha > 0.9), schizophrenia patients (N = 14) were unmedicated (>1 year), a comparable trend for increased “noise” was also seen in subjects at risk for schizophrenia (schizotypal subjects, N = 18), in those patients (N = 16) who had received antipsychotic drug treatment (typical and atypical) within a few days (>3 days) before investigation, “noise” was diminished as compared to longterm drug-free patients. In addition, it was found that patients with a diagnosis of depression (N = 62) do not show increased “noise” suggesting diagnostic specificity of the “noise” measure. Referring to computational neural network models, the authors then proposed that this (synchrony) deficit may lead to specific changes of the attractor properties within cortical networks, most notably a decrease of attractor stability, which ultimately may account for the clinical symptoms observed in schizophrenia (also see21). Since this initial report, Winterer and colleagues followed up on this finding in a series of electrophysiological and fMRI studies. As part of the NIMH schizophrenia sibling study, they explored whether this particular physiological abnormality predicts working memory and attentional performance (N-back task) and whether the “noise” is related to the genetic risk for schizophrenia.134 “Noise” was calculated for discrete frequency components across a broad frequency range (0.5–45.0 Hz) during processing of an auditory oddball paradigm in patients with schizophrenia, their clinically unaffected siblings, and healthy comparison subjects. As predicted, frontal “noise” was highest in patients (mostly medicated), intermediate in their siblings and lowest in the control subjects suggesting that increased prefrontal “noise” is associated with genetic risk for schizophrenia. Of note, increased “noise” in patients and their siblings was found across the entire frequency spectrum (0.5–45 Hz) suggesting that the synchronization deficit in schizophrenia may not be

G. Winterer

limited to low frequency oscillations but is a more general and perhaps frequency-independent feature – a finding which has been recently confirmed by several groups using different task conditions, i.e., visual steady-state evoked reponses.193,194 In the study of Winterer et al.,134 intraclass correlations (ICC) within sibpairs (schizophrenia patients versus unaffected siblings) were in the range between 0.6–0.9 but lower in the higher beta/gamma frequencies (0.05–0.1) which may indicate that high frequency oscillations are under less genetic control. In addition, it was observed that prefrontal “noise” (0.5–4.0 Hz) is negatively correlated with attention/working memory performance across all subjects (1-back r = −0.49, p < 0.0001; 2-back r = −0.34; p < 0.0002; 3-back r = −0.40, p < 0.0001). Interestingly, similar correlations were also seen for the alpha frequency band (8.5–12.5 Hz). On the basis of these results, the authors concluded that frontal lobe-related cognitive function may depend on the ability of a subject to synchronize cortical pyramidal neurons, which is in part genetically controlled and that increased prefrontal “noise” is an intermediate phenotype related to genetic susceptibility for schizophrenia. Since both the EEG and BOLD-response depend on postsynaptically generated field potentials (see Rolls et al.21), it was a logical step to investigate whether prefrontal BOLD-response variability is also increased in schizophrenia (Winterer et al.,)195. In this study, the authors used fMRI during a visual two-choice reaction task in order to measure, with higher topographic accuracy, signal stability in patients with schizophrenia compared to controls. They also assessed the relationship to more traditional measures of BOLD activation. In patients with schizophrenia, an increased cortical (prefrontal) BOLD-response variability (“BOLDnoise”) was found which predicted the level of prefrontal activation in these subjects. An additional Independent Component Analysis (ICA) revealed a “fractionized” and unfocussed pattern of activation in patients. In the left hemisphere, residual noise variance strongly correlated with psychotic symptoms (r = 0.7, p < .05). The authors proposed that these findings may suggest unstable cortical signal processing underlying classic abnormal cortical activation patterns as well as psychosis in schizophrenia. Molecular mechanisms and pharmacology of slowwave oscillations. For many years, theta oscillations

4

Translational Medicine

have been the topic of extensive investigations because of its association with memory processing and gating of neuronal discharge: it is believed to be critical for temporal coding/decoding of active neuronal ensembles and neuroplasticity, i.e., the modification of synaptic weights underlying memory (Long Term Potentiation and Long Term Depression173). In the rodent, a model of theta generation has been proposed (reviewed by Vertes175). In this model, neural activity underlying theta originates in tonic discharges in rostral pontine reticular formation that propagates to the supramammilary nucleus, where it is converted to a rhythmic discharge projecting to the medial septum, onto both hippocampal- projecting cholinergic and GABAergic neurons. Medial septal GABAergic neurons connect with and inhibit GABAergic cells of the hippocampus, thereby exerting a disinhibitory action on pyramidal neurons. Medial septal cholinergic pacemaking neurons simultaneously excite hippocampal pyramidal cells and GABAergic interneurons. In hippocampus, it is thought that the hippocampal GABAergic neurons are the chief theta pacemaker, and theta rhythm in field potentials is highly correlated with pyramidal cell discharges. Dopamine apparently also has an impact on slow wave oscillations. In rodents, Fitch et al.196 found that dopamine D1 receptor stimulation provides a statedependent bidirectional modulation of theta burst (4–7 Hz) occurrence. The effect of dopamine on taskrelated slow wave oscillations (0.5–4.6 Hz) was also investigated in humans by Winterer et al.197 In order to test the hypothesis whether dopamine affects synchrony of prefrontal slow wave oscillations during an auditory oddball task, the effect of the Val/Met COMT genotype was investigated. COMT is an enzyme critically involved in the catabolism of cortical dopamine. A number of neuroimaging investigations underpinned by a host of basic research studies (both reviewed by Tan et al.198) have suggested that Val allele carriers – as opposed to Met carriers – would have relatively greater inactivation of prefrontal synaptic DA and therefore less effective prefrontal DA signaling during working or short-term memory and attentional requiring tasks. Winterer et al.197 found in a Caucasian sample comprising unrelated normal subjects, schizophrenic probands, and their unaffected siblings that homozygous Valcarriers, who are thought to have less synaptic DA available in prefrontal cortex, have greatest prefrontal

79

electrophysiologic “noise” values during an auditory oddball task. In a subsequently conducted fMRI study,199 analogous results were obtained. In this study, event-related fMRI was conducted during a visual oddball task (checkerboard reversal). As compared to Val carriers, stronger and more extended BOLD responses were observed in homozygous Met carriers in left supplementary motor area extending to anterior cingulate cortex and dorsolateral prefrontal cortex. Vice versa, increased levels of “noise” were seen in Val carriers surrounding the peak activation maximum. Thus, this work suggests that dopamine is decreasing “noise”, i.e. increasing slow wave synchrony in prefrontal cortex. More generally, one could say that dopamine stabilizes cortical microcircuits by suppressing “noise” in cortical (prefrontal) networks.21,200 In summary, impaired (prefrontal) slow-wave oscillation (“noise”) currently emerge as a particular strong functional biomarker for schizophrenia and associated cognitive deficits. However, although findings appear to be promising, there is still considerable work ahead to be done. First of all, it will be necessary to further establish diagnostic specificity of “noise”. In addition, it will also be necessary to further explore the moleculargenetic and pharmacological basis of slow-wave synchrony in humans. Available data suggest that this biomarker may be useful to assess cortical dopamine signaling and synaptic plasticity and drugs that interfere with the involved molecular mechanisms.

Conclusions Several functional biomarkers for the assessment of cognition-related deficits in schizophrenia have currently achieved a degree of maturity that their application in drug research appears to be useful within certain limits. Ongoing research is promising since it can be expected that these biomarkers are further improved within the near future. Currently, the best validated fMRI biomarkers are related to working memory and attentional performance. Well established EEG biomarkers are related to pre-attentive and attentive processes. The available biomarkers are currently most suitable to measure drug effects that interfere with the dopaminergic nicotinergic/cholinergic and glutamatergic systems.

80

References 1. J DC, Spencer KM, Thaker GK, Winterer G, Hajós M. Neurophysiological biomarkers for drug development in schizophrenia. Nat Rev Drug Discov 2008;7:68–83 2. DiMasi JA, Hansen RW, Grabowski HG. The price of innovation: new estimates of drug development costs. J Health Econ 2003;22:151–185 3. Rawlins MD. Cutting the cost of drug development? Nat Rev Drug Discov 2004;3:360–364 4. Littman BH, Williams SH. The ultimate model organism: progress in experimental medicine. Nat Rev Drug Discov 2005;4:631–638 5. Borsook D, Becerra L, Hargreaves R. A role for fMRI in optimizing CNS drug development. Nat Rev Drug Discov 2006;5:411–424 6. Winterer G, Goldman D. Genetics of human prefrontal function. Brain Res Rev 2003;43:134–63 7. Sherwood CC, Subiaul F, Zawidzki TW. A natural history of the human mind: tracing evolutionary changes in brain and cognition. J Anat 2008;212:426–54 8. Winterer G, Carver FW, Musso F, Mattay V, Weinberger DR, Coppola R. Complex relationship between BOLD signal and synchronization/desynchronization of human brain MEG oscillations. Hum Brain Mapp 2007;28:805–816 9. Callicott JH, Mattay VS, Verchinski BA, Marenco S, Egan MF, Weinberger DR. Complexity of prefrontal cortical dysfunction in schizophrenia: more than up or down. Am J Psychiatry 2003;160:2209–2215 10. Bain LJ. A review of the “State of the Art” on mild cognitive impairment: the fourth annual symposium. Alzheimers Dement 2006;2:246–256 11. Miller DH. Biomarkers and surrogate outcomes in neurodegenerative disease: lessons from multiple sclerosis. NeuroRx 2004;1:284–294 12. Auer T, Schwarcz A, Horváth RA, Barsi P, Janszky J. Functional magnetic resonance imaging in neurology. Ideggyogy Sz 2008;61:16–23 13. McFarland HF, Barkhof F, Antel J, Miller DH. The role of MRI as a surrogate outcome measure in multiple sclerosis. Mult Scler 2002;8:40–51 14. Cho RY, Ford JM, Krystal JH, Laruelle M, Cuthbert B, Carter CS. Functional neuroimaging and electrophysiology biomarkers for clinical trials for cognition in schizophrenia. Schizophr Bull 2005;31:865–869 15. Prentice RL. Surrogate endpoints in clinical trials: definition and operational criteria. Stat Med 1989;8:431–440 16. Biomarkers Definitions Working Group. Biomarkers and surrogate endpoints: preferred definitions and conceptual framework. Clin Pharmacol Ther 2001;69:89–95 17. McGuire P, Howes OD, Stone J, Fusar-Poli P. Functional neuroimaging in schizophrenia: diagnosis and drug discovery. Trends Pharmacol Sci 2008;29:91–98 18. Logothetis NK, Pauls J, Augath M, Trinath T, Oeltermann A. Neurophysiological investigation of the basis of the fMRI signal. Nature 2001;412:150–157 19. Calhoun VD, Kiehl KA, Pearlson GD. Modulation of temporally coherent brain networks estimated using ICA at rest and during cognitive tasks. Hum Brain Mapp 2008 [epub ahead of print].

G. Winterer 20. Tallon-Baudry C, Bertrand O. Oscillatory gamma activity in humans and its role in object representation. Trends Cogn Sci 1999;3:151–162 21. Rolls ET, Loh M, Deco G, Winterer G. Computational models of schizophrenia and dopamine modulation in the prefrontal cortex. Nat Rev Neurosci 2008;9:696–708 22. Sehatpour P, Molholm S, Javitt DC, Foxe JJ. Spatiotemporal dynamics of human object recognition processing: an integrated high-density electrical mapping and functional imaging study of “closure” processes. Neuroimage 2006;29:605–618 23. Hermandez L, Badre D, Noll D, Jonides J. Temporal sensitivity of event-related fMRI. Neuroimage 2002; 17:1018–1026 24. Lin FH, Wald LL, Ahlfors SP, Hämäläinen MS, Kwong KK, Belliveau JW. Dynamic magnetic resonance inverse imaging of human brain function. Magn Reson Med 2006;56:787–802 25. Winterer G, McCarley RW. Electrophysiology of Schizophrenia. In: Weinberger DR, Harrison PJ, eds. Schizophrenia. 3rd ed. Oxford, Cambridge, MA: Blackwell, 2008. 26. Turetsky BI, Calkins ME, Light GA, Olincy A, Radant AD, Swerdlow NR. Neurophysiological endophenotypes of schizophrenia: the viability of selected candidate measures. Schizophrenia Bull 2007;33:69–94 27. Swallow KM, Braver TS, Snyder AZ, Speer NK, Zacks JM. Reliability of functional localization using fMRI. Neuroimage 2003;20:1561–1577 28. Vanduffel W, Fize D, Mandeville JB, Nelissen K, Van Hecke P, Rosen BR, Tootell RB, Orban GA. Visual motion processing investigated using contrast agent-enhanced fMRI in awake behaving monkeys. Neuron 2001;32:565–577 29. Brewer AA, Press WA, Logothetis NK, Wandell BA. Visual areas in macaque cortex measured using functional magnetic resonance imaging. J Neurosci 2002;22:10416–10426 30. Nelissen K, Luppino G, Vanduffel W, Rizzolatti G, Orban GA. Observing others: multiple action representation in the frontal lobe. Science 2005;310:332–336 31. Pinsk MA, DeSimone K, Moore T, Gross CG, Kastner S. Representations of faces and body parts in macaque temporal cortex: a functional MRI study. Proc Natl Acad Sci USA 2005;102:6996–7001 32. Op de Beeck HP, Deutsch JA, Vanduffel W, Kanwisher NG, Dicarlo JJ. A stable topography of selectivity for unfamiliar shape classes in monkey inferior temporal cortex. Cereb Cortex 2007 [epub ahead of print] 33. Febo M, Segarra AC, Nair G, Schmidt K, Duong TQ, Ferris CF. The neural consequences of repeated cocaine exposure revealed by functional MRI in awake rats. Neuropsychopharmacology 2005;30:936–943 34. Attwell D, Iadecola C. The neural basis of functional brain imaging signals. Trends Neurosci 2002;25:621–625 35. Iannetti GD, Wise RG. BOLD functional MRI in disease and pharmacological studies: room for improvement? Magn Reson Imaging 2007;25:978–988 36. Mobascher A, Brinkmeyer J, Warbrick T, Musso F, Wittsack HJ, Stoermer R, Saleh A, Schnitzler A, Winterer G. Fluctuations in electrodermal activity reveal variations in single trial brain responses to painful laser stimuli – A fMRI/ EEG study. Neuroimage 2009;144:1081–1092 37. Debener S, Ullsperger M, Siegel M, Fiehler K, von Cramon DY, Engel AK. Trial-by-trial coupling of concurrent electroencephalogram and functional magnetic resonance imaging

4

38.

39.

40.

41.

42.

43. 44. 45. 46. 47.

48. 49.

50.

51.

52.

53.

54.

Translational Medicine identifies the dynamics of performance monitoring. J Neurosci 2005;25:11730–11737 Wenz F, Schad LR, Knopp MV, Baudendistel KT, Flömer F, Schröder J, van Kaick G. Functional magnetic resonance imaging at 1.5 T: activation pattern in schizophrenic patients receiving neuroleptic medication. Magn Reson Imaging 1994;12:975–982 Andreasen NC, Flashman L, Flaum M, Arndt S, Swayze V 2nd, O’Leary DS, Ehrhardt JC, Yuh WT. Regional brain abnormalities in schizophrenia measured with magnetic resonance imaging. JAMA 1994;272:1763–1769 Renshaw PF, Yurgelun-Todd DA, Cohen BM. Greater hemodynamic response to photic stimulation in schizophrenic patients: an echo planar MRI study. Am J Psychiatry 1994;151:1493–1495 Ingvar DH, Franzén G. Abnormalities of cerebral blood flow distribution in patients with chronic schizophrenia. Acta Psychiatr Scand 1974;50:425–462 Weinberger DR, Berman KF, Illowsky BP. Physiological dysfunction of dorsolateral prefrontal cortex in schizophrenia. III. A new cohort and evidence for a monoaminergic mechanism. Arch Gen Psychiatry 1988;45:609–615 Desimone R, Duncan J. Neural mechanisms of selective visual attention. Annu Rev Neurosci 1995;18:193–222 Fuster JM. Memory in the cerebral cortex. MIT Press, Cambridge, MA; 1995 Fuster JM. Executive frontal functions. Exp Brain Res 2000;133:66–70 Deco G, Rolls ET. Attention, short-term memory, and action selection: a unifying theory. Prog Neurobiol 2005;76:236–256 Fuster JM. Unit-activity in prefrontal cortex during delayedresponse performance - neuronal correlates of transient memory. J Neurophysiol 1973;36:61–78 Smith EE, Jonides J. Storage and executive processes in the frontal lobes. Science 1999;283:1657–1661 Lebedev MA, Messinger A, Kralik JD, Wise SP. Representation of attended versus remembered locations in prefrontal cortex. PLoS Biology 2004;2:1919–1935 Tan HY, Sust S, Buckholtz JW, Mattay VS, MeyerLindenberg A, Egan MF, Weinberger DR, Callicott JH. Dysfunctional prefrontal regional specialization and compensation in schizophrenia. Am J Psychiatry 2006;163: 1969–1977 Kiehl KA, Stevens MC, Celone K, Kurtz M, Krystal JH. Abnormal hemodynamics in schizophrenia during an auditory oddball task. Biol Psychiatry 2005;57:1029–1040 Musso F, Konrad A, Vucurevic G, Schäffner C, Friedrich B, Frech P, Stoeter P, Winterer G. Distributed BOLD-response in association cortex vector state space predicts reaction time during selective attention. Neuroimage 2006;29: 1311–1318 Ariel RN, Golden CJ, Berg RA, Quaife MA, Dirksen JW, Forsell T, Wolson J, Graber B. Regional cerebral blood flow in schizophrenia with the 133-xenon inhalation method. Arch Gen Psychiatry 1983;40:258–263 Callicott JH, Ramsey NF, Tallent K, Bertolino A, Knable MB, Coppola R, Goldberg T, van Gelderen P, Mattay VS, Frank JA, Moonen CT, Weinberger DR. Functional magnetic resonance imaging brain mapping in psychiatry: methodological issues illustrated in a study of working memory in schizophrenia. Neuropsychopharmacology 1998;18:186–196

81 55. Curtis VA, Bullmore ET, Morris RG, Brammer MJ, Williams SC, Simmons A, Sharma T, Murray RM, McGuire PK. Attenuated frontal activation in schizophrenia may be task dependent. Schizophr Res 1999;37:35–44 56. Rasser PE, Johnston P, Lagopoulos J, Ward PB, Schall U, Thienel R, Bender S, Toga AW, Thompson PM. Functional MRI BOLD response to Tower of London performance of first-episode schizophrenia patients using cortical pattern matching. Neuroimage 2005;26:941–951 57. Volz H, Gaser C, Hager F, Rzanny R, Ponisch J, Mentzel H, Kaiser WA, Sauer H. Decreased frontal activation in schizophrenics during stimulation with the continuous performance test—a functional magnetic resonance imaging study. Eur Psychiatry 1999;14:17–24 58. Manoach DS, Press DZ, Thangaraj V, Searl MM, Goff DC, Halpern E, Saper CB, Warach S. Schizophrenic subjects activate dorsolateral prefrontal cortex during a working memory task, as measured by fMRI. Biol Psychiatry 1999;45:1128–1137 59. Manoach DS, Gollub RL, Benson ES, Seal MM, Goff DC, Halpern E, Saper CB, Rauch SL. Schizophrenic subjects show aberrant fMRI activation of dorsolateral prefrontal cortex and basal ganglia during working memory performance. Biol Psychiatry 2000;48:99–109 60. Callicott JH, Bertolino A, Mattay VS, Langheim FJ, Duyn J, Coppola R, Goldberg TE, Weinberger DR. Physiological dysfunction of the dorsolateral prefrontal cortex in schizophrenia revisited. Cereb Cortex 2000;10:1078–1092 61. Callicott JH, Mattay VS, Verchinski BA, Marenco S, Egan MF, Weinberger DR. Complexity of prefrontal cortical dysfunction in schizophrenia: more than up or down. Am J Psychiatry 2003;160:2209–2215 62. Gur RE, Turetsky BI, Loughead J, Snyder W, Kohler C, Elliott M, Pratiwadi R, Ragland JD, Bilker WB, Siegel SJ, Kanes SJ, Arnold SE, Gur RC. Visual attention circuitry in schizophrenia investigated with oddball event-related functional magnetic resonance imaging. Am J Psychiatry 2007;164:442–449 63. Liddle PF, Laurens KR, Kiehl KA, Ngan ET. Abnormal function of the brain system supporting motivated attention in medicated patients with schizophrenia: an fMRI study. Psychol Med 2006;36:1097–1108 64. Drobyshevsky A, Baumann SB, Schneider W. A rapid fMRI task battery for mapping of visual, motor, cognitive, and emotional function. Neuroimage 2006;31:732–744 65. Manoach DS, Halpern EF, Kramer TS, Chang Y, Goff DC, Rauch SL, Kennedy DN, Gollub RL. Test-retest reliability of a functional MRI working memory paradigm in normal and schizophrenic subjects. Am J Psychiatry 2001;158:955–958 66. Winterer G, Musso F, Beckmann C, Mattay V, Egan MF, Jones DW, Callicott JH, Coppola R, Weinberger DR. Instability of prefrontal signal processing in schizophrenia. Am J Psychiatry 2006;163:1960–1968 67. Fox MD, Snyder AZ, Vincent JL, Raichle ME. Intrinsic fluctuations within cortical systems account for intertrial variability in human behavior. Neuron 2007;56:171–184 68. Jafri MJ, Calhoun VD. Functional classification of schizophrenia using feed forward neural networks. Conf Proc IEEE Eng Med Biol Soc 2006;Suppl:6631–6634 69. Liu Y, Liang M, Zhou Y, He Y, Hao Y, Song M, Yu C, Liu H, Liu Z, Jiang T. Disrupted small-world networks in schizophrenia. Brain 2008;131:945–961

82 70. Winterer G, Weinberger DR. Genes, dopamine and cortical signal-to-noise ratio in schizophrenia. Trends Neurosci 2004;27:683–690 71. Sawaguchi T, Goldman-Rakic PS. D1 dopamine receptors in prefrontal cortex: involvement in working memory. Science 191;251:947–950 72. Williams GV, Goldman-Rakic PS. Modulation of memory fields by dopamine D1 receptors in prefrontal cortex. Nature 1995;376:572–575 73. Daniel DG, Weinberger DR, Jones DW, Zigun JR, Coppola R, Handel S, Bigelow LB, Goldberg TE, Berman KF, Kleinman JE. The effect of amphetamine on regional cerebral blood flow during cognitive activation in schizophrenia. J Neurosci 1991;11:1907–1917 74. Mattay VS, Berman KF, Ostrem JL, Esposito G, Van Horn JD, Bigelow LB, Weinberger DR. Dextroamphetamine enhances “neural network-specific” physiological signals: a positron-emission tomography rCBF study. J Neurosci 1996; 16:4816–4822 75. Mattay VS, Callicott JH, Bertolino A, Heaton I, Frank JA, Coppola R, Berman KF, Goldberg TE, Weinberger DR. Effects of dextroamphetamine on cognitive performance and cortical activation. Neuroimage 2000;12:268–275 76. Mattay VS, Goldberg TE, Fera F, Hariri AR, Tessitore A, Egan MF, Kolachana B, Callicott JH, Weinberger DR. Catechol O-methyltransferase val158-met genotype and individual variation in the brain response to amphetamine. Proc Natl Acad Sci USA 2003;100:6186–6191 77. Tipper CM, Cairo TA, Woodward TS, Phillips AG, Liddle PF, Ngan ET. Processing efficiency of a verbal working memory system is modulated by amphetamine: an fMRI investigation. Psychopharmacology 2005;180:634–643 78. Dixon AL, Prior M, Morris PM, Shah YB, Joseph MH, Young AM. Dopamine antagonist modulation of amphetamine response as detected using pharmacological MRI. Neuropsychopharmacology 2005;48:236–245 79. Lawrence NS, Ross TJ, Stein EA. Cognitive mechanisms of nicotine on visual attention. Neuron 2002;36:539–548 80. Jacobsen LK, D’Souza DC, Mencl WE, Pugh KR, Skudlarski P, Krystal JH. Nicotine effects on brain function and functional connectivity in schizophrenia. Biol Psychiatry 2004;55: 850–858 81. Thiel CM, Zilles K, Fink GR. Nicotine modulates reorienting of visuospatial attention and neural activity in human parietal cortex. Neuropsychopharmacology 2005;30:810–820 82. Musso F, Bettermann F, Vucurevic G, Stoeter P, Konrad A, Winterer G. Smoking impacts on prefrontal attentional network function in young adult brains. Psychopharmacology 2007;191:159–169 83. Vossel S, Thiel CM, Fink GR. Behavioral and neural effects of nicotine on visuospatial attentional reorienting in nonsmoking subjects. Neuropsychopharmacology 2008;33: 731–738 84. Winterer G, Musso F, Konrad A, Vucurevic G, Stoeter P, Sander T, Gallinat J. Association of attentional network function with exon 5 variations of the CHRNA4 gene. Hum Mol Genetics 2007;16:2165–2174 85. Näätänen R, Paavilainen P, Rinne T, Alho K. The mismatch negativity (MMN) in basic research of central auditory processing: a review. Clin Neurophysiol 2007;118:2544–2590

G. Winterer 86. Todd J, Michie PT, Schall U, Karayanidis F, Yabe H, Näätänen R. Deviant matters: duration, frequency, and intensity deviants reveal different patterns of mismatch negativity reduction in early and late schizophrenia. Biol Psychiatry 2008;63:58–64 87. Javitt DC, Shelley A-M, Silipo G, Lieberman JA. Deficits in auditory and visual context-dependent processing in schizophrenia: defining the pattern. Arch Gen Psychiatry 2000;57:1131–1137 88. Salisbury DF, Shenton ME, Griggs CB, Bonner-Jackson A, McCarley RW. Mismatch Negativity in chronic schizophrenia and first-episode schizophrenia. Arch Gen Psychiatry 2002;59:686–694 89. Salisbury DF, Kuroki N, Kasai K, Shenton ME, McCarley RW. Progressive and interrelated functional and structural evidence for post-onset brain reduction in schizophrenia. Arch Gen Psych 2007;64:521–529 90. Umbricht D, Koller R, Schmid L, Skrabo A, Grübel C., Huber T, Stassen H. How specific are deficits in mismatch negativity generation to schizophrenia? Biol Psychiatry 2003;53:1120–1131 91. Kasai K, Shenton ME, Salisbury DF, Hirayasu Y, Onitsuka T, Spencer M, Yurgelun-Todd DA, Kikinis R, Jolesz FA, McCarley RW. Progressive decrease of left Heschl’s gyrus and planum temporale gray matter volume in schizophrenia: a longitudinal MRI study of first-episode patients. Arch Gen Psychiatry 2003;60:766–775 92. Turetsky BI, Calkins ME, Light GA, Olincy A, Radant AD, Swerdlow NR. Neurophysiological endophenotypes of schizophrenia: the viability of selected candidate measures. Schizophrenia Bull 2007;33:69–94 93. Light GA, Braff DL. Stability of mismatch negativity deficits and their relationship to functional impairments in chronic schizophrenia. Am J Psychiatry 2005;162:9 94. Umbricht D, Javitt D, Novak G, Bates J, Pollack S, Lieberman J, Kane J. Effects of clozapine on auditory event-related potentials in schizophrenia. Biol Psychiatry 1998;44:716–725 95. Hall MH, Rijsdijk F, Picchioni M, Schulze K, Ettinger U, Toulopoulou T, Bramon E, Murray RM, Sham P. Substantial shared genetic influences on schizophrenia and event-related potentials. Am J Psychiatry 2007;164:84–82 96. Jessen F, Fries T, Kucharski C, Nishimura T, Hoenig K, Maier W, Falkai P, Heun R. Amplitude reduction of the mismatch negativity in first-degree relatives of patients with schizophrenia. Neurosci Lett 2001;309:185–188 97. Michie PT, Innes-Brown H, Todd J, Jablensky AV. Duration mismatch negativity in biological relatives of patients with schizophrenia spectrum disorders. Biol Psychiatry 2002;52: 749–758 98. Bramon E, Rabe-Hesketh S, Shama P, Murray RM, Frangou S. Meta-analysis of the P300 and P50 waveforms in schizophrenia. Schiz Res 2004;70:315–329 99. Stone JM, Pilowsky LS. Novel targets for drugs in schizophrenia. CNS Neurol Disord Drug Targets 2007;6:265–272 100. Javitt DC. Intracortical mechanisms of mismatch negativity dysfunction in schizophrenia. Audiol Neurootol 2000;5: 207–215 101. Javitt DC, Steinschneider M, Schroeder CE, Arezzo JC. Role of cortical N-methyl-D-aspartate receptors in auditory sensory memory and mismatch negativity generation:

4

Translational Medicine

102.

103.

104. 105.

106.

107.

108.

109.

110.

111.

112.

113.

114.

115.

116.

implications for schizophrenia. Proc Natl Acad Sci U S A 1996;93:11962–11967 Javitt DC, Jayachandra M, Lindsley RW, Specht CM, Schroeder CE. Schizophrenia-like deficits in auditory P1 and N1 refractoriness induced by the psychomimetic agent phencyclidine (PCP). Clin Neurophysiol 2000;111:833–836 Umbricht D, Schmid L, Koller R, Vollenweider FX, Hell D, Javitt DC. Ketamine-induced deficits in auditory and visual context-dependent processing in healthy volunteers: implications for models of cognitive deficits in schizophrenia. Arch Gen Psychiatry 2000;57:1139–1147 Tamminga CA. The neurobiology of cognition in schizophrenia. J Clin Psychiatry 2006;67Suppl9:9–13 Slawecki CJ, Thomas JD, Riley EP, Ehlers CL. Neonatal nicotine exposure alters hippocampal EEG and eventrelated potentials (ERPs) in rats. Pharmacol Biochem Behav 2000;65:711–718 Ehlers CL, Somes C. Long latency event-related potentials in mice: effects of stimulus characteristics and strain. Brain Res 2002;957:117–128 Halgren E, Baudena P, Clarke JM, Heit G, Liégeois C, Chauvel P, Musolino A. Intracerebral potentials to rare target and distractor auditory and visual stimuli. I. Superior temporal plane and parietal lobe. Electroencephalogr Clin Neurophysiol 1995;94:191–220 Halgren E, Baudena P, Clarke JM, Heit G, Marinkovic K, Devaux B, Vignal JP, Biraben A. Intracerebral potentials to rare target and distractor auditory and visual stimuli. II. Medial, lateral and posterior temporal lobe. Electroencephalogr Clin Neurophysiol 1995;94:229–250 Baudena P, Halgren E, Heit G, Clarke JM. Intracerebral potentials to rare target and distractor auditory and visual stimuli. III. Frontal cortex. Electroencephalogr Clin Neurophysiol 1995;94:251–264 Linden DE, Prvulovic D, Formisano E, Völlinger M, Zanella FE, Goebel R, Dierks T. The functional neuroanatomy of target detection: an fMRI study of visual and auditory oddball tasks. Cereb Cortex 1999;9:815–823 Winterer G, Musso F, Konrad A, Vucurevic G, Stoeter P, Sander T, Gallinat J. Association of attentional network function with exon 5 variations of the CHRNA4 gene. Hum Mol Genet 2007;16:2165–2174 Winterer G, Mulert C, Mientus S, Gallinat J, Schlattmann P, Dorn H, Herrmann WM. P300 and LORETA: comparison of normal subjects and schizophrenic patients. Brain Topogr 2001;13:299–313 Pae JS, Kwon JS, Youn T, Park HJ, Kim MS, Lee B, Park KS. LORETA imaging of P300 in schizophrenia with individual MRI and 128-channel EEG. Neuroimage 2003;20: 1552–1560 Roth WT, Cannon EH. Some features of the auditory evoked response in schizophrenics. Arch Gen Psychiatry 1972;27:466–471 Ford JM, White P, Lim KO, Pfefferbaum A. Schizophrenics have fewer and smaller P300s: a single-trial analysis. Biol Psychiatry 1994;35:96–103 Salisbury DF, O’Donnell BF, McCarley RW, Nestor PG, Faux SF, Smith, RS. Parametric manipulations of auditory stimuli differentially affect P3 amplitude in schizophrenics and controls. Psychophysiology 1994;31:29–36

83 117. McCarley RW, Shenton ME, O’Donnell BF, Faux SF, Kikinis R, Nestor PG, Jolesz FA. Auditory P300 abnormalities and left posterior superior temporal gyrus volume reduction in schizophrenia. Arch Gen Psychiatry 1993;50: 190–197 118. Turetsky B, Colbath EA, Gur RE. P300 subcomponent abnormalities in schizophrenia: II. Longitudinal stability and relationship to symptom change. Biol Psychiatry 1998;43: 31–39 119. Gallinat J, Riedel M, Juckel G, Sokullu S, Frodl T, Moukhtieva R, Mavrogiorgou P, Nisslé S, Müller N, Danker-Hopfe H, Hegerl U. P300 and symptom improvement in schizophrenia. Psychopharmacology (Berl) 2001;158:55–65 120. Frodl T, Meisenzahl EM, Müller D, Holder J, Juckel G, Möller HJ, Hegerl U. P300 subcomponents and clinical symptoms in schizophrenia. Int J Psychophysiol 2002;43: 237–246 121. Morstyn RM, Duffy FH, McCarley RW. Altered P300 topography in schizophrenia. Arch Gen Psych 1983;40: 729–734 122. Strik WK, Dierks T, Franzek E, Maurer K, Beckmann H. Differences in P300 amplitudes and topography between cycloid psychosis and schizophrenia in Leonhard’s classification. Acta Psychiatr Scand 1993;87:179–183 123. Salisbury DF, Shenton ME, McCarley RW. P300 topography differs in schizophrenia and manic psychosis. Biol Psychiatry 1999;45:98–106 124. O’Donnell BF, Vohs JL, Hetrick WP, Carroll CA, Shekhar A. Auditory event-related potential abnormalities in bipolar disorder and schizophrenia. Int J Psychophysiol 2004;53:45–55 125. Donchin E, Callaway E, Jones RT. Auditory evoked potential variability in schizophrenia. II. The application of discriminant analysis. Electroencephalogr Clin Neurophysiol 1970;29:429–440 126. Callaway E, Jones RT, Donchin E. Auditory evoked potential variability in schizophrenia. Electroencephalogr Clin Neurophysiol 1970;29:421–428 127. Roth WT, Pfefferbaum A, Kelly AF, Berger PA, Kopell BS. Auditory event-related potentials in schizophrenia and depression. Psychiatry Res 1981;4:199–212 128. Ford JM, White P, Lim KO, Pfefferbaum A. Schizophrenics have fewer and smaller P300s: a single-trial analysis. Biol Psychiatry 1994;35:96–103 129. Röschke J, Wagner P, Mann K, Fell J, Grözinger M, Frank C. Single trial analysis of event related potentials: a comparison between schizophrenics and depressives. Biol Psychiatry 1996;40:844–852 130. Roth A, Roesch-Ely D, Bender S, Weisbrod M, Kaiser S. Increased event-related potential latency and amplitude variability in schizophrenia detected through wavelet-based single trial analysis. Int J Psychophysiol;66:244–254 131. Spencer KM. Averaging, detection, and classification of single-trial ERPs. In: Handy T, ed. Event-Related Potentials: A Methods Handbook. Cambridge, MA: MIT Press, 2004, pp. 209–227 132. Winterer G, Egan MF, Raedler T, Sanchez C, Jones DW, Coppola R, Weinberger DR. P300 and genetic risk for schizophrenia. Arch Gen Psychiatry 2003;60:1158–1167 133. Mathalon DH, Ford JM, Pfefferbaum A. Trait and state aspects of P300 amplitude reduction in schizophrenia: a

84

134.

135.

136.

137.

138.

139.

140.

141.

142.

143.

144.

145.

146.

G. Winterer retrospective longitudinal study. Biol Psychiatry 2000;47: 434–449 Winterer G, Coppola R, Goldberg TE, Egan MF, Jones DW, Sanchez CE, Weinberger DR. Prefrontal broadband noise, working memory, and genetic risk for schizophrenia. Am J Psychiatry 2004;161:490–500 Gallinat J, Götz T, Kalus P, Bajbouj M, Sander T, Winterer G. Genetic variations of the NR3A subunit of the NMDA receptor modulate prefrontal cerebral activity in humans. J Cogn Neurosci 2007;19:59–68 Hill SY, Locke J, Zezza N, Kaplan B, Neiswanger K, Steinhauer SR, Wipprecht G, Xu J. Genetic association between reduced P300 amplitude and the DRD2 dopamine receptor A1 allele in children at high risk for alcoholism. Biol Psychiatry 1998:43;40–51 Anokhin AP, Todorov AA, Madden PA, Grant JD, Heath AC. Brain event-related potentials, dopamine D2 receptor gene polymorphism, and smoking. Genet Epidemiol 1999;17Suppl1:S37–S42 Mulert C, Juckel G, Giegling I, Pogarell O, Leicht G, Karch S, Mavrogiorgou P, Möller HJ, Hegerl U, Rujescu D. A Ser9Gly polymorphism in the dopamine D3 receptor gene (DRD3) and event-related P300 potentials. Neuropsychopharmacology 2004;31:1335–1344 Berman SM, Noble EP, Antolin T, Sheen C, Conner BT, Ritchie T. P300 development during adolescence: effects of DRD2 genotype. Clin Neurophysiol 2006;117:649–659 Gallinat J, Bajbouj M, Sander T, Schlattmann P, Xu K, Ferro EF, Goldman D, Winterer G. Association of the G1947A COMT (Val(108/158)Met) gene polymorphism with prefrontal P300 during information processing. Biol Psychiatry 2003;54:40–48 Golimbet V, Gritsenko I, Alfimova M, Lebedeva I, Lezheiko T, Abramova L, Kaleda V, Ebstein R. Association study of COMT gene Val158Met polymorphism with auditory P300 and performance on neurocognitive tests in patients with schizophrenia and their relatives. World J Biol Psychiatry 2006:7;238–245 Ehlis AC, Reif A, Herrmann MJ, Lesch KP, Fallgatter AJ. Impact of catechol-O-methyltransferase on prefrontal brain functioning in schizophrenia spectrum disorders. Neuropsychopharmacology 2007;32:162–170 Lin CH, Yu YW, Chen TJ, Tsa SJ, Hong CJ. Association analysis for dopamine D2 receptor Taq1 polymorphism with P300 event-related potential for normal young females. Psychiatr Genet 2001;11:165–168 Tsai SJ, Yu YW, Chen TJ, Chen MC, Hong CJ. Association analysis for dopamine D3 receptor, dopamine D4 receptor and dopamine transporter genetic polymorphisms and P300 event-related potentials for normal young females. Psychiatr Genet 2003;13:51–53 Bramon E, Dempster E, Frangou S, McDonald C, Schoenberg P, MacCabe JH, Walshe M, Sham P, Collier D, Murray RM. Is there an association between the COMT gene and P300 endophenotypes? Eur Psychiatry 2006;21:70–73 Oranje B, Gispen-de Wied CC, Westenberg HG, Kemner C, Verbaten MN, Kahn RS. No effects of l-dopa and bromocriptine on psychophysiological parameters of human selective attention. J Psychopharmacol 2006;20:789–798

147. López J, López V, Rojas D, Carrasco X, Rothhammer P, García R, Rothhammer F, Aboitiz F. Effect of psychostimulants on distinct attentional parameters in attentional deficit/ hyperactivity disorder. Biol Res 2004;37:461–468 148. Niznikiewicz MA, Patel JK, McCarley R, Sutton J, Chau DT, Wojcik J, Green AI. Clozapine action on auditory P3 response in schizophrenia. Schizophr Res 2005;76:19–21 149. Hollander E, Davidson M, Mohs RC, Horvath TB, Davis BM, Zemishlany Z, Davis KL. RS 86 in the treatment of Alzheimer’s disease: cognitive and biological effects. Biol Psychiatry 1987;22:1067–1078 150. Dierks T, Frölich L, Ihl R, Maurer K. Event-related potentials and psychopharmacology. Cholinergic modulation of P300. Pharmacopsychiatry 1994;27:72–74 151. Anokhin AP, Vedeniapin AB, Sirevaag EJ, Bauer LO, O’Connor SJ, Kuperman S, Porjesz B, Reich T, Begleiter H, Polich J, Rohrbaugh JW. The P300 brain potential is reduced in smokers. Psychopharmacology (Berl) 2000;149:409–413 152. Thomas A, Iacono D, Bonanni L, D’Andreamatteo G, Onofrj M. Donepezil, rivastigmine, and vitamin E in Alzheimer disease: a combined P300 event-related potentials/neuropsychologic evaluation over 6 months. Clin Neuropharmacol 2000;24:31–42 153. Knott V, Mohr E, Mahoney C, Engeland C, Ilivitsky V. Effects of acute nicotine administration on cognitive eventrelated potentials in tacrine-treated and non-treated patients with Alzheimer’s disease. Neuropsychobiology 2002;45: 156–160 154. Neuhaus A, Bajbouj M, Kienast T, Kalus P, von Haebler D, Winterer G, Gallinat J. Persistent dysfunctional frontal lobe activation in former smokers. Psychophamacology (Berl) 2006;186:191–200 155. Werber AE, Klein C, Rabey JM. Evaluation of cholinergic treatment in demented patients by P300 evoked related potentials. Neurol Neurochir Pol 2001;35 Suppl 3:37–43 156. Harrison JB, Buchwald JS, Kaga K, Woolf NJ, Butcher LL. ‘Cat P300’ disappears after septal lesions. Electroencephalogr Clin Neurophysiol 1988;69:55–64 157. Fries P, Nikolic D, Singer W. The gamma cycle. Trends Neurosci 2007;30:309–316 158. Spencer KM, Salisbury DF, Shenton ME, McCarley RW. Gamma-band steady-state responses are impaired in first episode psychosis. Soc Neurosci Abstr 2006;36:122 159. Ford JM, Roach BJ, Faustman WO, Mathalon DH. Out-ofsynch and Out-of-sorts: dysfunction of motor-sensory communicatio in schizophrenia. Biol Psychiatry [epub ahead of print] 160. Spencer KM, Nestor PG, Perlmutter R, Niznikiewicz MA, Klump MC, Frumin M, Shenton ME, McCarley RW. Neural synchrony indexes disordered perception and cognition in schizophrenia. Proc Natl Acad Sci USA 2004;101: 17288–17293 161. Selemon LD, Goldman-Rakic PS. The reduced neuropil hypothesis: a circuit based model of schizophrenia. Biol Psychiatry 1999;45:17–25 162. Kopell N, Ermentrout GB, Whittington MA, Traub RD. Gamma rhythms and beta rhythms have different synchronization properties. Proc Natl Acad Sci USA 2000;97: 1867–1872

4

Translational Medicine

163. Uhlhaas P, Silverstein SM, Phillips WA, Lovell PG. Evidence for impaired visual context processing in schizotypy with thought disorder. Schizophr Res 2004;68:249–260 164. Gallinat J, Winterer G, Herrmann CS, Senkowski D. Reduced oscillatory gamma-band responses in unmedicated schizophrenic patients indicate impaired frontal network processing. Clin Neurophysiol 2004; 115:1863–1874 165. Hong L, Summerfelt A, McMahon R, et al. Evoked gamma band synchronization and the liability for schizophrenia. Schizophr Res. 2004;70:293–302 166. Van der Stelt O, Belger A, Lieberman JA. Macroscopic fast neuronal oscillations and synchrony in schizophrenia. Proc Natl Acad Sci USA 2004;101:17567–17568 167. Herrmann CS, Demiralp T. Human EEG gamma oscillations in neuropsychiatric disorders. Clin Neurophysiol 2005;116:2719–2733 168. Brenner CA, Sporns O, Lysaker PH, O’Donnell BF. EEG synchronization to modulated auditory tones in schizophrenia, schizoaffective disorder, and schizotypal personality disorder. Am J Psychiatry 2003;160:2238–2240 169. Lewis DA, Hashimoto T, Volk DW. Cortical inhibitory neurons and schizophrenia. Nat Rev Neurosci 2005;6: 312–324 170. Cunningham MO, Hunt J, Middleton S, LeBeau FE, Gillies MG, Davies CH, Maycox PR, Whittington MA, Racca C. Region-specific reduction in entorhinal gamma oscillations and parvalbumin-immunoreactive neurons in animal models of psychiatric illness. J Neurosci 2006;26:2767–2776 171. Grunze HC, Rainnie DG, Hasselmo ME, Barkai E, Hearn EF, McCarley RW, Greene RW. NMDA-dependent modulation of CA1 local circuit inhibition. J Neurosci 1996;1: 2034–2043 172. Rujescu D, Bender A, Keck M, Hartmann AM, Ohl F, Raeder H, Giegling I, Genius J, McCarley RW, Möller H-Y, Grunze H. A pharmacological model for psychosis based on N-methyl-D-aspartate receptor hypofunction: molecular, cellular, functional and behavioral abnormalities. Biol Psychiatry 2006;59:721–729 173. Buzsaki G. Theta oscillations in the hippocampus. Neuron 2002;33:325–340 174. Hasselmo ME. What is the function of hippocampal theta rhythm?–Linking behavioral data to phasic properties of field potential and unit recording data. Hippocampus 2005;15:936–949 175. Vertes RP. Hippocampal theta rhythm: a tag for short-term memory. Hippocampus 2005;15:923–935 176. Miller R. Discovery and general behavioural correlates of the hippocampal theta rhythm in several species. In: Miller R, ed. Corticohippocampal Interplay and the Representation of Context in the Brain. Berlin: Springer, 1991 177. Leung L-WS, Borst JGG. Electrical activity of the cingulate cortex. 1. generating mechanisms and relations to behavior. Brain Res 1987;407:68–80 178. Borst JGG, Leung L-WS, MacFabe DF Electrical activity of the cingulate cortex. II. Cholinergic modulation. Brain Res 1987;407:81–93 179. Talk A, Kang E, Gabriel M. Independent generation of theta rhythm in the hippocampus and posterior cingulate cortex. Brain Res 2004;1015:15–24

85 180. Tsujimoto T, Shimazu H, Isomura Y. Direct recording of theta oscillations in primate prefrontal and anterior cingulate cortices. J Neurophysiol 2006;95:2987–3000 181. Cantero JL, Atienza M, Stickgold R, Kahana MJ, Madsen JR, Kocsis B. Sleep-dependent theta oscillations in the human hippocampus and neocortex. J Neurosci 2003;23:10897–10903 182. Guderian S, Düzel E. Induced theta oscillations mediate large-scale synchrony with mediotemporal areas during recollection in humans. Hippocampus 2005;15:901–912 183. Winterer G, Ziller M, Dorn H, Frick K, Mulert C, Dahhan N, Herrmann WM, Coppola R. Cortical activation, signal-tonoise ratio and stochastic resonance during information processing in man. Clin Neurophysiol 1999;110:1193–1203 184. Ishihara T, Yoshii N. Multivariate analytic study of EEG and mental activity in juvenile delinquents. Electroencephalogr Clin Neurophysiol 1972;33:71–80 185. Gevins A, Smith ME, McEvoy L, Yu D. High-resolution EEG mapping of cortical activation related to working memory: effects of task difficulty, type of processing, and practice. Cereb Cortex 1997;7:374–385 186. Gevins A, Smith ME. Neurophysiological measures of working memory and individual differences in cognitive ability and cognitive style. Cereb Cortex 2000;10:829–839 187. Jensen O, Lisman JE. Hippocampal sequence-encoding driven by a cortical multi-item working memory buffer. Trends Neurosci 2005;28:67–72 188. Jensen O, Colgin LL. Cross-frequency coupling between neuronal oscillations. Trends Cogn Sci 2007;11:267–269 189. Jones MW, Wilson MA. Theta rhythms coordinate hippocampal-prefrontal interactions in a spatial memory task. PLoS Biol 2005;3:e402 190. Jacobs J, Kahana MJ, Ekstrom AD, Fried I. Brain oscillations control timing of single-neuron activity in humans. J Neurosci 2007;27:3839–3844 191. Hyman JM, Zilli EA, Paley AM, Hasselmo ME. Medial prefrontal cortex cells show dynamic modulation with the hippocampal theta rhythm dependent on behavior. Hippocampus 2005;15:739–749 192. Winterer G, Ziller M, Dorn H, Frick K, Mulert C, Wuebben Y, Herrmann WM, Coppola R. Schizophrenia: reduced signalto-noise ratio and impaired phase-locking during information processing. Clin Neurophysiol 2000;111: 837–849 193. Krishnan GP, Vohs JL, Hetrick WP, Carroll CA, Shekhar A, Bockbrader MA, O’Donnell BF. Steady state visual evoked potential abnormalities in schizophrenia. Clin Neurophysiol 2004;116:614–624 194. Uhlhaas P, Rodriguez R, Roux F, Haenschel C, Maurer K, Singer W. Neuronal synchrony as a pathophysiological mechanism in schizophrenia. 1st European Conference on Schizophrenia Research (oral presentation) 2007. 195. Winterer G, Musso F, Beckmann C, Mattay V, Egan MF, Jones DW, Callicott JH, Coppola R, Weinberger DR. Instability of prefrontal signal processing in schizophrenia. Am J Psychiatry 2006;163:1960–1968 196. Fitch TE, Sahr RN, Eastwood BJ, Zhou FC, Yang CR. Dopamine D1/D5 receptor modulation of firing rate and bidirectional theta burst firing in medial septal/vertical limb of diagonal band neurons in vivo. J Neurophysiol 2006;95:2808–2820

86 197. Winterer G, Egan MF, Kolachana BS, Goldberg TE, Coppola R, Weinberger DR. Prefrontal electrophysiologic “noise” and catechol-O-methyltransferase genotype in schizophrenia. Biol Psychiatry 2006;60:578–584 198. Tan HY, Callicott JH, Weinberger DR. Dysfunctional and compensatory prefrontal cortical systems, genes and the pathogenesis of schizophrenia. Cereb Cortex 2007;Suppl1:171–181

G. Winterer 199. Winterer G, Musso F, Vucurevic G, Stoeter P, Konrad A, Seker B, Gallinat J, Dahmen N, Weinberger DR. COMT genotype predicts BOLD signal and noise characteristics in prefrontal circuits. Neuroimage 2006;32:1722–1732 200. Winterer G, Weinberger DR. Genes, dopamine and cortical signal-to-noise ratio in schizophrenia. Trends Neurosci 2004;27:683–690

Chapter 5

Leveraging High-Dimensional Neuroimaging Data in Genetic Studies of Neuropsychiatric Disease Cinnamon S. Bloss, Trygve E. Bakken, Alexander H. Joyner, and Nicholas J. Schork

Abstract The current state of biomedical science is such that both the number and sophistication of methods available to investigate the genetic determinants of disease is unprecedented. For example, the introduction of high-throughput technologies such as DNA microarrays, allow researchers to comprehensively assess the human genome for single nucleotide polymorphisms that confer genetic susceptibility. Indeed, while these, and other similarly sophisticated methods, have yielded notable findings with regard to identification of risk variants in diseases such as diabetes, obesity, and glaucoma, similar studies of neuropsychiatric diseases such as schizophrenia and bipolar disorder have been somewhat less successful in producing strong findings. The reasons why this is the case are numerous, but likely reflect the very complex genetic architecture of neuropsychiatric conditions. In this chapter, we consider an approach to addressing this complexity that involves the use of what are termed ‘endophenotypes’ (or alternatively ‘intermediate phenotypes’) in genetic studies of neuropsychiatric disorders. Endophenotypes are biological changes, such as

C.S. Bloss, T.E. Bakken, A.H. Joyner and N.J. Schork () Scripps Genomic Medicine and Scripps Translational Science Institute, and Scripps Health, La Jolla, CA, USA

brain structural differences, that are thought to represent underlying molecular, physiologic, or otherwise subclinical changes resulting directly from the genetic variations that mediate susceptibility to overt clinical disease. Furthermore, neuroimaging phenotypes are, for a variety of reasons, thought to represent good candidate endophenotypes for genetic association studies of neuropsychiatric disease. Like high-dimensional genome-wide data, however, these phenotyping technologies can produce hundreds to thousands of data points or more, when all neuroanatomic regions and tissue types of interest are considered. The question then becomes, how can two or more high-dimensional data types (i.e., in this case genomic and neuroimaging) be leveraged, integrated, and analyzed in order to make valid inferences about the genetic basis of neuropsychiatric disease? We comment on the analytic issues that arise when trying to leverage both genome-wide genetic data and neuroimaging data (e.g., problems related to multiple comparisons and false positives, as well as small sample sizes), and discuss four general approaches, each with its own set of advantages and disadvantages, that can be used in the analysis of imaging-genetics data. Finally, we provide a brief review of some of the recent studies that combine imaging and genetics, but note that the field, as a whole, is still very much in its infancy. We also provide suggestions for future directions.

T.E. Bakken Medical Scientist Training Program and Graduate Program in Neurosciences, University of California, San Diego, CA, USA

Keywords Neuroimaging • imaging-genetics • multivariate • genomics • statistical analysis

A.H. Joyner Graduate Program in Biomedical Sciences, University of California, San Diego, CA, USA

Abbreviations SNPs Single nucleotide polymorphisms; CNVs Copy number variants; GWA Studies Genome Wide Association Studies; DSM Diagnostic and Statistical Manual of Mental Disorders; NHGRI National Human Genome Research Institute; MRS

N.J. Schork Department of Molecular and Experimental Medicine, The Scripps Research Institute, La Jolla, CA, USA

M.S. Ritsner (ed.), The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes, © Springer Science + Business Media B.V. 2009

87

88

Magnetic resonance spectroscopy; NAA N-Acetyl aspartate; Cho Choline; DTI Diffusion tensor imaging; MD Diffusion; FA Fractional anisotropy; CT Computerized tomography; MRI Magnetic resonance imaging; fMRI Functional MRI; PET Positron emission tomography; SPECT Single photon emission computed tomography; BOLD Blood oxygen leveldependent signal; BP Binding potential; ANOVA Analysis of variance; PCA Principal component analysis; MDMR Multivariate distance matrix regression;

Introduction There has been a rapid increase in both the number and sophistication of methods available to investigate the genetic determinants of disease; in fact, the current research environment is virtually unprecedented in this regard. For example, the availability of highthroughput technologies such as DNA microarrays now enable scientists to scan the entire human genome for single nucleotide polymorphisms (SNPs), copy number variations (CNVs), and gene expression patterns that underlie disease risk. In fact, in their catalog of published genome-wide association (GWA) studies, the National Human Genome Research Institute (NHGRI) lists more than a hundred GWA studies published since 2007, versus only eight studies published during 2006 (http://www.genome. gov/26525384; May 29, 2008). Application of high-throughput genomic assays and technologies has yielded many important findings with respect to several human conditions (see Table 5.1). Prominent among these are identification of risk variants associated with diseases such as diabetes,1,2 obesity,3 and glaucoma,4 among others.5–12 Upon close examination of this literature, however, one can see that relative to other types of diseases (e.g., cardiovascular), the identification of genetic determinants of neuropsychiatric disorders has proved challenging.13,14 For example, although GWA studies of both schizophrenia15–17 and bipolar disorder13,18,19 have revealed some notable results (see Table 5.2), it has been difficult to detect signals that meet stringent criteria for genome-wide significance, and replication of findings has been difficult.

C.S. Bloss et al. Table 5.1 Selected recent whole genome association studies and discoveries Disease/Trait

Gene(s)/Loci

Celiac disease

RGS1, IL1RL1, IL18R1, IL18RAP, SLC9A4, IL12A, SCHIP1, LPP, TAGAP Colorectal cancer DQ515897, SMAD7 C-reactive protein LEPR, HNF1A, GCKR, IL6R Fetal hemoglobin levels HBB, BCL11A Glaucoma LOXL1 Nicotine dependence, CHRNA3, CHRNA5, lung cancer and CHRNB4 peripheral arterial disease Prostate cancer MSMB, KLK3, SLC22A3, LMTK2 Systemic lupus ITGAM, KIAA1542, erythematosus PXK Type 2 diabetes JAZF1, CDC123, CAMK1D, TSPAN8, LGR5, THADA, ADAMTS9

Reference [7]

[9] [8] [11] [4] [10]

[5] [6] [12]

Table 5.2 Selected recent whole genome association studies and discoveries in bipolar disorder and schizophrenia Disease/Trait

Gene(s)/Loci

Reference

Bipolar disorder Bipolar disorder Bipolar disorder Schizophrenia Schizophrenia Schizophrenia

MYO5B, TSPAN8, CACNA1C DGKH PALB2, NDUFAB1, DCTN5 CCDC60 RELN (in women) CSF2RA

[18] [19] [13] [17] [16] [15]

Genetic Associations in Neuropsychiatric Disorders There are many possible reasons for the lack of replicated, compelling findings in genetic studies of neuropsychiatric disorders. Some of these reasons include the following: inadequate power to detect alleles of modest effect sizes, population-specific locus effects, epistatic interactions of multiple genes each with a modest independent effect, gene-environment interactions, and effects of copy number variants or other

5

Leveraging High-Dimensional Neuroimaging Data in Genetic Studies

forms of genetic variation not well captured by the panels of common SNPs that have been used, including multiple rare disease alleles.18 In addition to these reasons, other possible explanations relate to phenotypic heterogeneity and the belief that there is simply an inherent imprecision associated with the diagnostic categories in neuropsychiatric disease.20 Indeed, traditional categorical diagnostic criteria for neuropsychiatric disorders, i.e., those delineated by the Diagnostic and Statistical Manual of Mental Disorders (DSM),21 lack more ‘objective’ or precise subclinical biomarkers in their definitions. This is in contrast, for example, with those subclinical measures used to make ‘medical’ diagnoses such as hypertension (i.e., diagnosed via a blood pressure test) or cancer (i.e., diagnosed via biopsy). In addition, another possible explanation for lack of compelling genetic associations in neuropsychiatric disorders is rooted in the notion that genes do not encode for ‘psychopathology’. Rather, as has been discussed widely in the neuropsychiatric literature, it is likely that the extent to which genes are associated with clusters of symptoms we define as representing a neuropsychiatric disorder (e.g., delusions and hallucinations in schizophrenia, depression and mania in bipolar disorder), they do so by affecting brain cells and neural systems that mediate these clinical features.22 Hence the search for genes that mediate heterogeneous, often vague and mechanistically imprecise, clinical diagnostic categorizations may be ill-conceived and should be supplanted, or at least supplemented, by genetic studies of underlying subclinical, mechanisticallyoriented phenotypes. Such phenotypes are referred to as ‘endophenotypes’ (or ‘intermediate phenotypes’) in the literature because they are thought to operate at a level that is intermediate in the chain of hierarchical biochemical and physiologic reactions that lead from gene action to overt, clinical phenotypic expression.

Endophenotypes for Neuropsychiatric Disorders Tan and colleagues22 provide a cogent review of the concept of intermediate, neurobiological endophenotypes and advocate their use in genetic association studies of neuropsychiatric disorders – the logic being that abnormal behavior reflects abnormal brain function. The

89

opposite notion, however, that abnormal brain function always results in abnormal behavior, is not necessarily the case since compensation by other brain systems occurs and is in fact, not uncommon (e.g., observing plaques and tangles at autopsy in an individual that did not carry a clinical diagnosis of Alzheimer’s disease). An extension of this argument is that differences in brain function associated with genetic variations may or may not result in observable behavior differences. Indeed, the changes in cognition, brain structure, and brain function that are observed in individuals with neuropsychiatric disorders are also found with greater frequency in their unaffected biological relatives than in unrelated control subjects. This suggests that these biological changes are related to disease susceptibility and are not sufficient (though possibly necessary) causes of disease. Hence these may be considered ‘intermediate’ or endophenotypes (see Fig. 5.1) of overt, clinical manifestations of disease.22–30 Although objections have been raised against the use of endophenotypes in genetic studies of neuropsychiatric disease31 there seems to be a growing consensus among members of the psychiatric community that the use of these measures may be preferable to categorical clinical diagnostic categories. One potential benefit of the use of endophenotypes is that they may increase statistical power to detect genetic associations and linkages.32 Neuroimaging phenotypes are considered to be paradigmatic candidate endophenotypes for neuropsychiatric disease as they are quantitative and can be used to identify replicable patterns and perturbations associated with particular diseases. There is also mounting evidence that phenotypes derived from both functional and structural neuroimaging studies23,33–35 are not only heritable, but also show clear associations with various neuropsychiatric diseases as well as increased risk of disease (i.e., among unaffected biological relatives of individuals with disease).24 Moreover, many brain functional and structural phenotypes have been shown to appear prior to the onset of clinical symptoms and the full expression of a given disease. In short, it is clear that there is likely substantial value to using neuroimaging-based endophenotypes in genetic association studies of neuropsychiatric disease. In this light, an important question arises as to what the best methods for integrating and analyzing neuroimaging endophenotypes in genetic studies might be.

90

C.S. Bloss et al.

Fig. 5.1 New approaches to traditional genetic linkage studies will be needed to unravel the connection between genes, cells, circuits and cognition. Neuroimaging data can increase the power of these studies by providing information about endophenotypes. These heritable markers measure biological processes that are closer to the genome than overt behavioral phenotypes, and so can be more sensitive to gene effects than behavioral or clinical measures. Note that the biological level interrogated does not necessarily correspond to the resolution of the neuroimaging technique. For example, PET gives lower spatial resolution than MRI even though it provides information about a microscopic feature of the brain

The field of imaging-genetics, still in its infancy, has largely been restricted to candidate gene approaches thus far. Although analysis of neuroimaging and candidate gene data is certainly nontrivial, a whole host of new issues are introduced when imaging-based phenotypes are used in GWA studies. Before we address this issue, however, we provide a brief description of the primary neuroimaging modalities in order to point out the unique challenges associated with each in the context of genetic studies of neuropsychiatric diseases.

Neuroimaging Modalities and Data Many imaging modalities are available to the researcher and clinician to better characterize brain structure and

function and their relationships to neuropsychiatric disease state and progression.36 These different modalities interrogate the brain and associated neurobiological systems at different levels. The primary neuroimaging modalities include structural, functional, magnetic resonance spectroscopy (MRS), and diffusion tensor imaging (DTI). We highlight the characteristics associated with the data generated by these techniques to point out issues in their use in genetic association studies. There are other imaging-based techniques in use (e.g., magnetoencephalography), as well as techniques that are under development, including methods that represent advances in post-processing such as voxel-based morphometry and techniques that aim to combine multiple methods already in use such as multimodal imaging.36 An in-depth discussion of all of these methods, however, is beyond the scope of this chapter.

5

Leveraging High-Dimensional Neuroimaging Data in Genetic Studies

Structural Neuroimaging Structural neuroimaging techniques are non-invasive procedures that map the anatomy of the brain. Although computerized tomography (CT) and magnetic resonance imaging (MRI) are both widely used in clinical settings to assess brain structure in neuropsychiatric disorders, we will focus our description on MRI as this method produces greater contrast than CT and thus enables visualization of a much greater range of tissue types and neural abnormalities than CT; MRI is also more commonly used in research studies. MRI is a technique that takes advantage of the magnetic potential and abundance of hydrogen in the human body in order to create high contrast images of normal and diseased tissue.36 At a very basic level, each hydrogen atom in the human body has a ‘spin’ with a positive charge that creates a small magnetic field. Typically the orientation of this field is random, but when in the large external magnetic field of an MRI scanner (i.e., more than 20,000 times the Earth’s magnetic field), these atoms align with the external magnetic field. Different structural imaging procedures and protocols (e.g., spin-echo, fast spin-echo, and gradient-echo imaging with varying echo times, flip angles, and repetition times, T1 weighted, T2 weighted, and proton density) produce different types of images. Postprocessing procedures can then be applied to the data, including segmentation algorithms, to produce volume measurements of different regions and neural tissues of interest. In any given study, volume measurements are typically generated for a very large number of neuroanatomic regions and subregions, which are then used in data analysis, e.g., see.37–40

Functional Neuroimaging Although there are many types of ‘functional’ imaging, we confine attention to functional MRI (fMRI) and emission tomography, which includes both positron emission tomography (PET) and single photon emission computed tomography (SPECT). Administered in conjunction with structural MRI, fMRI is a noninvasive procedure used to identify brain regions that are ‘active’ during a specific cognitive task, which is completed

91

while the subject is undergoing scanning. Notably, however, fMRI does not actually detect neural activity during a given task; rather, it tracks local changes in oxygen utilization during the task, and the assumption is that this activity is correlated with neuronal transmission. Specifically, fMRI leverages variable signal intensities that occur as a result of altered levels of cerebral blood oxygenation.36 The structural MRI images collected concurrently allow the functional data to be accurately mapped back onto specific neuroanatomic regions. In any given study, changes in blood oxygen level-dependent (BOLD) signal intensity values may be produced for hundreds or thousands of ‘voxels’ that represent activation levels for many different neuroanatomic points or regions and potentially across a number of cognitive tasks, e.g., see.41,42 PET and SPECT are both forms of emission tomography that involve nuclear imaging techniques and the injection of short half-life radioactive ligands to visualize neuroreceptors and enzymes. The assumption is that areas of high radioactivity indicate higher receptor binding potential, brain metabolism, and neural activity. In terms of the difference between the two technologies, SPECT detects single photons that are produced as radio-nucleotides decay, while PET detects dual photon emissions following the collision of positrons and electrons.36 This difference, for a variety of reasons, ultimately gives PET higher spatial resolution, but also makes this modality more costly than SPECT. These methods are mainly limited by the availability of viable radioactive ligands, and at present, ligands for dopamine, serotonin, GABA, opiate, and cholinergic receptor systems exist and are used often. In any given study, both PET and SPECT yield data that represent the extent to which the neuroreceptor(s) of interest (e.g., the 5-HT1A serotonin autoreceptor) are observed to bind to the injected radioactive ligand. This results in binding potential (BP) values/measurements across many different regions of interest and potentially for multiple ligands and neuroreceptors, depending on the scope of the study, e.g., see.43–45

Magnetic Resonance Spectroscopy (MRS) MRS is non-invasive and provides quantification of neurochemicals and metabolites in and across

92

specific brain regions. Like other modalities, MRS also measures hydrogen spin frequency. The information provided, however, relates to concentrations of brain neurochemicals and metabolites, and data are depicted as a spectrum rather than as an image.36 The type of MRS used for a particular study is dependent upon both the specific nuclei of interest and the metabolites to be measured. While it is possible to study several different nuclei, the most widely used is proton spectroscopy.36 Proton spectroscopy allows identification of several metabolites, including N-acetyl aspartate (NAA), which is considered to be a marker of neuronal integrity, and choline (Cho), which is thought to be associated with cell membrane turnover. In any given study, an MRS spectrum is generated across several regions of interest. These spectra consist of peak values of these metabolites plotted against frequency, where the area under the peak reflects relative concentration of a metabolite in a specific neuroanatomic region, e.g., see.46,47

Diffusion Tensor Imaging (DTI) DTI, a newer brain imaging modality, allows identification, localization, and orientation of white matter tracks in the brain, which is done via quantification of the diffusion of water in tissue.48,49 In gray matter, water diffusion is similar in all directions, but in white matter tracts there is increased diffusion along axons, which are in parallel bundles with myelin sheaths. DTI relies on the fact that water travels faster parallel versus perpendicular to axons, which is a concept referred to as anisotropic diffusion. With application of diffusion gradients in six directions, it is possible to reconstruct the brain’s underlying neuronal microstructure via calculation of a tensor for each voxel that describes the three-dimensional shape of diffusion. For any given study, Mean Diffusivity (MD) or Trace, a scalar measure of the total diffusion within a voxel, is generated, and analysis across voxels and regions provides information about white matter integrity and localization of white matter lesions, e.g., see.50,51 Fractional anisotropy (FA) is also commonly collected. In summary, the primary neuroimaging modalities currently used for investigating neuropsychiatric

C.S. Bloss et al.

disease are extremely complex and high-dimensional: Structural MRI can generate volume measurements across vast numbers of regions, subregions, and tissue types of interest; fMRI can yield BOLD signal intensity change values for hundreds or thousands of voxels that represent activation levels for many different neuroanatomic regions; PET and SPECT generate BP measurements across many different regions of interest and potentially for multiple ligands and neuroreceptors; MRS provides spectra that consist of neurochemical concentrations for different metabolites across different neuroanatomic regions; and DTI yields a scalar measure of water diffusion for hundreds or thousands of voxels across different regions. As emphasized, given that there is consensus that neuroimaging phenotypes should be used within the context of genetic association studies of neuropsychiatric disease, a question arises as to how to pursue relevant studies while avoiding pitfalls related to the need for data reduction and problems related to multiple comparisons and false positives. Furthermore, the availability and accessibility of whole genome assays that generate anywhere from 500,000 to 1 million genetic markers on individuals participating in a genetic association study, significantly intensifies these problems. In the next section, we comment on potential approaches to the analysis of combined genetic and neuroimaging data.

Issue in the Analysis of HighDimensional Neuroimaging and Genetic Data As outlined in the previous section, neuroimaging modalities provide an enormous amount of data for each subject in a study. This raises a host of analytic issues about the best way to integrate neuroimaging data with genetic data since, until now, genetic studies have typically been restricted to a more limited set of dependent variables (e.g., diagnostic status, candidate gene status). Prominent among these issues are the risk of false positives due to multiple comparisons if multiple univariate analyses are applied to each data point and/or genetic variation studied. Other issues relate to the small sample sizes traditionally associated with neuroimaging studies due to the high cost and complexity

5

Leveraging High-Dimensional Neuroimaging Data in Genetic Studies

of neuroimaging protocols, as well as the potential loss of information that results if standard multivariate strategies are used that rely on data reduction strategies (e.g., cluster and factor analyses). In regards to the issue of multiple comparisons and false positive results, one must be sensitive to the fact that the number of genetic variations that one might be interested in testing for association with a trait or condition is as important as the number of imaging-derived parameters (e.g., activation levels associated with individual voxels) one might want to test for association with genetic variations. Only a handful of genetic variations have been unequivocally linked to neuropsychiatric disorders.52 This makes the potential for spurious findings with high-dimensional neuroimaging data pronounced, especially if the neuroimaging parameters to be used in a study do not adequately capture elements of the true phenotype associated with a disorder, and one merely tests each parameter for an association with many genetic variations in a univariate manner.53 Recently this issue was studied empirically in an investigation by Meyer-Lindenberg et al.54 These authors selected 720 SNPs that failed to show association to psychiatric disorders and then tested them for association with gray matter content captured by over 2 million voxels, as well as BOLD response during a cognitive task in approximately 24,000 voxels.54 The authors assessed the extent to which associations exceeded a test-statistic threshold corrected for multiple comparisons and found that the number of SNPs showing false positive association was substantially below the acceptable experiment-wise error rate of 5%. Although these results suggest that the false positive rate for neuroimaging studies may be within acceptable limits with current methods of multiple comparisons corrections (e.g., family-wise error based on Gaussian Random Field analysis or the False Discovery Rate), this issue deserves further study and remains one to which researchers should be sensitive as the field of imaging-genetics grows. A second issue that is related to the multiple comparisons issue revolves around the fact that traditional sample sizes for neuroimaging studies are extremely small (e.g., the current practice in fMRI studies calls for a minimum of ten individuals in the smallest group under study). This is in stark contrast to current sample size expectations for genetic association studies, and in particular, GWA studies, which call for thousands to tensof-thousands of individuals in order to be adequately

93

powered to detect associations of weak to moderate effect given multiple the comparisons issues. The reasons for the typical use of small samples sizes in neuroimaging studies relate to the time and expense associated with collection of neuroimaging data. While there is thought to be increased statistical power as a result of utilizing endophenotypes as dependent variables in genetic association studies, limited sample sizes are nonetheless an issue to be acknowledged and considered. Certainly, analytic methods that maximally leverage high-dimensional data collected on small samples would be of great benefit. A third issue relates to consideration of the units of analysis. For example, one could consider each individual imaging-derived phenotypic element (e.g., voxel) and/or genetic variation as a unit of analysis. This can be problematic if, e.g., the voxels in one individual do not correspond to exactly the same neuroanatomic location in another individual or if the genetic variations being tested are not causally-related to the phenotype of relevance but are rather merely in linkage disequilibrium with causal variations. In both of these contexts it would make sense to group the relevant data points into more coherent units. For imaging data, this may amount to grouping voxels or particular image locations by anatomical proximity and working with the data points within these groups collectively, either by averaging them or using a multivariate analysis technique. One could vary the size and nature of the groupings and repeat the analysis. This approach has been labeled the ‘searchlight’ approach.55 For genetic data, grouping genetic variations is typically done by exploiting linkage disequilibrium patterns and knowledge of which variations are known to ‘tag’ other variations by virtue of their being in linkage disequilibrium with them.56 In the absence of direct knowledge of which variations tag others, one could use a ‘moving (or sliding) window’ approach in which some number of physically adjacent variations are used to form haplotypes that are then related to the phenotypes in question – the basic idea being that unobserved variations are likely to be unevenly distributed across the different haplotypes so that haplotype associations with a phenotype may reflect the effect of an unobserved variation on the chromosomes that have particular haplotypes. The window is then moved to the next set of variations and the analysis is repeated until one has reached the end of a set of physically linked variations or the end of the genome.57,58 Whatever method of

94

grouping is used, if any, an additional issue involves the choice of a method for relating the variations (or sets of variations) to the imaging phenotypes (or sets of imaging phenotypes), as described below.

Analytical Approaches to Combined Neuroimaging and Genetic Data Although there are a number of analytical approaches one could exploit in relating a large number of genetic variations to high-dimensional imaging-derived endophenotypic measures, the vast majority of these approaches can be grouped into four basic classes, each with its own set of advantages and disadvantages, which we briefly describe below.

Multiple Univariate Approaches The simplest and most widely used approach for analyzing high-dimensional genetic variation information with phenotypes of all sorts involves relating each genetic variation to each phenotype in a univariate manner. In fact, all GWA studies performed to date have exploited this approach.14 For example, if one is seeking to test the association of 500,000 SNPs with a quantitative phenotype such as activation levels at a single voxel from an fMRI study, then one could perform 500,000 analysis-of-variance (ANOVA) tests which contrast average activation levels across the three genotypic categories for each SNP. One could then pursue this analysis for each voxel assayed from the fMRI studies. The advantages of this approach are that the results of each test are intuitive and every element of the data is tested, allowing a researcher to pinpoint exactly where relationships between the genetic variant and imaging phenotypes reside. The problems with this approach are obvious in that multiple comparisons issues loom large. To combat multiple comparisons, one could exploit methodologies such as false discovery rate analysis to determine how many analyses might have produced results worth considering further in light of the multiple comparisons issues.54,59,60 Another disadvantage of the multiple univariate analysis approach is that it does not, by itself, provide more ‘holistic’ insights into how the individual data points

C.S. Bloss et al.

may cohere in more fundamental ways, although methodologies for grouping the results of multiple independent statistical analyses have been devised.61–63

Data Reduction Approaches Another approach to the analysis of large-scale data sets involves reducing (or ‘decomposing’) the data points into a smaller set of homogeneous data points that essentially capture the information or variation in the entire data set. These reduced data points can then be scrutinized for patterns or associations with other data points. Specific analysis methods that fall into this class include factor analysis, principal components analysis (PCA), and latent structure analysis.64,65 The advantage of these approaches are that they can provide ways of empirically organizing large data sets and can reveal patterns that may have been imperceptible from the results of multiple univariate analyses. Disadvantages of these approaches for combined imaging and genetic data is that one would have to apply the data reduction strategy to the genetic data and imaging data independently and then ask how the reduced data points from both are associated. The results of such an analysis could be very difficult to interpret from a biological standpoint, especially in the context of the genetic variation data. A better strategy might be to reduce the neuroimaging data and then relate the reduced data points (e.g., factors, clusters, principal components, etc.) to each of the genetic variations (or groups of variations in the form of haplotypes). The interpretability, however, of the ‘factors’ identified in the neuroimaging data may still be problematic to interpret. A further problem with many multivariate approaches that are rooted in data reduction or decomposition is their computational demand.

Regression and Regression-Like Approaches An alternative approach to data reduction or decomposition approaches to the analysis of multivariate data are approaches rooted in multivariate regression and regression-like methodologies in which, e.g., the genetic variations provided are treated as ‘predictors’

5

Leveraging High-Dimensional Neuroimaging Data in Genetic Studies

or independent variables for the imaging data, which are treated as a set of dependent variables. Partial least squares and canonical correlation analyses are approaches similar to multivariate regression approaches.66–68 The advantages of these approaches are that they allow one to assess the effect of each data element simultaneously with the others, which could lead to insights into which variables have effects that are essentially independent of the others. Also, these approaches allow one to consider interactions between predictor variables. Some disadvantages include interpretability, computational efficiency, and an inability to easily identify redundancies among the variables used. In addition, it would be difficult at best to relate a very large number of genetic variations to a large number of neuroimaging phenotypic elements. To combat this, one might consider taking advantage of the moving window and/or searchlight approach in which sets of adjacent genetic variations are related to sets of neuroimaging phenotypes in order to identify genomic regions that harbor (potentially multiple) genetic variations that influence specific sets of neuroimaging phenotypes.

Multivariate Profile Similarity Analysis Another approach to the analysis of multivariate data with great applicability to combined neuroimaging and genetic data involves treating the imaging data as providing multivariate ‘profiles’ of each individual in a study whose similarity to other individuals’ profiles can be assessed. Factors that explain or influence the greater or lesser similarity observed between the profiles of the individuals in the study can then be tested. Schork and colleagues have considered this approach in many non-imaging contexts and have termed the approach Multivariate Distance Matrix Regression (MDMR) because testing of variables for association with profiles can be done in a regression-like context.69–71 A hypothetical example of the MDMR approach with imaging data could, for example, involve brain activation levels in schizophrenia subjects during a specific task. In this scenario, it may be the case that activation mechanisms are perturbed in the brains of individuals with schizophrenia such that the profile or overall pattern of activation is more similar among individuals with schizophrenia versus individuals without the disorder. This situation may not be identifiable

95

with univariate analysis methods if the contribution of each voxel’s activation level to the overall profile difference between the schizophrenics and controls is slight. A detailed illustration of the proposed method was presented by Zapala and Schork71 and involved analysis of gene expression patterns in the human frontal cortex among individuals who died at various ages.72 Initially, these authors performed correlational analyses to identify 463 genes that correlated with age, and then calculated a Pearson correlation-based heat map matrix that covered all pair-wise comparisons of individuals. They then analyzed the distance matrix based on the pair-wise correlations using the proposed MDMR-based method. Within the context of this method, the values in the distance matrix served as the dependent variables and age and sex served as independent variables. Although sex was not a significant predictor of the gene expression patterns in the frontal cortex, as would be expected, age was significantly associated with gene expression profile similarity. Specifically, age appeared to explain approximately 35% of the variance in profile similarity among the individuals in expression patterns of the previously determined age-related genes. Thus, through the use of this methodology, similarity/dissimilarity in patterns of variables of interest were assessed for association with a set of predictor variables. We have already applied the MDMR method to imaging data73 and are currently considering applications to imaging-genetics studies. For instance, assume that one has collected some number of neuroimaging outcome variables (e.g., BOLD response values across approximately 24,000 voxels) and some number of genomic predictor variables (e.g., 500,000 SNPs) on 20 subjects. In this situation, although highly computationally intensive, one could consider how the 500,000 SNP predictor variables relate to the similarity or dissimilarity of the subjects under study with respect to activation levels across the 24,000 voxels. This would involve construction of a similarity or distance matrix, which reflects the similarity/dissimilarity of each pair of individuals with respect to the BOLD response values obtained on them and testing each SNP for association with variation in the matrix. In this context, one can also rely on permutation tests to evaluate the probabilistic significance of an observed association,74 and could also pursue step-wise SNP (i.e., in this case the independent variables) selection procedures with the technique, as with standard multiple regression analysis.75

96

C.S. Bloss et al.

The Schork Laboratory has developed a website that can be used to carry out this analysis in genetic association study contexts (http://polymorphism.scripps.edu/). Advantages of the analysis approaches rooted in multivariate profile similarity, like MDMR, are that they can be used with large amounts of data collected on relatively few subjects, and they provide a method for examining the data as whole. Disadvantages of the approach are that, by treating all the data points equally, the ‘signal’ provided by a few data points may be confounded by the ‘noise’ given off from the others. However, this may be overcome by confining the construction of the profiles to imaging phenotypes defined by an appropriate ‘searchlight.’

Example Imaging-Genetics Studies As noted, the field of imaging genetics is still in its infancy. However, a number of studies limited to testing a few ‘candidate’ variations or genes for association with neuroimaging endophenotypes have been pursued in both healthy population samples and among individuals with neuropsychiatric, neurodegenerative, and neurodevelopmental disorders. We touch on some examples of these studies, with a more in-depth review of studies that have been done in schizophrenia and bipolar disease (Table 5.3) given that these two diseases are neuropsychiatric in nature, and they also represent diseases that have been studied with each of the

Table 5.3 Selected imaging-genetics studies of bipolar disorder (BP) and schizophrenia (SZ) Disease Method

Gene(s)

BP

DTI

NRG1

BP

MRS/MRI

BP

fMRI

BP

MRI

SZ

DTI/MRI

SZ

fMRI

SZ

fMRI

SZ

fMRI

Description

Subjects with the risk-associated TT genotype of NRG1 had reduced white matter density in the anterior limb of the internal capsule and evidence of reduced structural connectivity in the same region using DTI Mitochondrial The 5178 polymorphism in mitochondrial DNA may regulate vulnerability DNA to bipolar disorder via alteration of brain energy metabolism CLOCK A single nucleotide polymorphism (SNP) in the 3′ flanking region of CLOCK (rs1801260) influenced diurnal preference in healthy humans and caused sleep phase delay and insomnia in patients affected by bipolar disorder Heritability Genes that raise the likelihood of developing schizophrenia may exert their (families) effects by diminishing grey matter volume in the DLPFC and VLPFC and their associated white matter connections. Genes for bipolar illness might have subtle effects on brain structure, which may need particularly large samples to detect NRG1 This is the first demonstration that variation of NRG1 affects medial frontal white matter microstructure in humans and possibly the degree of neuronal myelination. By extension, NRG1 may contribute to a risk for schizophrenia via its impact on myelination in frontal lobe white matter DRD1 Multivariate analyses (partial least squares) revealed an inverse relationship between covariance patterns of different brain regions including the dorsolateral prefrontal cortex and two genotypes of DRD1, suggesting that distinct dopamine receptor genotypes may use distinct neural systems to retrieve information 384 SNPs Parallel ICA identifies an fMRI component (parietal lobe activation) with a SNP component (in AADC, ADRA2A, CHRNA7, DISC1, CHRNA7). Both fMRI and SNP components showed significant differences in loading parameters between the schizophrenia and control groups COMT COMT genotype was related in allele dosage fashion to performance on the Wisconsin Card Sorting Test of executive cognition. This study suggests that the COMT Val allele, because it increases prefrontal dopamine catabolism, impairs prefrontal cognition and physiology, and by this mechanism slightly increases risk for schizophrenia

Reference [87]

[105] [95]

[89]

[104]

[67]

[98]

[96]

(continued)

5

Leveraging High-Dimensional Neuroimaging Data in Genetic Studies

97

Table 5.3 (continued) Disease Method

Gene(s)

Description

Reference

SZ

fMRI

GAD1

99

SZ

fMRI/MRI

COMT, NRG1

SZ

MRI

AKT1, COMT

SZ

MRI

COMT, PRODH

SZ

MRI

DISC1

SZ

MRI

Heritability (families)

SZ

MRI

Heritability (siblings)

SZ

MRI

Heritability (siblings)

SZ

MRI

NRG1

SZ

PET/MRI

COMT

SZ

PET/MRI

IL-1B

SZ

SPECT

DAT

A 5′ flanking SNP affecting cognition in the families was associated in unrelated healthy individuals with inefficient BOLD activation of dorsal prefrontal cortex during a working memory task. This study implicates GAD1 in the etiology of schizophrenia via altered cortical GABA inhibitory activity, perhaps modulated by dopaminergic function. Subjects with COMT Val allele had increased fMRI activation in lateral prefrontal cortex and anterior and posterior cingulate. NRG1 promoter risk allele was associated with decreased activation of pre-frontal and temporal lobe regions These data implicate AKT1 in modulating human prefrontal-striatal structure and function and suggest that the mechanism of this effect may be coupled to dopaminergic signaling and relevant to the expression of psychosis Two nonsynonymous SNPs in the PRODH gene were associated with bilateral frontal WM density reductions and a SNP in the P2 promoter region of COMT (rs2097603) was associated with GM increase in the right superior temporal gyrus. There was also evidence for COMT and PRODH epistasis; in patients with a COMT Val allele (rs4680) and with one or two mutated PRODH alleles, there was an observed increased WM density in the left inferior frontal lobe These findings implicate DISC1 in variation of prefrontal cortical volume and positive symptoms, thus providing a potential mechanism through which DISC1 may confer increased risk for schizophrenia or schizoaffective disorder A reduced volume of ALIC in affected families supports the hypothesis of disturbed frontothalamic connectivity in schizophrenia and demonstrates functional relevance by an association with reduced neurocognitive performance Patients with schizophrenia had significant regional gray matter decreases in the frontal, temporal, and parietal cortices compared with healthy volunteers These findings support evidence of genetic control of brain volume even in adults, particularly of hippocampal and neocortical volume and of cortical volumetric reductions being familial, but do not support measures of subcortical volumes per se as representing intermediate biologic phenotypes In the child-onset schizophrenia (COS) group, risk allele carriers had greater total gray and white matter volume in childhood and a steeper rate of subsequent decline in volume into adolescence. By contrast, in healthy children, possession of the risk allele was associated with different trajectories in gray matter only and was confined to frontotemporal regions, likely reflecting epistatic or other illness-specific effects mediating NRG1 influence on brain development in COS Randomization analyses using [(15)O]H(2)O positron emission tomography (PET) cerebral blood flow data found Val/Val patients had higher frontal lobe activation than Met/Met patients while performing the one-back task. Overall, findings do not support a major role for COMT in increasing susceptibility for schizophrenia or in mediating frontal lobe function Those patients who were allele 2 (-511 T) carriers showed a lower metabolic activity in the left DLPFC with respect to patients homozygous for allele 1 (-511 C) (U = 16, z = −2.32, P = 0.02). Results suggest that hypofrontality reported in some schizophrenic patients might be explained, at least in part, by this functional polymorphism at IL-1B gene The VNTR polymorphism did not influence AP-induced EPS and did not affect DAT gene expression or protein function

[97]

[93]

[94]

[92]

[90]

[88]

[24]

[91]

[102]

[101]

[100]

98

modalities that have been discussed in this chapter (i.e., MRI, fMRI, PET/SPECT, MRS, and DTI). Among the most well-studied genetic variants in imaging-based neurophenotype studies is the ε4 allele of the apolipoprotein E gene (APOE), which is a known risk factor for the development of Alzheimer’s disease.76 A large number of studies have confirmed, for example, that whole brain, as well as region-specific (e.g., in the hippocampus) atrophy rates are strongly correlated with APOE-ε4 allelic dose.77,78 Interaction effects of other genes with APOE have also been reported in association studies leveraging neuroimaging data, such as between APOE-ε4 and a variant of the nicotinic receptor subunit CHRNA4.79 Examples of structural MRI studies looking at other genes include a study by Pezawas et al., which showed that a variation in the brain-derived growth factor (BDNF) gene was associated with reduced hippocampal gray matter volume among normal controls.80 Neurodevelopmental disorders have also been studied. For instance, among children with attention deficithyperactivity disorder (ADHD), a polymorphism in the dopamine D4 receptor (DRD4) gene has been shown to be associated with region-specific brain volumes.81 In addition to structural MRI, other imaging modalities have been used in a number of studies to try to discover the neural circuits involved in geneticallymediated neuropathology associated with particular clinical disorders. For example, individuals that are homozygous for the short allele of the serotonin transporter gene have been shown to exhibit more anxioustype responses to certain stimuli during functional imaging and to be at higher risk for depression.82,83 Among normal controls, Pohjalainen and colleagues utilized PET and found no difference in D2 receptor density between individuals carrying the Del allele of the -141C (Ins/Del) polymorphism located in the immediate 5′-flanking region of the dopamine D2 receptor gene.84 SPECT has been used to examine an association among normal controls between a polymorphism in a noncoding region of the D2 receptor gene (i.e., the A1 allele of the Taq1 ‘A’ system) and lower D2 receptor density, although findings have been inconsistent across studies.85 MRS was utilized to examine a possible association between a SNP in the metabotropic glutamate receptor 3 (GRM3) gene, a putative risk factor for schizophrenia, and N-acetyl aspartate in healthy controls, and it was found that

C.S. Bloss et al.

the A/A genotype group showed significant reduction in the right dorsolateral prefrontal cortex relative to G carriers.86 Finally DTI revealed reduced white matter density and connectivity in controls with the neuregulin 1 (NRG1) psychosis risk genotype.87 A representative sample of imaging-genetics studies in schizophrenia and bipolar disorder completed to date can be found in Table 5.3. Clearly, a greater number of studies have been done in schizophrenia, and indeed, nearly all imaging modalities are represented with respect to this disorder. Many structural MRI studies look have looked at heritability,24,88–90 as well as specific candidate genes91–94 (see Table 5.3). A number of studies that utilize fMRI67,95–99 and PET/SPECT100,101 have also been completed. Specifically, in schizophrenia, PET studies have focused on correlating genetic variants (e.g., COMT) with activation in frontal regions,101,102 and SPECT studies have looked at striatal DAT binding as a function of variable number tandem repeats (VNTR) in the DAT gene (SLC6A3).100,103 DTI studies have also been done,104 and in addition, an association between the 5178 polymorphism in mitochondrial DNA and brain intracellular pH measured by (31)P-MRS in bipolar disorder was examined and findings showed significantly higher levels in patients with 5178A versus 5178C.105

Conclusions and Future Directions There is an unprecedented array of sophisticated technologies a researcher can exploit to investigate the genetic determinants of neuropsychiatric disease. High-throughput DNA microarrays for genotyping and high-dimensional neuroimaging modalities are prominent among these technologies, and as emphasized throughout this review, neuroimaging phenotypes are thought to be excellent candidate phenotypes for use in identifying susceptibility genes for neuropsychiatric disease. The high-dimensional nature of the data produced with these technologies, however, raises serious questions with respect to how these data are best integrated and analyzed for drawing probabilistically compelling inferences. In this chapter we have described some of the analytic issues and challenges that arise when trying to leverage both large-scale or genome-wide genetic variation data and neuroimaging data (i.e., problems related to multiple comparisons and false positives, as well as small sample sizes).

5

Leveraging High-Dimensional Neuroimaging Data in Genetic Studies

Finally, we have reviewed some of the candidate gene imaging-genetics studies in the literature, with a focus on studies of neuropsychiatric diseases, including schizophrenia and bipolar disorder. In terms of future directions, given that imaginggenetics studies can, in many instances, provide an early indicator of disease (e.g., in Alzheimer’s disease), the pursuit of these studies will have an enormous impact on the identification of factors that could lead to early diagnosis and the potential administration of interventions that may slow the rate of progression, or even lead to a cure.106 Furthermore, additional biomedical applications of the combined assessment and analysis of genetic variation and neuroimaging endophenotypes, include the identification of biomarkers that can improve the efficiency of pharmaceutical trials by providing a less biased assessment of drug efficacy and toxicity (e.g., by specifying more objective, quantifiable treatment endpoints relative to DSM diagnostic category), that could be used as an aid in the determination of dosing regimens and understanding of drug mechanisms.107 In short, this burgeoning field has the enormous potential to usher in a new era of treatment development for neuropsychiatric diseases and pave the way for personalized medicine. Acknowledgments The authors would like to thank Dr. Natalia Kleinhans for her comments on parts of this manuscript. Aspects of the research represented in the chapter were supported by the following grants: The National Heart Lung and Blood Institute Family Blood Pressure Program [FBPP; U01 HL064777-06]; The National Institute on Aging Longevity Consortium [U19 AG023122-01]; The National Institute of Mental Health Consortium on the Genetics of Schizophrenia [COGS; 5 RO1 HLMH065571-02]; National Institute of Health [RO1s: HL074730-02 and HL070137-01]; Scripps Genomic Medicine and the Scripps Translational Science Institute.

References 1. Todd JA, Walker NM, Cooper JD, et al. Robust associations of four new chromosome regions from genome-wide analyses of type 1 diabetes. Nature Genetics. 2007;39:857–864 2. Hakonarson H, Grant SF, Bradfield JP, et al. A genome-wide association study identifies KIAA0350 as a type 1 diabetes gene. Nature. 2007;448:591–594 3. Frayling TM, Timpson NJ, Weedon MN, et al. A common variant in the FTO gene is associated with body mass index and predisposes to childhood and adult obesity. Science (New York) 2007;316:889–894

99

4. Thorleifsson G, Magnusson KP, Sulem, P et al. Common sequence variants in the LOXL1 gene confer susceptibility to exfoliation glaucoma. Science (New York) 2007;317: 1397–1400 5. Eeles RA, Kote-Jarai Z, Giles GG, et al. Multiple newly identified loci associated with prostate cancer susceptibility. Nature Genetics. 2008;40:316–321 6. Harley JB, Alarcon-Riquelme ME, Criswell LA, et al. Genome-wide association scan in women with systemic lupus erythematosus identifies susceptibility variants in ITGAM, PXK, KIAA1542 and other loci. Nature Genetics. 2008;40:204–210 7. Hunt KA, Zhernakova A, Turner G, et al. Newly identified genetic risk variants for celiac disease related to the immune response. Nature Genetics. 2008;40:395–402 8. Ridker PM, Pare G, Parker A, et al. Loci related to metabolic-syndrome pathways including LEPR, HNF1A, IL6R, and GCKR associate with plasma C-reactive protein: the women’s genome health study. American Journal of Human Genetics. 2008;82:1185–1192 9. Tenesa A, Farrington SM, Prendergast JG, et al. Genomewide association scan identifies a colorectal cancer susceptibility locus on 11q23 and replicates risk loci at 8q24 and 18q21. Nature Genetics. 2008;40:631–637 10. Thorgeirsson TE, Geller F, Sulem P, et al. A variant associated with nicotine dependence, lung cancer and peripheral arterial disease. Nature. 2008;452:638–642 11. Uda M, Galanello R, Sanna S, et al. Genome-wide association study shows BCL11A associated with persistent fetal hemoglobin and amelioration of the phenotype of beta-thalassemia. Proceedings of the National Academy of Sciences of the United States of America. 2008;105:1620–1625 12. Zeggini E, Scott LJ, Saxena R, et al. Meta-analysis of genome-wide association data and large-scale replication identifies additional susceptibility loci for type 2 diabetes. Nature Genetics. 2008;40:638–645 13. WTCCC. Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature. 2007;447:661–678 14. Manolio TA, Brooks LD, Collins FS. A HapMap harvest of insights into the genetics of common disease. The Journal of Clinical Investigation. 2008;118:1590–1605 15. Lencz T, Morgan TV, Athanasiou M, et al. Converging evidence for a pseudoautosomal cytokine receptor gene locus in schizophrenia. Molecular Psychiatry. 2007;12: 572–580 16. Shifman S, Johannesson M, Bronstein M, et al. Genomewide association identifies a common variant in the reelin gene that increases the risk of schizophrenia only in women. PLoS Genetics. 2008;4:e28 17. Kirov G, Zaharieva I, Georgieva L, et al. A genome-wide association study in 574 schizophrenia trios using DNA pooling. Molecular Psychiatry. 2008 advance online publication, II March 2008 (DOI 10.1038/mp.2008.33) 18. Sklar P, Smoller JW, Fan J, et al. Whole-genome association study of bipolar disorder. Molecular Psychiatry. 2008;13: 558–569 19. Baum AE, Akula N, Cabanero M, et al. A genome-wide association study implicates diacylglycerol kinase eta (DGKH) and several other genes in the etiology of bipolar disorder. Molecular Psychiatry. 2008;13:197–207

100 20. Bearden CE, Freimer NB. Endophenotypes for psychiatric disorders: ready for primetime? Trends in Genetics. 2006; 22:306–313 21. APA. Diagnostic and Statistical Manual of Mental Disorders. 4-Text Revision ed. Washington, DC: APA, 2000 22. Tan HY, Callicott JH, Weinberger DR. Endophenotypes in schizophrenia genetics redux: is it a no brainer? Molecular Psychiatry. 2008;13:233–238 23. Callicott JH, Egan MF, Mattay VS, et al. Abnormal fMRI response of the dorsolateral prefrontal cortex in cognitively intact siblings of patients with schizophrenia. The American Journal of Psychiatry. 2003;160:709–719 24. Goldman AL, Pezawas L, Mattay VS, et al. Heritability of brain morphology related to schizophrenia: a large-scale automated magnetic resonance imaging segmentation study. Biological Psychiatry. 2008;63:475–483 25. Gottesman, II, Shields J. Genetic theorizing and schizophrenia. British Journal of Psychiatry. 1973;122:15–30 26. Toulopoulou T, Picchioni M, Rijsdijk F, et al. Substantial genetic overlap between neurocognition and schizophrenia: genetic modeling in twin samples. Archives of General Psychiatry. 2007;64:1348–1355 27. Braff DL, Greenwood TA, Swerdlow NR, et al. Advances in endophenotyping schizophrenia. World Psychiatry. 2008;7: 11–18 28. Braff D, Schork NJ, Gottesman, II. Endophenotyping schizophrenia. The American Journal of Psychiatry. 2007;164: 705–707 29. Braff DL, Freedman R, Schork NJ, Gottesman, II. Deconstructing schizophrenia: an overview of the use of endophenotypes in order to understand a complex disorder. Schizophrenia Bulletin. 2007;33:21–32 30. Braff DL, Light GA. The use of neurophysiological endophenotypes to understand the genetic basis of schizophrenia. Dialogues in Clinical Neuroscience. 2005;7:125–135 31. Flint J, Munafo MR. The endophenotype concept in psychiatric genetics. Psychological Medicine. 2007;37:163–180 32. Gottesman, II, Gould TD. The endophenotype concept in psychiatry: etymology and strategic intentions. The American Journal of Psychiatry. 2003;160:636–645 33. Fusar-Poli P, Perez J, Broome M, et al. Neurofunctional correlates of vulnerability to psychosis: a systematic review and meta-analysis. Neuroscience and Biobehavioral Reviews. 2007;31:465–484 34. Bartley AJ, Jones DW, Weinberger DR. Genetic variability of human brain size and cortical gyral patterns. Brain. 1997;120(Pt 2):257–269 35. MacDonald AW, 3rd, Carter CS. Event-related FMRI study of context processing in dorsolateral prefrontal cortex of patients with schizophrenia. Journal of Abnormal Psychology. 2003;112:689–697 36. Malhi GS, Lagopoulos J. Making sense of neuroimaging in psychiatry. Acta Psychiatrica Scandinavica. 2008;117: 100–117 37. Bloss CS, Courchesne E. MRI neuroanatomy in young girls with autism: a preliminary study. Journal of the American Academy of Child and Adolescent Psychiatry. 2007;46: 515–523 38. Carper RA, Courchesne E. Localized enlargement of the frontal cortex in early autism. Biological Psychiatry. 2005; 57:126–133

C.S. Bloss et al. 39. Carper RA, Moses P, Tigue ZD, Courchesne E. Cerebral lobes in autism: early hyperplasia and abnormal age effects. NeuroImage. 2002;16:1038–1051 40. Courchesne E, Karns CM, Davis HR, et al. Unusual brain growth patterns in early life in patients with autistic disorder: an MRI study. Neurology. 2001;57:245–254 41. Kleinhans NM, Richards T, Sterling L, et al. Abnormal functional connectivity in autism spectrum disorders during face processing. Brain. 2008;131:1000–1012 42. Kleinhans NM, Johnson LC, Mahurin R, et al. Increased amygdala activation to neutral faces is associated with better face memory performance. Neuroreport. 2007;18:987–991 43. Bailer UF, Frank GK, Henry SE, et al. Altered brain serotonin 5-HT1A receptor binding after recovery from anorexia nervosa measured by positron emission tomography and [carbonyl11C]WAY-100635. Archives of General Psychiatry. 2005;62:1032–1041 44. Frank GK, Kaye WH. Positron emission tomography studies in eating disorders: multireceptor brain imaging, correlates with behavior and implications for pharmacotherapy. Nuclear Medicine and Biology. 2005;32:755–761 45. Kaye WH, Bailer UF, Frank GK, et al. Brain imaging of serotonin after recovery from anorexia and bulimia nervosa. Physiology & Behavior. 2005;86:15–17 46. Bertolino A, Weinberger DR. Proton magnetic resonance spectroscopy in schizophrenia. European Journal of Radiology. 1999;30:132–141 47. Bertolino A, Esposito G, Callicott JH, et al. Specific relationship between prefrontal neuronal N-acetylaspartate and activation of the working memory cortical network in schizophrenia. The American Journal of Psychiatry. 2000; 157:26–33 48. Conturo TE, Lori NF, Cull TS, et al. Tracking neuronal fiber pathways in the living human brain. Proceedings of the National Academy of Sciences of the United States of America. 1999;96:10422–10427 49. Mori S, Crain BJ, Chacko VP, van Zijl PC. Threedimensional tracking of axonal projections in the brain by magnetic resonance imaging. Annals of Neurology. 1999; 45:265–269 50. Pfefferbaum A, Rosenbloom MJ, Adalsteinsson E, Sullivan EV. Diffusion tensor imaging with quantitative fibre tracking in HIV infection and alcoholism comorbidity: synergistic white matter damage. Brain. 2007;130:48–64 51. Sullivan EV, Pfefferbaum A. Diffusion tensor imaging and aging. Neuroscience and Biobehavioral Reviews. 2006; 30:749–761 52. Straub RE, Weinberger DR. Schizophrenia genes – famine to feast. Biological Psychiatry. 2006;60:81–83 53. Meyer-Lindenberg A, Weinberger DR. Endophenotypes and genetic mechanisms of psychiatric disorders. Nature Reviews Neuroscience. 2006;7:818–827 54. Meyer-Lindenberg A, Nicodemus KK, Egan MF, et al. False positives in imaging genetics. NeuroImage. 2008;40: 655–661 55. Kriegeskorte N, Goebel R, Bandettini P. Information-based functional brain mapping. Proceedings of the National Academy of Sciences of the United States of America. 2006;103:3863–3868 56. Balding DJ. A tutorial on statistical methods for population association studies. Nature Reviews. 2006;7:781–791

5

Leveraging High-Dimensional Neuroimaging Data in Genetic Studies

57. Fallin D, Cohen A, Essioux L, et al. Genetic analysis of case/control data using estimated haplotype frequencies: application to APOE locus variation and Alzheimer’s disease. Genome Research. 2001;11:143–151 58. Li Y, Sung WK, Liu JJ. Association mapping via regularized regression analysis of single-nucleotide-polymorphism haplotypes in variable-sized sliding windows. American Journal of Human Genetics. 2007;80:705–715 59. Tang Y, Ghosal S, Roy A. Nonparametric bayesian estimation of positive false discovery rates. Biometrics. 2007;63: 1126–1134 60. Gadbury GL, Xiang Q, Yang L, et al. Evaluating statistical methods using plasmode data sets in the age of massive public databases: an illustration using false discovery rates. PLoS Genetics. 2008;4:e1000098 61. Schmidt S, Shao Y, Hauser ER, et al. Life after the screen: making sense of many P-values. Genetic Epidemiology. 2001;21 Suppl 1:S546–S551 62. Cucala L. A hypothesis-free multiple scan statistic with variable window. Biometrical Journal. 2008;50: 299–310 63. Gordon D, Hoh J, Finch SJ, et al. Two approaches for consolidating results from genome scans of complex traits: selection methods and scan statistics. Genetic Epidemiology. 2001;21 Suppl 1:S396–402 64. Anderson TW. An Introduction to Multivariate Statistical Analysis. New York: Wiley, 1984 65. Tuncer Y, Tanik MM, Allison DB. An overview of statistical decomposition techniques applied to complex systems. Computational Statistics and Data Analysis. 2008;52: 2292–2310 66. McIntosh AR, Bookstein FL, Haxby JV, Grady CL. Spatial pattern analysis of functional brain images using partial least squares. NeuroImage. 1996;3:143–157 67. Tura E, Turner JA, Fallon JH, et al. Multivariate analyses suggest genetic impacts on neurocircuitry in schizophrenia. Neuroreport. 2008;19:603–607 68. Nandy RR, Cordes D. Novel nonparametric approach to canonical correlation analysis with applications to low CNR functional MRI data. Magnetic Resonance in Medicine. 2003;50:354–365 69. Nievergelt CM, Libiger O, Schork NJ. Generalized analysis of molecular variance. PLoS Genetics. 2007;3:e51 70. Wessel J, Schork NJ. Generalized genomic distance-based regression methodology for multilocus association analysis. American Journal of Human Genetics. 2006;79:792–806 71. Zapala MA, Schork NJ. Multivariate regression analysis of distance matrices for testing associations between gene expression patterns and related variables. Proceedings of the National Academy of Sciences of the United States of America. 2006; 103:19430–19435 72. Lu T, Pan Y, Kao SY, et al. Gene regulation and DNA damage in the ageing human brain. Nature. 2004;429:883–891 73. Kaye WH, Bailer UF, Bloss CS, et al. 5-HT1A receptor binding is increased after recovery from bulimia nervosa compared to control women and is associated with behavioral inhibition in both groups (submitted) 74. Manly B. Randomization, Bootstrap, and Monte Carlo Methods in Biology. London: Chapman & Hall, 1997 75. Neter J, Wasserman W, Kutner MH. Applied Linear Statistical Models. Homewood, IL: Richard D. Irwin, 1985

101

76. Strittmatter WJ, Saunders AM, Schmechel D, et al. Apolipoprotein E: high-avidity binding to beta-amyloid and increased frequency of type 4 allele in late-onset familial Alzheimer disease. Proceedings of the National Academy of Sciences of the United States of America. 1993;90:1977–1981 77. Chen K, Reiman EM, Alexander GE, et al. Correlations between apolipoprotein E epsilon4 gene dose and whole brain atrophy rates. The American Journal of Psychiatry. 2007;164:916–921 78. Jak AJ, Houston WS, Nagel BJ, et al. Differential crosssectional and longitudinal impact of APOE genotype on hippocampal volumes in nondemented older adults. Dementia and Geriatric Cognitive Disorders. 2007;23: 382–389 79. Espeseth T, Greenwood PM, Reinvang I, et al. Interactive effects of APOE and CHRNA4 on attention and white matter volume in healthy middle-aged and older adults. Cognitive, Affective & Behavioral Neuroscience. 2006;6: 31–43 80. Pezawas L, Verchinski BA, Mattay VS, et al. The brain-derived neurotrophic factor val66met polymorphism and variation in human cortical morphology. Journal of Neuroscience. 2004;24:10099–10102 81. Durston S, Hulshoff Pol HE, Schnack HG, et al. Magnetic resonance imaging of boys with attention-deficit/hyperactivity disorder and their unaffected siblings. Journal of the American Academy of Child and Adolescent Psychiatry. 2004;43:332–340 82. Pezawas L, Meyer-Lindenberg A, Drabant EM, et al. 5HTTLPR polymorphism impacts human cingulate-amygdala interactions: a genetic susceptibility mechanism for depression. Nature Neuroscience. 2005;8:828–834 83. Surguladze SA, Elkin A, Ecker C, et al. Genetic variation in the serotonin transporter modulates neural system-wide response to fearful faces. Genes, Brain, and Behavior. 2008 84. Pohjalainen T, Nagren K, Syvalahti EK, Hietala J. The dopamine D2 receptor 5′-flanking variant, -141C Ins/Del, is not associated with reduced dopamine D2 receptor density in vivo. Pharmacogenetics. 1999;9:505–509 85. Laruelle M, Gelernter J, Innis RB. D2 receptors binding potential is not affected by Taq1 polymorphism at the D2 receptor gene. Molecular Psychiatry. 1998;3:261–265 86. Marenco S, Steele SU, Egan MF, et al. Effect of metabotropic glutamate receptor 3 genotype on N-acetylaspartate measures in the dorsolateral prefrontal cortex. The American Journal of Psychiatry. 2006;163:740–742 87. McIntosh AM, Moorhead TW, Job D, et al. The effects of a neuregulin 1 variant on white matter density and integrity. Molecular Psychiatry. 2007 88. Honea RA, Meyer-Lindenberg A, Hobbs KB, et al. Is gray matter volume an intermediate phenotype for schizophrenia? A voxel-based morphometry study of patients with schizophrenia and their healthy siblings. Biological Psychiatry. 2008;63:465–474 89. McIntosh AM, Job DE, Moorhead WJ, et al. Genetic liability to schizophrenia or bipolar disorder and its relationship to brain structure. American Journal of Medical Genetics Part B Neuropsychiatric Genetics. 2006;141B:76–83 90. Wobrock T, Kamer T, Roy A, et al. Reduction of the internal capsule in families affected with schizophrenia. Biological Psychiatry. 2008;63:65–71

102 91. Addington AM, Gornick MC, Shaw P, et al. Neuregulin 1 (8p12) and childhood-onset schizophrenia: susceptibility haplotypes for diagnosis and brain developmental trajectories. Molecular Psychiatry. 2007;12:195–205 92. Szeszko PR, Hodgkinson CA, Robinson DG, et al. DISC1 is associated with prefrontal cortical gray matter and positive symptoms in schizophrenia. 2008;79:103–110 93. Tan HY, Nicodemus KK, Chen Q, et al. Genetic variation in AKT1 is linked to dopamine-associated prefrontal cortical structure and function in humans. The Journal of Clinical Investigation. 2008;118:2200–2208 94. Zinkstok J, Schmitz N, van Amelsvoort T, et al. Genetic variation in COMT and PRODH is associated with brain anatomy in patients with schizophrenia. Genes, Brain, and Behavior. 2008;7:61–69 95. Benedetti F, Radaelli D, Bernasconi A, et al. Clock genes beyond the clock: CLOCK genotype biases neural correlates of moral valence decision in depressed patients. Genes, Brain, and Behavior. 2008;7:20–25 96. Egan MF, Goldberg TE, Kolachana BS, et al. Effect of COMT Val108/158 met genotype on frontal lobe function and risk for schizophrenia. Proceedings of the National Academy of Sciences of the United States of America. 2001;98:6917–6922 97. Lawrie SM, Hall J, McIntosh AM, et al. Neuroimaging and molecular genetics of schizophrenia: pathophysiological advances and therapeutic potential. British Journal of Pharmacology. 2008;153 Suppl 1:S120–S124 98. Liu J, Pearlson G, Windemuth A, et al. Combining fMRI and SNP data to investigate connections between brain function and genetics using parallel ICA. Human Brain Mapping. 2008 advance online publication, 10 Dec 2007 (DOI 10.1002/nbm.20508) 99. Straub RE, Lipska BK, Egan MF, et al. Allelic variation in GAD1 (GAD67) is associated with schizophrenia and influences cortical function and gene expression. Molecular Psychiatry. 2007;12:854–869

C.S. Bloss et al. 100. Lafuente A, Bernardo M, Mas S, et al. Dopamine transporter (DAT) genotype (VNTR) and phenotype in extrapyramidal symptoms induced by antipsychotics. Schizophrenia Research. 2007;90:115–122 101. Papiol S, Molina V, Rosa A, et al. Effect of interleukin1beta gene functional polymorphism on dorsolateral prefrontal cortex activity in schizophrenic patients. American Journal of Medical Genetics Part B Neuropsychiatric Genetics. 2007;144B:1090–1093 102. Ho BC, Wassink TH, O’Leary DS, et al. Catechol-O-methyl transferase Val158Met gene polymorphism in schizophrenia: working memory, frontal lobe MRI morphology and frontal cerebral blood flow. Molecular Psychiatry. 2005; 10:229, 287–298 103. Martinez D, Gelernter J, Abi-Dargham A, et al. The variable number of tandem repeats polymorphism of the dopamine transporter gene is not associated with significant change in dopamine transporter phenotype in humans. Neuropsychopharmacology. 2001;24:553–560 104. Winterer G, Konrad A, Vucurevic G, et al. Association of 5′ end neuregulin-1 (NRG1) gene variation with subcortical medial frontal microstructure in humans. NeuroImage. 2008;40:712–718 105. Kato T, Kunugi H, Nanko S, Kato N. Association of bipolar disorder with the 5178 polymorphism in mitochondrial DNA. American Journal of Medical Genetics. 2000;96: 182–186 106. Roffman JL, Weiss AP, Goff DC, et al. Neuroimaginggenetic paradigms: a new approach to investigate the pathophysiology and treatment of cognitive deficits in schizophrenia. Harvard Review of Psychiatry. 2006;14: 78–91 107. Wang YX. Medical imaging in pharmaceutical clinical trials: what radiologists should know. Clinical Radiology. 2005;60:1051–1057

Chapter 6

Proteomics as a New Tool for Biomarker-Discovery in Neuropsychiatric Disorders Thomas J. Raedler, Harald Mischak, Holger Jahn, and Klaus Wiedemann

Abstract Despite recent advances in our understanding of the neurobiology of neuropsychiatric disorders, most neuropsychiatric disorders remain clinical diagnoses that are based on the presence of a typical symptomconstellation as well as a typical time-course. Technical and laboratory examinations are frequently used to exclude other CNS-etiologies, such as tumor, infection, intoxication or epilepsy. However, only few biomarkers exist to assist in the differential diagnosis of neuropsychiatric disorders. Proteomics describes the study of the structure and function of proteins with respect to the complement of proteins in a cell or an organism. New technical developments have made it possible to identify thousands of proteins in different tissues and body-fluids, including blood, urine and CSF. Thus proteomics offers a new approach for the identification of biomarkers for neuropsychiatric disorders. Potential applications of proteomics include the differential diagnosis of neuropsychiatric disorders, the identification of subtypes of neuropsychiatric disorders, predictors of treatment response as well as early measures of treatment response. Based on preliminary studies using proteomics in neuropsychiatric disorders (in particular schizophrenia

T.J. Raedler University of Calgary, Faculty of Medicine, Department of Psychiatry, Calgary, Alberta, Canada H. Mischak University of Hamburg, Department of Psychiatry, Hamburg, Germany H. Jahn Mosaiques Diagnostics and Therapeutics AG, Hannover, Germany K. Wiedemann University of Hamburg, Department of Psychiatry, Hamburg, Germany

and Alzheimer Disease), we will review the potential impact of this new technology for research into neuropsychiatric disorders. Keywords Proteomics • biomarkers • schizophrenia • Alzheimer disease • differential diagnosis • neuropsychiatric disorders

Abbreviations CE: Capillary electrophoresis; CSF: Cerebrospinal fluid; MS: Mass spectrometry; TOF: Time-of-flight

Introduction The last decade has helped to advance our understanding of the neurobiology of neuropsychiatric disorders. Still, many open questions remain.1 Most neuropsychiatric disorders remain clinical diagnoses that are based on a typical symptom constellation as well as a typical timecourse of symptoms in the absence of other aetiologies (e.g. tumour, infection, trauma, intoxication). Technical or laboratory examinations in the diagnostic work-up of neuropsychiatric disorders serve primarily to exclude other CNS-aetiologies such as tumour or infection. Only few markers have so far been identified that can reliably assist in the differential diagnosis of neuropsychiatric disorders. So far, no technical or laboratory tool exists to identify relevant subgroups of neuropsychiatric disorders (e.g. responders to specific treatments, subjects at risk of treatment failure, subjects at risk of chronification). This diagnostic dilemma is further complicated by the fact that clinical symptoms are frequently not specific for only one neuropsychiatric disorder but

M.S. Ritsner (ed.), The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes, © Springer Science + Business Media B.V. 2009

103

104

show significant overlap between different disorders. In addition, many neuropsychiatric disorders often occur co-morbid with other neuropsychiatric disorders (e.g. mood disorders, anxiety disorders and substance use disorders). This diagnostic dilemma is even more pronounced in early or pre-symptomatic stages of illness (e.g. prodromal phase of schizophrenia; early cognitive impairment in Alzheimer Disease). Over the years to come, this diagnostic dilemma will become even more relevant with the advent of potentially prophylactic treatments. However, the potential effects of early treatment (disease modification versus potential cure) must be weighed against the potential side effects of exposing subjects to psychotropic agents. Potential problems of an overinclusive treatment approach include financial expenditure, unnecessary anxiety and unnecessary exposure of subjects to risks associated with medication, who would have done well without treatment.2 Specialised treatment services (e.g. memory clinics, early psychosis programs) are superior to normal clinical services in the evaluation of subjects at risk or in early stages of illness. For example, a specific symptom-constellation including clinical symptoms and genetic risk can be used to identify subjects with prodromal psychosis at risk of conversion to psychosis.3 However, these specialised services are not readily available to many subjects at risk. In summary, our current diagnostic procedures rely heavily on clinical presentations. Psychiatric diagnoses frequently depend on the examiner and are subject to bias reflecting training and diagnostic preferences. Additional diagnostic tools will be very helpful to assist with the diagnosis and differential diagnosis of neuropsychiatric disorders, the identification of subtypes of neuropsychiatric disorders, predictors of treatment response as well as early indicators for treatment response. Over the past decades, large-scale genetic studies have been conducted for many psychiatric disorders. The initial goal of identifying a limited number of genes that serve as disease specific markers or predictors of psychiatric illness had to be abandoned. Instead of being characterized by a handful of specific genes, most psychiatric disorders are complex genetic disorders and are characterised by a myriad of different candidate genes, which contribute to the overall load of illness. These candidate genes may carry diagnostic and predictive relevance for a subgroup of patients with a specific

T.J. Raedler et al.

illness but cannot be used as biomarkers.4 In addition, many of these candidate genes are not specific for one neuropsychiatric disorder but show frequent overlap between different disorders.5 One strategy to identify the role of risk gene variants in psychiatric disorders is the intermediate phenotype concept.6 To further complicate these issues, the relevance of epigenetics,7 including interactions between genes and environment8,9 as well as post-translational modifications (e.g. through proteolysis, glycosylation and phosphorylation) has been recognised over the past years. In contrast to genetic studies, studies of the protein composition are more likely to identify epigenetic effects and posttranslational modifications of proteins.

Proteomics The term proteome refers to the entire protein complement of a given cell or tissue. Proteomics, the studies of the proteome, offers an exciting new approach for the study of neuropsychiatric disorders.10–12 Different approaches have been used to study the composition of proteins in neuroscience.13 Like other ‘-omics’ (e.g. lipidomics, metabolomics), proteomics focus on the comprehensive analysis of proteins in biological fluids and tissues (e.g. blood, CSF and urine). A variety of different methods and technical approaches exist for proteomic studies.14 Early approaches used chromatography, a labour-intensive process that frequently required the breakdown of proteins prior to analysis. Over the past years, chromatography has been replaced by the wide-spread use of newer technologies (e.g. mass spectrometry). These newly developed technologies vastly enhanced the possibilities of protein analysis. Recent developments in analytical methods and datahandling have made high-throughput studies possible, allowing the screening of a high number of proteins in a very short period of time. Thanks to these advances it is now possible to analyse several thousand proteins in one single time limited step. Recent advances have further improved the sensitivity of this method, allowing the detection of polypeptides within the femtomole (fmol) range. Proteomic studies consist of different steps including sample preparation, sample separation, measurement and data-analysis. Proteins, which are present in very

6

Proteomics as a New Tool for Biomarker-Discovery in Neuropsychiatric Disorders

high concentrations (e.g. albumin) are frequently removed prior to analysis to improve the detection of less abundant proteins. Different techniques, including electrospray ionization (ESI), matrix-assisted laser desorption/ionization (MALDI) and surface-enhanced laser desorption/ionization (SELDI) are used for the ionization of molecules as the basis of molecule separation. Mass spectrometry (MS) is nowadays the most frequently used analytical tool for protein identification. Based on their charge and molecular weight, molecules can then be separated according to their mass-to-charge ratio (m/z). Proteins of interest can then be specifically identified and analysed in further steps. A detailed discussion of the underlying principles for proteomics is beyond the scope of this article (for more detailed information see12,15,16). Figure 6.1 shows the analytical set-up that was used in our studies.17,18 Proteomics-based approaches can be used to identify, characterize and validate biomarkers for different diseases. Proteomic studies can be performed on a variety of different body fluids.19 Proteomic approaches do not rely upon a preset hypothesis of the underlying pathophysiology. Instead, proteomics uses a ‘shotgun approach’ by analysing a large number of proteins for group differences. While this proteomic approach has been viewed sceptically by some groups,20 this approach

Fig. 6.1 Schematic depiction of an analytical set-up used in proteomic studies in neuropsychiatric disorders.17,18 Capillary electrophoresis is coupled on-line to a mass spectrometer utilized to separate proteins and polypeptides of body fluids by their charge and size. After the separation, the polypeptides are ionised through the application of high voltage and analyzed in the mass spectrometer (TOF, time of flight). The combination of the two instruments yields a mass spectrogram of mass per charge plotted against retention time

105

carries great potential for improved diagnostics of different disorders. The process of using proteomics for biomarker discovery can be broken down into a discovery phase followed by a validation phase. In the initial discovery phase, potential biomarkers for further large-scale studies are identified in small clinical samples. In the validation phase, these potential biomarkers are then evaluated in larger samples for their specificity and clinical usefulness.21 Blood, CSF and urine have been most frequently used for proteomic studies. For the study of neuropsychiatric disorders, CSF has the advantage of the closest proximity to the brain and may therefore be more likely to reflect underlying pathophysiology. However, this advantage must be carefully weighed against the difficulties and risks associated with obtaining CSF in clinical samples. This applies in particular to the use of CSF in routine diagnostics. Compared to plasma and serum, urine may have the advantage of a smaller degree of base-line variability in sample-composition and pose less technological challenges and may thus be more promising for proteomic studies.21 The advent of clinical proteomics over the past years has lead to the publication of numerous proteomic studies. Unfortunately not all of these studies fully satisfy rigorous scientific standards. In order to improve the quality

106

of future proteomic studies, a standardised set of requirements was recommended for future proteomic studies.21 These requirements include19: • A clear definition of a clinical need and the resulting aim of the proteomic study • A clear definition of the sample source, both for the initial phase and the validation phase • Assessment of reproducibility and comparability of the analysis platform (including assessment of variability of the sample material and a proper sampling protocol) • Protocols for quality control, data handling, and statistical evaluation The full set of recommended requirements for proteomic studies is listed in Table 6.1. The analysis of proteomic studies is further complicated by the large number of proteins present in biological fluids and tissues. Specific guidelines were issued for studies reporting protein identification data.22,23 In order to assist with better analysis of proteomic studies, different data-bases are being compiled to combine the currently available proteomic information.24,25 A human proteinpedia is currently being created that will gather information on the human proteome. This proteinpedia will improve the availability of protein data and thus facilitate the identification of proteins.26 With regards to at proteomic studies in CNS disorders, these studies are further complicated by the large number of several thousand proteins that are expressed in brain tissue.27 The proteomics approach has been used to develop disease-specific markers as well as disease specific

T.J. Raedler et al.

predictors in different somatic conditions. Examples include oncologic as well as nephrologic, endocrine and cardiologic disorders.28–31 Only few studies have so far applied this approach to the study of CNS-disorders. Looking at CNS-disorders, proteomic approaches have been increasingly used in animal research and have helped to clarify the aetiology of psychiatric disorders.32–34 Proteomic approaches are also increasingly used as an analytical tool in post mortem studies of human brains.35 The focus of this review will be on potential applications of this novel approach for the studies of neuropsychiatric disorders.

Application of Proteomics to Neuropsychiatric Disorders Alzheimer Disease As the populations are aging world-wide, the global burden of dementia and Alzheimer Disease in particular is rising.36 The diagnosis of Alzheimer Disease remains based primarily on a typical clinical presentation. In particular in the early stages of illness, the clinical symptoms are frequently non-specific and Alzheimer Disease shares considerable symptomatic overlap with normal aging, other neuropsychiatric disorders (e.g. depression) or other forms of dementia (e.g. vascular dementia, frontotemporal dementia, dementia associated with Parkinson’s Disease). However, the correct identification of early stages of Alzheimer Disease (including

Table 6.1 Recommended requirements for future clinical proteomics studies21 1. Define a clear clinical question and how the outcome of the study would improve the diagnosis and/or treatment of the disease 2. Define the patient and control populations, clinical data to be collected, as well as protocols for sampling and sample preparation 3. Define the type of samples needed for the discovery and validation phases 4. Define and validate the analytical platforms for discovery (those for validation may well differ) 5. Obtain IRB approval and written informed consent from the participant 6. Perform a pilot study on a validated discovery platform 7. Statistically evaluate data from the pilot study to calculate the number of cases and controls for the training set 8. Perform study of the training set on the validated platform based on the calculated number of cases and controls 9. Evaluate findings from the training set on blinded samples 10. Deposit datasets in a public database 11. Using these results, transfer the assay to the application platform and test using a training set (if applicable) and subsequently a blinded set 12. Apply towards clinical use to show whether the findings improve the current clinical situation

6

Proteomics as a New Tool for Biomarker-Discovery in Neuropsychiatric Disorders

pre-symptomatic stages) will become increasingly important as novel treatments promise a delayed onset of symptoms. Both relevant genetic markers (e.g. familial Alzheimer Disease) as well as neurochemical biomarkers (tau, phosphotau and amyloid fragments like A-β1-42) have been identified for Alzheimer Disease. For more detailed information on recent advances in neurochemical and imaging-based markers for Alzheimer Disease we recommend a review by Hampel et al.37 These established biomarkers can be used as a standard against with which newly developed methods can be compared.36 Given the interest in Alzheimer Disease it is not surprising that most clinical proteomic studies in neuropsychiatric disorders have so far been conducted in Alzheimer Disease.12,38,39 Several studies using proteomics in CSF in Alzheimer Disease have now been published, resulting in a panel of potential biomarkers.12 Markers were found to distinguish with good sensitivity and specificity between Alzheimer Disease, other forms of dementia and healthy controls.40 Zhang identified an Alzheimerspecific polypeptide pattern in CSF that can distinguish between Alzheimer Disease and Parkinson’s Disease. The following proteins contributed to this pattern: Tau, brain-derived neurotrophic factor (BDNF), interleukin 8, A-β42, β2-microglobulin, vitamin D binding protein, apolipoprotein (apo) AII, and apoE.41 Another panel of 15 biomarkers was able to differentiate between Alzheimer Disease and healthy controls. A subset of these biomarkers distinguished between Alzheimer Disease and other forms of dementia.42 A related panel of CSF biomarkers can be used to identify subjects with mild cognitive dysfunction, who will progress to Alzheimer Disease.43,44 Our group found specific patterns of CSF-proteins, which differentiated between subjects with Alzheimer Disease, mild cognitive impairment and healthy controls (Jahn, manuscript submitted). Other findings in CSF include differences in post-translational modifications of thransthyretin in Alzheimer Disease.45 Proteomic approaches are more likely to identify these posttranslational modifications than other approaches. As CSF is difficult to obtain, proteomic studies have also been performed in other body fluids. The protein distribution in plasma could be used to distinguish between Alzheimer Disease, other forms of dementia and healthy controls. The relevant proteins (complement factor H (CFH) precursor and alpha-2-macroglobulin

107

(alpha-2M) ) include proteins that were previously implicated in the etiology of Alzheimer Disease. However, the sensitivity of this initial study was not very high (56%).46 Liu used proteomics to look for differences in proteinexpression in the serum of subjects with Alzheimer Disease and controls. They found that apolipoprotein A-I (ApoA-I) was significantly less expressed in the serum of subjects with Alzheimer Disease.47

Schizophrenia Several years ago, we performed a small pilot-study to assess the feasibility of the proteomic approach in subjects with schizophrenia. We used proteomics to analyse CSF in a small sample of subjects with schizophrenia as well as controls. We were able to identify specific protein patterns that separated subjects with schizophrenia from healthy controls as well as subjects with Alzheimer Disease.17 These patterns were not further specified. These results seemed very promising and led to a larger-scale study in Alzheimer Disease. In a series of sophisticated studies, Huang et al used surface-enhanced laser desorption ionization (SELDI) mass spectrometry to study proteins in CSF from 179 subjects, including normal controls as well as subjects suffering from different psychiatric disorders. This method allows measuring and quantifying proteins with a molecular weight from less than 1,000 Da up to 500 kDa. Protein patterns differed significantly between drug-naïve first episode patients with schizophrenia and matched healthy controls. Further analysis of these patterns revealed an up-regulation of a 40 amino acid VGF-derived protein as well as the down-regulation of transthyretin. These differences had high sensitivity and specificity and were replicated in an independent second sample. Both the up-regulation of VGF as well as the down-regulation of transthyretin in schizophrenia were also confirmed in a post-mortem study in the prefrontal cortex. However, the authors also found that a panel of biomarkers may be more helpful to achieve disease-specificity than isolated biomarkers.48 In a follow-up study using a similar proteomic approach, the same group found a significant decrease in the concentrations of apolipoprotein A1 (apoA1) in CSF and serum in schizophrenia.49 Applying this analysis to a group of patients with early or prodromal schizophrenia,

108

the proteomic profile identified only a small proportion of the patients with prodromal symptoms, who progressed to develop full symptoms of schizophrenia.50 Craddock used SELDI-TOF MS to analyse blood samples from minimally medicated patients with schizophrenia as well as healthy controls. Alpha defensins are a group of small proteins that are a crucial part of the innate immune system. Alpha defensins were significantly increased in patients with schizophrenia as well as their unaffected twins.51 Alpha defensins were seen as a potential biomarker for the early diagnosis of schizophrenia.

T.J. Raedler et al.

and social interactions as well as repetitive behaviours. The aetiology of this frequently debilitating illness remains unclear. Corbett used a proteomics approach to analyse blood samples from children with autism as well as matched controls. Four proteins (Apolipoprotein (apo) B-100, Complement Factor H Related Protein (FHR1), Complement C1q and Fibronectin 1 (FN1) ) were differentially expressed in the serum of children with autism. Apo B-100 differed between higher- and lower-functioning children with autism.55 In a post mortem study in brains from subjects with autism, a proteomic approach found an increase of glyoxalase I in autism.56

Bipolar Disorder Conclusion So far, most proteomic studies in bipolar disorder have used post-mortem tissue. While schizophrenia was associated with a variety of changes in synaptic proteins, patients with bipolar disorder had changes in metabolic and mitochondria-associated proteins.52,53 Lithium remains the gold-standard for clinical treatment of bipolar disorder. However, lithium is associated with a risk of lithium-induced nephropathies.54 So far, no tools exist for the early recognition of subjects at risk of lithium-induced nephropathies. In a small pilot-study we obtained urine from 15 subjects who had been continuously treated with lithium for several years. All subjects had normal kidney-function as assessed with routine clinical monitoring. Using a general nephropathy screen, we found a clearly abnormal result in one subject and borderline-results in two more subjects.18 In further studies we will try to develop a specific lithium-screen with the goal of early identification of subjects at risk of lithium-induced nephropathies. Thus, our studies may be a step towards individualised medicine as an attempt to identify subjects at risk of potentially life-threatening side-effects at a very early stage.

Autism Autism is a severe neuropsychiatric disorder with an onset in early childhood. Autism is characterised by severe impairment in verbal and non-verbal communications

Proteomics is a novel analytical approach that is just starting to be applied to neuropsychiatric disorders. The underlying principle of proteomic studies consists of an analysis of a large number of proteins in a given body-fluid or tissue. First studies have shown that this approach is feasible in neuropsychiatric disorders. While the initial studies have focused on CSF, more recent studies have also used other body-fluids (plasma, urine). These initial studies have yielded very exciting and promising results that justify further studies over the years to come. So far, this novel approach is only used as a research tool. Further studies will establish the role of proteomics in neuropsychiatric disorders outside of research settings. Based on currently unmet clinical needs, the following applications of proteomics seem worthy of further investigation: • Proteomics as a tool to identify diagnostic markers for neuropsychiatric disorders • Proteomics as a tool to identify diagnostic markers for specific subtypes of neuropsychiatric disorders • Proteomics as a tool to identify responders/nonresponders to specific pharmacological treatments • Proteomics as a tool to identify subjects at risk of serious side-effects of pharmacological treatment • Proteomics as a tool to identify response (nonresponse) to a specific treatment at a very early stage As outlined above, neuropsychiatric diagnoses remain for the most part clinical diagnoses. The importance of a

6

Proteomics as a New Tool for Biomarker-Discovery in Neuropsychiatric Disorders

thorough clinical evaluation by an experienced clinician cannot be overstated. Still, many clinical presentations, especially in the early stages of illness, are non-specific with symptoms overlapping between different disorders. This issue will become even more relevant with the introduction of disease-specific treatments (e.g. amyloid modifying treatments in Alzheimer Disease). At the same time, we do not have reliable predictors available to prospectively identify the optimal treatment for an individual subject. Nor can we identify subjects at risk of severe side-effects from a given treatment (e.g. weightgain under antipsychotics; agranulocytosis under clozapine; nephropathy under lithium). Conventional thinking postulates a delayed onset of action of antipsychotics and antidepressants that occurs after several weeks. This concept has recently been challenged and an early onset of action has been shown for antipsychotics57,58 and antidepressants59 that occurs within the first week of treatment. However, the early response to pharmacological treatment can be difficult to assess unequivocally if solely based on clinical features. Nonresponders may be exposed to several weeks of unsuccessful treatment. So far, no biomarkers exist to assist with the assessment of early response to treatment. Proteomic studies carry the potential to help to address many of these issues. Still, proteomic studies are associated with a variety of methodological issues and potential pit-falls that will need to be clarified over the next few years. Different methods have been used for proteomics that may be associated with specific advantages or limitations. The currently available proteomic studies have been performed on fairly small samples. It has not yet been shown if these results can be applied to the general population. In particular, little is known about confounding factors (e.g. age, gender, ethnicity, smoking status, medication status, co-morbid illnesses, co-morbid medications) that may have an impact on proteomic studies. These issues will need to be resolved before a wide-spread use of proteomics. Another major limitation of proteomics is the current cost of analysis, which may decrease with a more widespread use of this method. In conclusion, proteomics is a new and so far investigational approach that can be used to identify biomarkers for neuropsychiatric disorders. Future studies will help to determine the role of this approach in the diagnosis and differential diagnosis of neuropsychiatric disorders.

109

References 1. Hyman SE. Can neuroscience be integrated into the DSMV? Nat Rev Neurosci 2007;8:725–732. 2. Warner R. Problems with early and very early intervention in psychosis. Br J Psychiatry Suppl. 2005;48:s104–s107. 3. Cannon TD, Cadenhead K, Cornblatt B, et al. Prediction of psychosis in youth at high clinical risk: a multisite longitudinal study in North America. Arch Gen Psychiatry 2008; 65:28–37. 4. Owen MJ, Craddock N, Jablensky A. The genetic deconstruction of psychosis. Schizophr Bull 2007;33:905–911. 5. Craddock N, O’Donovan MC, Owen MJ. Genes for schizophrenia and bipolar disorder? Implications for psychiatric nosology. Schizophr Bull 2006;32:9–16. 6. Meyer-Lindenberg A, Weinberger DR. Intermediate phenotypes and genetic mechanisms of psychiatric disorders. Nat Rev Neurosci 2006;7:818–827. 7. Isles AR, Wilkinson LS. Epigenetics: what is it and why is it important to mental disease? Br Med Bull 2008;85:35–45. 8. Moffitt TE, Caspi A, Rutter M. Strategy for investigating interactions between measured genes and measured environments. Arch Gen Psychiatry 2005;62:473–481. 9. Caspi A, Moffitt TE. Gene-environment interactions in psychiatry: joining forces with neuroscience. Nat Rev Neurosci 2006;7:583–590. 10. Pennington K, Cotter D, Dunn MJ. The role of proteomics in investigating psychiatric disorders. Br J Psychiatry 2005;187:4–6. 11. Andrade EC, Krueger DD, Nairn AC. Recent advances in neuroproteomics. Curr Opin Mol Ther 2007;9:270–281. 12. Zetterberg H, Rüetschi U, Portelius E, et al. Clinical proteomics in neurodegenerative disorders. Acta Neurol Scand 2008;118:1–11. 13. Choudhary J, Grant SG. Proteomics in postgenomic neuroscience: the end of the beginning. Nat Neurosci 2004;7: 440–445. 14. Fonteh AN, Harrington RJ, Huhmer AF, et al. Identification of disease markers in human cerebrospinal fluid using lipidomic and proteomic methods. Dis Markers 2006;22: 39–64. 15. Tannu NS, Hemby SE. Methods for proteomics in neuroscience. Prog Brain Res 2006;158:41–82. 16. Paik YK, Kim H, Lee EY, et al. Overview and introduction to clinical proteomics. Methods Mol Biol 2008;428:1–31. 17. Wittke S, Mischak H, Walden M, et al. Discovery of biomarkers in human urine and cerebrospinal fluid by capillary electrophoresis coupled to mass spectrometry: towards new diagnostic and therapeutic approaches. Electrophoresis 2005;26:1476–1487. 18. Raedler TJ, Wittke S, Jahn H, et al. Capillary electrophoresis mass spectrometry as a potential tool to detect lithium-induced nephropathy:preliminaryresults.ProgNeuropsychopharmacol Biol Psychiatry 2008;32:673–678. 19. Good DM, Thongboonkerd V, Novak J, et al. Body fluid proteomics for biomarker discovery: lessons from the past hold the key to success in the future. J Proteome Res 2007;6:4549–4555.

110 20. Lescuyer P, Hochstrasser D, Rabilloud T. How shall we use the proteomics toolbox for biomarker discovery? J Proteome Res 2007;6:3371–3376. 21. Mischak H, Apweiler R, Banks RE, et al. Clinical proteomics: a need to define the field and to begin to set adequate standards. Proteomics Clin Appl 2007;1:148–156. 22. Carr S, Aebersold R, Baldwin M, et al. The need for guidelines in publication of peptide and protein identification data: working group on publication guidelines for peptide and protein identification data. Mol Cell Proteomics 2004;3:531–533. 23. Bradshaw RA, Burlingame AL, Carr S, et al. Reporting protein identification data: the next generation of guidelines. Mol Cell Proteomics 2006;5:787–788. 24. Hamacher M, Apweiler R, Arnold G, et al. HUPO Brain Proteome Project: summary of the pilot phase and introduction of a comprehensive data reprocessing strategy. Proteomics 2006;6:4890–4898. 25. Hamacher M, Stephan C, Eisenacher M, et al. Maintaining standardization: an update of the HUPO Brain Proteome Project. Expert Rev Proteomics 2008;5:165–173. 26. Mathivanan S, Ahmed M, Ahn NG, et al. Human Proteinpedia enables sharing of human protein data. Nat Biotechnol 2008;26:164–167. 27. Wang H, Qian WJ, Chin MH, et al. Characterization of the mouse brain proteome using global proteomic analysis complemented with cysteinyl-peptide enrichment. J Proteome Res 2006;5:361–369. 28. Koomen JM, Haura EB, Bepler G, et al. Proteomic contributions to personalized cancer care. Mol Cell Proteomics 2008;7:1780–1794 29. Unwin RD, Whetton AD. How will haematologists use proteomics? Blood Rev 2007;21:315–326. 30. Decramer S, Wittke S, Mischak H, et al. Predicting the clinical outcome of congenital unilateral ureteropelvic junction obstruction in newborn by urinary proteome analysis. Nat Med 2006;12:398–400. 31. Rossing K, Mischak H, Dakna M, et al. Urinary proteomics in diabetes and CKD. J Am Soc Nephrol 2008;19: 1283–1290. 32. Kobeissy FH, Ottens AK, Zhang Z, et al. Novel differential neuroproteomics analysis of traumatic brain injury in rats. Mol Cell Proteomics 2006;5:1887–1898. 33. Vercauteren FG, Flores G, Ma W, et al. An organelle proteomic method to study neurotransmission-related proteins, applied to a neurodevelopmental model of schizophrenia. Proteomics 2007;7:3569–3579. 34. Tannu NS, Howell LL, Hemby SE. Integrative proteomic analysis of the nucleus accumbens in rhesus monkeys following cocaine self-administration. Mol Psychiatry 2008 (in press). 35. Beasley CL, Pennington K, Behan A, et al. Proteomic analysis of the anterior cingulate cortex in the major psychiatric disorders: evidence for disease-associated changes. Proteomics 2006;6:3414–3425. 36. Blennow K, de Leon MJ, Zetterberg H. Alzheimer’s disease. Lancet 2006;368:387–403. 37. Hampel H, Bürger K, Teipel SJ, et al. Core candidate neurochemical and imaging biomarkers of Alzheimer’s disease. Alzheimers Dement 2008;4:38–48.

T.J. Raedler et al. 38. Papassotiropoulos A, Fountoulakis M, Dunckley T, et al. Genetics, transcriptomics, and proteomics of Alzheimer’s disease. J Clin Psychiatry 2006;67:652–670. 39. Lovestone S, Güntert A, Hye A, et al. Proteomics of Alzheimer’s disease: understanding mechanisms and seeking biomarkers. Expert Rev Proteomics 2007;4:227–238. 40. Abdi F, Quinn JF, Jankovic J, et al. Detection of biomarkers with a multiplex quantitative proteomic platform in cerebrospinal fluid of patients with neurodegenerative disorders. J Alzheimers Dis 2006;9:293–348. 41. Zhang J, Sokal I, Peskind ER, et al. CSF multianalyte profile distinguishes Alzheimer and Parkinson diseases. Am J Clin Pathol 2008;129:526–529. 42. Simonsen AH, McGuire J, Podust VN, et al. A novel panel of cerebrospinal fluid biomarkers for the differential diagnosis of Alzheimer’s disease versus normal aging and frontotemporal dementia. Dement Geriatr Cogn Disord 2007;24:434–440. 43. Simonsen AH, McGuire J, Hansson O, et al. Novel panel of cerebrospinal fluid biomarkers for the prediction of progression to Alzheimer dementia in patients with mild cognitive impairment. Arch Neurol 2007;64:366–370. 44. Hansson O, Zetterberg H, Buchhave P, et al. Association between CSF biomarkers and incipient Alzheimer’s disease in patients with mild cognitive impairment: a follow-up study. Lancet Neurol 2006;5:228–234. 45. Biroccio A, Del Boccio P, Panella M, et al. Differential post-translational modifications of transthyretin in Alzheimer’s disease: a study of the cerebral spinal fluid. Proteomics 2006;6:2305–2313. 46. Hye A, Lynham S, Thambisetty M, et al. Proteome-based plasma biomarkers for Alzheimer’s disease. Brain 2006;129: 3042–3050. 47. Liu HC, Hu CJ, Chang JG, et al. Proteomic identification of lower apolipoprotein A-I in Alzheimer’s disease. Dement Geriatr Cogn Disord 2006;21:155–161. 48. Huang JT, Leweke FM, Oxley D, et al. Disease biomarkers in cerebrospinal fluid of patients with first-onset psychosis. PLoS Med 2006;3:e428. 49. Huang JT, Wang L, Prabakaran S, et al. Independent proteinprofiling studies show a decrease in apolipoprotein A1 levels in schizophrenia CSF, brain and peripheral tissues. Mol Psychiatry 2000;13:1118–1128. 50. Huang JT, Leweke FM, Tsang TM, et al. CSF metabolic and proteomic profiles in patients prodromal for psychosis. PLoS ONE 2007;2:e756. 51. Craddock RM, Huang JT, Jackson E, et al. Increased alpha defensins as a blood marker for schizophrenia susceptibility. Mol Cell Proteomics 2008;7:1204–1213. 52. Pennington K, Beasley CL, Dicker P, et al. Prominent synaptic and metabolic abnormalities revealed by proteomic analysis of the dorsolateral prefrontal cortex in schizophrenia and bipolar disorder. Mol Psychiatry 2008;13:1102–1117. 53. Behan A, Byrne C, Dunn MJ, et al. Proteomic analysis of membrane microdomain-associated proteins in the dorsolateral prefrontal cortex in schizophrenia and bipolar disorder reveals alterations in LAMP, STXBP1 and BASP1 protein expression. Mol Psychiatry 2008 (in print). 54. Raedler TJ, Wiedemann K. Lithium-induced nephropathies. Psychopharmacol Bull 2007;40:134–149.

6

Proteomics as a New Tool for Biomarker-Discovery in Neuropsychiatric Disorders

55. Corbett BA, Kantor AB, Schulman H, et al. A proteomic study of serum from children with autism showing differential expression of apolipoproteins and complement proteins. Mol Psychiatry 2007;12:292–306. 56. Junaid MA, Kowal D, Barua M, et al. Proteomic studies identified a single nucleotide polymorphism in glyoxalase I as autism susceptibility factor. Am J Med Genet A 2004;131:11–17.

111

57. Agid O, Kapur S, Arenovich T, et al. Delayed-onset hypothesis of antipsychotic action: a hypothesis tested and rejected. Arch Gen Psychiatry 2003;60:1228–1235. 58. Raedler TJ, Schreiner A, Naber D, et al. Early onset of treatment effects with oral risperidone. BMC Psychiatry 2007;7:4. 59. Posternak MA, Zimmerman M. Is there a delay in the antidepressant effect? A meta-analysis. J Clin Psychiatry 2005;66:148–158.

Chapter 7

Schizophrenia Endophenotypes as Treatment Targets Stephen I. Deutsch, Barbara L. Schwartz, Richard B. Rosse, John Mastropaolo, Ayman H. Fanous, Abraham Weizman, Jessica A. Burket and Brooke L. Gaskins

Abstract The concept of the “endophenotype” will be reviewed and its potential to elucidate genetic variations that are causally related to schizophrenia and gene-environment interactions that may be obscured with dependence on the categorical diagnosis of this disorder or reliance on symptoms and symptom clusters to detect these associations. The identification of endophenotypes that reflect genetic variations may also lead to refinement of diagnosis, especially subtyping of the disorder. Conceivable, a spectrum of schizophrenias may exist reflecting a cluster of critical endophenotypes that must be present in all individuals diagnosed with the disorder and/or discrete, nonoverlapping endophenotypes that are found in geneticallydistinct groups of individuals; nonetheless, all patients that are diagnosed with the disorder using operationallydefined diagnostic criteria, such as DSM-IV-TR, share common phenotypic characteristics even though they may have different endophenotypic profiles. Moreover, because endophenotypes may reflect pathophysiological processes that contribute to the emergence of the clinical syndrome, they may serve as “targets” for pharmacotherapeutic interventions that treat a fundamental pathophysiological disturbance, as opposed to “voices.” Also, because endophenotypes may be

S.I. Deutsch, B.L. Schwartz, R.B. Rosse, J. Mastropaolo, A.H. Fanous, J. A. Burket, and B. L. Gaskins Mental Health Service Line, Department of Veterans Affairs Medical Center, Washington, USA S.I. Deutsch, B.L. Schwartz, R.B. Rosse, and A.H. Fanous Department of Psychiatry, Georgetown University School of Medicine, Washington, USA A. Weizman Sackler Faculty of Medicine, Tel-Aviv University, Ramat-Aviv, Tel-Aviv, Israel

detected in unaffected relatives and patients with subsyndromal disorder or spectrum disorder, these targeted interventions may prove beneficial to the functionality of these persons as well (including prevention of emergence of the clinical syndrome); presumably, subtle deficits in socialization and cognition in these seemingly unaffected or “less-affected” persons may be related to the genetic variations that will be elucidated by the endophenotypic approach. Finally, a heuristic example will be given of an evolving medication strategy that addresses a presumptive endophenotype of schizophrenia that is associated with a precise genetic locus on chromosome 15, whose region contains the promoter and gene for the α7 subunit of the nicotinic acetylcholine receptor. Because of the importance of the construct of the endophenotype to the elucidation of genetic variations and neurobiological mechanisms relevant to the clinical syndrome of schizophrenia, their segregation in seemingly unaffected family members and patients with schizophrenia spectrum disorders (e.g., schizotypy), utility in clarifying geneenvironment interactions that can precipitate overt disorder, and potential role as targets for rationale pharmacotherapeutic strategies, the NIMH has funded a multi-site collaboration, i.e., the Consortium on the Genetics of Schizophrenia (COGS), whose goal is to examine “the genetic architecture of quantitative endophenotypes in families with schizophrenia.”1–3 Hopefully, the COGS will validate several “candidate” neurocognitive and neurophysiological endophenotypes, and clarify their potential to refine our diagnosis of the disorder and its subtypes, identify persons at-risk before overt expression of the disorder, clarify environmental and other “second-hit” risk factors that increase liability of expression of overt illness, and stimulate development of more effective pharmacotherapies.

M.S. Ritsner (ed.), The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes, © Springer Science + Business Media B.V. 2009

113

114

Keywords Schizophrenia susceptibility genes • endophenotypes • epistasis • P50 sensory abnormality • therapeutic target Abbreviations DSM-IV-TR: Diagnostic and Statistical Manual of Mental Disorders-IV-Text Revision; NIMH: National Institute of Mental Health; COGS: Consortium of the Genetics of Schizophrenia; DNA: Deoxyribonucleic acid; PPI: Prepulse Inhibition; GAD67: Glutamate decar-boxylase; GABA: Gammaaminobutyric acid; fMRI: Functional magnetic resonance imaging; COMT: Catechol-O-methyl transferase; RGS4: Regulator of G protein signaling 4; DISC1: Disrupted in Schizophrenia 1; nAChR: Nicotinic acetylcholine receptor; MMN: Mismatch negativity; CHRNA7: Cholinergic receptor, nicotinic, alpha 7; MK-801: Dizocilpine; PCP: Phencyclidine; PANSS: Positive and Negative Syndrome Scale

Introduction The results of twin, adoption and family studies support a significant heritable component to the complex disorder that is schizophrenia. However, this component differs from what is found in neuropsychiatric disorders that follow Mendelian modes of genetic transmission (e.g., Huntington’s disease), whose clinical changes may be unambiguously linked to changes in the genetic coding of the sequence of amino acids of single proteins.1,4,5 Complex genetic disorders do not reflect the effect of single genes alone, especially syndromal psychiatric disorders that are diagnosed on the basis of the presence of variable constellations of signs and symptoms. These disorders may result from interactions between several genes (the concept of epistasis), some of which may be normally-occurring polymorphisms and others truly different from the unaffected population, environmental influences and epigenetic factors.6 The genetic variations that are referred to most commonly are polymorphisms or naturallyoccurring, stable inherited sequences of DNA, as opposed to mutations. Given the fact that the human genome contains approximately ten million polymorphic sites and the likely possibility that interactions between genes (or endophenotypes), as opposed to the effect of a single gene, are responsible for disease liability in

S.I. Deutsch et al.

most instances, large samples of families (for linkage analysis) and unrelated individuals (for most association studies) are needed to determine (causal) relationships between genes and disease phenotypes.6 Epigenetic factors include stable covalent modifications of proteins that interact with DNA (e.g., acetylation of histone proteins within the nucleosome) and of the DNA itself (e.g., methylation of cytosine residues in promoter regions) that influence transcription; these stable, heritable organic modifications may come about as a result of complex feedback pathways that include exposure to environmental influences.7 Moreover, the combination of genes that contributes to the emergence of seemingly similar complex genetic disorders may differ among clinically-affected individuals. With the elucidation of the approximately 30,000 discrete human structural genes, representing about 3,000,000,000 base pairs, of which about 16,000 are expressed in brain, and the ability to identify mutant genes, normal polymorphisms and quantify their expression using gene chips, microexpression arrays and powerful bioinformatic and statistical support, it may soon be possible to resolve genetically-distinctive disorders within the DSM-IV-TR diagnostic categories. This genetically-determined nosology will be based on different combinations or sets of genes among affected individuals within the categorical diagnostic assignments that are currently employed. The geneticallydistinctive group of schizophrenias may account for differences in onset, predominant symptom constellation, course, outcome and response to treatment. Given the likely existence of genetically-distinctive disorders, it is remarkable that existing pharmacotherapies for schizophrenia are as effective as they are for so many patients, although treatment-refractoriness and a paucity or absence of medication strategies to treat domains of schizophrenia psychopathology that contribute to poor functional outcomes, such as cognitive and negative symptom domains, still exist. Some of the endophenotypes/intermediate phenotypes may not be narrowly associated with schizophrenia per se, but, rather reflect disruption along complex circuits involving frontal/temporal cortex and limbic structures that may also be pathologically involved in bipolar disorder. Thus, it may not be surprising that some of these endophenotypic abnormalities have been found in patients with bipolar disorder and their biological relatives. Ideally, elucidation of genetic architecture will lead to a nosology that more accu-

7 Schizophrenia Endophenotypes as Treatment Targets

rately reflects pathophysiological differences and more readily suggests pharmacotherapeutic targets than existing categorical syndromal approaches.8

What Is an Endophenotype? The concept of the endophenotype (or intermediate phenotype) may facilitate the identification of genetic variations that are associated with schizophrenia and clarify the pathways from genetic variations to alterations of specific neurocognitive and neurophysiological functions and, eventually, the emergence of the clinical syndrome (that is recorded in categorical fashion as present or absent). The deciphering (or “deconstructing”) of the genetic architecture of the endophenotypes may also reveal a “network” of genes that must act in concert to produce the endophenotype or clinical syndrome, clarify interactions with environmental influences that lead from endophenotype to illness expression and, perhaps, modifying genetic and environmental influences that prevent progression from endophenotype to an overt clinical syndrome.1,6,9 Early attempts to identify associations between genetic loci and the categorical, quantitative diagnosis of schizophrenia were “atheoretical;” in addition, because of the large number of possible associations that would have to be tested (i.e., about 30,000 genes), the likelihood of finding association was low. However, the statistical likelihood of finding associations between endophenotypes and genetic loci is increased because of their functional relevance to the disease process itself and “quantitative” nature, as opposed to the categorical presence or absence of schizophrenia. Also, the inclusion of unaffected relatives that manifest the endophenotype in the search for possible associations increases the number of subjects and, thus, the power of the investigation.9

Challenges Associated with Endophenotypic Studies in Schizophrenia As opposed to clinical signs and symptoms, endophenotypes are often not visible to the “naked eye,” requiring laboratory-based or specialized testing procedures to detect them (e.g., electrophysiological

115

and neurocognitive measures, respectively). Further, in contrast to the heterogeneity of the clinical syndrome of schizophrenia, which occurs even when the disorder is diagnosed using rigorous definitional criteria, sample selection using endophenotypes, which can be measured quantitatively, can lead to homogeneity with respect to the presence and severity of the endophenotype.1 In addition to their heritability, the value of the endophenotype is its stability (i.e., they are not statedependent); thus, they are detectable in both exacerbated and remitted patients. Their heritability implies that they segregate in biological family members of index patients with schizophrenia, most of whom may be unaffected with the disorder. Moreover, while it is very improbable that the effects of a single abnormal gene or genetic variation on illness will be discernable, there is greater likelihood that relationships can be found between endophenotypes, which reflect functional neurophysiological and neurocognitive abnormalities, and single genes.1–6,10–12 Although certain endophenotypes may be necessary but not sufficient for the manifestation of overt illness in most patients, complex interactions between multiple genetically-determined endophenotypes and environmental “second-hit” influences are more likely to result in the clinical disorder.13 Endophenotypes are proximal to genetic variations that cosegregate with, and contribute to, the clinical syndrome of schizophrenia, relative to the clinical syndrome itself; presumably, the expression of these genetic variations begins early in development and, by the time the neurobiological function is measured in the adult, their earliest developmental expression may be dramatically modified by nongenetic influences.1 Thus, the theoretically tight correlation between genetic polymorphisms and presumptive quantitative endophenotypes measured in adulthood or the usual age of illness recognition may not be found. Also, some phenotypic manifestations of illness may reflect the consequences of adaptive changes and expression of genes that do not represent disease alleles. Further, if the deleterious effect of a disease allele occurs during the development of the brain (i.e., a “first-hit” that may be necessary but not sufficient for illness expression), it may not be possible to detect this allele using statistical approaches that seek associations between (endo)phenotype and genotype in large samples. Curiously, if actual illness reflects interactions between multiple endophenotypes, given the independent segregation of endophenotypes

116

within families, it may be easier to link a quantitative functional endophenotype to a gene within unaffected biological relatives than the patients themselves. Again, the prevalence of the endophenotype in biological relatives of index patients with schizophrenia is higher than the general population. The interactions between multiple endophenotypes in patients may obscure an association between a single quantitative endophenotype and a specific genetic variation. Endophenotypes have already illustrated the heterogeneity of schizophrenia and divergence of individual clinically-diagnosed patients from each other. For example, although many schizophrenia patients have deficits of both P50 suppression and PPI, there are individual patients that show only one of these independently segregating and dissimilar gating deficits. As discussed, in the search for disease genes and their causal relation to pathophysiological processes, endophenotypes offer significant advantages in complex diseases characterized by genetic and phenotypic heterogeneity. For example, associations between blood glucose levels and genetic variations are more likely to be found in diabetes mellitus than associations between elements of the clinical syndrome (such as excessive urination, thirst, increased appetite or fatigue) and disease genes. Also, the clinical syndrome of diabetes mellitus illustrates the challenge to finding disease genes in disorders that are phenotypically similar but genetically heterogeneous, nonoverlapping disorders; for example, the etiology and pathogenesis of decreased insulin secretion and diminished insulin sensitivity may differ, but, nonetheless, result in many clinical similarities.5 The persistence of schizophrenia across cultures, countries, racial and ethnic groups and the stability of its prevalence (i.e., a basal rate of occurrence of about 1% of the population) suggest that the schizophrenia vulnerability genes offer some functional advantage(s) for the human species; certainly, overt illness is not associated with a reproductive advantage.6 Perhaps, an attenuated form of the “wanderlust” that is seen in some patients in association with good, as opposed to impaired, planning and problem solving skills and general functioning of the frontal lobes enabled migration and settling of new and, oftentimes, hostile environments across the planet by our ancestral human population. Whereas associations between “wanderlust” and genes may be difficult to demonstrate, associations between relevant neurocognitive endophenotypes and

S.I. Deutsch et al.

genes may be possible utilizing specialized statistical methodologies.6 Selective advantage of at least some traits during human history may be reflected in an enrichment of some coding regions of putative schizophrenia susceptibility genes within the general population. Alternatively, incomplete or variable penetrance of some genes or inheritance of less than complete sets of susceptibility genes among seemingly “unaffected” persons could confer selective advantage. In addition to “wanderlust,” the advantages to the human species may be reflected in creativity, language and religious expression.14,15 Linkage analysis is performed on biologicallyrelated individuals and is based on the assumption that genetic variations contributing to disease are derived from a common ancestor.6 Marker loci close to the genetic variation contributing to overt illness segregate with the illness in biological family members; the latter is evidence of linkage. Associations are often sought in large populations of unrelated individuals wherein the linkage disequilibrium of marker loci (i.e., loci that segregate with the putative disease genes) are shown in patients affected by illness (cases), as compared to controls.6 The location of these marker loci could lead to the identification of disease genes in the same chromosomal location. A powerful statistical genetic strategy for identifying disease genes is the “transmission-disequilibrium test.” In this “test,” associations of marker loci occurring at greater than chance levels are sought in affected patients and their heterozygous parents (i.e., above 50%).6 Confounding the search for genetic variations that confer risk for schizophrenia and are associated with quantifiable endophenotypes is the epigenetic regulation of gene expression; early “nongenetic” insults to the developing brain may result in persistent, heritable covalent modifications of histone proteins (e.g., acetylation) and cytosine residues within promoter regions of genes (i.e., methylation) that alter gene transcription.7 The epigenetic changes affect the packaging of chromatin (i.e., chromosomal remodeling) and influence the accessibility of transcriptional factors to promoter regions of DNA and, thereby, affect gene expression. There are provocative data suggesting that epigenetic regulation is responsible for the diminished expression of the 67 kDa isoform of glutamic acid decarboxylase (GAD67), an enzyme responsible for GABA synthesis, and reelin, a signaling molecule necessary for the normal lamination of the developing

7 Schizophrenia Endophenotypes as Treatment Targets

brain, that has been reported in some postmortem studies of brains from patients with schizophrenia.7

Candidate Schizophrenia Endophenotypes Neuroimaging Markers Neuroimaging studies of normal twins showed that the heritability of the anatomy of localized middle frontal cortical regions near Brodmann’s areas 9 and 46 is at least 0.9; thus, genetic factors contribute significantly to the development of frontocortical regions implicated in the pathophysiology of schizophrenia.16 Moreover, these studies revealed that not only are monozygotic co-twins virtually identical in terms of grey matter concentrations in these frontal brain regions, associations were found between frontal grey matter differences and test performance across multiple cognitive domains.16 Thus, brain structure, in addition to neurophysiological and neurocognitive measures, may serve as endophenotypes that are closely linked to schizophrenia susceptibility genes that have functional relevance (i.e., these genes influence brain structures that, in turn, underlie cognitive processes). fMRI studies showed inefficient prefrontal information processing during performance of a working memory task in unaffected siblings of patients with schizophrenia that resembled the fMRI and information processing deficits seen in patients.17 These functional imaging data in nonaffected siblings support the use of neuroimaging measures as endophenotypes of the disorder. The corpus callosum is the brain structure responsible for the interhemispheric transfer of information. Similar to their affected siblings with schizophrenia, unaffected monozygotic co-twins showed significant callosal displacements, especially an upward bowing of the callosum.16 Various provocative associations have been reported between functional and anatomic neuroimaging abnormalities in schizophrenia and specific genes and genetic loci, including the “valine” polymorphism of the catechol-Omethyltransferase (COMT) gene leading to the more rapid breakdown of dopamine on chromosome 22, the “G-protein signaling subtype gene (RGS4)” on chromosome 1q21, and the “disrupted-in-schizophrenia 1 (DISC1) gene” on chromosome 1q421.16

117

Neurocognitive and Neurophysiological Several “candidate” endophenotypes that are associated with liability to schizophrenia are emerging; they include neurophysiological and neurocognitive measures that reflect output of neuronal systems that may be disrupted as a consequence of disease genes (i.e., promotion, suppression or altered rates of their expression). We will mention briefly some of the more promising or intensely studied candidate endophenotypes before focusing on one that has implicated the gene for the α7 subunit of the nicotinic acetylcholine receptor (nAChR), which confers unique pharmacological properties on the receptor. The diminished expression of the α7 nAChR in the brains of at least some patients with schizophrenia, the possible linkage of the gene for the α7 subunit with P50 sensory gating abnormalities and schizophrenia, and the increased prevalence of the P50 sensory gating abnormality in seemingly-unaffected, biological relatives of index patients with schizophrenia led to the development and testing of a pharmacotherapeutic strategy for the sustained stimulation of this receptor.13,18,19 The strategy exploits the combination of a positive allosteric modulator of nAChRs in general (i.e., galantamine) and selective α7 nAChR agonist (i.e., choline derived from orally-administered CDP-choline) to improve the efficiency of coupling between the binding of choline and channel opening. The results of a promising pilot investigation will be presented.18,19 To review, the ideal candidate for an endophenotype is one that can be measured in a quantitative fashion, is distributed continuously in the population, is proximal to the action of the gene (as opposed to the clinical syndrome), reflects a function that, when disturbed, is relevant to the illness, is state independent and artifactually affected by medication, and shows high heritability such that its prevalence among closely-related biological relatives of probands with schizophrenia is higher than the basal rate in the general population.2,5 Using these criteria, several promising candidate neurocognitive and neurophysiological endophenotypes have been proposed and under active investigation (e.g., the multisite Consortium on the Genetics of Schizophrenia [COGS] funded by NIMH); however, no measure has yet been shown to satisfy all criteria unambiguously. The list of promising cognitive endophenotypes include deficits of sustained focused attention, verbal declarative memory, working

118

memory, antisaccade task performance (i.e., ability to inhibit the prepotent visual grasp reflex) and, perhaps, processing of facial information.2,5 The antisaccade task assesses aspects of attention, frontocortical inhibition and working memory (i.e., noting the position of the stimulus on the mental sketchpad and using this positional information to guide the correct response). The candidate neurophysiological endophenotypes reflect deficits in information processing that extend from early preattentive processes to much later higher cortical ones (i.e., from as early as 50 ms to later than about 300 ms after stimulus presentation) and fall into two broad categories: measures of inhibitory failure and measures of impaired deviance detection.3 These fundamental processes regulate the “inflow of information from the environment,” inhibiting activation of cortical pathways in response to redundant or irrelevant stimuli, while facilitating this activation in response to “deviant, novel, or salient stimuli”.3 Prominent candidate endophenotypic measures of inhibitory failure include prepulse inhibition (PPI) of the startle reflex and suppression of the P50 auditory evoked potential, and measures of impaired deviance detection are mismatch negativity and the P300 eventrelated potential. Normally, the startle response is attenuated, which is most often measured electromyographically in humans as contraction of the orbicularis oculi muscle, when a weak prestimulus is presented 30 to 300 ms before the startling stimulus. The circuitry involved in mediation of PPI in rodents includes mesial temporal cortex, medial prefrontal cortex, striatum, pallidum and pontine tegmentum; thus, the neuroanatomic circuitry is relevant to pathophysiological mechanisms in schizophrenia. The PPI phenotype in rodents is (largely) genetically-determined as heritable strain differences have been shown; also, heritability estimates of greater than 0.5 were reported in studies of human twin pairs.3,20,21 Higher prevalence of impaired PPI has been determined in unaffected siblings of schizophrenia probands, compared to control subjects. Thus, PPI of the (acoustic) startle response is an attractive candidate phenotype. Mismatch Negativity (MMN) is the electrophysiological response, beginning as early as 50 ms and peaking as late as 200 ms after stimulus presentation, to an “oddball” auditory stimulus that abruptly interrupts a sequence of repetitive standard sounds. The “oddball” or deviant stimulus can be distinguished from the stan-

S.I. Deutsch et al.

dard repetitive one by changes in duration, pitch or loudness. The evoked MMN response is a preattentive and highly stable characteristic over time that reflects some automatic “sensorimemory” process; the response is elicited in sleeping infants and brain-injured and comatose patients. Schizophrenia patients show deficits in deviance detection, as reflected in a reduced response to an “oddball” auditory stimulus that may be correlated with functional disability, such as ability to live independently. In contrast to the early preattentive response to an “oddball” stimulus in the MMN paradigm, the P300 event-related potential reflects a variety of (later) cognitive processes, including “directed attention, the contextual updating of working memory, and the attribution of salience to a deviant stimulus.”3 The P300 is a multicomponent response that reflects the complementary actions of discrete anatomic brain regions (e.g., hippocampus, thalamus, inferior parietal lobe, superior temporal gyrus, and frontal lobe). Schizophrenia patients have been reported to show reduced amplitude and prolonged latency of the evoked P300 response to an oddball stimulus; reduced amplitude is the more robust and reliable abnormal response. Correlations have been reported between P300 abnormalities and measures of decreased volume in left superior temporal gyrus and frontal lobe in schizophrenia; the possibility of linkage of P300 abnormalities to the region on chromosome 6 containing the dysbindin gene, a candidate gene in schizophrenia, was shown in a collaborative study on the genetics of alcoholism, and the DISC1 gene in a family pedigree that included schizophrenia patients and unaffected family members with a balanced translocation of the long arm of chromosome 1 and the short arm of chromosome 11.3 In any event, MMN and the P300 abnormality are promising neurophysiological endophenotypes that may reflect genetic variations occurring among patients with schizophrenia and closely-related, unaffected biological family members.

P50 Suppression Deficit: An Endophenotypic Target in Schizophrenia Ordinarily, the amplitude of the P50 waveform evoked in response to the second of an identical pair of auditory stimuli presented 500 ms apart is reduced, when

7 Schizophrenia Endophenotypes as Treatment Targets

compared with the amplitude of the response evoked by the first stimulus of the pair. The amplitudes of the two evoked potentials are averaged over multiple trials. A common outcome measure that is used to quantify the magnitude of sensory gating is the ratio of the average evoked amplitude to the second or test stimulus compared to the average evoked potential to the first or conditioning stimulus, referred to as the T/C ratio; ratios above 50% are taken as evidence of impaired sensory gating. When examining the T/C or P50 suppression ratio in longitudinal studies or clinical trials, a decrease in the amplitude of the second evoked potential, as opposed to increased amplitude of the first, is interpreted as better evidence of improved sensory gating.12 Using the “P50 Suppression” paradigm, a sensory gating abnormality has been demonstrated in a large population of patients with schizophrenia and their closely related biological relatives, many of whom are unaffected with the illness. Moreover, the sensory gating deficit, reflected in an abnormality of P50 suppression, that segregates within families of probands with schizophrenia appears to be inherited in an autosomal dominant fashion.12,22 The failure to suppress P50 after repeated exposure to identical stimuli in clinically-unaffected, closely-related biological relatives suggests that this endophenotype may be a “necessary but not sufficient” condition for overt expression of at least some presentations of schizophrenia. Sensory gating is not an example of learning, rather it is a preattentional, pre-conscious neurophysiological process that allows the individual to “ignore” repetitive stimuli in the environment; this facilitates the direction of attentional resources toward novel and less predictable stimuli. P50 is an electroencephalographic wave form that is evoked and measured approximately 50 ms after the presentation of a stimulus; acoustic stimuli are usually used to evoke this electrophysiological response. The integrity of the CA3 and CA4 region of the hippocampus and cholinergic projection pathways into this region is necessary for normal suppression of the evoked P50 response to the second of a pair of identical auditory stimuli presented 500 ms apart. Importantly, the density of nAChRs in the hippocampus containing the α7 subunit appear to contribute prominently to normal suppression of the P50 response; an inverse relation between density of hippocampal α7 nAChRs and magnitude of suppression of the homologous P20-N40 complex in rodents has been reported among inbred

119

mouse strains.23 Pharmacological, receptor binding and genetic data converge to implicate diminished expression and impaired signal transduction by the α7 nAChR as an important pathogenetic mechanism of the disorder. For example, abnormalities of P50 suppression have been transiently normalized in patients with schizophrenia and their biological relatives with nicotinic acetylcholine agonist interventions.24,25 Postmortem studies of brains of patients with schizophrenia revealed decreased expression of the α7 nAChR (e.g., frontal cortex, hippocampus and thalamus). Genetic studies have also shown evidence of “linkages” between impaired P50 sensory gating abnormalities, the locus on chromosome 15q13–14 that contains the gene for the α7 nAChR (CHRNA7), abnormal promoter variants of CHRNA7 and schizophrenia.26 The existence of (selective) disturbances of nAChR-mediated neurotransmission should not be surprising in view of the approximately threefold to fourfold higher prevalence of cigarette smoking among schizophrenia patients, compared with the general population. From a descriptive perspective, P50 gating deficits and “hippocampal” pathology may be reflected in disturbances of focused attention, sensory flooding and disturbance of associations that are commonlymanifested in this disorder. There has been much speculation about the relationship of the P50 abnormality to descriptive phenomena of schizophrenia (e.g., sensory flooding), as well as positive, negative and cognitive symptoms. Also, although the “generator” of the P50 response is not known with certainty (or even if thee is only one generator), evidence consistent with a (primary) hippocampal location suggests that the P50 abnormality could be involved in the disturbance of associations (i.e., inability to gate “loose” or contextually-inappropriate associations and, thereby, prevent their emergence into consciousness). In addition to the hippocampus, other potential sites of this generator include Heschl’s gyrus, superior temporal gyrus, prefrontal cortex and rhinal cortex.27 Although they are often confounded by small sample sizes, to date, widespread correlations have either not been found, or not examined, between P50 abnormalities and other potential cognitive endophenotypes, positive and negative symptoms, and measures of illness severity in schizophrenia patients.27 Possible interpretations of these (preliminary) data could include the obscuring of any contributions of the P50 abnormality to the phenotypic manifestations of cognitive, positive and negative symptoms and illness severity by the

120

clinical syndrome itself and (genetic) heterogeneity of the disorder (i.e., the group of schizophrenias may represent different admixtures of endophenotypes that affect presence and severity of clinical manifestations that are inconsistently present and may or may not be related to specific underlying endophenotypes). Nonetheless, a recent review of this area referenced some work reporting possible correlations between P50 abnormalities and measures of attention, cognitive processing speed, a visual implicit memory task, forward digit span that may be related to working memory, and a semantic priming task in patients with schizophrenia.27 Clearly, more research needs to be conducted that could elucidate possible pathways leading from the P50 abnormality, a proposed endophenotype, to qualitative clinical manifestations of the illness. However, P50 abnormalities were shown prior to overt illness in a study of persons at high genetic risk and those manifesting prodromal symptoms. Thus, the P50 endophenotype segregates with, and is demonstrable before, the earliest manifestations of clinical disorder, supporting its potentially important role in pathophysiology and as a treatment target. Nonetheless, in spite of all the limitations, the promising data on impaired P50 suppression and diminished expression of the α7 nAChR in schizophrenia has led to the identification of the α7 nAChRas a pharmacotherapeutic target in this disorder.13 Unfortunately, there are both a paucity of safe selective α7 nAChR agonists that can be and nAChRs in general rapidly desensitize upon exposure to agonist. Thus, sustained exposure to an agonist may result in “functional antagonism,” further exacerbating the problems associated with the already diminished expression of this receptor in schizophrenia.13,18,19

Example of “Endophenotypic Data” Informing a Treatment Strategy Clearly, the high heritability of the P50 gating abnormality among patients with schizophrenia and their closely-related biological relatives, including those unaffected with schizophrenia, the relationship of decreased hippocampal density of α7 nAChRs and impaired P50 suppression, diminished expression of the α7 nAChR in post-mortem brains from patients with schizophrenia, and the linkages between the deficit of P50 suppression, schizophrenia and the q13–14

S.I. Deutsch et al.

region of chromosome 15 implicate defective or deficient signal-transduction by this selective subtype of nAChR.13 Thus, a strategy was sought that would improve the efficiency of coupling between the binding of a relatively selective α7 nAChR agonist and channel opening; this would, perhaps, improve transduction of the nicotinic cholinergic signal in spite of the diminished density of a α7 nAChRs in schizophrenia. Also, in order to permit sustained stimulation of the α7 nAChR that is necessary in a chronic disorder like schizophrenia, a therapeutic strategy was necessary that would oppose transition of the receptor to a refractory, as opposed to a responsive, state upon continued exposure to agonist; unfortunately, rapid desensitization of the receptor occurs as a general characteristic of nAChRs. Choline, the hydrolytic split product and precursor of acetylcholine, has been shown to mimic acetylcholine at the α7 nAChR. In fact, local generation of choline in areas surrounding α7 nAChRs may be an important regulator of α7 nAChR-mediated neurotransmission.28 Importantly, choline is a naturally-occurring metabolite and would, thus, be expected to avoid many potential toxic or medical adverse side effects that may be associated with synthetic analogues. However, earlier studies of choline administration to patients with schizophrenia, usually administered in the context of treating neuroleptic-associated tardive dyskinesia, were without significant beneficial clinical effect.13 This lack of effect of choline may have been due to the rapid desensitization kinetics of the α7 nAChR. The experimental strategy that is under investigation in our laboratory is the combination of CDP-choline, a dietary source of exogenous choline with good bloodbrain barrier penetrability, and galantamine, a positive allosteric modulator of nAChRs in addition to its action as an inhibitor of acetylcholinesterase. Using this combination strategy, it is hoped that the agonist effect of choline would be potentiated by galantamine’s ability to improve the efficiency of coupling between choline’s binding to the receptor and channel opening, while also preserving the receptor in a sensitive state.13,18,19 In a preclinical mouse model of schizophrenia, galantamine was shown to modulate the effect of CDP-choline on mouse popping behavior (i.e., irregular episodes of intense jumping behavior) elicited by MK-801 (dizocilpine), a noncompetitive NMDA receptor antagonist that binds to the same hydrophobic channel domain as phencyclidine (PCP).18

7 Schizophrenia Endophenotypes as Treatment Targets

The combination of CDP-choline, titrated up to 2 g per day in divided doses, and galantamine, titrated up to 24 mg per day in divided doses, was added to the stable antipsychotic medication regimens of six patients with schizophrenia in an open-lable, 12-week pilot investigation.19 All of the patients had residual symptoms and were attending an intensive, all-day structured outpatient program that was operational Monday through Friday. At the end of 12-weeks, all of the patients tolerated the combination of CDPcholine and galantamine and completed the trial. Transient GI disturbance, including diarrhea, that resolved in a few days was the most commonlyobserved side effect. Three of the patients responded with significant reductions of clinical global severity scores and both total and subscale scores on the Positive and negative Symptom Scale (PANSS). Importantly, three of the patients requested to continue on the combination at the conclusion of the trial. Symptoms across several domains of psychopathology improved in at least some of the patients, including hallucinations, social withdrawal, flattened affect and poor personal hygiene.19 Based on the promising findings of the pilot study, a placebo-controlled, double-blind investigation of the adjuvant therapeutic efficacy of the combination has been implemented. In addition to the clinical measures of treatment efficacy and safety, the ability of the combination of CDPcholine and galantamine to improve the P50 deficit of sensory gating will be measured.

Conclusions Endophenotypes are heritable abnormalities of structure and function, including neurocognitive and neurophysiological functions that may be more closely “linked” to genetic variations, which may be normallyoccurring polymorphisms, than the categorical presence of illness. The endophenotypes would be quantitative laboratory-based measures that are normally-distributed within the population, allowing for the application of “powerful” statistical methodologies to confirm their associations/linkages to genetic variations. The presence of endophenotypes in unaffected biological relatives support the view that they may not be “sufficient” for manifestation of overt illness, but the functions that they reflect may, nonetheless, be

121

involved in the pathophysiology of the disorder. Endophenotypes may be a method for subtyping the illness that has a biological basis, as opposed to a descriptive one, as well as resolving members of the “group” of schizophrenias from each other. Very importantly, a genetically determined disruption of a function may be an ideal (and manageable) target for a strategically-designed intervention. The efficacy of the endophenotypically-targeted intervention could be measured in a quantitative and objective way, avoiding many of the pitfalls associated with subjective rating instruments that require staff training and practice to achieve reliability and avoid “definitional drift” between raters. To this end, we have designed a novel pharmacological strategy to target the presumed basis of the P50 sensory gating deficit in schizophrenia (i.e., deficient expression of the α7 nAChR in the brains of patients with schizophrenia). The intervention is an adjuvant one and involves adding the combination of CDP-choline, a dietary source of choline, a selective α7 nAChR agonist, and galantamine, a positive allosteric modulator of nAChRs in general. It is hoped that galantamine will improve the efficiency of coupling between the binding of choline and channel opening (addressing the deficient or impaired transduction of the cholinergic signal by the receptor), while preserving the receptor in a sensitive, as opposed to refractory, state that would avoid the further exacerbation of the α7 nAChR deficiency that exists in schizophrenia associated with prolonged exposure to agonist. The strategy is promising and serves as a paradigm for developing targeted endophenotypic interventions for the treatment of schizophrenia.

Future Directions The characterization of endophenotypes will refine the nosology of psychiatric disorders in general, leading to the “lumping” of patients based on their genetically determined biological similarities, as opposed to the descriptive similarities that are currently employed. Hopefully, genetic investigations will not only improve diagnostic accuracy and estimates of “risk,” but lead to definitive interventions that correct the pathophysiological abnormalities mediating functional disability. Perhaps, these investigations will lead to interventions that actually prevent emergence of overt illness.

122 Acknowledgment The authors acknowledge the generous support provided by the VISN 5 Mental Illness Research, Education and Clinical Center (MIRECC) of the Department of Veterans Affairs, and the guidance and continued friendship of its Director, Alan S. Bellack, Ph.D.

References 1. Braff DL, Freedman R, Schork NJ, Gottesman II. Deconstructing schizophrenia: an overview of the use of endophenotypes in order to understand a complex disorder. Schizophr Bull 2007; 33(1): 21–32. 2. Gur RE, Calkins ME, Gur RC, Horan WP, Nuechterlein KH, Seidman LJ, Stone WS. The consortium on the genetics of schizophrenia: neurocognitive endophenotypes. Schizophr Bull 2007; 33(1): 49–68. 3. Turetsky BI, Calkins ME, Light GA, Olincy A, Radant AD, Swerdlow NR. Neurophysiological endophenotypes of schizophrenia: the viability of selected candidate measures. Schizophr Bull 2007; 33(1): 69–94. 4. Cannon TD, Keller MC. Endophenotypes in the genetic analyses of mental disorders. Annu Rev Clin Psychol 2006; 2: 267–290. 5. Thaker GK. Schizophrenia endophenotypes as treatment targets. Expert Opin Ther Targets 2007; 11(9): 1189–1206. 6. Schork NJ, Greenwood TA, Braff DL. Statistical genetics concepts and approaches in schizophrenia and erlated neuropsychiatric research. Schizophr Bull 2007; 33(1): 95–104. 7. Deutsch SI, Rosse RB, Mastropaolo J, Long KD, Gaskins BL. Epigenetic therapeutic strategies of neuropsychiatric disorders: ready for prime time? Clin Neuropharmacol 2008;31(2):104–119 . 8. Berrettini WH. Genetics of psychiatric disease. Annu Rev Med 2000; 51: 465–479. 9. Braff DL. Introduction: the use of endophenotypes to deconstruct and understand the genetic architecture, neurobiology, and guide future treatments of the group of schizophrenias. Schizophr Bull 2007; 33(1): 19–20. 10. Benes FM. Searching for unique endophenotypes for schizophrenia and bipolar disorder within neural circuits and their molecular regulatory mechanisms. Schizophr Bull 2007; 33(4): 932–936. 11. Joyce EM, Roiser JP. Cognitive heterogeneity in schizophrenia. Curr Opin Psychiatry 2007; 20: 268–272. 12. Adler LE, Olincy A, Waldo M, Harris JG, Griffith J, Stevens K, Flach K, Nagamoto H, Bickford P, Leonard S, Freedman R. Schizophrenia, sensory gating, and nicotinic receptors. Schizophr Bull 1998; 24: 189–202. 13. Deutsch SI, Rosse RB, Schwartz BL, Weizman A, Chilton M, Arnold DS, Mastropaolo J. Therapeutic implications of a selective α7 nicotinic receptor abnormality in schizophrenia. Isr J Psychiatry 2005; 42(1): 33–44. 14. Crespi B, Summers K, Dorus S. Adaptive evolution of genes underlying schizophrenia. Proc Biol Sci 2007 Nov 22; 274(1627): 2801–2810.

S.I. Deutsch et al. 15. Pearlson GD, Folley BS. Schizophrenia, psychiatric genetics, and Darwinian psychiatry: an evolutionary framework. Schizophr Bull 2007 Nov 21 [epub ahead of print]. 16. Bearden CE, van Erp TGM, Thompson PM, Toga AW, Cannon TD. Cortical mapping of genotype-phenotype relationships in schizophrenia. Hum Brain Mapp 2007; 28: 519–532. 17. Callicott JH, Egan MF, Mattay VS, Bertolino A, Bone AD, Verchinski B, Weinberger DR. Abnormal fMRI response of the dorsolateral prefrontal cortex in cognitively intact siblings of patients with schizophrenia. Am J Psychiatry 2003; 160: 709–719. 18. Deutsch SI, Rosse RB, Schwartz BL, Schooler NR, Gaskins BL, Long KD, Mastropaolo J. Effects of CDP-choline and the combination of CDP-choline and galantamine differ in an animal model of schizophrenia: Development of a selective α7 nicotinic acetylcholine receptor agonist strategy. Eur Neuropsychopharmacol 2008;18:147–151. 19. Deutsch SI, Schwartz BL, Schooler NR, Rosse RB, Mastropaolo J, Gaskins B. First administration of cytidine diphosphocholine and galantamine in schizophrenia: a sustained alpha7 nicotinic agonist strategy. Clin Neuropharmacol 2008;31(1):34–39. 20. Swerdlow NR, Geyer MA. Using an animal model of deficient sensorimotor gating to study the pathophysiology and new treatments of schizophrenia. Schizophr Bull 1998; 24: 285–301. 21. Cadenhead KS, Geyer MA, Braff DL. Impaired startle prepulse inhibition and habituation in schizotypal patients. Am J Psychiatry 1993; 150: 1862–1867. 22. Leonard S, Adams C, Breese CR, Adler LE, Bickford P, Byerley W, Coon H, Griffith JM, Miller C, Myles-Worsley M, et al. Nicotine receptor function in schizophrenia. Schizophr Bull 1996; 22: 421–445. 23. Stevens KE, Freedman R, Collins AC, Hall M, Leonard S, Marks MJ, Rose GM. Genetic correlation of hippocampal auditory evoked response and α-bungarotoxin binding in inbred mouse strains. Neuropsychopharmacology 1996; 15: 152–162. 24. Adler LE, Hoffer LD, Griffith J, Waldo MC, Freedman R. Normalization by nicotine of deficient auditory sensory gating in the relatives of schizophrenics. Biol Psychiatry 1992; 32: 607–616. 25. Adler LE, Hoffer LD, Wiser A, Freedman R. Normalization of auditory physiology by cigarette smoking in schizophrenic patients. Am J Psychiatry 1993; 150: 1856–1861. 26. Leonard S, Freedman R. Genetics of chromosome 15q13q14 in schizophrenia. Biol Psychiatry 2006; 60: 115–122. 27. Potter D, Summerfelt A, Gold J, Buchanan RW. Review of clinical correlates of P50 sensory gating abnormalities in patients with schizophrenia. Schizophr Bull 2006; 32(4): 692–700. 28. Albuquerque EX, Alkondon M, Pereira EFR, Castro NG, Schrattenholz A, Barbosa CTF, Bonfante-Cabarcas R, Aracava Y, Eisenberg HM, Maelicke A. Properties of neuronal nicotinic acetylcholine receptors: pharmacological characterization and modulation of synaptic function. J Pharmacol Exp Ther 1997; 280: 1117–1136.

Part II

Neuropsychological, Neurocognitive and Neurophysiological Domains

Chapter 8

Neuropsychological Endophenotypes in Schizophrenia and Bipolar I Disorder: Yields from the Finnish Family and Twin Studies Annamari Tuulio-Henriksson, Jonna Perälä, Irving I. Gottesman, and Jaana Suvisaari

Abstract This chapter summarizes findings related to the neuropsychology of schizophrenia and bipolar disorders and the use of neuropsychological functions as endophenotypes in genetic analyses in Finnish family and twin studies. The endophenotypes are the intermediate factors between the phenotype and genotype, and they are assumed to be contributed to by fewer genes than the clinical phenotypes of the complex psychiatric disorders. There is a long tradition of research both on schizophrenia and bipolar disorder in Finland, and large population based family and twin samples for clinical, neuropsychological, and genetic studies have been collected. In these samples, we have used the endophenotype strategy, first evaluating whether the neuropsychological test variables fulfill the criteria for valid endophenotypes and then using them in genomewide linkage and candidate gene association analyses. In the genomewide linkage analysis, several variables related to learning and memory were shown to enhance the linkage as A. T.-Henriksson Department of Mental Health and Substance Abuse Services, National Institute for Health and Welfare (former National Public Health Institute), Helsinki, Finland Department of Psychology, University of Helsinki, Helsinki, Finland J. Perälä Department of Mental Health and Substance Abuse Services, National Institute for Health and Welfare (former National Public Health Institute), Helsinki, Finland I. I. Gottesman Departments of Psychiatry and Psychology, University of Minnesota, Minneapolis, MN, USA J. Suvisaari Department of Mental Health and Substance Abuse Services, National Institute for Health and Welfare (former National Public Health Institute), Helsinki, Finland Department of Public Health, University of Tampere, Tampere, Finland

compared with using the categorical clinical phenotype. Moreover, working memory, and several variables from the learning and memory process were found to be heritable and to show significant associations with DISC1, reelin, and AKT1 in schizophrenia, and with DISC1 in bipolar disorder, in which DAOA also associated with visuospatial ability. Although the associations are modest, our results support the validity of the endophenotype approach in investigating the genetic etiology of severe mental illnesses. Future studies should further evaluate endophenotypes derived from neuroscience to obtain optimal variables that help in tracking multiple genes of small effect predisposing to developing the complex psychiatric disorders. Keywords Neuropsychology • endophenotype • cognition • schizophrenia • bipolar disorder • family • twin • association analysis Abbreviations AKT1 v-akt murine thymoma viral oncogene homolog 1; BF Bipolar family study; BT Bipolar twin study; cAMP Cyclic adenosine monophosphate; COMT Catechol-O-methyl transferase; COWA Controlled Oral Word Association; CVLT California Verbal Learning Test; DAOA d-amino acid oxidase activator; DISC1 Disrupted in Schizophrenia Gene 1; DSM-IV Diagnostic and Statistical Manual of Mental Disorders, version four; DZ Dizygotic; EEG Electroencephalogram; GABA Gamma-aminobutyric acid; IQ Intelligence quotient; MEG Magnetoencephalogram; MRI Magnetic resonance imaging; MZ Monozygotic; NMDA N-methyl d-aspartate; SCID-I Structured Clinical Interview for DSM-IV Axis I Disorders; SCID-II Structured Clinical Interview for DSM-IV Axis II Personality Disorders; SF Schizophrenia family study;

M.S. Ritsner (ed.), The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes, © Springer Science + Business Media B.V. 2009

125

126

ST Schizophrenia twin study; WAIS-R Wechsler Adult Intelligence Scale-Revised; WMS-R Wechsler Memory Scale-Revised

Introduction This chapter summarizes findings related to the neuropsychology of schizophrenia and bipolar I disorder, and to the use of neuropsychological functions as endophenotypes in genetic analyses in Finnish family and twin studies. Using nationwide population and health care registers, we have formed population-based cohorts of patients with schizophrenia and bipolar disorder and their relatives, and established family and twin studies of schizophrenia and bipolar disorder. Key aspects in all the studies have been investigating neuropsychological deficits related to schizophrenia and bipolar disorder and to genetic liability to these disorders, examining whether neuropsychological deficits could be considered as endophenotypes of schizophrenia and bipolar disorder, and using neuropsychological test results in genetic analyses.

Cognitive Deficits Among Patients with Schizophrenia An important component of schizophrenia involves compromised performance in most of the cognitive domains: attention, working memory, learning and memory, and executive function. Some patients do not show impairments compared with controls1 but it may be that their cognitive functioning has deteriorated from the premorbid level.2 However, no specific neuropsychological profile has been detected for schizophrenia. In their meta-analysis, Heinrichs and Zakzanis3 reported that individuals with schizophrenia score between one-half to one-and-a-half standard deviations below the control mean in a wide variety of neuropsychological domains including attention, memory, intelligence, motor speed, spatial ability, executive functioning and verbal fluency, thus showing a generalized deficit. The largest effect sizes have been detected in verbal memory.3,4 Recently, Dickinson et al.5 have emphasized the importance of slowed information processing against the generalized

A. Tuulio-Henriksson et al.

cognitive impairment in schizophrenia. In a recent study applying structural equation modeling to neuropsychological variables, approximately 63.6% of the schizophrenia-related variance in cognitive performance was mediated through the general factor, but verbal memory (13.8%) and processing speed (9.1%) also had direct effects.6 Many subjects with prodromal symptoms show cognitive deficits, particularly in attention,7 and although there is not one specific deficit that predicts the onset of the disorder, some dysfunction in cognition often precedes it.8 Usually schizophrenia symptoms emerge together with a marked reduction in several critical cognitive domains. The impairments do not directly associate with age, severity of symptoms, length of the illness, or medication, and usually they do not progress but remain stable.9,10 Recently, Szöke et al.11 conducted a meta-analysis on longitudinal studies of cognition in schizophrenia. Including studies with more than 1 month between the baseline test and retest sessions, they found improvement in several domains of cognition. However, they concluded that the improvement could be mostly accounted for by the effect of practice. Cognitive functions in schizophrenia seem to show a similar impairment pattern to those that accompany ageing. Executive functions, however, may show a more accelerated decline.12 Impairments in cognitive functioning may prevent patients from attaining an optimal adaptation in their everyday life.13 Generally, impaired cognition has been linked to community outcome, social problem solving and social skill acquisition more closely than the clinical symptoms. Dysfunction in executive processes, including planning, problem solving and enterprising, play a relevant role in restricting patients’ ability to retain, acquire, or relearn skills that are needed for psychosocial functioning.14 Furthermore, the psychosocial and other functional consequences of the disorder seem to be most clearly related to impairments in verbal learning and memory functions.15 In several studies, cognitive deficits in relatives of schizophrenia patients have been found to parallel those observed in the patients, although less severe.16–18 Twin studies have shown that the deficits correlate more strongly in discordant monozygotic than in discordant dizygotic twins.19,20 Dysfunction in cognition has also been observed to appear in subjects at high risk for schizophrenia, and deficits in attention during childhood may predict schizophrenia in subjects with

8

Neuropsychological Endophenotypes in Schizophrenia and Bipolar I Disorder

at least one affected parent.8,21,22 Many subjects with prodromal symptoms show cognitive deficits, observed in attention, verbal learning and memory, working memory, and processing speed.23 –25

Cognitive Functioning in Bipolar Disorder Research on the nature and the neuropsychological basis of cognitive deficits in bipolar disorder, and their relationship with symptoms has grown rapidly, and evidence from many recent studies speaks for objective impairment in cognition in these disorders.26 Both patients in depressed and manic phases of the illness have shown deficits in attention, memory and executive functions.27 It is not well known whether cognitive deficits precede the onset of bipolar disorder. In a feasibility study on 25 adolescents,28 the children of bipolar parents were found to have abnormal coping styles and dysregulation of sleep compared with children of healthy parents. In a study on high risk offspring of parents with schizophrenia or bipolar disorder, Maziade et al.24 found that both offspring groups present dysfunction in several cognitive domains; with 45 offspring in each group, most prominent impairments were found in verbal learning and memory functions, and in executive functioning. In the executive domain of problem solving, offspring with bipolar parents scored higher than those from the schizophrenia group. Further studies in bipolar high-risk populations on cognitive functioning are called for. It has been argued that recurrent episodes of depression and mania may cause a cumulative pathological effect, reflected in a more pronounced cognitive dysfunction in chronic and elderly patients with bipolar disorder.29 It is unknown whether this effect is medication-based, or reflects the neurotoxicity of the illness episodes, or whether it is related to other factors, e.g. comorbid alcoholism.30 In Zubieta et al.,31 the patients had recently suffered a manic episode with psychotic symptoms (delusions or hallucinations), which may indicate the presence of a more severe form of the disorder. The patients showed deficits in verbal learning and executive functioning, which both correlated with number of episodes and psychosocial functioning. The extent of cognitive deficits may be related to either the severity of the symptoms during acute phases of the illness, or to a cumulative effect of repeated episodes

127

of mania and depression.32 In line with earlier studies, Martinez-Aran et al.33 showed that a history of psychotic symptoms was associated with greater cognitive impairment in euthymic bipolar patients particularly in verbal memory and executive functioning. Recently, many studies have shown that patients with bipolar disorder do not always make a full interepisode recovery in cognitive functioning, but the dysfunction may persist beyond episodes of illness.34 In the meta-analyses of cognitive functioning in euthymic bipolar patients and their first-degree relatives, Arts et al.26 noted large effect sizes (d > 0.8) for executive functions (working memory, executive control, fluency) and verbal memory. Medium effect sizes (0.5 < d < 0.8) were reported for aspects of executive function (concept shifting, executive control), mental speed, visual memory, and sustained attention. Although possible medication effects should not be overlooked, the findings provide evidence for the existence of neuropsychological impairment in patients with euthymic bipolar disorder. There are fewer studies and less evidence of impairments among unaffected relatives of bipolar patients than among schizophrenia families, and the findings have been controversial. The discrepancies may result from methodological differences and relatively small sample sizes in these studies. In recent meta-analyses, however, first-degree relatives have been shown to differ from controls particularly in executive function and verbal memory. These functions have been suggested as valid candidate bipolar endophenotypes.26,35,36

Neuropsychological Endophenotypes An endophenotype is proposed as an intermediate factor between the phenotype and genotype (Fig. 8.1). The main justification for using endophenotypes in studies aimed at gene finding is their assumed genetic simplicity in comparison to complex disease phenotypes. The endophenotype refers to an internal construct that can be quantified by measurements, but is not seen with an “unaided eye”.37,38 The endophenotypes may be any neurobiological measures related to the underlying molecular genetics of the illness, including neurophysiological,39 neuroanatomical,40 neurological,41 or neuropsychological markers.

128

A. Tuulio-Henriksson et al.

Fig. 8.1 Gene regions, genes, and candidate endophenotypes are implicated in a biological systems approach to psychosis research. Endophenotypes are theoretically characterized as having simpler structural antecedents than the psychiatric phenotypes themselves. This schema (genes to endophenotypes

to symptoms to disease), allowing for epigenetic, environmental, and stochastic influences, can be applied to other complex diseases as well. None of the components in the figure can be definitive; many more elements exist or await discovery (© I.I. Gottesman and T.D. Gould, 2008, used by permission)

The endophenotypes should fulfill several criteria before they can be considered valuable in furthering a genetically-informed research strategy: (1) the endophenotype is associated with illness in the population, (2) the endophenotype is significantly heritable, (3) the

endophenotype is present in individuals with and without an active phase of the illness, (4) in families with the illness, many of the unaffected relatives have the same endophenotypic trait, and (5) the endophenotype that is present in the affecteds, is more prevalent in the

8

Neuropsychological Endophenotypes in Schizophrenia and Bipolar I Disorder

unaffecteds in the family than in the general population.37 Further, some endophenotypes may require a challenge to be elicited (e.g. an abnormal glucose tolerance test in unaffected relatives of diabetics). The endophenotype strategy is supposed to provide a large increase in statistical power for showing differences as the undiagnosed but tested relatives can be included in the analyses. The search for endophenotypes for severe mental illnesses “to inform genetic studies of psychiatric disorders is never as productive and exciting as it is in neuropsychological studies”.42 This is because the cognitive deficits observed in these disorders are trait-like core features closely related to clinical and functional outcome. Furthermore, the neurobiology of some of the cognitive functions, such as working memory and verbal memory, are increasingly well understood, with some already shown to have genetic connections.

Neuropsychological Tests Used in the Studies Neuropsychological test batteries used in the Finnish schizophrenia and bipolar family and twin studies are described in Table 8.1. There is good overlap in neuropsychological test methods among studies. The methods are valid and used internationally, assessing the core cognitive functions, attention, short term and long term memory, working memory, executive functioning, processing speed, and basic verbal and psychomotor abilities.

129

Table 8.1 Neuropsychological tests used in the Finnish schizophrenia and bipolar family and twin studies SF

BF

ST

BT

x x x x

x x x x

x x x x

x x x x

x x x x

x x x x

x x x x x

x x x x x

Working memory WMS-R44 Digit span backward Visual span backward

x x

x x

x x

x x

Verbal memory functions WMS-R44 Logical memory stories California verbal learning test

x x

x x

x x

x x

x

x

x x

x x

x x

x x

x

x

x x x x x

x x x x x

x

x

Intelligence WAIS-R43 Vocabulary Similarities Block design Digit symbol Attention WMS-R44 Digit span forward Visual span forward Trail making test A45 Continuous performance test46 Choice reaction time47

Visual memory functions WMS-R44 Visual reproduction Face recognition48 Executive functioning Stroop49 COWA50 Wisconsin card sorting test51 Trail making test B45 Dual task52 Motor speed Finger tapping53

SF = Schizophrenia family study, BF = bipolar family study, ST = schizophrenia twin study, BT = bipolar twin study

The Finnish Schizophrenia Family Study The Genetic Epidemiology and Molecular Genetics of Severe Mental Disorders-study was started by professors Leena Peltonen-Palotie and Jouko Lönnqvist as early as 1988. For practical reasons, the study was divided into two, one focusing on families with schizophrenia, the other on bipolar families. When the schizophrenia family study was started, 33,731 individuals with a diagnosis of schizophrenia were first identified from a cohort of all people born in Finland from 1940 to 1976 using data derived from three nationwide computerized health care registers for 1969–

1998. Persons with hospital treatment and a diagnosis of schizophrenia were identified from the National Hospital Discharge Register, persons with free outpatient antipsychotic medication for schizophrenia from the Medication Reimbursement Register of the Finnish Social Insurance Institution, and persons with disability pension because of schizophrenia from the Pension Register of the Social Insurance Institution (nowadays kept by the Finnish Center for Pensions). The firstdegree relatives of these individuals were identified from the Population Register Center, and information

130

on relatives was linked back to the health care registers to obtain information on their treatments and diagnoses. Two samples were collected for this populationbased genetic study of schizophrenia. The first sample consisted of families with at least two siblings with schizophrenia from the entire geographical area of Finland. The second sample comprised patients and their parents and siblings from families with at least one member with schizophrenia from an isolated region resulting from a genetic ‘bottleneck’ in the northern part of the country with an exceptionally high lifetime risk (3.2%) of schizophrenia.54 About 3,500 subjects from 1,000 families have participated in the study and have given blood samples for DNA analysis. Approximately one third of them have also been interviewed using the Structured Clinical Interview for DSM-IV (SCID), and were administered a neuropsychological test battery. The family data and careful diagnostic and neuropsychological assessment allowed us to put the criteria of the topical neuropsychological endophenotypes37 to test. We first investigated the heritability of cognitive functions.55 Significant additive heritability estimates were detected in several cognitive functions that reflect the encoding stage during verbal learning. The significantly heritable functions were the following: Using semantic clusters as a learning strategy while recalling words from a word list, providing words that were not in the list (intrusive recall errors), recognition memory, verbal working memory, and verbal ability. In addition, the heritability estimate of visual working memory was marginally significant. Impairment in the functions during the encoding phase affects maintenance and retrieval of information, which in turn may cause poor recognition memory producing recall errors. Slowed encoding affects working memory and impairs the activation of mental representations essential for efficient working memory. The study showed that the cognitive traits related to working memory and such verbal memory functions that associate with the primary encoding phase of learning, are at least modestly heritable in schizophrenia families and could be valuable endophenotypic traits in genetic analyses. Traits measuring working memory functions were considered of particular interest as they were contributed by a very restricted number of loci in the estimation of their theoretical number. The result suggested that genetic

A. Tuulio-Henriksson et al.

linkage analyses could benefit particularly from using the quantitative data derived from working memory test scores. Tuulio-Henriksson et al.56 further examined whether the number of affected family members, i.e. family loading, would exacerbate the cognitive impairment among healthy siblings as expected for a heritable quantitative trait. In that study, 94 unaffected relatives of schizophrenia patients were included in the analysis. The subjects were divided into two groups according to the number of first-degree relatives with schizophrenia. The singleton group included healthy relatives with only one family member with schizophrenia, and the multiplex group included those with at least one relative with schizophrenia plus at least another with schizophrenia or other non-affective psychosis. Only visual working memory was found to associate with increased load for schizophrenia. Unaffected relatives from multiply affected families scored worse in this function than those with only one affected family member. We also investigated the effect of age of onset on cognitive functions among 237 patients with schizophrenia.57 Worse performance associated with earlier age of onset in verbal memory, in using semantic clusters as a learning strategy, in producing intrusive recall errors, and in recognition memory. Verbal memory impairment has consistently been found to associate with earlier onset,58 and impairment in this function should be taken into account in the neuropsychological evaluation and efforts at remediation in patients with early-onset disorder. To explore possibilities of reducing heterogeneity among the families with schizophrenia, we conducted a novel family based cluster analysis including the neuropsychological variables in the clustering process.59 The process aimed to find families with similar neuropsychological test performance. In all 54 families included, there were at least two siblings with schizophrenia, schizoaffective psychosis or schizophreniform disorder, and their unaffected family members. In addition, some family members with affective psychotic disorders were included. Altogether 17 neuropsychological test variables, plus age and sex were incorporated in a visually aided clustering algorithm. This process identified three clusters of families. The first cluster included 17 families with well-performing subjects, cluster two comprised an impaired cluster with 12 families, and the third was intermediate with a

8

Neuropsychological Endophenotypes in Schizophrenia and Bipolar I Disorder

less clear performance profile in 25 families. The clusters did not differ from each other in age, sex distribution, and, regarding the affected subjects, in the age of onset. The well-performing cluster was better educated than the other two, but controlling for education did not change the main results in comparing the cognitive test differences between the clusters. The diagnostic distribution among the patients was interesting, as all the patients in the impaired cluster suffered from schizophrenia, and no family members with affective psychotic disorders appeared in this cluster. It can thus be considered to represent a subsample of core schizophrenia with the most impaired cognitive functioning. This cluster included the same proportion of unaffected subjects as the two other clusters, and by the logic of the clustering algorithm, the unaffected subjects in this cluster also performed poorly. In a more recent study, we examined neuropsychological functioning of patients, first-degree relatives, and population based controls, among families with familial schizophrenia.60 The patients had a generalized cognitive dysfunction, while the relatives differed mainly in information processing speed and executive functioning.

Genetic Analyses In the first phase of the study, genomewide scans were conducted using psychiatric diagnoses only.61–65 We then included several quantitative neuropsychological measures in a genomewide linkage analysis.66 In this analysis, evidence was found for a locus for verbal learning and memory on 4q21 and suggestive evidence for visual working memory on 2q36. In addition, some evidence emerged for a locus for recognition memory on 10p13, visual attention on 15q22 and executive function on 9p22 in the complete sample, as well as for delayed memory on 8q12, semantic clustering and intrusions on 1q42 and visual attention on 3p25 in the genealogically distinctive sample subsets. Our results revealed initial information on the effect of the loci associated with schizophrenia in multiple studies, and evidenced the value of neuropsychological endophenotypes in the search for susceptibility loci for schizophrenia. The importance of the results was further emphasized by previous evidence of linkage, using the same study population, to the same region with the clinical diagnosis as the phenotype.65

131

Based on findings from the linkage studies, genetic areas of particular interest have been studied further. Most notably, Disrupted in Schizophrenia Gene 1 (DISC1) turned out to be associated with both schizophrenia67 and bipolar I disorder68 in Finland. DISC1 was initially identified in a large Scottish pedigree, where a balanced (1;11)(q42.1;q14.3) translocation co-segregates with schizophrenia and major affective disorders.69 After determining that an allelic haplotype that spans from intron 1 to exon 2 of the DISC1 gene is associated with increased risk for schizophrenia spectrum disorders in Finnish families,67 we showed that it was also associated with visual working memory function deficits in males.70 We re-analysed our data conditioning on the presence of the DISC1 risk haplotype, and found an association between schizophrenia and the NDE1 gene in females in families carrying the DISC1 risk haplotype. Contrary to DISC1, however, NDE1 was not associated with visual working memory deficits.71 Another area showing evidence of linkage to schizophrenia is 7q22.64 Four candidate genes in the 7q22 area have been studied further: metabotropic glutamate receptor 3, semaphorin 3 A, nerve growth factor inducible, and reelin. None of these families was significantly associated with a schizophrenia diagnosis. Instead, we found evidence of association of the reelin gene with verbal and visual working memory, learning, executive functioning, and visual attention.72

The Finnish Bipolar Family Study The Genetic Epidemiology and Molecular Genetics of Severe Mental Disorders-study targeted also in detailed investigation among families with bipolar disorder. From a large sample of families of whom blood samples had previously been drawn, we examined 153 individuals from 57 families with the SCID interview and administered a neuropsychological test battery. Analogously to the schizophrenia studies, we had the opportunity to evaluate the feasibility of the endophenotype strategy in bipolar disorder. In a series of studies, we first compared the neuropsychological functioning of patients with bipolar I disorder, their first-degree relatives, and a population based control group, and found that both patients and relatives showed impairments in executive functioning and in processing

132

speed compared with the controls.73 In the second study, we pioneered in computing the heritability estimates of the neuropsychological test variables among 52 bipolar I families. Significant additive heritabilities were found in verbal ability, executive functioning, and psychomotor processing speed. Genetic contribution was low to verbal learning functions. High heritability in executive functioning and psychomotor processing speed suggest that these may be valid endophenotypic traits for genetic studies of bipolar disorder.74

Genetic Analyses As previous studies had linked DISC1 gene with both schizophrenia and bipolar disorder,75,76 our next interest was to investigate whether the DISC1 gene associated with bipolar spectrum disorders in the Finnish bipolar family sample, and whether it was also associated with neuropsychological impairments within the families. The same risk haplotype at the 5′end that had been identified from Finnish schizophrenia families turned out to be a risk factor for the presence of psychotic disorder in Finnish bipolar families.68 This risk haplotype, or alleles within it, was also associated with perseverative errors in the CVLT and poorer long-term memory. Another allele from the same region was associated with auditory attention. These associations were mostly observed in males, as in the schizophrenia family study. However, several haplotypes at the 3′end of DISC1 gene were associated to bipolar spectrum disorders, particularly one protective haplotype. These haplotypes, or alleles within them, also showed association with neuropsychological test performance, particularly verbal fluency, visuospatial ability, and psychomotor processing speed, some associating with better neuropsychological performance. Thus, haplotypes within the 5′end of the DISC1 gene are related to schizophrenia spectrum disorders, psychosis, and poor performance in visual and verbal memory tests particularly among males, while haplotypes at the 3′end are associated with bipolar spectrum disorders.68,70 The endophenotype approach was successful also when other candidate genes for bipolar and psychotic disorders and their endophenotypes, including d-amino acid oxidase activator (DAOA), catechol-O-methyl

A. Tuulio-Henriksson et al.

transferase (COMT), dystrobrevin binding protein 1 (dysbindin), neuregulin 1, and v-akt murine thymoma viral oncogene homolog 1 (AKT1), were investigated.77 Only COMT showed evidence for association to bipolar spectrum disorders, but both COMT and DAOA were associated with visuospatial ability, as measured by the Block Design test. DAOA showed evidence of association to general ability, psychomotor speed, auditory attention and visual memory, as well.

The Finnish Schizophrenia Twin Study The Finnish Schizophrenia Twin Study is a collaborative study between the National Public Health Institute and the University of California, Los Angeles, started by professors Jouko Lönnqvist and Tyrone Cannon. Families including twin pairs in which at least one of the pairs had schizophrenia were identified using the same register data as in the family study, and Finnish twin cohort data were used to check that the ascertainment of twin pairs had been complete. A representative sample of monozygotic (MZ) and same-sex dizygotic (DZ) twin pairs concordant and discordant for schizophrenia and demographically similar control pairs from Finnish twins born 1940 through 1957 were assessed first, and the study was later extended to cover twins born from 1958 to 1990. The twins have been evaluated using structured psychiatric interview and symptom assessment (SCID I–II, SANS, SAPS), neuropsychological testing, and magnetic resonance imaging (MRI), and blood samples have been collected for genetic analyses. Also positron emission tomography imaging, functional MRI, electroencephalogram (EEG) and magnetoencephalogram (MEG) have been conducted for some twin pairs. Altogether some 200 twin pairs were evaluated in the study. Cannon et al. first investigated the inheritance of neuropsychological dysfunction in twins discordant for schizophrenia.19 The aims were to identify neurocognitive deficits that increase with genetic relationship to a person with schizophrenia and to assess the extent to which neurocognitive deficits were caused by nongenetic factors. A canonical discriminant analysis was performed to identify neuropsychological functions that most accurately discriminated twins with

8

Neuropsychological Endophenotypes in Schizophrenia and Bipolar I Disorder

schizophrenia, unaffected monozygotic and dizygotic twins, and controls. The functions that contributed uniquely to the discrimination of genetic loading for schizophrenia were visual working memory, divided attention, recall intrusions, and choice reaction time to visual targets. When comparing discordant MZ twins, patients scored significantly lower than their unaffected co-twins on verbal memory and learning, semantic clustering, visual reproduction, and overall IQ, but not on visual working memory, divided attention, choice reaction time, recall intrusions, or psychomotor speed, suggesting that the former deficits are also influenced by disease-related nongenetic factors, while deficits in the latter in both patients and their MZ unaffected co-twins are accounted for by genetic factors. The task that was used to assess visual working memory was Visual span, a complex visuospatial memory task requiring encoding, maintenance, manipulation, and a complex motor response. Glahn et al.78 developed a novel spatial delayed-response task more specifically testing visuospatial working memory. They found that spatial working memory performance decreased with increasing genetic liability to schizophrenia, and all groups declined equally when the task became more demanding. The findings suggested that deficits in the encoding or storage aspects of short-term spatial memory processing might be an effective endophenotypic marker for schizophrenia. Johnson et al.79 investigated the relationship between neuropsychological performance and schizotypal symptoms in unaffected co-twins of persons with schizophrenia and control twins. She found that while schizotypal symptoms were not associated with neuropsychological performance in controls, schizotypal symptoms were related to deficits in most domains of neuropsychological functioning in co-twins of persons with schizophrenia. In the absence of schizotypy symptoms, only working memory performance differed between co-twins of persons with schizophrenia and controls. More recently, van Erp et al.80 investigated genetic and disease-specific effects on free recall, cued recall, and recognition in verbal learning and memory tasks. As expected, the largest differences were found in the most demanding task, free recall, in which patients performed the worst, and the performance improved linearly with decreasing genetic relationship to an affected individual (unaffected MZ co-twin < unaffected DZ co-twin < controls). The same differences

133

were evident but somewhat smaller in the cued recall task. In the recognition task, in contrast, patients performed worse than the other groups, but there was no significant difference in performance between unaffected MZ and DZ twins and controls. Intra-pair differences in verbal memory performance correlated with intrapair differences in left hippocampal volume. The study suggests that verbal memory tasks that require active retrieval may be useful endophenotypic measures, because they are compromised in unaffected relatives, with increasing performance deficit associated with increased genetic relationship to proband with schizophrenia. However, verbal memory deficits in patients with schizophrenia are also influenced by disease-specific influences, partly mediated by hippocampal pathology.

Genetic Analyses The use of neuropsychological functions as endophenotypes has been tested in genetic analyses. A region on the distal portion of chromosome 1 that has showed evidence for linkage in the family study61,63 was investigated first.81 The Visual Span test, particularly the more demanding backward span showed evidence for linkage at a marker lying fairly close to the DISC1 gene. When haplotypes within the DISC1 gene were later shown to be associated with schizophrenia in our family study,67 the association between neuropsychological test results and these haplotypes was also investigated in the twin study sample by Cannon et al.82 Two of the haplotypes were significantly associated with verbal learning and memory, and one was associated with visual working memory and choice reaction time. The same haplotypes were associated with small, focal reductions of gray matter density in the superior, middle and inferior frontal gyri, and portions of superior temporal gyrus and superior parietal cortex. Another gene that has been investigated in the schizophrenia and bipolar twin samples is AKT1 (v-akt murine thymoma viral oncogene homolog 1). While AKT1 was not associated with diagnosis of schizophrenia and bipolar I disorder, it was associated with verbal learning and memory, examined by CVLT, and with decreased gray matter density in medial and dorsolateral prefrontal cortex.83

134

A. Tuulio-Henriksson et al.

Finnish Bipolar Twin Study

Discussion

The Finnish Bipolar Twin Study was started shortly after the schizophrenia twin study by Dr. Tuula Kieseppä. The twins were born between 1940 and 1969. As in the schizophrenia twin study, the twins were interviewed with SCID-I and SCID-II interviews, and were administered the same neuropsychological test battery as in the schizophrenia twin study.84 Twins with bipolar I disorder and their unaffected co-twins were not found to differ from controls in general intelligence, as estimated by the Vocabulary subtest of the WAIS-R.84 Twins with bipolar I disorder performed worse than controls in tests measuring information processing speed, and in both verbal and visual memory tasks. Impairment in memory tasks was not explained by current or lifetime psychotic symptoms, duration of illness, or the number of episodes. After taking antipsychotic medication into account, the effect of diagnosis was no longer significant in CVLT total recall, but otherwise the results remained unchanged. The only significant difference between co-twins and controls emerged among female co-twins in long-term verbal memory.84 Kieseppä et al.84 and Pirkola et al.85 investigated memory deficits further. Kieseppä et al. investigated whether poor performance in visual and verbal memory tests was related to slowed information processing speed. After taking information processing speed into account, twins with bipolar disorder did not differ significantly from controls in verbal and working memory tasks, but they were still impaired in more demanding long-term memory tasks story recall and visual reproduction.84 A parallel result was obtained by Pirkola et al.85 who found that after adjusting for the effect of parental socioeconomic status, general ability, and educational level, twins with bipolar disorder no longer differed from controls in visual and verbal working memory tasks. In contrast, twins with schizophrenia still performed worse that controls in both visual and verbal working memory tasks, and their unaffected twin pairs performed worse in visual working memory tasks. The results suggest that working memory deficits, and visual working memory deficits in particular, may be more related to the genetic liability to schizophrenia than to bipolar disorder, but both disorders are associated with deficits in long-term memory.

In Finland, we have been in the frontline in taking advantage of the endophenotype strategy, both in evaluating and investigating the criteria for these intermediate phenotypes and in applying them in genetic linkage and association analyses. Although our studies may deserve criticism for including a rather narrow selection of cognitive variables in the analyses, we have based our selection on large scale background work to choose those variables that best fulfill the suggested features for valid endophenotypes. In this process, we have not only explored our own samples but also taken into account the findings in other studies conducted in family samples of schizophrenia and bipolar disorder around the world. It is well established that cognitive impairments are core features of schizophrenia, and that particularly working memory, and verbal learning and memory, show an impairment pattern also in the relatives of the patients, and that genetic, or familial, loading associates with these impairments. Studies on their heritability show that they also fulfill this critical criterion for valid endophenotypes. Moreover, studies on cognitive functioning among families with bipolar disorders have shown that cognitive impairments, particularly in executive functions, are a part of the clinical picture and may be an inherited vulnerability indicator of the illness.

Genetic Associations of Cognitive Functions Our most consistent genetic findings of neuropsychological functions as endophenotypes relate to the DISC1 gene. Haplotypes at the 5′ end of the gene were consistently associated with poor performance in visual working memory and verbal learning and memory tests particularly among males. The associations were originally found in the schizophrenia family study,70 and were replicated in schizophrenia twin study82 and bipolar family study.68 These findings are consistent with other reports concerning the effects of DISC1 variants on cognitive functions. SNPs within the DISC1 gene have been associated with impairment in attention (TMT-A), verbal working memory (Digits backward), and long-term memory (Logical Memory

8

Neuropsychological Endophenotypes in Schizophrenia and Bipolar I Disorder

II WMS) in patients with schizophrenia.86,87 DISC1 has also been associated with cognitive aging in a population-based longitudinal study.88 Findings associating DISC1 with particularly impaired memory functions are complemented by studies linking DISC1 haplotypes or SNPs to reduced gray matter density in the prefrontal cortex and smaller hippocampal volume.82,89 The DISC1 protein is a multifunctional scaffold protein which has many binding partners and which seems to link different functional complexes within the brain. Through these interactions, DISC1 is involved in neuronal migration, cAMP signalling, centrosomal and microtubule-based functions, kinesin-mediated intracellular transport and neurite extension. These multiple functions may also explain why mutations in DISC1 have so diverse effects. Altogether, there is strong evidence that DISC1 is implicated in psychiatric illness and in cerebral cortex development.90 Our findings relating reelin to many cognitive processes, including attention, memory, learning, and executive functioning fit well with reelin’s biological functions. Reelin controls several essential steps during the embryonic phase of brain development, particularly neuronal migration.91 In the adult brain, reelin is expressed primarily in GABA-containing interneurons and participates in the regulation of GABA-mediated inhibitory circuits in the brain. It functions in the control of synaptic transmission, memory, and learning, and participates in the controlling of NMDA receptors.91 Therefore, the association between reelin and neurocognitive functions is quite plausible. A recent genome-wide association scan using DNA pooling found and replicated a female-specific association between reelin and schizophrenia,92 and previous studies have found lower reelin expression levels in patients with schizophrenia.93 Together these results encourage further research on the link between reelin, schizophrenia, and cognitive functions, but also suggest that there may be other genes predisposing to schizophrenia in the chromosome 7q22 area. Although AKT1 did not show evidence of association to schizophrenia94 or bipolar I disorder,77 we found it to be linked with verbal learning and memory and gray matter density in medial and dorsolateral prefrontal cortex.83 While finding no association with schizophrenia contradicts some other recent studies,95–97 they fit well with other recent research linking AKT1 with

135

inefficient prefrontal cortical activation and deficits in executive functioning in humans,96 and impaired working memory in AKT1 deficient mice.98 DAOA was associated with a wide spectrum of cognitive functions in families with bipolar disorder, most prominently with visuospatial ability.77 Thus far, we have not investigated DAOA in our schizophrenia study samples. One recent study found an association between semantic fluency and DAOA risk haplotype in patients with schizophrenia,99 and another found that DAOA was associated with hippocampal activation during a sentence completion test in subjects at high risk for schizophrenia.100 These studies suggest that DAOA is linked with a variety of cognitive functions, which fits its role as a regulator of NMDA-type glutamate receptors. However, each of these studies has used different SNPs, and there appears to be much allelic and locus heterogeneity.99 Thus, while there is evidence that DAOA may be linked to cognitive endophenotypes in schizophrenia and bipolar disorder, more work is clearly needed. The COMT gene is one of the most widely studied candidate genes of severe mental disorders. The gene lies in chromosome 22q11 in the chromosomal area that is deleted in the Velo-Cardio-Facial syndrome which is associated with 25% risk of developing schizophrenia in the adulthood.101 COMT is also a candidate gene based on its biological function. It has an important role in the extracellular degradation of dopamine and other catecholamines. There is a functional polymorphism within the COMT gene causing a Valine to Methionine transition at position 158. The Met allele codes for a thermo-labile enzyme displaying lower enzymatic activity, causing increased synaptic dopamine concentration. The Met allele has been associated with poorer performance in tests measuring executive functioning, processing speed, attention, and working memory.102–104 The role of rs4680 (Val158Met) or other COMT alleles in increasing the risk of schizophrenia or bipolar disorder is less clear,105,106 with possibly more evidence for its association with bipolar I disorder86 than schizophrenia.105 We found suggestive evidence that rs4680 increases the risk of bipolar disorder, while another variant, rs165599, was associated with visuospatial ability.77 There was also suggestive evidence of association between the Val158Met variation and verbal learning and memory, but these associations were not significant after correction for

136

multiple testing.77 We are currently investigating the role of COMT in our schizophrenia family study. Altogether, our results support previous findings that allelic variation in the COMT gene affects cognitive functioning in families with severe mental illness.

Conclusions and Suggestions for Further Research The impact that endophenotypes add into genetic analyses has thus far been modest. It may be that using single measures derived from neuropsychological test data is not the optimal way to apply this methodology. Some investigators have suggested that a better way could be using composite endophenotypes collected from neurophysiological, neuroanatomical, neuropsychological, and possibly some other assessments from the area of neuroscience.40 Future research will show whether this will prove fruitful. However, the rationale for the endophenotype strategy lies in reducing genetic heterogeneity. The endophenotypes are assumed to be contributed to by fewer genes than the clinical phenotypes of the complex psychiatric disorders themselves. Building composite phenotypes of parts into circuits that are complex themselves, while neurobiologically more understandable than the psychiatric phenotypes, may add heterogeneity. On the other hand, combining neurobiological indices, such as brain morphometry and cognitive deficits with neurodevelopmental indicators, such as minor physical abnormalities and neurologic soft signs, might provide “cleaner” macro-endophenotypes for genetic studies that help in tracking multiple genes of small effect.41,107 Methodical research is required to further develop the endophenotype approach. At the same time, definitions of the clinical phenotypes dependent on DSM and ICD obviously need refinement. To summarize, our studies support the validity particularly of working memory and learning and memory functions as endophenotypes for schizophrenia and bipolar disorder. They are heritable characteristics present in both affected individuals and their unaffected relatives, with increasing severity associated with increasing genetic relatedness to an affected individual, particularly among relatives of patients with schizophrenia. They are strongly associated with genetic variation in some candidate genes for these disorders, particularly with reelin, DISC1, and DAOA. However,

A. Tuulio-Henriksson et al.

much work is still needed to understand the pathophysiological mechanisms leading to cognitive impairments, and to the development of major mental illness in some of the individuals in whom these neuropsychological endophenotypes confer liability to the illness. Acknowledgements Professors Jouko Lönnqvist and Leena Peltonen-Palotie have a fundamental role in the investigation of the genetics and genetic epidemiology of severe mental disorders at the National Public Health Institute (currently National Institute for Health and Welfare). We gratefully acknowledge their vast impact on all our studies. Professor Tyrone Cannon is the leader and principal investigator of the schizophrenia twin study, but his expertise has benefited us all, for which he is warmly thanked. Professor Jaakko Kaprio has a central role in the schizophrenia and bipolar twin studies, which are only a small part of the famous Finnish twin cohort studies lead by him. The bipolar twin study sample was collected by Dr. Tuula Kieseppä, and her expertise and effort have been irreplaceable in the bipolar family study, as well. Many researchers working at the National Public Health Institute have contributed substantially to these studies during the past 20 years: Mervi Antila, Ritva Arajärvi, Jesper Ekelund, Jenny Ekholm, Jari Haukka, William Hennah, Iiris Hovatta, Matti Huttunen, Susanna Juselius, Hannu Juvonen, Marja-Liisa Kokko-Sahin, Annamaria Kuha, Dirk Lichtermann, Anu Loukola, Marko Manninen, Marika Palo, Timo Partonen, Tiina Paunio, Petra Pekkarinen-Ijäs, Olli Pietiläinen, Tiia Pirkola, Pia Soronen, Sebastian Therman, Liisa Tomppo, Minna Torniainen, Joni Turunen, Teppo Varilo, Leena Väisänen, and Juho Wedenoja. Their role, as well as the work done by our international collaborators, is thankfully acknowledged. We warmly thank Merja Blom, Marjukka Heikkinen, Saara Heusala, Margit Keinänen-Guillaume, Pilvi Kujala, Helena Kurru, Pirkko Levon, Outi McDonald, Liisa Moilanen, Tuula Mononen, Merja Nissi, and Silva Ruoppila for skillfully conducted interviews; Marjut Schreck for data management; and Tuula Koski, Kirsi Niinistö and Maija Norilo for administrative work. Finally, we sincerely thank all the participants.

References 1. Reichenberg A, Harvey PD, Bowie CR, Mojtabai R, Rabinowitz J, Heaton RK, Bromet E. Neuropsychological Function and Dysfunction in Schizophrenia and Psychotic Affective Disorders. Schizophr Bull., in press. 2. Keefe RS, Eesley CE, Poe MP. Defining a cognitive function decrement in schizophrenia. Biol Psychiatry 2005;57:688–691 3. Heinrichs RW, Zakzanis KK. Neurocognitive deficit in schizophrenia: a quantitative review of the evidence. Neuropsychology 1998;12:426–445 4. Cirillo MA, Seidman LJ. Verbal declarative memory dysfunction in schizophrenia: from clinical assessment to genetics and brain mechanisms. Neuropsychol Rev 2003;13:43–77 5. Dickinson D, Ramsey ME, Gold JM. Overlooking the obvious: a meta-analytic comparison of digit symbol coding

8

Neuropsychological Endophenotypes in Schizophrenia and Bipolar I Disorder

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

tasks and other cognitive measures in schizophrenia. Arch Gen Psychiatry 2007;64:532–542 Dickinson D, Ragland JD, Gold JM, Gur RC. General and specific cognitive deficits in schizophrenia: Goliath defeats David? Biol Psychiatry 2008;64:823–827 Cornblatt BA, Malhotra AK. Impaired attention as an endophenotype for molecular genetic studies of schizophrenia. Am J Med Genet 2001;105:11–15. Eastvold AD, Heaton RK, Cadenhead KS. Neurocognitive deficits in the (putative) prodrome and first episode of psychosis. Schizophr Res 2007;93:266–277 Heaton RK, Gladsjo JA, Palmer BW, Kuck J, Marcotte TD, Jeste DV. Stability and course of neuropsychological deficits in schizophrenia. Arch Gen Psychiatry 2001;58:24–32 Hoff AL, Svetina C, Shields G, Stewart J, DeLisi LE. Ten year longitudinal study of neuropsychological functioning subsequent to a first episode of schizophrenia. Schizophr Res 2005;78:27–34 Szöke A, Trandafir A, Dupont ME, Méary A, Schürhoff F, Leboyer M. Longitudinal studies of cognition in schizophrenia: meta-analysis. Br J Psychiatry 2008;192:248–257 Fucetola R, Seidman LJ, Kremen WS, Faraone SV, Goldstein JM, Tsuang MT. Age and neuropsychologic function in schizophrenia: a decline in executive abilities beyond that observed in healthy volunteers. Biol Psychiatry 2000;48:137–146 Green MF. What are the functional consequences of neurocognitive deficits in schizophrenia? Am J Psychiatry 1996;153:321–330. Review Green MF. Cognitive impairment and functional outcome in schizophrenia and bipolar disorder. J Clin Psychiatry 2006;67 Suppl 9:3–8; discussion 36–42 Kurtz MM, Mueser KT. A meta-analysis of controlled research on social skills training for schizophrenia. J Consult Clin Psychol 2008;76:491–504 Gur RE, Nimgaonkar VL, Almasy L, Calkins ME, Ragland JD, Pogue-Geile MF, Kanes S, Blangero J, Gur RC. Neurocognitive endophenotypes in a multiplex multigenerational family study of schizophrenia. Am J Psychiatry 2007;164:813–819 Ma X, Wang Q, Sham PC, Liu X, Rabe-Hesketh S, Sun X, Hu J, Meng H, Chen W, Chen EY, Deng W, Chan RC, Murray RM, Collier DA, Li T. Neurocognitive deficits in first-episode schizophrenic patients and their first-degree relatives. Am J Med Genet B Neuropsychiatr Genet 2007;144B:407–416 Sitskoorn MM, Aleman A, Ebisch SJ, Appels MC, Kahn RS. Cognitive deficits in relatives of patients with schizophrenia: a meta-analysis. Schizophr Res 2004;71:285–295 Cannon TD, Huttunen MO, Lönnqvist J, Tuulio-Henriksson A, Pirkola T, Glahn D, Finkelstein J, Hietanen M, Kaprio J, Koskenvuo M. The inheritance of neuropsychological dysfunction in twins discordant for schizophrenia. Am J Hum Genet 2000;67:369–382 Goldberg TE, Ragland JD, Torrey EF, Gold JM, Bigelow LB, Weinberger DR. Neuropsychological assessment of monozygotic twins discordant for schizophrenia. Arch Gen Psychiatry 1990;47:1066–1072 Erlenmeyer-Kimling L, Rock D, Roberts SA, Janal M, Kestenbaum C, Cornblatt B, Adamo UH, Gottesman II.

22.

23.

24. 25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

137

Attention, memory, and motor skills as childhood predictors of schizophrenia-related psychoses: the New York high-risk project. Am J Psychiatry 2000;157:1416–1422 Gottesman II, Erlenmeyer-Kimling L. Family and twin strategies as a head start in defining prodromes and endophenotypes for hypothetical early-interventions in schizophrenia. Schizophr Res 2001;51:93–102 Cornblatt BA, Malhotra AK. Impaired attention as an endophenotype for molecular genetic studies of schizophrenia. Am J Med Genet 2001;105:11–15 Maziade et al., still in press. Niendam TA, Bearden CE, Zinberg J, Johnson JK, O’Brien M, Cannon TD. The course of neurocognition and social functioning in individuals at ultra high risk for psychosis. Schizophr Bull 2007;33:772–781 Arts B, Jabben N, Krabbendam L, van Os J. Metaanalyses of cognitive functioning in euthymic bipolar patients and their first-degree relatives. Psychol Med 2008;38:771–785 Malhi GS, Ivanovski B, Hadzi-Pavlovic D, Mitchell PB, Vieta E, Sachdev P. Neuropsychological deficits and functional impairment in bipolar depression, hypomania and euthymia. Bipolar Disord 2007;9:114–125 Jones SH, Tai S, Evershed K, Knowles R, Bentall R. Early detection of bipolar disorder: a pilot familial high-risk study of parents with bipolar disorder and their adolescent children. Bipolar Disord 2006;8:362–372 Burt T, Prudic J, Peyser S, Clark J, Sackeim HA. Learning and memory in bipolar and unipolar major depression: effects of aging. Neuropsychiatr Neuropsychol Behav Neurol 2000;13:246–253 Salloum IM, Thase ME. Impact of substance abuse on the course and treatment of bipolar disorder. Bipolar Disord 2000;2:269–280. Review Zubieta JK, Huguelet P, O’Neil RL, Giordani BJ. Cognitive function in euthymic bipolar I disorder. Psychiatry Res 2001;102:9–20 Gruber S, Rathgeber K, Bräunig P, Gauggel S. Stability and course of neuropsychological deficits in manic and depressed bipolar patients compared to patients with Major Depression. J Affect Disord 2007;104:61–71 Martinez-Aran A, Torrent C, Tabares-Seisdedos R, Salamero M, Daban C, Balanza-Martinez V, Sanchez-Moreno J, Manuel Goikolea J, Benabarre A, Colom F, Vieta E. Neurocognitive impairment in bipolar patients with and without history of psychosis. J Clin Psychiatry 2008; 69:233–239 Mur M, Portella MJ, Martínez-Arán A, Pifarré J, Vieta E. Persistent neuropsychological deficit in euthymic bipolar patients: executive function as a core deficit. J Clin Psychiatry 2007;68:1078–1086 Robinson LJ, Thompson JM, Gallagher P, Goswami U, Young AH, Ferrier IN, Moore PB. A meta-analysis of cognitive deficits in euthymic patients with bipolar disorder. J Affect Disord 2006;93:105–115 Hasler G, Drevets WC, Gould TD, Gottesman II, Manji HK. Toward constructing an endophenotype strategy for bipolar disorders. Biol Psychiatry 2006;60:93–105 Gottesman II, Gould TD. The endophenotype concept in psychiatry: etymology and strategic intentions. Am J Psychiatry 2003;160:636–45. Review

138 38. Gottesman II, Shields J. Schizophrenia and Genetics: A Twin Study Vantage Point. New York: Academic, 1972 39. Hall MH, Rijsdijk F. Validating endophenotypes for schizophrenia using statistical modeling of twin data. Clin EEG Neurosci 2008;39:78–81 40. Prasad KM, Keshavan MS. Structural cerebral variations as useful endophenotypes in schizophrenia: do they help construct “extended endophenotypes”? Schizophr Bull 2008;34(4):774–790 41. Chan RC, Gottesman II. Neurological soft signs as candidate endophenotypes for schizophrenia: a shooting star or a Northern star? Neurosci Biobehav Rev 2008;32:957–971 42. Egan MF, Goldberg TE. Intermediate cognitive phenotypes associated with schizophrenia. Methods Mol Med 2003;77:163–197, 163 43. Wechsler D. Wechsler Adult Intelligence Scale – Revised (WAIS-R), Manual. The Psychological Corporation. Cleveland, OH: Harcourt Brace Jovanovich, 1981 44. Wechsler D. Wechsler Memory Scale – Revised (WMS-R), Manual. The Psychological Corporation. San Antonio, TX: Harcourt Brace Jovanovich, 1987 45. Reitan RM. The Halstead-Reitan Neuropsychological Test Battery. Tucson, AZ: Neuropsychology Press 46. Rosvold HE, Mirsky A, Sarason I, Bransome EDJ, Beck LH. A continuous perfomace test of brain damage. J Consult Psychology 1956;20:343–350 47. Posner M, Petersen SE. The attention system of the human brain. Annu Rev Neurosci 1990;13:25–42 48. Gur RC, Jaggi JL, Ragland JD, Resnick SM, Shtasel D, Muenz L, Gur RE. Effects of memory processing on regional brain activation: cerebral blood flow in normal subjects. Int J Neurosci 1993;72:31–44 49. Golden C. Stroop Color and Word Test: Manual for Clinical and Experimental Uses. Chicago, IL: Stoelting, 1978 50. Benton AL, Hamsher K. Multilingual Aphasia Examination Manual. Iowa City, IA: University of Iowa, 1989 51. Heaton RK. Wisconsin Card Sorting Test. Odessa, TX: Psychological Assessment Resources, 1981 52. Vilkki J, Virtanen S, Surma-Aho O, Servo A. Dual task performance after focal cerebral lesions and closed head injuries. Neuropsychologia 1996;34:1051–1056 53. Reitan R, Wolfson D. The Halstead-Reitan neuropsychological test battery: theory and clinical interpretation. Tucson, AZ: Neuropsychology Press, 1985 54. Hovatta I, Terwilliger JD, Lichtermann D, Mäkikyrö T, Suvisaari J, Peltonen L, Lönnqvist J. Schizophrenia in the genetic isolate of Finland. Am J Med Genet B Neuropsychiatr Genet 1997;74:353–360 55. Tuulio-Henriksson A, Haukka J, Partonen T, Varilo T, Paunio T, Ekelund J, Cannon TD, Lönnqvist J. Heritability of neurocognitive functions and number of quantitative trait loci contributing to them in families with schizophrenia, Am J Med Genet (Neuropsych Genet), 2002;114:483-490 56. Tuulio-Henriksson A, Arajärvi R, Partonen T, Haukka J, Varilo T, Schreck M, Cannon TD, Lönnqvist J. Familial loading associates with impaired visual span among healthy siblings of schizophrenia patients. Biol Psychiatry, 2003; 54:623–628 57. Tuulio-Henriksson A, Partonen T, Suvisaari J, Haukka J, Lönnqvist J. Age of onset and cognitive functioning in schizophrenia. Br J Psychiatry 2004;185:215–219

A. Tuulio-Henriksson et al. 58. Hambrecht M, Lammertink M, Klosterkotter J, Matuschek E, Pukrop R. Subjective and objective neuropsychological abnormalities in a psychosis prodrome clinic. Br J Psychiatry 2002;181(S43):30–37 59. Hoti F, Tuulio-Henriksson A, Haukka J, Partonen T, Holmström L, Lönnqvist J. Family-based clusters of cognitive test performance in familial schizophrenia. BMC Psychiatry 2004;22;4:20 60. Kuha A, Tuulio-Henriksson A, Eerola M, Perälä J, Suvisaari J, Partonen T, Lönnqvist J. Impaired executive performance in healthy siblings of schizophrenia patients in a populationbased study. Schizophr Res 2007;92:142–150 61. Hovatta I, Varilo T, Suvisaari J, Terwilliger JD, Ollikainen V, Arajärvi R, Juvonen H, Kokko-Sahin ML, Väisänen L, Mannila H, Lönnqvist J, Peltonen L. A genomewide screen for schizophrenia genes in an isolated Finnish subpopulation, suggesting multiple susceptibility loci. Am J Hum Genet 1999;65:1114–1124 62. Ekelund J, Hennah W, Hiekkalinna T, Parker A, Meyer J, Lönnqvist J, Peltonen L. Replication of 1q42 linkage in Finnish schizophrenia pedigrees. Mol Psychiatry 2004;9:1037–1041 63. Ekelund J, Hovatta I, Parker A, Paunio T, Varilo T, Martin R, Suhonen J, Ellonen P, Chan G, Sinsheimer JS, Sobel E, Juvonen H, Arajärvi R, Partonen T, Suvisaari J, Lönnqvist J, Meyer J, Peltonen L. Chromosome 1 loci in Finnish schizophrenia families. Hum Mol Genet 2001;15:1611–1617 64. Ekelund J, Lichtermann D, Hovatta I, Ellonen P, Suvisaari J, Terwilliger JD, Juvonen H, Varilo T, Arajarvi R, Kokko-Sahin ML, Lonnqvist J, Peltonen L. Genomewide scan for schizophrenia in the Finnish population: evidence for a locus on chromosome 7q22. Hum Mol Genet 2000;9:1049–1057 65. Paunio T, Ekelund J, Varilo T, et al. Genome-wide scan in the nationwide study sample of schizophrenia families in Finland. Hum Mol Genet 2001;10:3037–3048 66. Paunio T, Tuulio-Henriksson A, Hiekkalinna T, Perola M, Varilo T, Partonen T, Cannon TD, Lönnqvist J, Peltonen L. Search for cognitive trait components of schizophrenia reveals a locus for verbal learning and memory on 4q and for visual working memory on 2q. Hum Mol Genet 2004;13:1693–1702 67. Hennah W, Varilo T, Kestilä M, Paunio T, Arajärvi R, Haukka J, Parker A, Martin R, Levitzky S, Partonen T, Meyer J, Lönnqvist J, Peltonen L, Ekelund J. Haplotype transmission analysis provides evidence of association for DISC1 to schizophrenia and suggests sex-dependent effects. Hum Mol Genet 2003;12:3151–3159 68. Palo OM, Antila M, Silander K, Hennah W, Kilpinen H, Soronen P, Tuulio-Henriksson A, Kieseppä T, Partonen T, Lönnqvist J, Peltonen L, Paunio T. Association of distinct allelic haplotypes of DISC1 with psychotic and bipolar spectrum disorders and with underlying cognitive impairments. Hum Mol Genet 2007;16:2517–2528 69. Blackwood DH, Fordyce A, Walker MT, St Clair DM, Porteous DJ, Muir WJ. Schizophrenia and affective disorders - cosegregation with a translocation at chromosome 1q42 that directly disrupts brain-expressed genes: clinical and P300 findings in a family. Am J Hum Genet 2001;69:428–433 70. Hennah W, Tuulio-Henriksson A, Paunio T, Ekelund J, Varilo T, Partonen T, Cannon TD, Lönnqvist J, Peltonen L.

8

71.

72.

73.

74.

75.

76.

77.

78.

79.

80.

81.

82.

Neuropsychological Endophenotypes in Schizophrenia and Bipolar I Disorder A haplotype within the DISC1 gene is associated with visual memory functions in families with a high density of schizophrenia. Mol Psychiatry 2005;10:1097–1103 Hennah W, Tomppo L, Hiekkalinna T, Palo OM, Kilpinen H, Ekelund J, Tuulio-Henriksson A, Silander K, Partonen T, Paunio T, Terwilliger JD, Lönnqvist J, Peltonen L. Families with the risk allele of DISC1 reveal a link between schizophrenia and another component of the same molecular pathway, NDE1. Hum Mol Genet 2007;16:453–462 Wedenoja J, Loukola A, Tuulio-Henriksson A, Paunio T, Ekelund J, Silander K, Varilo T, Heikkilä K, Suvisaari J, Partonen T, Lönnqvist J, Peltonen L. Replication of linkage on chromosome 7q22 and association of the regional Reelin gene with working memory in schizophrenia families. Mol Psychiatry 2008;13:673–684 Antila M, Tuulio-Henriksson A, Kieseppä T, Eerola M, Partonen T, Lönnqvist J. Cognitive functioning in patients with familial bipolar I disorder and their unaffected relatives. Psychol Med 2007 May;37(5):679–687 Antila M, Tuulio-Henriksson A, Kieseppä T, Soronen P, Palo M, Paunio T, Haukka J, Partonen T, Lönnqvist J. Heritability of cognitive traits in families with bipolar disorder. Am J Med Genet (Neuropsych Genet), 2007;144:802–808 Hodgkinson CA, Goldman D, Jaeger J, Persaud S, Kane JM, Lipsky RH, Malhotra AK. Disrupted in schizophrenia 1 (DISC1): association with schizophrenia, schizoaffective disorder, and bipolar disorder. Am J Hum Genet 2004;75:862–872 Thomson PA, Wray NR, Millar JK, Evans KL, Hellard SL, Condie A, Muir WJ, Blackwood DH, Porteous DJ. Association between the TRAX/DISC locus and both bipolar disorder and schizophrenia in the Scottish population. Mol Psychiatry 2005;10:657–668 Soronen P, Silander K, Antila M, Palo OM, TuulioHenriksson A, Kieseppä T, Ellonen P, Wedenoja J, Turunen JA, Pietiläinen OP, Hennah W, Lönnqvist J, Peltonen L, Partonen T, Paunio T. Association of a nonsynonymous variant of DAOA with visuospatial ability in a bipolar family sample. Biol Psychiatry 2008;64:438–442 Glahn DC, Therman S, Manninen M, Huttunen M, Kaprio J, Lönnqvist J, Cannon TD. Spatial working memory as an endophenotype for schizophrenia. Biol Psychiatry 2003;53:624–626 Johnson JK, Tuulio-Henriksson A, Pirkola T, Huttunen MO, Lönnqvist J, Kaprio J, Cannon TD. Do schizotypal symptoms mediate the relationship between genetic risk for schizophrenia and impaired neuropsychological performance in co-twin of schizophrenic patients. Biol Psychiatry 2003;54:1200–1204 van Erp TGM, Therman S, Pirkola T, Tuulio-Henriksson A, Glahn DC, Bachman P, Huttunen MO, Lönnqvist J, Hietanen M, Kaprio J, Koskenvuo M, Cannon TD. Verbal recall and recognition in twins discordant for schizophrenia. Psychiatry Res 2008;159:271–280 Gasperoni TL, Ekelund J, Huttunen M, Palmer CGS, TuulioHenriksson A, Lönnqvist J, Kaprio J, Peltonen L, Cannon TD. Genetic linkage and association between chromosome 1q and working memory function in schizophrenia. Am J Med Genet B Neuropsychiatr Genet 2003;116:8–16 Cannon TD, Hennah W, van Erp TG, Thompson PM, Lonnqvist J, Huttunen M, Gasperoni T, Tuulio-Henriksson A,

83.

84.

85.

86.

87.

88.

89.

90.

91. 92.

93.

94.

95.

139

Pirkola T, Toga AW, Kaprio J, Mazziotta J, Peltonen L. Association of DISC1/TRAX haplotypes with schizophrenia, reduced prefrontal gray matter, and impaired short- and longterm memory. Arch Gen Psychiatry 2005;62:1205–1213 Pietiläinen OP, Paunio T, Loukola A, Tuulio-Henriksson A, Kieseppä T, Thompson P, Toga AW, van Erp TG, Silventoinen K, Soronen P, Hennah W, Turunen JA, Wedenoja J, Palo OM, Silander K, Lönnqvist J, Kaprio J, Cannon TD, Peltonen L. Association of AKT1 with verbal learning, verbal memory, and regional cortical gray matter density in twins. Am J Med Genet B Neuropsychiatr Genet., in press. Kieseppä T, Tuulio-Henriksson A, Haukka J, van Erp T, Glahn D, Cannon TD, Kaprio J, Lönnqvist J. Memory and verbal learning functions in twins with bipolar I disorder, and the role of information processing speed. Psychol Med 2005;35:205–215 Pirkola T, Tuulio-Henriksson A, Glahn D, Kieseppä T, Haukka J, Kaprio J, Lönnqvist J, Cannon TD. Spatial working memory function in twins with schizophrenia and bipolar disorder. Biol Psychiatry 2005;58:930–936 Burdick KE, Hodgkinson CA, Szeszko PR, Lencz T, Ekholm JM, Kane JM, Goldman D, Malhotra AK. DISC1 and neurocognitive function in schizophrenia. Neuroreport 2005;16:1399–1402 Callicott JH, Straub RE, Pezawas L, et al. Variation in DISC1 affects hippocampal structure and function and increases risk for schizophrenia. PNAS 2005;102:8627–8632 Thomson PA, Harris SE, Starr JM, Whalley LJ, Porteous DJ, Deary IJ. Association between genotype at an exonic SNP in DISC1 and normal cognitive aging. Neurosci Lett 2005;389:41–45 Szeszko PR, Christian C, Macmaster F, Lencz T, Mirza Y, Taormina SP, Easter P, Rose M, Michalopoulou GA, Rosenberg DR. Gray matter structural alterations in psychotropic drug-naive pediatric obsessive-compulsive disorder: an optimized voxel-based morphometry study. Am J Psychiatry 2008;165:1299–1307 Chubb JE, Bradshaw NJ, Soares DC, Porteous DJ, Millar JK. The DISC locus in psychiatric illness. Mol Psychiatry 2008;13:36–64 Herz J, Chen Y. Reelin, lipoprotein receptors and synaptic plasticity. Nature Rev Neurosci 2006;7:850–859 Shifman S, Johannesson M, Bronstein M, Chen SX, Collier DA, Craddock NJ, Kendler KS, Li T, O’Donovan M, O’Neill FA, Owen MJ, Walsh D, Weinberger DR, Sun C, Flint J, Darvasi A. Genome-wide association identifies a common variant in the reelin gene that increases the risk of schizophrenia only in women. PLoS Genetics 2008;4:e28 Fatemi SH, Earle JA, McMenomy T. Reduction in Reelin immunoreactivity in hippocampus of subjects with schizophrenia, bipolar disorder and major depression. Mol Psychiatry 2000;5:654–663 Turunen JA, Peltonen JO, Pietiläinen OP, Hennah W, Loukola A, Paunio T, Silander K, Ekelund J, Varilo T, Partonen T, Lönnqvist J, Peltonen L. The role of DTNBP1, NRG1, and AKT1 in the genetics of schizophrenia in Finland. Schizophr Res 2007;91:27–36 Emamian ES, Hall D, Birnbaum MJ, Karayiorgou M, Gogos JA. Convergent evidence for impaired AKT1-GSK3 signaling in schizophrenia. Nature Gen 2004;36:131–137

140 96. Tan HY, Nicodemus KK, Chen Q, Li Z, Brooke JK, Honea R, Kolachana BS, Straub RE, Meyer-Lindenberg A, Sei Y, Mattay VS, Callicott JH, Weinberger DR. Genetic variation in AKT1 is linked to dopamine-associated prefrontal cortical structure and function in humans. J Clin Invest 2008;118:2200–2208 97. Thiselton DL, Vladimirov VI, Kuo PH, McClay J, Wormley B, Fanous A, O’Neill FA, Walsh D, Van den Oord EJ, Kendler KS, Riley BP. AKT1 is associated with schizophrenia across multiple symptom dimensions in the Irish study of high density schizophrenia families. Biol Psychiatry 2008;63:449–457 98. Lai WS, Xu B, Westphal KG, Paterlini M, Olivier B, Pavlidis P, Karayiorgou M, Gogos JA. Akt1 deficiency affects neuronal morphology and predisposes to abnormalities in prefrontal cortex functioning. Proc Natl Acad Sci U S A 2006;103:16906–16911 99. Opgen-Rhein C, Lencz T, Burdick KE, Neuhaus AH, Derosse P, Goldberg TE, Malhotra AK. Genetic variation in the DAOA gene complex: impact on susceptibility for schizophrenia and on cognitive performance. Schizophr Res 2008;103:169–177 100. Hall J, Whalley HC, Moorhead TW, Baig BJ, McIntosh AM, Job DE, Owens DG, Lawrie SM, Johnstone EC. Genetic variation in the DAOA (G72) gene modulates hippocampal function in subjects at high risk of schizophrenia. Biol Psychiatry 2008;64:428–433 101. Gothelf D, Feinstein C, Thompson T, Gu E, Penniman L, Van Stone E, Kwon H, Eliez S, Reiss AL. Risk factors for the emergence of psychotic disorders in adolescents with 22q11.2 deletion syndrome. Am J Psychiatry 2007;164:663–669

A. Tuulio-Henriksson et al. 102. Bilder RM, Volavka J, Czobor P, Malhotra AK, Kennedy JL, Ni X, Goldman RS, Hoptman MJ, Sheitman B, Lindenmayer JP, Citrome L, McEvoy JP, Kunz M, Chakos M, Cooper TB, Lieberman JA. Neurocognitive correlates of the COMT Val(158)Met polymorphism in chronic schizophrenia. Biol Psychiatry 2002;52:701–707 103. Egan MF, Goldberg TE, Kolachana BS, Callicott JH, Mazzanti CM, Straub RE, Goldman D, Weinberger DR. Effect of COMT Val108/158 Met genotype on frontal lobe function and risk for schizophrenia. Proc Natl Acad Sci U S A 2001;98:6917–6922 104. Ehlis AC, Reif A, Herrmann MJ, Lesch KP, Fallgatter AJ. Impact of catechol-O-methyltransferase on prefrontal brain functioning in schizophrenia spectrum disorders. Neuropsychopharmacology 2007;32:162–170 105. Fan JB, Zhang CS, Gu NF, Li XW, Sun WW, Wang HY, Feng GY, St Clair D, He L. Catechol-O-methyltransferase gene Val/Met functional polymorphism and risk of schizophrenia: a large-scale association study plus meta-analysis. Biol Psychiatry 2005;57:139–144 106. Sanders AR, Duan J, Levinson DF, Shi J, He D, Hou C, Burrell GJ, Rice JP, Nertney DA, Olincy A, Rozic P, Vinogradov S, Buccola NG, Mowry BJ, Freedman R, Amin F, Black DW, Silverman JM, Byerley WF, Crowe RR, Cloninger CR, Martinez M, Gejman PV. No significant association of 14 candidate genes with schizophrenia in a large European ancestry sample: implications for psychiatric genetics. Am J Psychiatry 2008;165:497–506 107. John JP, Arunachalam V, Ratnam B, Isaac MK. Expanding the schizophrenia phenotype: a composite evaluation of neurodevelopmental markers. Compr Psychiatry 2008;49:78–86

Chapter 9

Is More Cognitive Experimental Psychopathology of Schizophrenia Really Necessary? Challenges and Opportunities Angus W. MacDonald, III

Abstract After a promising start, the cognitive experimental psychopathological approach to schizophrenia has run into difficulties. Despite the observation that behavioral impairments constitute the largest effect sizes in schizophrenia, the wide-spread nature of these impairments suggest that they are likely the downstream effects of undifferentiated, upstream neuronal causes. In addition to the prominence of generalized deficits, experimental psychopathology faces the combined challenge that (1) few behavioral paradigms can mitigate between competing theories of schizophrenia, and (2) according to some reviewers, behavioral measures have not helped to identify genes associated with schizophrenia. Thus, much of schizophrenia science considers the elegant behavioral experiments of experimental psychopathology to be a sideshow to the real work of understanding the pathophysiology of schizophrenia. Meanwhile, this “real work” is proceeding in the absence of intermediate behavioral measures of any kind. For example, functional imaging data from rest and default networks reliably differentiates between schizophrenia patients and patients with other diagnoses. The beginning of the chapter describes these challenges to the field of cognitive experimental psychopathology. The remainder of the chapter addresses these objections to illustrate how cognitive experimental methods can adapt and continue to advance the theory and practice of schizophrenia research.

A.W. MacDonald, III Translational Research in Cognitive and Affective Mechanisms Laboratory, Departments of Psychology and Psychiatry, University of Minnesota, MN, USA

Keywords Schizophrenia • methodology • experimental psychology • generalized deficit • theory testing • endophenotypes • neuroinformatics • neurodiagnosis

Introduction: The Gathering Storm There is a chill wind blowing through a half a century of efforts aimed at understanding mental disorders. Just as storms tend to arise at the edge of weather systems, so this storm is gathering as new techniques for understanding mental illness, such as independent components analysis, evolutionary genetic algorithms and genome-wide association studies, rise in prominence. Though still young, an approach to mental illness known as cognitive experimental psychopathology, can already be seen as quaint. Guided by theories from experimental psychology about how cognitive mechanisms function, this approach is showing cracks just as it is becoming a more widely accepted and mainstream method for studying mental disorders. One indicator of the modern era of cognitive experimental psychopathology was a special issue of the reputable Journal of Clinical and Experimental Neuropsychology that was published in 2002. The human genome had just been sequenced and decade of the brain had just ended. Informed by these heady developments, contributors focused on the importance of a growing trend in neuropsychology and experimental psychopathology to utilize the paradigms and theories of cognitive experimental psychology to better understand deficits in psychiatric patients. This issue included a number of articles by eminent scientists and also included a optimistic piece by the current author and my mentor, psychiatrist Cameron Carter.1 Our contribution foresaw an era in which the behavioral study of mental

M.S. Ritsner (ed.), The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes, © Springer Science + Business Media B.V. 2009

141

142

disorders, and in particular schizophrenia, moved away from clinical neuropsychological testing batteries of the kind that had been used for decades. We predicted that experimental cognitive psychopathology which had guided only a minority of studies in patients up to that time, would “become the principal method through which cognitive deficits are investigated in schizophrenia.”1 Since the study of schizophrenia is a bellwether for many techniques used to study psychiatric disorders this was a bold, and somewhat cheeky, claim. To better understand the stakes at that time, there is a small matter of nomenclature that I should first address. In this chapter, I will use the term clinical neuropsychology, to describe an approach to understanding mental disorders based on patients’ deficits across a battery of well-normed neuropsychological tasks.2 It is so named because people with different forms of brain damage have known patterns of performance on such testing batteries. In this case, the mechanisms underlying mental disorders are inferred based on the similarity of performance between patients with the mental disorder and patients with known types of brain damage. Clinical psychologists have been trained in this tradition since World War II, and even more commonly since the 1980s. Experimental psychology describes a tradition that is largely ignorant of mental disorders, but attempts to understand mechanisms of normative mental functioning, for example in healthy populations. While experimental psychology covers a broad swathe of researches, I will use the term cognitive experimental psychology to refer to the large subset of that field dedicated to understanding the mechanisms that underlie perception, information processing and motor behaviors – a designation which in this case includes both hot, or emotional, and cold cognitive mechanisms. Cognitive psychologists have been training in this tradition since the eras of Wilhelm Wundt and Ivan Pavlov. Finally, by cognitive experimental psychopathology I mean to invoke the application of these theoretical models developed in healthy populations to understanding impairments in patients with mental disorders. I make only the smallest distinction between this designation and another common term, clinical cognitive neuroscience, which highlights the biological levels of analysis and data collection common in this approach. It was this approach that I once so loudly espoused. Many of the reasons for that belief remain quite compelling. The first reason for my zeal for cognitive experimental psychopathology is alluded to above. That is,

A.W. MacDonald, III

the concordance between the methods of this approach and those of human electrophysiology and imaging, pharmacological challenge, animal recording and lesion experiments allowed investigators to build a more cumulative science of psychopathology. It appeared almost axiomatic that we should use our best understanding about how the brain operates normatively to derive theories of how pathology develops. Exhortations at the time from the National Institute of Mental Health, the major funding agency for research on mental disorders in the United States, also advocated this transfer of technology and theory. The second reason for embracing cognitive experimental psychopathology was the problem of failures to replicate linkage and early association findings in schizophrenia. The need for continuous measures of liability to schizophrenia, known as endophenotypes, was increasingly being cited as a way to address this problem of unreliable findings.3,4 The behavioral paradigms of cognitive experimental psychology provided something that looked like a continuous measures of the degree to which a specific brain mechanism had failed on the way to developing psychosis. Thus behavioral deficits on such paradigms might map more closely to particular neurotransmitter systems, proteins and genes than the more global deficits that could be detected using clinical neuropsychological tasks. Since that time, the pace and expense of developments in the science of psychopathology has been marked. At the 2007 meeting of the Society for Research in Psychopathology in Iowa City, Iowa I returned to this topic to present a more jaundiced perspective. At that time, the author participated in a symposium with three august figures in the field of clinical cognitive neuroscience, those being Deanna Barch of Washington University of St. Louis, James Gold of the Maryland Psychiatric Research Center, and Stephen Silverstein of the New Jersey School of Medicine and Dentistry. In this symposium I was caste as Judas, or perhaps Cato. I was assigned the intellectually stimulating exercise of pointing out all the flies in the ointment of cognitive experimental psychopathology. This chapter explores those concerns more fully in the fashion of a trial lawyer making the case against cognitive experimental psychopathology. This perspective illuminates the formidable conceptual problems that have cropped-up as cognitive experimental psychopathology has grown in popularity in recent years. The first of these concerns is the challenge of understanding the large generalized

9

Is More Cognitive Experimental Psychopathology of Schizophrenia Really Necessary?

deficit in patients with schizophrenia. The second is the distance between performance on experimental tasks and the current biological framework for conceptualizing the causes of schizophrenia. The last is the observation that brute force empiricism, in the absence of any cognitive theory at all, is making remarkable progress in revealing core features about schizophrenia. I would implore the reader not to simply stop there, however, for the story is only half-told. At Iowa City, I was obliged to leave my accusations unanswered. The answering was left for others. But here I will retort. The case against cognitive experimental psychopathology is followed by answers to these accusations. It is hoped that this format will illuminate several ways in which cognitive experimental psychopathology can anticipate and overcome the foreseeable challenges of the coming decade.

The Case Against Cognitive Experimental Psychopathology The Generalized Deficit in Schizophrenia The case against cognitive experimental psychopathology begins with a meta-analysis. Surprisingly, this meta-analysis demonstrated how important, rather than how unimportant, cognitive deficits are in schizophrenia. How could this be a problem for cognitive experimental psychopathology, which directs itself at the very question of cognitions in mental disorders? Indeed, we shall see. In 1998 R. Walter Heinrichs, a professor of psychology at York University in Toronto, published along with one of his students Konstantine Zakzanis, the first of a collection of meta-analyses on cognition in schizophrenia.5 While the use of neuropsychological testing in patients could no longer be called a young methodology, there had been few systematic reviews up to that time. Some informed observers believed schizophrenia patients might be spared the kind of performance deficits observed in neurological disorders.2 There even remained pockets of theorists who believed schizophrenia was akin to creative genius.6,7 Heinrichs and Zakzanis’ efforts to quantify the collective wisdom of the field buried these presumptions. They demonstrated that across every domain of performance included in their

143

wide-ranging review of psychological and clinical neuropsychological tests, patients with schizophrenia performed worse. In the event, patients performed the worst on global verbal memory (Cohen’s effect size = −1.41) and performed best on block design (Cohen’s effect size = −0.46).1 However, between these two bookends, there was very little order to patients’ measured impairments. For example, non-verbal IQ measures other than Block Design – such as motor skills and performance IQ – had effect sizes of −1.30 and −1.26, respectively. Thus, while no cognitive domain appeared to be spared by the disorder, there was no particular domain that was clearly differentially impaired. The overall effect size of the deficit in patients across all tasks was observed to be about −0.92.8 More recent work has also demonstrated that such cognitive impairments precede the onset of psychosis by many years. A meta-analysis of general intellectual functioning in patients that would later develop schizophrenia suggested these impairments (on a much narrower battery of tests) have a Cohen’s effect size of −0.54.9 What’s more, patients’ firstdegree relatives also show a decline in performance across a broad range of cognitive tests with a Cohen’s effect size of about −0.35.10 This pattern of findings, that of wide-spread behavioral impairments has come to be called the “generalized” deficit.11 It would appear that generalized deficits are associated not only with schizophrenia, but with the genetic predisposition to schizophrenia. Heinrichs has also asserted that in his hands the magnitude of this deficit is about twice as great as the magnitude of deficits found in neuroimaging, electrophysiology or even molecular biological studies of schizophrenia.8 The magnitude of impairments in schizophrenia patients on such tasks does not vindicate the efforts of cognitive experimental psychopathologists. This is ironic because large effect sizes on behavioral tasks are the coin of the realm of experimental psychopathology. However, the pattern of generalized deficits poses a challenge to the field in two ways. First, the largest effect sizes should guide the scientist as to the domain most affected by the disorder, which then, a few studies later, can be identified with a neural substrate. 1

Effect size measures can be thought of measuring the difference between two groups using the common metric of standard deviations. That is an effect size of 1.0 means there is a 1.0 standard deviation difference between the means of the two groups.

144

However, when all domains of cognitive functioning are impaired, there is no clear starting place. The multiplicity of cognitive domains to examine has led to a Babel of studies all leading off in different directions and implicating nearly every brain network. If every cognitive process and every brain region is under suspicion, then it is unlikely that any one cognitive mechanism stands at the core of the pathology. Second, these meta-analyses generally examine traditional psychological tests – such as intelligence subtests and total scores – and clinical neuropsychological tests – such as the Trail Making and the Wisconsin Card Sort Tests. These are off-the-shelf tests and generally have excellent psychometric properties. They are easily standardized across studies and are therefore easy for meta-analysts to compile and score. Though putatively measures of distinct cognitive domains, it is widely understood that such tests rely on a number of overlapping and uncontrolled cognitive processes. The approach also has the simple advantage of looking for a statistical main effect. The sensitivity that comes from good psychometrics, the magnitude of impairment that comes from casting a wide net across many cognitive processes and the statistical power that comes from measuring a main effect add-up to a winning combination for such traditional tests. We can contrast these advantages of traditional tests with paradigms adapted from cognitive experimental psychology. To isolate subtle impairments in specific processes, cognitive paradigms general invoke a variant of the subtractive method. For example, the subtractive method is often used in the Stroop Color-Word paradigm to understand how taxing it is to control attention. In this paradigm, the instruction is to name the ink color in which a word is printed. Attentional control is required if the word is another color word, for example RED printed in green. According to experimentalists, knowing that the average reaction time is 0.5 s for controls and 1 s for patients on such stimuli tells you little about patients’ capacity for attentional control. Such a difference may simply be due to patients’ capacity for color perception, their capacity to respond quickly, or even their motivation or comprehension of the instructions. More telling would be a difference score, say the difference in reaction time between RED printed in green and DOG printed in green. The former is referred to as an incongruent trial whereas the later is a neutral trial. Further understanding can be gained by including congruent trials, such as GREEN printed in green on

A.W. MacDonald, III

which someone lacking in attentional control may have a distinct advantage.12 In the context of basic experimental cognitive psychology, these are generally withinsubject factors and power is rarely an issue. But when comparing groups on these subtle differences between conditions, effect sizes are much smaller than for clinical neuropsychological tasks. In addition to this purely statistical limitation, there are a number of other shortcomings of tests adapted from experimental psychology. Such tests generally have poor psychometric properties when the psychometric properties are known at all. This is because as a field, experimental psychology focuses on developing paradigms rather than tests. Paradigms highlight the kinds manipulations that are relevant for observing a particular phenomenon, but they do not dictate the number of trials, the inter-trial intervals or any other aspect of the test thought to be irrelevant to observing the phenomenon. This is simply because in building a cumulative science, experimentalists continually modify their paradigms to address new questions. But it means there has been no emphasis on pedestrian problems of reliability, stability, ceiling or floor effects. This further limits the capacity for theoretically-informed tasks to compete head-to-head with their off-the-shelf clinical neuropsychological brethren. We know less about experimental task performance for two other, related reasons. Because experimental psychology relies on paradigms rather than tasks, it is more difficult to combine the findings from different studies into a principled meta-analysis because each instance is likely somewhat different. Finally, because such tasks generally require many more-or-less similar trials in two or more conditions, the tasks are long and often strain the tolerance of patients. As a result, it is rare for more than one such test to be measured in any given experiment, which further limits our knowledge base about these tests relative to traditional psychological and clinical neuropsychological ones. So the first fly in the ointment is that cognitive experimental psychopathologists have fallen into a trap of their own making. Psychopathology should address itself to a better understanding, and alleviation of, the gaping generalized deficit. It is cognitive dysfunctions across a broad landscape of life that impair patients’ ability to maintain strong, positive relationships, fulfilling work and financial security. In the face of this generalized deficit, the study of specific cognitive deficits measured with tests of unknown psychometric providence using impoverished statistical tests

9

Is More Cognitive Experimental Psychopathology of Schizophrenia Really Necessary?

increasingly appears to be a parochial distraction. If psychopathology is going to address the problems of mental illness, addressing the generalized deficit, rather than fixating on specialized mechanisms, would be a good place to start.

Weakness of Inferring Causes from Behavior Cognitive experimental psychopathology is vulnerable from another angle, as well. This is that cognitive experimental psychopathology does not contribute to understanding the causes of schizophrenia because experimental tasks cannot adjudicate between theories of the causes of schizophrenia. Furthermore, experimental tasks have no proven advantage in helping to identify causal genes. Two brief narratives serve to illustrate these two limitations of the cognitive experimental approach. Weakness of inferring pathophysiology from behavior. While there are many competing theories of schizophrenia, one of the most prominent theories features an imbalance in the neurotransmitter dopamine. In the late 1980s, two cross-trained psychiatrists and cognitive psychologists at Western Psychiatric Institute and Clinic and Carnegie Mellon University in Pittsburgh joined forces to demonstrate that a very particular pattern of errors across a number of cognitive tasks might be accounted for by a failure of a dopamine-linked mechanism they called context processing. Carnegie Mellon being a school for computer wonks, these two investigators, Jonathan Cohen and David ServanSchreiber, went so far as to build a simulation of patients’ deficits using connectionist modeling.13 Models are particularly elegant ways to illustrate a theory. In a tour-de-force widely recognized as a breakthrough within the field, these simulations showed that a change in the “gain function,” which enabled the executive component in the model to distinguish signal from noise, accounted for patients’ impairments across a number of tasks. The change in the gain function, they cogently argued, modeled what occurred when dopamine ran low in prefrontal cortex, and was related to a specific deficit in performance on an expectancy manipulation of a task that required attentional control.14,15 While Cohen and Servan-Schreiber were not the first to implicate dopamine in the pathophysiology

145

of schizophrenia, they elegantly used the tools of cognitive experimental psychopathology to describe it’s role, and to fill-in an important intermediate level of analysis – the moment-to-moment experiences that stand between the global presentation of symptoms and the molecular level of neurotransmission. Thus, their program of research provided compelling evidence for the role of dopamine in a particular aspect of schizophrenia: the inability to organize thoughts and actions. A second prominent theory is that schizophrenia is caused by an imbalance in another neurotransmitter, glutamate. Glutamate theories have a more recent origin than dopamine theories of schizophrenia,16,17 but are growing in popularity.2 18 In the late 1990s, Bill Phillips, a computationally-minded neuropsychologist at the University of Stirling began a trans-Atlantic collaboration when he sought-out the expertise of Steve Silverstein, a clinical psychologist then at the University of Rochester. After several years of wrangling with editors, their tightly argued thesis was published along with 27 peer commentaries (including one from the current author), and their careful responses to any objections their peers had cared to voice. In their review they proposed that schizophrenia was characterized by a failure of NMDA-based neurotransmission, which is a primary function of the glutamate system. This failure was evident in patient’s inability to perceive the gestalt shapes within a field of noise. Like Cohen and Servan-Schreiber’s work before them, their argument was based on careful computational modeling work and supported by experimental data. These data, drawn from a task that required perceiving shapes in a field of noise, showed that patients had an impairment in perceptual organization or gestalt perception (see also the elegant experiments of Place and Gilmore19). Thus, their program of research provided compelling evidence for the role of glutamate in a particular aspect of schizophrenia: the inability to extract regularities from the environment. One thing that is important to understand about the study of schizophrenia is that it is a very complicated illness. It has many different symptoms whose relationship to each other is not well understood. So, is 2

Ironically, whereas dopamine theories are based on the pharmacology of what helps to reduce psychosis (anti-psychotics), glutamate theories are based on what seems to cause psychosis in otherwise healthy people (phencyclidine or PCP).

146

there not room for multiple theories to describe different aspects of the illness? Indeed, there was no obvious overlap between the constructs of top-down attentional control and bottom-up perceptual organization. Nor did there appear to be overlap in the brain between the deficits revealed by such tasks: functional MRI experiments of Cohen’s and Servan-Schreiber’s task revealed patients’ impairments in prefrontal cortex,20,21 whereas Silverstein’s task revealed patients’ impairments in occipital cortex.22 It might appear then that the two models conveniently addressed different aspects of psychotic heterogeneity. However this was not the case. Patients’ deficits on this expectancy AX task were selectively related to a particularly important dimension of symptoms known as “disorganization” or “thought disorder” (r = 0.41,23 see also20). Disorganization is of particular interest because it is more heritable than the other prominent dimensions of the disorder, reality distortions (such as delusions and hallucinations) and deficit symptoms (such as blunted affect and impoverished speech).24 Just so, patients deficits on the perceptual organization task were selectively related to this same dimension of psychopathology (r = 0.47,25 see also26). Both the top-down, prefrontal cortex demanding AX task and the bottom-up, occipital cortex-demanding perceptual organization tasks related to a common aspect of the illness, that is the symptoms of disorganization. This despite the fact that their theoretical basis lay in two entirely different neurotransmitter systems. An optimist would suggest that this is an isolated problem of two theories that just happen to attempt to account for different performance deficits related to a single aspect of the disorder. But it may also reveal a more fundamental flaw in the cognitive experimental armamentarium: that behavioral tasks are generally ill suited for the task of probing and distinguishing between different neurotransmitter systems. For example, it may be that a number of different pharmacologic challenges affect performance on the AX and perceptual organization (for a specific example, see27). A pessimist might even observe that patients with the greatest generalized deficit are also likely to be those with the most disorganization symptoms. Given these limitations of behavioral paradigms, it is compelling to conclude that behavior is too far removed from the pathophysiology of schizophrenia to aid in falsifying theories. Weakness of inferring genetic vulnerability from behavior. If elegant programs of research inspired by cognitive experimental psychopathology do not

A.W. MacDonald, III

adjudicate between competing theories, then perhaps they can help us understand the causes of a disorder in other ways. For example, perhaps the approach can help us to understand the cognitive and affective mechanisms that fill the chasm between the genes that cause schizophrenia and the downstream manifestation of symptoms. Irving Gottesman, a pioneer in studying the genetic basis of schizophrenia and an emeritus Professor at the University of Minnesota, has long argued for the importance of such an approach. Along with British psychiatrist James Shields, he coined the term “endophenotype” to describe simpler behavioral and physiological traits that may result most directly from schizophrenia liability genes.4,28 As many readers will already know, the idea of an endophenotype is to increase power for detecting genes by reducing the proportion of “false negative” measurements in a study – that is, reducing the likelihood that someone carrying a real liability gene will be considered a non-case simply because they have not yet or may never decompensate. Many studies have subsequently used an endophenotype-informed approach for studying the genetic liability to schizophrenia. Recently, two British scientists, Jonathan Flint and Marcus Munafo, compared four measures of genetic liability to schizophrenia.29 The first measure was the diagnosis itself. The researchers used only genes with established relationships to schizophrenia to establish a benchmark magnitude, and estimated that five genes with successful associations (DRD2, 5-HT2A, SLC6A4, G72/G30, RGS4) accounted for approximately 0.2% of variance. Adequate power to detect such a small amount of variance would require 900 cases and 900 controls under ideal circumstances. Then they asked, how much additional power does the use of an endophenotype buy? Their meta-analysis of studies relating the Wisconsin Card Sorting Test and the catechol-O-methyl transferase (COMT) gene showed the gene accounted for 0.5% of variance in performance. Such a study would require 1,700 subjects to reliably detect a significant effect, only 100 less than the diagnosis itself. The relationship between the N-back and COMT was nearly identical. Finally, they considered the P300 waveform derived from scalp electrodes and COMT. Here the results were even less striking, with COMT accounting for (a non-significant) 0.01% of variance in P300 magnitude and latency, which would of course necessitate an astronomical sample size. The authors are led to the conclusion that endophenotypes remain too genetically complex, or multi-determined, to improve much on the diagnosis itself. Thus they write, there is “no support

9

Is More Cognitive Experimental Psychopathology of Schizophrenia Really Necessary?

for the view that the effect sizes of loci contributing to [endo]phenotypes is closer to the biological basis of disease” (p. 163). While the work of Flint and Munafo is directed at the utility of endophenotypes, it has implications for the utility of the cognitive experimental approach. This is because some cognitive experimental psychopathologists will hold out the hope that even if their tools are not useful for understanding the pathophysiology of schizophrenia, they can at least help determine what genes are involved. This work suggests that this, too, may be a false hope. Thus, at least two important applications of cognitive experimental psychopathology are under threat largely due to the weakness of behavioral findings for constraining our understanding of more physiological levels of analysis. If there were no alternatives to this approach, the field would simply have to muddle through the best it could. Fortunately (or unfortunately, for the cognitive experimentally inclined), new neuroinformatics studies suggest that it is possible to leap-frog the cognitive level of analysis altogether to link symptoms and diagnosis to functional brain anatomy. In effect, these studies use brute force empiricism rather than elegant cognitive theories to advance our understanding of the pathophysiology of schizophrenia.

Successes of Brute Force Empiricism We navigate an era of whole-brain imaging, genomewide association studies and computers capable of petaflops (that is 1015 calculations per second) and unlimited data storage. The exploitation of this potentially rich source of resources for understanding psychopathology is in its infancy, but already a number of promising studies have emerged in the schizophrenia literature. This work, in which data rather than theory drive the science, threaten to make experimental cognitive psychopathology appear to be a quaint footnote in the study of mental illnesses. Two studies published in 2007 serve to illustrate this point. Apostolos Georgopoulos is a neuroscientist who became internationally established several decades ago by demonstrating how motor neurons in macaque monkeys worked in concert to make gross movements. Since that time, his work has drawn him increasingly into studies of mental illness in humans using, among other things, magnetoencephalography, mercifully

147

shorted to MEG. If MEG is really good at one thing, it is detecting activity in sulcal gray matter all over the brain on a very fine-grained time scale. Therefore, Georgopoulos reasoned, if patients with schizophrenia show a particular pattern of impairment in the connections between different cortical regions, then this might be detected by examining the correlations between this brain-wide activity over time. This impairment in connectivity would be evident even if they were not performing any particular cognitive task. In the event, participants lay in an MEG machine for a minute simply staring at a fixation cross.30 Participants came from six different diagnostic groups, including schizophrenia, alcoholism, Alzheimer’s disease, multiple sclerosis and Sjogren’s syndrome, which is an autoimmune disorder. At the end of just 1 min, the investigators had 248 time series from different locations around the brain with about 60,000 time points each. In the past, the problem with collecting so much data would be the problem of sifting through it to look for patterns without using many, many statistical tests. In the modern era of informatics, this problem can be addressed by using any of a number of machine learning techniques. In this case, Georgopoulos’ team used several, including an evolutionary algorithm which worked overnight to find subsets of observations that effectively separated each group from the others. When the algorithm finished its job with one subset of data, it began with another. The purpose of this effort was to test whether the number of good predictors exceeded the number expected by chance (that is, it allowed for a form of inference), and to generate a classification algorithm that could be crossvalidated in a new set of patients. In the event, it served both purposes. In thousands of runs, the algorithm was able to separate a sample of 52 participants into the correct six groups with 100% accuracy. More remarkable still, many of the algorithms developed in this initial sample were capable of classifying, with 95–100% accuracy, a new group of 46 subjects from 5 of the original 6 different diagnostic categories. There was no elegant theory of the different psychological mechanisms that distinguished schizophrenia patients from Alzheimer’s patients. There was no elaborately simple experimental task. The brilliance of this work was finding a way to let a rich dataset speak for itself. Georgopoulos summed-up this atheoretical approach to me over dinner: “You know there is gold in America, but all you have is a map of the United States. The computer algorithm is there to show you where the gold is” (A.P. Georgopoulos, October 2007, personal communication).

148

MEG and evolutionary algorithms are not the only arrows in the machine learning quiver. The work of Vincent Calhoun and colleagues31,32 at the MIND Institute in New Mexico has shown that theory-poor, computer-intensive analyses can be also be applied to Functional magnetic resonance imaging data to diagnose schizophrenia. Whereas Georgopoulos’ group utilized a technique with high temporal resolution and calculated correlations in the brain across time, Calhoun’s work used the spatial-informative data from MRI and relied on independent components analysis (ICA). In independent components analysis, there is no independent variable, per se. Instead, it analyzes the relationships between the different voxels of the brain over time. Thus, ICA results in brain maps that show the extent to which different regions share patterns of activity or inactivity. In this case, Calhoun evaluated two such maps that emerged when 21 schizophrenia patients, 14 bipolar patients and 26 healthy controls were asked to listen to a string of beeps. The two maps of interest here were the default mode and temporal lobe networks from all of the participants. These networks are made-up of a number of diverse regions linked by common activity. The default mode network was typified by activity in the posterior anterior cingulate, precuneus and frontopolar cortex. The temporal lobe network was characterized by activity in the inferior temporal lobe and temporal poles. The observation of these networks, it might be noted, has little to do with the task chosen and could be extracted when the brain performed any number of tasks. Calhoun then used these networks in a leave-one-out statistical procedure, in which he allowed the computer to distinguish between the default and temporal networks of the different diagnostic groups. (Thus the algorithm also learned which within-group differences in maps were irrelevant to classification.) The resulting classification algorithm had 90% sensitivity and an average specificity of 95%. Controls were accurately classified 95%, schizophrenia patients 92%, and bipolar patients 83% of the time. As a point of reference, this machine-based discrimination compares favorably to the agreement between OPCRIT diagnosis of schizophrenia and that of an experienced psychiatrist (93% sensitivity and 62% specificity33), and to emergency room and discharge diagnoses (88% sensitivity and 85% specificity34). Data-driven approaches to understanding the brain and psychopathology are now being published regularly. These two studies illustrate a central feature of this informatics approach: there is little need for an

A.W. MacDonald, III

intermediate theory of brain mechanisms to detect and classify patients’ pathophysiology. In the absence of a theoretical basis, there is no need for elegant behavioral paradigms that characterize the sin qua non of cognitive experimental psychopathology.

Summarizing the Case Against Experimental Psychopathology Thus the case against cognitive experimental psychopathology can be summed up as follows. Past theories of schizophrenia were psychological, but modern explanations are multi-determined and neuronal in nature. Therefore, cognitive experimental psychopathology, which explains moment-by-moment thoughts and feelings, is no longer directly addressing the disorder at the level of primary importance. In addition to being off target in this regard, this approach focuses on narrow cognitive mechanisms, whereas in the case of schizophrenia the generalized deficit is most prominent and most in need of a solution. Still, cognitive experimental psychopathology might be useful if it contributed to understanding the biological roots of the disorder. But as several examples suggest, the approach has not proved useful in falsifying or distinguishing between different neurotransmitter theories, nor has it helped in the effort to identify liability genes. Finally, the approach could be forgiven if it were the only game in town. But, as it happens, atheoretical approaches that are better able to take advantage of rich sources of data are beginning to show their strengths. I sincerely hope you will finish reading this chapter, however. In the next section I will undertake to address these concerns, and to highlight ways in which a modern version of cognitive experimental psychopathology might emerge from today’s challenges.

Answering the Accusations: Updating Cognitive Experimental Psychopathology for the Modern Era of Neuroscience and Genomics In this final section, I will answer the objections to cognitive experimental psychopathology in turn. I do not anticipate that these answers will be complete.

9

Is More Cognitive Experimental Psychopathology of Schizophrenia Really Necessary?

They are instead designed to point the way toward answering the accusations outlined above.

Offering a Systematic Approach to the Generalized Deficit in Schizophrenia The generalized deficit has either (1) many independent causes, or (2) a single (or a few) common causes. These independent or common causes may be most perspicuously described at the cognitive or neurobiological level of analysis. For example, it may be that the generalized deficit is largely a function of reduced motivation. This reduction in motivation may have many biological precursors that express themselves in a common psychological process. Alternatively, it may be that the generalized deficit is largely a function of dopamine dysregulation. This may affect a number of different cognitive processes, including motivation, attentional control and perceptual organization, which in turn cascade into problems on numerous different tasks in cognitive domains even further a field. These are both examples of the generalized deficit stemming from a single cause. One could also make piecemeal predictions to test whether the deficit stemmed from numerous independent causes. The point is simply that when stripped of its monolithic mystique, the problem of the generalized deficit really appears as a number of ripe apples to be picked. Should they chose, cognitive experimental psychopathologists could be well-positioned to do this picking. From this perspective, cognitive theories of the deficits in schizophrenia are far from a “parochial distraction”. They provide a systematic approach to analyzing the problem. The image of a ripe apple suggests a low hanging fruit, however. This aspect of the metaphor is misleading, because the generalized deficit of schizophrenia is a jumbled mess indeed. The complexity of the problem of untangling the generalized deficit directly contributes to a habit of compartmentalizing the whole issue rather than subjecting it to rigorous analysis. The problem of the generalized deficit is that it is really two problems, a psychometric one and a psychopathological one. They are so well mixed together that it is like an accident between that proverbial apple cart and the muffin man; the muffins and apples look like one big mess. The solution starts with separating the mess into its original sources.

149

Much has been written about the psychometric confound,35–38 so I will keep my description brief. The psychometric confound can occur whenever one asks whether patients are more impaired on one task that on another. It’s clear why this is an important way to think about the problem in the face of the large generalized deficit associated with schizophrenia. The psychometric confound need not always happen, but it can if the tasks being compared have different degrees of discriminating power. Discriminating power is a property of a task that refers to its ability distinguish between two groups. It derives from several test properties. Discriminating power increases linearly with reliability, loglinearly with variance and the number of items, and as an inverted u-shaped curve with accuracy.39 In the face of the psychometric confound, we must be cautious in interpreting conclusions, though written by eminent scientists in the highest quality journals, such as this: “We present a quantitative evaluation of the literature demonstrating that the most severe impairments [in schizophrenia] are apparent in episodic memory and executive control processes, evident on a background of a generalized cognitive deficit”40 (p. 833). I very much endorse this conclusion, but I might be drawn to it for the wrong reasons. In the absence of evidence to the contrary, tasks that evaluate episodic memory and executive control processing may simply tend to have the most discriminating power (optimal difficulty combined with high reliability, variance and length). Of course there are solutions to the psychometric confound that allow one to sort out the generalized deficit. First, the procedures that do not work include using standardized scores, demographic norming, or the standardized residual scores procedure.38 Other solutions that are more promising include matching for psychometric properties,35 the use of disordinal interactions, psychometric functions or other process-oriented approaches,36,37 or the modeling of latent constructs using an item-response theoretical approach.41 Currently, most tasks adapted from cognitive experimental psychology are not amenable to these solutions. The context in which cognitive theories are generated and tasks are developed is rarely attuned to the problem of studying group differences. This kind of psychometric sophistication is a special ingredient to be added by the conscientious psychopathologist. In addition to constructing and piloting tasks that control for differences in discriminating power or allow for the direct estimation of latent constructs using item-response

150

theory, it may eventually be possible to directly estimate the discriminating power of a test (although true-score variance42 alone may not provide a sufficient estimate39). Another means to obtain some leverage over the psychometric confound is to examine cognitive processes that are spared by schizophrenia.43 These strategies are, in effect, ways to improve the interpretability and yield of the cognitive experimental approach. However, the field will remain hobbled in the absence of replication and clinical applications. One way to further facilitate replication and application is to more thoroughly evaluate the psychometric properties of a specific version of a cognitive paradigm with an eye to maximizing its reliability and sensitivity to deficits, while at the same time condensing it into a time-efficient instrument. To the extent that this can be accomplished through a few additional analyses as part of the original study, it is of great service to the discipline. Such tasks then need to circulate for purposes of replication and extension. These efforts may result in a great database of cognitive experiments in patients, but it is likely that clinical neuropsychological tests will still appear to be more sensitive to patients’ deficit, making them more compelling targets for inclusion in, for example, treatment studies.45 One reason for this, alluded to above, is that the interpretability of cognitive experimental paradigms is derived from interaction effects (performance on condition A is relatively worse than condition B in patients). Interaction effects always lose to main effects in the statistical long run. A passive way to mitigate this cruel statistical fact is to provide the raw means and standard deviations for each task condition for future metaanalysts. A more active approach is to compare the effect sizes we derive from a novel task to those derived from more comprehensive reviews of the behavioral literature (e.g.5,40,46). This may seem a psychometrically unsophisticated public relations stunt unrelated to interpretability. But, given proper caveats, there is no shame in providing a clear-headed evaluation of the magnitude of a cognitive deficit relative to a general benchmark. In the race to understand the causes and cure for schizophrenia, no single solution is likely to increase the viability of cognitive experimental psychopathology. These are refinements that, if taken together, might just increase the appeal of this approach founded in cognitive theory. However, making cognitive experimental psychopathology viable and appealing is a poor goal

A.W. MacDonald, III

indeed, compared to the goal of understanding the causes and cure for schizophrenia. Thus, it is not enough to make the approach appealing. It needs to be able to better address the rift between behavior and the biological roots of psychopathology.

Strengthening the Capacity for Behavioral Tasks to Reveal Causes and Falsify Theories As illustrated above, one of the great challenges of the cognitive experimental approach to understanding schizophrenia are the chasms between global descriptors, moment-to-moment cognitive functions, neurotransmission and genes. Global descriptors, including diagnoses and symptom severity, reflect the cumulative weight of many dynamic cognitive and affective processes that may or may not be present at a given moment in time. Cognitive and affective processes are multi-determined and reflect the public expression of a baffling number of synapses and neurons, which are in turn built from numerous proteins guided by DNA. The challenge of linking one level of analysis to another is reflected in the very structure of disciplines in academia and their corresponding methodologies. As theories of schizophrenia have increasingly retreated from being primarily psychological in nature to being primarily biological in nature, the need for translating from behavioral to biological measurements has become increasingly important. Indeed, the theme of this volume speaks to this progression. As the prosecution has argued, it is one thing to find correlations between variables at different levels of analysis, and quite another to use those relationships to constrain theories about the causes of a disorder. This reflects on the health of cognitive experimental psychopathology only to the extent that the field is unable to raise the bar on what translation between levels of analysis means. For example, it is increasingly common for tasks used in neuroimaging to be designed so that they can be used in neural recording studies in non-human primates and, more rarely, rats and even mice. Such closer connections open important avenues not only for testing the effects of different neurotransmitter agonists and antagonists, but also for early phases of drug development. It is nearly impossible for me to conceive of a neuroscientist approaching this task without a strong sense of how to interpret any results in light of theory.

9

Is More Cognitive Experimental Psychopathology of Schizophrenia Really Necessary?

While this is an important avenue for exploring the correspondence between human and non-human brains, there has been very little work in any species comparing the impact of agonists and antagonists of different neurotransmitter systems. If the glutamate antagonist ketamine mimics the symptoms of schizophrenia47 and disrupts performance on an attentional control task,27 how do these deficits compare to the deficits observed with dopamine agonists (for reviews, see48,49). A question closely linked to the above is the diagnostic specificity of specific cognitive deficits. Diagnostic specificity has to be considered in the light of the greater generalized deficits in patients with schizophrenia. For example, demonstrating that bipolar patients have half as much of an impairment on a given condition does not advance our understanding of the disorder. In contrast, showing the bipolar patients have a deficit on condition A, whereas schizophrenia patients have a deficit on condition B is informative. Searching for the later pattern of results is the kind of challenge that cognitive experimental psychopathologists are especially qualified to search for. Finally, some preliminary reviews suggests that behavioral phenotypes have not facilitated the search for genes associated with schizophrenia.29 However, the tasks reviewed (the Wisconsin Sorting Test and the N-Back Continuous Performance Task) target a number of complicated brain systems. One might instead ask if specifically targeting a cognitive process increases the effect size associated with gene-behavioral relationships. That is, are more precisely measured cognitive processes more closely associated with given genetic polymorphisms? I know of only one study that has undertaken the question in manner.50 In this case, the COMT polymorphism was genotyped in about 450 healthy Caucasians who’d been tested on the N-back task, the AX task and a variant of the AX task called the dot pattern expectancy (DPX) task. The DPX task had been optimized to be sensitive to deficits in patients’ healthy relatives and in the general population.51 The results of this study showed that, despite the fact the N-back showed the greatest difference between patients and controls, individuals with the genotype that conveyed a slightly elevated risk for schizophrenia, performed no differently from individuals with other genotypes. On the AX task, which showed a slightly smaller patient-control difference, genotype was also irrelevant, perhaps due to ceiling effects. Instead, it was the combination of sensitivity to a specific deficit and avoidance

151

of ceiling effects afforded by the DPX task that demonstrated impaired attentional control was associated with the val/val genotype. Such findings must be considered preliminary indicators. More evidence about the utility of measuring specific deficits and associating them with genotype is needed to determine whether this is an odd instance or a general property of psychometrically-optimized cognitive experimental tasks.

Adding Value to Brute Force Empiricism There is no defense against the progress of brute force empiricism. It does not benefit the cognitive experimentalist to deny the impressive advances that can be wrought by combining a mountain of data (generally of a biological nature) with sophisticated statistical tools that provide a level of prediction capable of diagnosis and “mind-reading”. But in the face of such advances, how can it be useful to cling to the elegant paradigms of cognitive experimental psychology and their supposed relevance to theory? There are two strategies that cognitive experimentalists can use to build upon these advances. The first is to integrate these findings into theories of psychopathology. First example, both studies used to illustrate the advance of brute force empiricism suggest that, irrespective of what they are thinking about, patients with schizophrenia have a reduced level of functional connectivity relative to the general population.30,32 These observations are directly relevant to theories of the pathophysiology and symptoms of schiozphrenia.52,53 The findings from these studies do not, however, offer hard proof and are difficult to wield in such a way as to falsify competing ideas. Thus another way to build upon these advances is to incorporate them in to experiments. One first-generation example of this approach is to study patients’ brains in different states. This is being done by my graduate student Kristen Haut in collaboration with Vincent Calhoun. For her dissertation Haut is asking whether patients are more distinct from controls when engaging a specific process (in this case working memory), or at rest (that is, when the brain’s default network is active). Do patients’ impairments in functions like working memory provide important leverage for understanding the biological basis of the illness, or are these impairments an epiphenomenal manifestation of an underlying

152

problem that exists even when the brain is at rest? A difference in the capacity of brain activity to discriminate between the two groups, one way or the other, may have important implications for understanding schizophrenia. Another technique to be learned from these advances is that predicting an individual’s status (diagnosis or subsequent behavior) can be far more compelling that the rejection of a null hypothesis. Broader application of such approaches must certainly increase the pace of scientific progress. In this case, there is not a choice between brute force empiricism and theory-based approaches. The informatics horses are leaving the barn. The question is simply whether theoreticians can harness these methods to probe more deeply into the roots of psychopathology.

Investigating Complex Pathology in an Era of Informatics The preeminent German psychologist Hans Eysenck, who worked for most of his career in Britain, spent a great deal of thought and ink railing against the encroachment of meta-analysis into the psychological sciences. He recalled Newton, who wrote in a letter in 1676, “It is not number of experiments, but weight to be regarded; where one will do, what need of many?”54 This is a very compelling argument. Cognitive experimental psychology leans toward this Eysenckian and Newtonian model wherein a single experiment attempts to be definitive. But in an era where data proliferate, no single study of a complicated phenomenon like a mental disease can be considered definitive. To paraphrase Newton, it is not the number of experiments but the size, replicability and interpretability of the effects that lend weight. Or, to quote the seamanlike wisdom of a map on my wall, “The prudent mariner will not rely on any single aid to navigation.” To this end, there is now a greater emphasis on replicating findings and extending, rather than replacing, tasks to ask new questions. This within-laboratory replication is an important aspect of building the weight of evidence. Even more importantly, the last several years have seen a move toward a greater cooperative spirit among psychopathologists. This has facilitated an increase in cross-laboratory replication, and occasionally, failures to replicate. In addition, this collaborative spirit allows studies of the relationships between tasks

A.W. MacDonald, III

driven by one’s own theories and tasks driven by others’ theories. One aspect of this cross-fertilization is the increasing ability of investigators to test competing theories more rigorously. Ultimately, support for a theory does not come from rejecting the null hypothesis, but from testing its predictions against other interesting predictions. Finally, in an era of informatics, it is incumbent upon cognitive experimentalists to regularly review and meta-analyze their own field. Reviews are highly cited papers that play an important role in shaping how people think about the field. The synthesis of findings from cognitive experimental psychopathology is a tricky endeavor that requires careful consideration. I can do no better than provide one recent example,44 although there need to be more.

Conclusion My reflections on experimental cognitive psychopathology have not led me to abandon it; nor should they have this effect on students who are choosing what methods to embrace for their careers. The advances of biological and statistical methods, the worthy strengths of neuropsychological tests, should be a call-to-arms rather than a call to quit the field. However, this is a time of rapid change and the jury is still out. Either the coming decade will see the field learn from the competing approaches and incorporate their strengths into the kinds of theory-guided questions that we can ask. Or the kind of theory-guided approach to understanding behavior – brain disease relationships will whither. Let the prudent mariner be your guide, and rely on no single measure to find your way. Acknowledgements This work was supported by grants from National Alliance for Research in Schizophrenia and Affective Disorders and NIMH grants 60662, 66629, 69675 and 79262. Thanks to Deanna Barch, James Gold, Steven Silverstein and Michael Pogue-Geile for discussions and feedback.

References 1. MacDonald AW, III, Carter CS. Cognitive experimental approaches to investigating impaired cognition in schizophrenia: a paradigm shift. Journal of Clinical and Experimental Neuropsychology. 2002;24:873–882.

9

Is More Cognitive Experimental Psychopathology of Schizophrenia Really Necessary?

2. Lezak MD. Neuropsychological Assessment. 3rd ed. New York: Oxford University Press; 1995. 3. Moldin S. Indicators of liability to schizophrenia: perspectives from genetic epidemiology. Schizophrenia Bulletin. 1994;20(1):169–184. 4. Gottesman, II, Gould TD. The endophenotype concept in psychiatry: etymology and strategic intentions. The American Journal of Psychiatry. Apr 2003;160(4):636–645. 5. Heinrichs RW, Zakzanis KK. Neurocognitive deficit in schizophrenia: a quantitative review of the evidence. Neuropsychology. 1998;12:426–445. 6. Karlsson JL. Heterozygous inheritance of schizophrenia. Hereditas. 1987;107:59–64. 7. Luwig AW. The Price of Greatness: Resolving the Creativity and Madness Controversy. New York: Guilford; 1996. 8. Heinrichs RW. The primacy of cognition in schizophrenia. American Psychologist. 2005;60(3):229–242. 9. Woodberry KA, Giuliano AJ, Seidman LJ. Premorbid IQ in schiziophrenia: a meta-analytic review. The American Journal of Psychiatry. 2008;165(5):579–587. 10. Snitz BE, MacDonald AW, III, Carter CS. Cognitive deficits in unaffected first-degree relatives of schizophrenia patients: a meta-analytic review of putative endophenotypes. Schizophrenia Bulletin. 2006;32(1):179–194. 11. Chapman LJ, Chapman JP. Disordered Thought in Schizophrenia. Englewood Cliffs, NJ: Prentice-Hall; 1973. 12. Barch DM, Carter CS, Cohen JD. Factors influencing Stroop performance in schizophrenia. Neuropsychology. 2004;18(3):477–484. 13. Cohen JD, Servan-Schreiber D. Context, cortex and dopamine: a connectionist approach to behavior and biology in schizophrenia. Psychological Review. 1992 1992;99(1):45–77. 14. Servan-Schreiber D, Cohen J, Steingard S. Schizophrenic deficits in the processing of context: a test of a theoretical model. Archives of General Psychiatry. 1996 1996; 53:1105–1112. 15. MacDonald AW, III. Building a clinically relevant cognitive task: case study of the AX paradigm. Schizophrenia Bulletin. 2008:doi:10.1093/schbul/sbn1038. 16. Javitt DC, Jotkowitz A, Sircar R, Zukin SR. Non-competitive regulation of phencyclidine/sigma-receptors by the N-methyl-D-aspartate receptor antagonist D-(-)-2-amino-5phosphonovaleric acid. Neuroscience Letters. July 22 1987;78(2):193–198. 17. Javitt DC, Zukin SR. Recent advances in the phencyclidine model of schizophrenia. The American Journal of Psychiatry. 1991;148:1301–1308. 18. Moghaddam B. Bringing order to the glutamate chaos in schizophrenia. Neuron. 2003;40:881–884. 19. Place EJS, Gilmore GC. Perceptual organization in schizophrenia. Journal of Abnormal Psychology. 1980;89:409–418. 20. MacDonald AW, III, Carter CS, Kerns JG, et al. Specificity of prefrontal dysfunction and context processing deficts to schizophrenia in a never medicated first-episode psychotic sample. The American Journal of Psychiatry. 2005; 162:475–484. 21. Barch D, Sheline YI, Csernansky J, Snyder A. Working memory and prefrontal cortex dysfunction: specificity to schizophrenia compared with major depression. Biological Psychiatry. 2003(53):376–384.

153

22. Silverstein S, Berten S, Essex B, Kovacs I. An fMRI examination of visual integration in schizophrenia (submitted). 23. Cohen JD, Barch DM, Carter CS, Servan-Schreiber D. Context-processing deficits in schizophrenia: converging evidence from three theoretically motivated cognitive tasks. Journal of Abnormal Psychology. 1999 1999;108(1): 120–133. 24. Cardno AG, Sham PC, Murray RM, McGuffin P. Twin study of symptom dimensions in psychosis. British Journal of Psychiatry. 2001;179:39–45. 25. Silverstein SM, Kovacs I, Corry R, Valone C. Perceptual organization, the disorganizatino syndrome and context processing in chronic schizophrenia. Schizophrenia Research. 2000;43(1):11–20. 26. Uhlhaas PJ, Phillips WA, Silverstein SM. The course and clinical correlates of dysfunctions in visual perceptual organization in schizophrenia during the remission of psychotic symptoms. Schizophrenia Research. June 15 2005;75(2–3):183–192. 27. Umbricht D, Schmid L, Koller R, Vollenweider FX, Hell D, Javitt DC. Ketamine-induced deficits in auditory and visual context-dependent processing in healthy volunteers. Archives of General Psychiatry. 2000;57:1139–1147. 28. Gottesman II, Shields J. Schizophrenia and Genetics: A Twin Study Vantage Point. New York: Academic; 1972. 29. Flint J, Munafo MR. The endophenotype concept in psychiatric genetics. Psychological Medicine. 2007;37:163–180. 30. Georgopoulos AP, Elissaios K, Arthur CL, Scott ML, Joshua KL, Aurelio AA. Synchronous neural interactions assessed by magnetoencephalography: a functional biomarker for brain disorders. Journal of Neural Engineering. 2007;4:349. 31. Calhoun VD, Adali T, Giuliani NR, Pekar JJ, Kiehl KA, Pearlson GD. Method for multimodal analysis of independent source differences in schizophrenia: combining gray matter structural and auditory oddball functional data. Human Brain Mapping. 2006;27:47–62. 32. Calhoun VD, Maciejewski PK, Pearlson GD, Kiehl KA. Temporal lobe and ‘ ‘default’ ’ hemodynamic brain modes discriminate between schizophrenia and bipolar disorder. Human Brain Mapping. 2008;29:1265–1275. 33. Jakobsen KD, Frederiksen JN, Hansen T, Jansson LB, Parnas J, Werge T. Reliability of clinical ICD-10 schizophrenia diagnoses. Nordic Journal of Psychiatry. 2005;59:209–212. 34. Woo KP, Sevilla CC, Obrocea GV. Factors influencing the stability of psychiatric diagnoses in the emergency setting: review of 934 consecutively inpatient admissions. General Hospital Psychiatry. 2006;28:434–436. 35. Chapman LJ, Chapman JP. Problems in the measurement of cognitive deficit. Psychological Bulletin. 1973;79:380–385. 36. Strauss ME. Demonstrating specific cognitive deficits: a psychometric perspective. Journal of Abnormal Psychology. 2001;110(1):6–14. 37. Knight RA, Silverstein SM. A process-oriented approach for averting confounds resulting from general performance deficiencies in schizophrenia. Journal of Abnormal Psychology. 2001;110:15–30. 38. MacDonald AW, III, Kang SS. Cassandra’s calculations: simulation studies of the psychometric confound. In: French

154

39.

40.

41.

42. 43.

44. 45.

46.

A.W. MacDonald, III DP, ed. Schizophrenia Psychology: New Research. Hauppauge, NY: Nova Science; 2006, pp. 281–301. Kang SS, MacDonald AW, III. Measuring discriminating power: beyond true-score variance. In: Degregorio RA, ed. Psychological Tests and Testing. New York: Nova; 2007, pp. 91–117. Reichenberg A, Harvey PD. Neuropsychological impairments in schizophrenia: integration of performance-based and brain imaging findings. Psychological Bulletin. 2007;133(5):833–858. Coleman MJ, Cook S, Matthysse S, et al. Spatial and object working memory impairments in schizophrenia patients: a bayesian item-response theory analysis. Journal of Abnormal Psychology. 2002;111(3):425–435. Chapman LJ, Chapman JP. The measurement of differential deficit. Journal of Psychiatric Research. 1978;14:303–311. Gold JM, Fuller RL, Robinson BM, McMahon RP, Braun EL, Luck SJ. Intact attentional control of working memory encoding in schizophrenia. Journal of Abnormal Psychology. 2006;115(4):658–673. Luck SJ, Gold JM. The Construct of Attention in Schizophrenia. Biological Psychiatry. 2008;64:34–39. Nuechterlein KH, Green MF, Kern RS, et al. The MATRICS consensus cognitive battery: Part 1: test selection, reliability, and validity. The American Journal of Psychiatry. 2008;165:203–213. Dickinson D, Ramsey MB, Gold JM. Overlooking the obvious: a meta-analytic comparison of digit symbol coding tasks and other cognitive measures in schizophrenia. Archives of General Psychiatry. 2007;64:532–542.

47. Javitt D, Zukin S. Recent advances in the phencyclidine model of schizophrenia. The American Journal of Psychiatry. 1991;148(10):1301–1308. 48. Kapur S, Remington G. Dopamine D2 receptors and their role in atypical antipsychotic action: still necessary and may even be sufficient. Biological Psychiatry. 2001; 50:873–883. 49. Laviolette SR. Dopamine modulation of emotional processing in cortical and subcortical neural circuits: evidence for a final common pathway in schizophrenia? Schizophrenia Bulletin. 2007:doi:10.1093/schbul/sbm1048. 50. MacDonald AW, III, Carter CS, Flory JD, Ferrell RE, Manuck SB. COMT Val158Met and executive control: a test of the benefit of specific deficits to translational research. Journal of Abnormal Psychology. 2007;116(306–312). 51. MacDonald AW, III, Goghari VM, Hicks BM, Flory JD, Carter CS, Manuck SB. A convergent-divergent approach to context processing, general intellectual functioning and the genetic liability to schizophrenia. Neuropsychology. 2005;19:814–821. 52. Frith CD, Blakemore S, Wolpert DM. Explaining the symptoms of schizophrenia: abnormalities in the awareness of action. Brain Research – Brain Research Reviews. 2000;31(2–3):357–363. 53. Phillips WA, Silverstein SM. Convergence of biological and psychological perspectives on cognitive coordination in schizophrenia. Behavioral and Brain Sciences. 2003;26:65–138. 54. Eysenck HJ. Meta-analysis and its problems. British Medical Journal. 1994;309:789–792.

Chapter 10

Intellectual Functioning as an Endophenotype for Schizophrenia Odette de Wilde

Abstract Different studies have shown that abnormal cognitive functioning is present in patients with schizophrenia even in the early phase of the disease. Cognitive deficits in patients with schizophrenia are characterised by a generalised intellectual deficit, coupled with abnormalities in specific neuropsychological domains such as working memory, attention and executive functioning. Intellectual deficits have been suggested to precede psychosis in patients with schizophrenia and also to be a risk factor for developing schizophrenia. If the presence of intellectual deficits would coincide with increased risk to develop schizophrenia they should be present in excess of general population levels in the biological relatives of individuals with schizophrenia. Studies investigating intellectual functioning in first-degree relatives of patients with schizophrenia indeed have shown deficits in intellectual functioning paralleling the degree of genetic liability in that performance of the relatives fell between the performance of the patients and the controls. In this chapter recent results in literature investigating the evidence, that a generalized intellectual deficit coupled with deficits in specific domains is also present in first-degree relatives of patients with schizophrenia, is reviewed and discussed in the light of its usefulness in identifying schizophrenia susceptibility genes. Keywords Schizophrenia • endophenotype • intelligence

O. de Wilde AMC de Meren, Amsterdam, The Netherlands

Abbreviations GE: Gene-environment; IQ: Intelligence Quotient; Gf: Fluid intelligence; Gc: Crystallized intelligence; KAIT: The Kaufman Adolescent and Adult Intelli-gence Test; WAIS: Wechsler Adult Intelligence Scale; RAPM: The Raven’s Advanced Progressive Matrices; NART: National Adult Reading Test; WASI: Wechsler Abbreviated Scale of Intelligence; AFQT: Armed Forces Qualification Test; WCST: Wisconsin Card Sorting Test; NIMH: National Institute of Mental Health; MATRICS: Measurement and Treatment Research to Improve Cognition in Schizophrenia; MZ: Monozygotic; DZ: Dizygotic; COMT: Catechol-omethyltransferase; DTNBP1: Dystrobrevinbinding protein 1; DISC1: Disrupted-in-Schizophrenia-1; NRG1: Neuregulin 1

Introduction Schizophrenia is a complex disorder which is not inherited because of a deficit in a single gene, but in fact is thought to be polygenic, i.e. is determined by multiple genes’ interaction. In addition to genetic effects, environmental factors presumably also play a role in the development of schizophrenia. Since environmental and genetic effects on the development of schizophrenia may not always be independent of each other, understanding the specific causal pathways of gene action is crucial when evaluating any genetic account of schizophrenia. Gene effects may be direct, and depend on environmental factors (gene-environment interaction, GE), and they may act indirectly through correlated environments (GE correlation) i.e. people susceptible to developing schizophrenia might create or evoke situations that further enhance their schizophrenia proneness

M.S. Ritsner (ed.), The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes, © Springer Science + Business Media B.V. 2009

155

156

(active and reactive GE correlation respectively).1 If environments are not randomly assigned to each individual but are, partly, selected by the individual based on genetically influenced preferences (GE autocorrelation), it becomes merely impossible to discern which genetic effects are directly responsible for schizophrenia susceptibility and which result from the action of environmental causally linked to genetic differences.1 Because the environment can determine the impact of genetic variation (GE interaction), high schizophrenia heritability does not necessarily lead to the development of schizophrenia in an individual. Finally, one must bear in mind that heritability within a group does not imply that group differences are due to genetic factors. Environmental factors could completely explain group differences, even when genetic factors completely explain within-group differences.1 One way to unravel the genetic architecture of schizophrenia is to employ the endophenotype strategy as outlined in the first chapter of this book. To qualify as a schizophrenia endophenotype any measured physiological, biochemical, neuroanatomical, molecular, neuropsychological, or perceptual internal marker should be heritable, co-segregate with schizophrenia, yet be present even when the disease is not (i.e. state-independent),

Fig. 10.1 Watershed model of the pathway between upstream genes and downstream phenotypes. Specific genes (1a, 1b) contribute variation to narrowly defined endophenotypes such as dopaminergic regulation in the prefrontal cortex (2b). This and other narrowly defined endophenotypes affect more broadly defined endophenotypes, such as working memory (3c). Working memory in conjunction with several other endophenotypes (3a, 3b, 3d) affects phenotypically observable phenotypes, such as symptoms of schizophrenia (4) (Reprinted with permission from the Annual Review of Clinical Psychology, Volume 2 (c)2006 by Annual Reviews www. annualreviews.org; adopted from Cannon and Keller3)

O. de Wilde

and be found in non-affected family members at a higher rate then in the normal population.2 The endophenotype strategy is proposed to investigate the relationship between genes and preclinical intermediate phentoypes not visually observable. These endophenotypes then are hypothesized to reflect the effects of a specific gene more directly then the illness itself. This however may not be through a causal route from phenotype to gene. Cannon and Keller3 propose a watershed model between ‘upstream’ genes and ‘downstream’ phenotypes (see Fig. 10.1). They hypothesized specific genes, for example COMT, to contribute variation to narrowly defined endophenotypes such as dopaminergic regulation in the prefrontal cortex. This and other narrowly defined endophenotypes would then affect more broadly defined endophenotypes, such as working memory. Further ‘downstream’ working memory would in conjunction with several other endophenotypes affect phenotypically observed phenotypes, such as symptoms of schizophrenia.3 Different studies4,5 have shown that abnormal cognitive functioning is present in patients with schizophrenia even in the early phase of the disease. Cognitive deficits in patients with schizophrenia are characterised by a generalised intellectual deficit, coupled with

10

Intellectual Functioning as an Endophenotype for Schizophrenia

abnormalities in specific neuropsychological domains such as working memory, attention and executive functioning.4,5 Weickert and Goldberg6 have suggested cognitive deficits to develop along three different pathways based on the timing, and degree of impairment of general intellectual capacities. The first course they propose suggests the disease process to become manifest as a widespread cognitive impairment that may be relatively profound at an early stage and is present subsequent to the development of positive symptoms of schizophrenia. This group of patients, which is premorbidly comprised, may have encountered early developmental stressors and/or a genetic predisposition which have let to the observed cognitive deficits. The second suggested pathway assumes the cognitive deficits to become apparent concurrent to the onset of psychotic symptoms, which results in an encapsulated pattern of deficits that encompasses the cognitive domains of executive functioning, attention and longterm memory. Weickert and Goldberg6 state that this second pathway is self-limiting since patients do not go on to develop a full-blown dementia. The cognitive deficits however do occur in the presence of intellectual decline. The third course they describe proposes cognitive deficits to be concurrent with the onset of psychotic symptoms, however the invalidating cognitive deficits associated with the ongoing disease process may be relatively subtle and are restricted to the domains of executive functioning/working memory and attention. In this last group it remains unclear whether the cognitive deficits arise concurrent to the positive symptoms or are present before disease onset. Reichenberg et al.7 showed that healthy adolescents with within-normal-range IQ scores who later in life developed schizophrenia already had intellectual decline years before the disease onset. In a recent meta-analysis this finding was supported since intellectual deficits were found to precede psychosis in patients with schizophrenia years before disease onset, and were found to be approximately one-half of a standard deviation below that of healthy comparison subjects.8 This suggests deficits in IQ scores to be a risk factor for developing schizophrenia. If the presence of intellectual deficits coincides with increased risk to develop schizophrenia then they should be present in excess of general population levels in the biological relatives of individuals with schizophrenia. Studies investigating intellectual functioning in first-degree

157

relatives of patients with schizophrenia should then show deficits in intellectual functioning parallel to the degree of genetic liability in that performance of the relatives falls between the performance of the patients and the controls. In this chapter recent results in literature investigating the evidence, that a generalized intellectual deficit combined with deficits in specific domains is also present in first-degree relatives of patients with schizophrenia, is reviewed and discussed in the light of its usefulness in identifying schizophrenia susceptibility genes.

The Definition and Measurement of Intelligence Various and different concepts of intelligence have been proposed over the years. A task force convened by the American Psychological Association9 agreed on the following description of intelligence: ‘… individuals differ from one another in their ability to understand complex ideas, to adapt effectively to the environment, to learn from experience, to engage in various forms of reasoning, to overcome obstacles by taking thought. Although these individual differences can be substantial, they are never entirely consistent: a given person’s intellectual performance will vary on different occasions, in different domains, as judged by different criteria. Concepts of ‘intelligence’ are attempts to clarify and organize this complex set of phenomena.’ Different expressions or concepts of intelligence have indeed been hypothesized. Spearman’s ‘g’, a measure for general cognitive ability which can be calculated by factor analyzing scores on various tests, is thought to be made up of a small number of (interrelated) subfactors that represent more specific abilities such as working memory.10 Fluid intelligence (Gf) refers to reasoning and novel problem-solving ability1 and is distinct from Crystallized intelligence (Gc), which refers to overlearned skills and static knowledge such as vocabulary. Based on these different concepts, many different tests have been developed to measure intelligence. The Kaufman Adolescent and Adult Intelligence Test (KAIT)11 for instance, specifically aims to measure Gf and Gc. The results of the Wechsler Adult Intelligence Scale (WAIS) know in its third edition,12,13 however are

158

expressed in intelligence quotients (IQ) and are thought to represent Gc. The Raven’s Advanced Progressive Matrices (RAPM)14 is a non-verbal intelligence test measuring Gf, whereas the National Adult Reading test (NART)15 is a verbal intelligence test mostly used to measure the premorbid level of intellectual functioning and Gc. Individuals differ distinctly in performance on these various measures, but individuals who tend to do well on one test, tend to do well on many other tests as well, suggesting a common underlying factor for all different measures of intelligence. Most international studies on intelligence in general and on intelligence in schizophrenia specifically, have used the WAIS in its different editions. However it is necessary that other intelligence tests will be used to compare the different test to each other and reach a more clear picture of the criterion, construct and concurrent validity of the different tests not in the last place with respect to usefulness as possible endophenotypes.

Intellectual Heritability It is a commonly excepted idea that intelligence is heritable. De Geus and Boomsma16 stated that in their longitudinal sample of 209 Dutch twin pairs they found a intellectual heritability of 80% in adults. They stress that a heritability of 80% does not mean that environmental factors (GE interaction) do not play any role in level of intelligence but it means that the variation between individuals in the examined population is due to genetic variation, for example, because environmental conditions and the educational system are above the minimum level required to maximize genetic potential. By decomposing the variance in IQ measures by structural equation modeling into genetic, shared and non-shared environmental components they confirmed and extended their previous finding showing that verbal and performance IQ are highly heritable, 85% and 69%, respectively.17 De Geus and Boomsma16 state processes of working memory, inhibitory control and attention to be highly plausible sources of individual differences in cognitive ability. Although cognitive ability (or IQ) in itself is highly heritable, it is likely to be influenced by a number of genes of small effect. These genes might also be related to the above named sources and are

O. de Wilde

therefore more easily uncovered by focusing on elementary aspects of cognition, such as processing speed or resistance to interference. So in itself intellectual ability might need an endophenotype approach to be uncovered genetically.

Intellectual Deficits in Relatives A few studies to date have explicitly investigated intellectual deficits in relatives of patients with schizophrenia. Using a short form of the WAIS-R Kravarity et al.18 tested intellectual asymmetry and found a relative superiority for verbal skills over spatials skills to be present in non-affected relatives of patients with schizophrenia in a degree that was proportional to the individuals genetic susceptibility for developing schizophrenia. They suggested intellectual asymmetry, with a relative superiority of verbal skills over spatial skills to represent a probable endophenotype of schizophrenia. McIntosh et al.19 used the Wechsler Abbreviated Scale of Intelligence (WASI) and found verbal and premorbid IQ to be impaired in patients with schizophrenia and their relatives. To a lesser extend they also found relatives of patients with schizophrenia to have lower performance IQ’s. They found verbal IQ to be related to genetic liability for schizophrenia. Performance IQ however was found to be inversely related to genetic liability. The positive association they found between IQ and genetic liability was suggested to provide evidence that genes for schizophrenia may convey an advantage for unaffected family members. Toulopoulou et al.20 investigated performance on the WAIS-III and its indexes in a large sample of 267 monozygotic (MZ) and dizygotic (DZ) twins concordant and disconcordant for schizophrenia and healthy MZ and DZ twins. They found patients with schizophrenia and their unaffected co-twins to perform significantly worse when compared to healthy controls. The MZ sample performed worse on all measures analysed while the DZ sample only performed worse on a single measure: perceptual organization. Toulopoulou et al.20 found full scale IQ as well as the working memory index to have the highest proportion of interindividual differences associated with genetic effects and also to share the largest genetic variance with schizophrenia. Perceptual organization was found

10

Intellectual Functioning as an Endophenotype for Schizophrenia

to be moderately related to schizophrenia whereas verbal comprehension and processing speed were found to have the weakest genetic link to schizophrenia. Kremen et al.21 used the Armed Forces Qualification Test (AFQT) administered at the time of entrance into the services as an indication of ‘g’ or general cognitive ability. They did not find a significant difference between non-schizofrenic co-twins when compared to normal controls although the mean AFQT score was found to be almost midway between control twins and probands and was characterised as a mild cognitive deficit. Since the healthy co-twins were found to be midway of controls and schizophrenic probands this might suggest general cognitive ability as measured with the AFQT may still be a possible endophenotype of schizophrenia. In the study of de Wilde et al.5 97 patients and 34 young full siblings were compared to 36 healthy controls on WAIS-III performance. We found the performance of healthy siblings of patients with schizophrenia to be worse than that of education-adjusted unrelated healthy controls. The pattern of deficits was similar to that of the patients except for the subtests picture completion and picture arrangement and the processing speed index. The degree of intellectual performance deficits in siblings was found to parallel the degree of genetic liability in that performance of the relatives fell between the performance of patients and controls. Based on the findings we suggested that there is clear evidence of shared genetic and/or environmental factors on IQ variation with the exception of processing speed. Based on siblings’ results we suggested complementary visual perception and verbal mediation as a way of solving a problem could be seen as possible protective factors when on a normal level, but when lowered they could also be indicative for heightened risk for developing schizophrenia. The findings of McIntosh et al.19 and Toulopoulou et al.20 might be influenced by the control sample having an above mean IQ of nearly 1 sd above the general population mean (mean = 100, sd = 15). This limitation however does not obviate the fact that some relatives with schizophrenia show obvious cognitive impairments without suffering from schizophrenia symptoms. Next to studies focussing on intellectual functioning in relatives of patients with schizohrenia, many studies on neuro-cognitive deficits in relatives of patients with schizophrenia also report on IQ-measures. Cannon et al.22 for instance used a short form of the WAIS-R (Vocabulary, Similarities, Block Design, and Digit Symbol subtests) and found significantly lower IQ

159

scores in MZ as well as DZ co-twins of patients with schizophrenia. Barrantes-Vidal et al.23 used the subtests Block Design and Information from the WAIS-III to estimate intelligence and found healthy siblings of patients with schizophrenia to have a significantly lower estimated IQ when compared to healthy controls. Kremen et al.24 also found a lower mean IQ in relatives of patients with schizophrenia when compared to healthy controls. However in these, and other, studies the main goal was to measure cognitive deficits and therefore the IQ scores may have been comprised by none matched controls when considering years of education and or premorbid educational levels. However one can not ignore the repeatedly shown lower estimated IQ-scores in relatives of patients with schizophrenia when compared to healthy unrelated controls. The study of Toulopoulou et al.20 explicitly shows intellectual functioning and schizophrenia to cosegregate in families. Across the different studies described in this chapter it is clear that intellectual deficits are found in affected family members and in nonaffected family members at a higher rate than in the general population.

The Relationship Between Intellectual and Cognitive Abilities and Cognitive Deficits Since intellectual functioning as measured by most intelligence test is built up from different cognitive tasks. The assumption is made that intellectual functioning is causally related to cognitive functioning: if performance on one of these cognitive tasks is lowered, overall IQ score will also become lower. However, even though the two are related, different cognitive functions are more or less closely related to intelligence. For instance the subtest block design of the WAIS-III is known to be closely related to the ‘g’- factor or overall intelligence.12 The working memory function, on the other hand, is known to be influenced by different factors and might therefore be less related to overall intellectual functioning. In the development of intellectual tests like the NART)15 measuring premorbid level of intellectual functioning, the assumption has been made that verbal reading ability is the least likely to be affected by neurological decline (excluding aphasia and such

160

disorders) and therefore a reliable measure of premorbid intellectual functioning. Many studies have investigated performance on different neurocognitive tasks21–28 in relatives of patients with schizophrenia and found many deficits. Kremen et al.21 found verbal and visual memory differences to remain significant after adjustment for level of intelligence. Egan et al.26 found relatives to perform worse on a test of working memory/executive function (WCST) and occulomotor scanning/psychomotor speed (Trails B) even though they had a matching Fullscale IQ when compared to a control group. Birkett et al.25 used the NART to control for differences in intellectual functioning and still found relatives of patients with schizophrenia to be impaired on the same measures of executive functioning as the investigated patients and suggested this to be consistent with an endophenotype of executive impairment. Krabbendam et al.28 found speed, working memory and episodic memory to be predictive of group membership independent of each other and independent of IQ-level. And these cognitive functions discriminated patients and relatives significantly from controls. Barrantes-Vidal et al.23 found the effect size for IQ differences between healthy non-psychotic relatives and healthy controls to be larger then the effect sizes found for several test measuring working memory, verbal fluency and attention and suggested IQ deficits to be more than the sum of all parts.

Specificity The NIMH Measurement and Treatment Research to Improve Cognition in Schizophrenia (MATRICS) Initiative29,30 has identified several cognitive domains that are dysfunctional in schizophrenia and propose these cognitive domains should be considered in the context of clinical trials. Cognitive impairments in these domains are also seen in bipolar disorder, but they tend to be milder.31 These deficits therefore seem to be a core feature of schizophrenia and to a lesser extend also exist outside of mood episodes in bipolar disorder. Since the deficits are less in patients with mood disorders when compared to patients with schizophrenia one would also expect the relatives of patients with a bipolar disorder to perform better then relatives of patients with schizophrenia. Reichenberg

O. de Wilde

et al.7 indeed only found premorbid intellectual decline to be related to the manifestation of schizophrenia and not to bipolar disorder, depression or any anxiety disorder. McIntosh et al.19 also found patients with bipolar disorder and their relatives not to differ from controls on WAIS-III-IQ while patients with schizophrenia and their relatives did significantly differ. Kremen et al.24 even found relatives of patients with a bipolar disorder to have a higher mean IQ when compared to controls.

Conclusions and Future Directions Gottesman and Gould2 have suggested five criteria for any measure to qualify as an endophenotype (see Chapter 1 by Ritsner and Gottesman in this volume). We will now discuss the different results presented in this chapter using these criteria. As was discussed in the introduction to this chapter many different studies have found intellectual decline in patients with schizophrenia before as well as after disease onset. Intellectual deficits have repeatedly been associated with schizophrenia in the population4,5,7,8 and would therefore meet the first endophenotype criterion.2 Concerning the second criterion, heritability, intelligence itself is found to be heritable and has an heritability of approximately 80% comparable to the estimated 83% of the liability to schizophrenia attributable to additive genetic effects.32 Toulopoulou et al.20 even suggested that the same genes underlying individual differences in intelligence also contribute to the liability of schizophrenia. It has been shown that the intellectual deficits are in part state independent since they have been found present before disease onset, and found to progress during the course of the illness, and therefore would also meet the third criterion for an endophenotype. With respect to the fourth criterion, the studies discussed show that intellectual deficits and schizophrenia co-segregate within families. It is clear from the discussed studies that intellectual deficits are found in affected family members and in nonaffected family members at a higher rate than in the general population, also meeting the last criterion for an endophenotype. Next to these five criteria we also discussed the specificity of intellectual deficits for schizophrenia with respect to other psychotic disorders. Literature

10

Intellectual Functioning as an Endophenotype for Schizophrenia

showed that family members of patients with bipolar disorder did not show intellectual deficits7,19,24 and were also not found in relatives of patients with a major depressive disorder or anxiety disorders.7 This suggests that intellectual deficits are more specific to a vulnerability to schizophrenia then to any other psychiatric psychotic or non-psychotic disorder. As was described in the introduction to this chapter Weickert and Goldberg6 have suggested different pathways for cognitive deficits associated with schizophrenia to emerge. Review of the literature showed intellectual deficits to be present separate from other cognitive deficits such as working memory, processing speed, executive functioning and verbal tasks and vice versa. Intellectual decline was found to be present before disease onset in patients, but also to be present in healthy first-degree relatives in a widespread and relatively profound manner.5,18–20 These findings suggest the first pathway as proposed by Weickert and Goldberg6 to be the most fitting hypothesis in family and genetic studies of intelligence and indeed suggests a genetic predisposition to cognitive deficits and schizophrenia. Intellectual deficits therefore seem to qualify as a strong and important endophenotype for schizophrenia. In itself, as was suggested by de Geus an Boomsma,16 intellectual abilities might also need to be broken down in endophenotypes to be linked to underlying genes. The division in verbal versus performal IQ and/or Gf versus Gc might be important in this respect, but also the indices as present in the WAIS III. With regard to these indices it must be considered that the working memory and processing speed index might be less related to overall intellectual capacities than the verbal comprehension and perceptual orientation index. As was stated before most international studies have used the WAIS in its different editions to measure intellectual functioning. Of course any other measure of intelligence such as the KAIT11 or the RAPM14 could be very useful and might even give way to new insights in genetic studies in schizophrenia. The different cognitive abilities are of course related to the level of intelligence in the sense that a higher level of IQ is related to for instance higher working memory function or more verbal knowledge. However deficits in these cognitive functions seem to be affected independently of intellectual functioning or decline and vice versa.21–28 For endophenotype research it might be important to discover these additive effects. Following the watershed model of Cannon and Keller3

161

it might be suggested that the level of intellectual functioning might be closer to the schizophrenia phenotype then for instance working memory, which might be further ‘down stream’ to another more biological intermediate phenotype and to the gene. There is growing evidence that cognitive functioning in schizophrenia is modulated by genetic variation. The catechol-o-methyltransferase (COMT) gene33 for instance has been found to be related to different measures of working memory functioning. Burdick et al.34 found genetic variation in dystrobrevinbinding protein 1 (DTNBP1 or Dysbindin) to be associated with spatial working memory function and Donnohoe et al.35 found the same gene to be related to general intellectual function. Variations in the DISC 1 gene have been found to be related to structure and function of the hippocampus,36 and a functional variant in the Neuregulin 1 (NRG1) gene has been associated with impaired prefrontal function and decreased IQ.37 Zinkstok et al.38 were the first to investigate the relationship between intellectual functioning in patients with schizophrenia and their relatives and a schizophrenia susceptibility gene and indeed found a covariation between IQ scores and the DTNBP1 gene. We did however not study the (inter)dependence of the subscales and indices of the WAIS-III with this gene. For future research purposes it would be important to investigate the interdependence and or additive effects of intellectual decline and underlying cognitive functions in schizophrenia to come to a better understanding of the genetic architecture of schizophrenia, intellectual functioning and intellectual functioning in schizophrenia. To study this one must keep in mind that the underlying genes might have different effects in different people or subgroups (relatives versus patients) and that different genes might interact with each other or the environment to affect intelligence and schizophrenia. Acknowledgments The author would like to thank Peter M. Dingemans for his critical review of the text.

References 1. Gray JR, Thompson PM. Neurobiology of intelligence: science and ethics. Nat Rev Neurosci 2004; 5: 471–482. 2. Gottesman II, Gould TD. The endophenotype concept in psychiatry: etymology and strategic intentions. Am J Psychiatry 2003; 160: 636–645.

162 3. Cannon TD, Keller MC. Endophenotypes in the genetic analyses of mental disorders. Ann Rev Clin Psychology 2006; 2: 267–290. 4. Bilder RM, Goldman RS, Robinson D, et al. Neuropsychology of first-episode schizophrenia: initial characterization and clinical correlates. Am J Psychiatry 2000; 157: 549–559. 5. de Wilde OM, Dingemans PM, Bour LJ, et al. Generalized intellectual deficit as an endophenotype in young patients with recent onset schizophrenia and unaffected young siblings 2008 (in press). 6. Weickert TW, Goldberg TE. The course of cognitive deficits in schizophrenia. In: Sharma T, Harvey P, eds. Cognition in Schizophrenia; Impairments, Importance and Treatment Strategies. New York: Oxford University Press, 2000, pp. 3–15. 7. Reichenberg A, Weiser M, Rapp MA, et al. Elaboration on premorbid intellectual performance in schizophrenia; premorbid intellectual decline and risk for schizophrenia. Arch Gen Psychiatry 2005; 62: 1297–1304. 8. Woodberry KA, Giuliano AJ, Seidman LJ. Premorbid IQ in schizophrenia: a meta-analytic review. Am J Psychiatry 2008; doi: 10.1176/appi.ajp.2008.07081242. 9. Neisser U, Boodoo G, Bouchard TJ Jr, et al. Ingelligence: knows and unknowns. Am Psychol 1996; 51: 77–101. 10. Caroll, J. Human Cognitive Abilities: A Survey of Factor Analytic Studies. Cambridge: Cambridge University Press, 1993. 11. Kaufman AS, Kaufman NL. Manual for the Kaufman Adolescent and Adult Intelligence Test (KAIT). Circle Pines, MN: American Guidance Service, 1993. 12. Wechsler D. Manual for the Wechsler Adult Intelligence Scale (WAIS). New York: The Psychological Corporation, 1955. 13. Tulsky DS, Chelune GJ, Ivnik RJ, et al. Clinical interpretation of the WAIS-III and WMS-III. Academic Press, New York, 2003. 14. Raven J, Raven JC, Court JH. Manual for Raven’s Progressive Matrices and Vocabulary Scales. Section 4: The Advanced Progressive Matrices. San Antonio, TX: Harcourt Assessment, 1998. 15. Nelson HE. The National Adult Reading Test (NART): Test Manual. Windsor, Canada: NFER-Nelson, 1982. 16. De Geus EJC, Boomsma DI. A genetic neuroscience approach to human cognition. Eur Psychol 2001; 6 (4): 241–253. 17. Posthuma D, Mulder EJCM, Boomsma DI, et al. Genetic analysis of IQ, processing speed and stimulus response incongruency effects. Biol Psychology 2002; 61: 157–182. 18. Kravariti E, Toulopoulou T, Mapua-Filbey F, et al. Intellectual asymmetry and genetic liability in first-degree relatives of probands with schizophrenia. Brit J Psychiatry 2006; 188: 186–187. 19. McIntosh AM, Harrison LK, Forrester K, et al. Neuropsychological impairments in people with schizophrenia or bipolar disorder and their unaffected relatives. Brit J Psychiatry 2005; 186: 378–385. 20. Toulopoulou T, Picchioni M, Rijsdijk F, et al. Substantial genetic overlap between neurocognition and schizophrenia; genetic modeling in twin samples. Arch Gen Psychiatry 2007; 64 (12): 1348–1355.

O. de Wilde 21. Kremen WS, Lyons MJ, Boake C, et al. A discordant twin study of premorbid cognitive ability in schizophrenia. J Clin Exp Neuropsychol 2006; 28: 208–224. 22. Cannon TD, Huttunen MO, Lonnqvist J, et al. The inheritance of neuropsychological dysfunction in twins discordant for schizophrenia. Am J Hum Gen 2000; 67(2): 369–382. 23. Barrantes-Vidal N, Aguilera M, Campanera S, et al. Working memory in siblings of schizophrenia patients. Schizophr Res 2007; 95: 70–75. 24. Kremen WS, Faraone SV, Seidman LJ, et al. Neuropsychological risk indicators for schizophrenia: a preliminary study of female relatives of schizophrenic and bipolar probands. Psychiatry Res 1998; 79: 227–240. 25. Birkett P, Sigmundsson T, Sharma T, et al. Executive function and genetic predisposition to schizophrenia–the Maudsley family study. Am J Med Genet B Neuropsychiatr Genet 2008; 147(3):285–293. 26. Egan MF, Goldberg TE, Gscheidle T, et al. Relative risk for cognitive impairments in siblings of patients with schizophrenia. Biol Psychiatry 2000; 50(2): 98–107. 27. Burdick KE, Goldberg JF, Harrow M, et al. Neurocognition as a stable endophenotype in bipolar disorder and schizophrenia. J Nerv Ment Dis 2006; 194: 255–260. 28. Krabbendam L, Marcelis M, Delespaul P, et al. Single or multiple familial cognitive risk factors in schizophrenia? Am J Med Gen 2001; 105: 183–188. 29. Nuechterlein KH, Green MF, Kern RS, et al. The MATRICS consensus battery, part 1: test selection, reliability and validity. Am J Psychiatry 2008; 165(2): 203–213. 30. Kern RS, Nuechterlein KH, Green MF, et al. The MATRICS consensus cognitive battery, part 2: co-norming and standardization. Am J Psychiatry 2008; 165(2): 214–220. 31. Green MF. Cognitive impairment and functional outcome in schizophrenia and bipolar disorder. J Clin Psychiatry 2006; 67(10); e12. 32. Cannon TD, Kaprio J, Lönnqvist J, et al. The genetic epidemiology of schizophreniain a Finnish Twin cohort: a population-based modelling study. Arch Gen Psychiatry 1998; 55: 67–74. 33. Tunbridge EM, Harrison PJ, Weinberger DR. Catechol-omethyltransferase, cognition, and psychosis: Val158Met and beyond. Biol Psychiatry 2006; 60: 141–151. 34. Burdick KE, Lencz T, Funke B, et al. Genetic variation in DTNBP1 influences general cognitive ability. Hum Mol Genet 2006; 15: 1563–1568. 35. Donohoe G, Morris DW, Clarke S, et al. Variance in neurocognitive performance is associated with dysbindin-1 in schizophrenia: a preliminary study. Neuropsychologia 2007; 45: 454–458. 36. Callicott JH, Straub RE, Pezawas L, et al. Variation in DISC1 affects hippocampal structure and function and increases risk for schizophrenia. Proc Natl Acad Sci USA 2005; 102: 8627–8632. 37. Hall J, Whalley HC, Job DE, et al. A neuregulin 1 variant associated with abnormal cortical function and psychotic symptoms. Nat Neurosc 2006; 9: 1477–1478. 38. Zinkstok J, de Wilde O, van Amelsfoort T, et al. Association between the DTNBP1 gene and intelligence: a case-control study in young patients with schizophrenia and related disorders and unaffected siblings. Behav Brain funct 2007; 3: 19.

Chapter 11

Emotion Recognition Deficits as a Neurocognitive Marker of Schizophrenia Liability Renata Schoeman, Dana J.H. Niehaus, Liezl Koen, and Jukka M. Leppänen

Abstract Schizophrenia is a heterogeneous psychiatric disorder with a strong heritable component. However, to date, no single gene, cellular or molecular marker has consistently been identified to be associated with this illness. Rather, it appears more likely that schizophrenia develops as a consequence of contributions from multiple susceptibility genes acting in concert with environmental factors. It is therefore argued that the use of endophenotypes could play a valuable role in the understanding of this complex illness. The endophenotype approach seeks to identify very specific aspects/syndromes of a disorder, link them to candidate genes, and thereby explicate the more complex phenomenon by breaking it down into salient units that may be more amenable to rigorous scientific investigation. In this chapter a potential role for deficiency of facial affect recognition as a candidate neurocognitive marker and endophenotype in schizophrenia is discussed. Key components of endophenotypes are the heritability and stability of the trait under investigation. Considerable evidence linking deficits in face and emotion processing to possible heritability exists, with schizophrenia sufferers as well as close genetic relatives exhibiting similar deficiencies. Furthermore, the trait appears to be stable over the course of the illness and fairly specific for the schizophrenia-spectrum disorders. The neural basis for emotion recognition deficits, such as structural and functional abnormalities in occipital-temporal cortex (fusiform gyrus) and the amygdala, some of which

R. Schoeman, D. J. H. Niehaus, and L. Koen Department of Psychiatry, University of Stellenbosch, South Africa J. M. Leppänen () Department of Psychology, University of Tampere, Finland

have been found to be present even during the first episode of schizophrenia or in unaffected relatives of schizophrenia patients, is also summarized. In conclusion, the potential utility of this and other neurocognitive markers of schizophrenia in the investigation of genetic determinants of the illness and identification of individuals at risk is discussed. Keywords Schizophrenia • endophenotype • neurocognition • affect recognition • facial expressions Abbreviations EEG Electroencephalograph; ERP Event-related potential; fMRI Functional magnetic resonance imaging; MEG Magnetoencephalograph; PET Positron emission tomography; STS Superior temporal sulcus

Introduction It has been argued that the use of endophenotypes could play a valuable role in the understanding of complex illnesses such as schizophrenia.1 Endophenotypes refer to any neurobiological measure that can be identified and validated as such. The key criteria include heritability, stability of the trait, family association, co-segregation and the ability to reliably measure the marker. As such the endophenotype approach seeks to identify very specific aspects/syndromes of a disorder, link them to candidate genes, and thereby explicate the more complex phenomenon by breaking it down into salient units that may be more amenable to rigorous scientific investigation. There has been a growing interest in the use of different cognitive deficits as potential endophenotypic

M.S. Ritsner (ed.), The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes, © Springer Science + Business Media B.V. 2009

163

164

markers of schizophrenia liability.2 Cognitive impairment is recognised as an integral part of the psychopathology of schizophrenia. It represents a major barrier to functional recovery with up to 79% of the variance in improvement in occupational ability ascribable to deficits in neurocognitive functioning.3 Cognition is also hypothesized to be a critical component of social skills and an important determinant of social competence and functioning. The tripartite model of social skills proposes that social behaviour consists of three components.4,5 Social perception refers to the ability to appraise the relevant information in a situation, such as the context and location of a conversation, the relationship between the participators, and the affect conveyed (i.e. whether it is a friendly conversation with family members at home, or a formal exchange with a business colleague at a meeting). Social cognition refers to the formulation of a goal and plan of action, after the necessary information has been obtained. Finally, behavioural skills, such as non-verbal communication, paralinguistic skills and turn-taking, enable a person to implement this plan. Social cognition can also be explained as our ability to infer and represent the mental state of others (“theory of mind”). According to Borthers et al.6 social cognition depends on perceiving faces, perceiving and recognizing expressions of emotions and engaging in social interactions. Deficits in social cognition and facial information processing including the ability to recognize the identity (“who the person is”) and emotional state (“how does the person feel”) of another person, have been identified as one of the key impairments in schizophrenia. Impaired emotional functioning, core to Eugene Bleuler’s description of schizophrenia in 1911, includes deficits in social cognition and its impact on social behaviour.7 A 1-year prospective study by Kee et al.,8 examined cross-sectional and longitudinal relationships between perception of emotion and aspects of social relationships, work functioning an independent living/selfcare in 94 clinically stable outpatients with schizophrenia. They found that emotion processing is a key determinant of work functioning/independent living for individuals with serious mental illness, independent of psychopathological symptoms such as conceptual disorganization. Another study by Hooker and Park9 confirmed the importance of face processing deficits and found facial affect recognition deficits to be the most predictive of social behaviour problems. There is also

R. Schoeman et al.

evidence that social perception and social cognition impairments predict the odds of suffering from schizophrenia significantly better than impairments in nonsocial cognition.10 Given that impairments in social cognition and facial affect processing figure prominently in schizophrenia, understanding the nature and mechanism of these impairments may assist in diagnosis and management of this disorder. In this chapter, a potential role for facial affect processing as a candidate neurocognitive marker and endophenotype of schizophrenia is discussed. A brief overview of behavioral studies of facial affect recognition deficits in schizophrenia patients is provided. Neural mechanisms underlying face-based affect processing and its deficits as well as the value of facial affect processing deficits as a trait marker in schizophrenia and high-risk individuals is summarized.

Deficient Facial Expression Recognition in Schizophrenia The ability to perceive and interpret communicative signals provided by others’ faces is fundamental to normal social interaction. It is important for us to not only recognize a face as a face but also to recognize the identity of the face (“who the person is”) and the emotional state of the person involved (“is he/she happy, sad or angry?”). Throughout this chapter the different aspects of face processing will be referred to as “face perception”, “facial identity recognition”, and “facial affect recognition”, respectively. Certain facial expressions are universally recognized to act as external markers of specific emotional states. The most consistently expressed and recognized emotions in various literate and preliterate cultures include anger, disgust, fear, sadness, and happiness.11 Several studies have shown deficits in facial affect recognition to exist in schizophrenia and these findings have been extensively reviewed elsewhere.12,13 On balance, the available evidence would suggest that individuals with schizophrenia are less accurate than controls at perceiving and recognizing facial expressions; however, the specificity, extent, and nature of the deficits are unclear. Possibly, the deficits could just be part of generalized cognitive impairment, reflecting wider underlying impairments of perception, memory,

11

Emotion Recognition Deficits as a Neurocognitive Marker of Schizophrenia Liability

language processing, attention and executive functioning.14 However, some findings support a disproportionate deficit in facial affect recognition tasks.15,16 Furthermore, although facial identification and age recognition seems to be impaired in schizophrenia,17 facial affect recognition deficits seem to correlate with severity of psychopathology and neurocognitive deficits, and contribute to social dysfunction. Finally, schizophrenia patients appear to have greater difficulty with the identification and discrimination of negative emotions,18–20 although a specific deficit in the recognition of negative affect has not been demonstrated in all studies.21 Also, Johnston et al.22,23 have argued that the specific impairment in the recognition of negative emotions may reflect methodological problems such as psychometric properties of material used and a priori differences in task difficulties. As such, they demonstrated that performance degraded differentially across emotion categories when difficulty increased, with the greatest deterioration present on negative valence stimuli. Deficits in facial affect recognition have been demonstrated at an early age and stage of illness. Habel et al.24 compared mood induction and facial affect recognition in 20 children and adolescents with schizophrenia and 20 matched healthy controls. A reduced ability to discriminate negative valenced facial expressions was present in the patient group, with the difference approaching statistical significance. Edwards et al.19 examined facial affect recognition in first episode psychosis and demonstrated that schizophrenia spectrumdisordered patients were more impaired on the fear and sadness subscales of computerized emotional labeling tasks than patients with affective psychosis and normal controls.

165

Although current evidence is conflicting it has been suggested that both the presence of negative symptoms and the subtype of schizophrenia could act as confounding factors with regard to facial affect recognition deficits. Some studies25–27 have reported patients with blunted affect and other negative symptoms to be more sensitive in recognizing negatively valenced faces. On the other hand, Van’t et al.28 linked negative symptoms to greater impairment in the recognition of fearful faces. Associations between late onset and non-paranoid schizophrenia with deficits in the recognition of neutral faces have also been demonstrated.29

Neural Networks Underlying Face and Emotion Processing Neuropsychological studies of patients with focal brain lesions as well as functional imaging (fMRI, PET, EEG/ERP) studies of healthy individuals have provided researchers with valuable information about the neural correlates of face-based affect recognition. (For more detailed information please see reviews.30–32) Distinct brain structures have been linked to the perceptual processing of emotional stimuli such as facial expressions (fusiform gyrus and superior temporal sulcus in the occipitotemporal cortex), affect recognition and generation of emotional reactions in response to an affective stimulus (amygdala, anterior insula, orbitofrontal cortex, and ventral striatum), and in the regulation of emotional reactions evoked in response to the stimulus (anterior cingulate cortex, prefrontal cortex, see Fig. 11.1).

Emotion regulation

Fig. 11.1 A diagram showing the psychological processes involved in the processing of (visual) emotional signals and the neural structures that support these processes (modified from models presented in Adolphs, 200230 and Phillips et al., 200332). ACC, anterior cingulate cortex; DLPFC, dorsolateral prefrontal cortex; DMPFC, dorsomedial prefrontal cortex; FG, fusiform gyrus; OFC, orbitofrontal cortex; STS, superior temporal sulcus

ACC, DMPFC, DLPFC

Stimulus

Analysis of stimulus

Emotion recognition and

features

emotional response

V1/V2, FG, STS

Amygdala, Insula, OFC, Ventral Striatum

166

Structures that have been closely associated with perception and recognition of facial expressions include regions in the occipitotemporal cortex and the amygdala. Certain higher-level visual areas, notably those involving the fusiform gyrus and superior temporal sulcus respond selectively to faces and are thought to play an essential role in the construction of structural representations of those facial attributes that are important for the recognition of facial identity and facial expressions.30,31 The fusiform gyrus in the posterior region of the inferior temporal lobe and the occipital lobe appears to be especially important in recognizing a face as a face and in analyzing those invariant aspects of faces that are critical for recognizing facial identity.31 These conclusions are supported by several studies. Firstly, single-cell recordings in monkeys have identified neurons in the inferior temporal cortex that fire selectively at the sight of faces and, in particular, in response to facial identity.33,34 Secondly, there is consistent evidence from fMRI (functional Magnetic Resonance Imaging) studies in humans showing that the right fusiform gyrus – the so-called fusiform face area – responds stronger to passive viewing of intact faces than to scrambled faces and in response to other human (e.g. hands) and non-human objects (e.g. houses).31,35,36 Finally, lesions to the ventral occipitotemporal and temporal cortices result in a condition known as prosopagnosia where patients present with an inability to recognize familiar faces whilst other visual objects are spared.37,38 Whereas the fusiform cortex is most prominently associated with processing invariant aspects of faces (cues related to facial identity) other areas of the occipitotemporal cortex (i.e., STS) may play a major role in the processing of the changeable aspects of faces, such as eye movements, lip movements, and facial expressions.31,36 In primates, neurons responsive to facial gestures have been demonstrated primarily in the STS,33 whilst functional imaging studies in humans have also shown that active attention to emotional expression activates the STS.39,40 Higher-level visual areas in the occipitotemporal cortex feed forward to areas that serve to link the visual stimulus to its motivational significance. The amygdala has been consistently associated with the processing of certain negative emotions such as expressions of anxiety and fear,41–43 although recent functional imaging studies suggest a broader role in the processing of different negative and positive affects (see Winston et al.40).

R. Schoeman et al.

The amygdala projects to several other subcortical and cortical regions which are important with regard to regulating different aspects of an emotional response to a stimulus such as autonomic nervous system activation, changes in endocrine secretion and modulation of vigilance level and attention.43–46 Other areas that have been linked to the processing of affective information from social cues include the orbitofrontal cortex (especially for the processing of positive and negative reinforcers, see Rolls47), the nucleus accumbens and the striatum/basal ganglia. (especially in the processing of positive emotions, see Phillips et al.32)

Neural Bases of Deficits in Facial Affect Recognition in Schizophrenia Structural abnormalities: Structural brain changes such as gray matter reduction in the left superior temporal gyrus, the medial prefrontal cortex, the right anterior cingulate, the right insula and the prefrontal cortex, have all been linked to impaired social cognition.48 Ventricular enlargement of up to 33% is present in schizophrenia and this is most clearly seen in the temporal horns of the lateral ventricles, reflecting underlying tissue loss in the medial temporal lobes, including the amygdala, hippocampus and parahippocampal areas.49,50 Furthermore, postmortem and in vivo structural studies51 show bilateral reduction in fusiform gray matter volume that is evident at the onset of the psychosis. A meta-analysis by Wright et al.52 concluded that the average volume of the amygdala in schizophrenia is 6% smaller than those of healthy controls. Available evidence would thus appear to support that the neural structures involved with face and affect perception are relatively smaller in schizophrenia sufferers. Functional abnormalities: Indirect evidence for functional abnormalities in the brain networks underlying facial affect processing comes from eyetracking studies showing deficient visual scanning of faces in schizophrenia patients. Specifically, it has been suggested that the amygdala is important in directing attention to emotionally salient aspects of the visual input.46 From this perspective, it is interesting to note that there is consistent evidence that both facial affect recognition and visual scanpaths (the path that the eyes follow to extract information form a stimulus)

11

Emotion Recognition Deficits as a Neurocognitive Marker of Schizophrenia Liability

to facial expressions are impaired in schizophrenia. Using an infrared-corneal-reflection technique to determine scanpaths,53–55 schizophrenia patients were shown to differ from controls with regard to several eye movement parameters such as length of mean scanpath, mean duration of fixation as well as inefficient and more restricted search strategies. Although patients demonstrated a significant deficit in facial affect recognition, performance and eye movement abnormalities did not correlate. Both the facial affect recognition deficits and the eye movement abnormalities remained stable over time. In addition to behavioral and eye-tracking studies, functional abnormalities in face and emotion processing mechanisms have also been investigated more directly. Both event-related brain potentials (ERPs) and different functional imaging technologies (fMRI, PET) have been employed to study the neural correlates of facial affect recognition in schizophrenia. In the next few paragraphs a brief overview of the recent ERP and fMRI studies will follow. ERP studies: Although ERP methodology provides very useful information with regard to the timing of mental events (in the order of milliseconds), it lacks the localization accuracy of the functional imaging technologies such as fMRI and PET. ERPs are small voltage fluctuations resulting from evoked neural activity in response to discrete events. A discrete stimulus typically invokes a series of ERP components, whose latency, amplitude, and scalp topography provide information about the timing of a particular mental event, the amount of neural activity involved in a particular cognitive process and the neural generators of the observed activity. ERP studies have revealed abnormalities in neural processes associated with the evaluation of stimulus information in schizophrenia patients. Streit et al.15,56 recorded ERPs in 16 schizophrenia patients and 16 healthy controls during a facial affect categorization and an object categorization task. Reduced amplitudes (hypoactivation) over frontal recording sites 180–250 ms post-stimulus were demonstrated during affect categorization tasks for patients. Another study57 failed to demonstrate an increase in N200 (negative deflection in approximately 200 ms after the stimulus is presented) over frontal scalp regions in schizophrenia patients, suggesting diminished prefrontal EEG response during facial affect recognition. There is also evidence

167

that the P300 (positive amplitude at 300 ms) over parietal scalp regions is enhanced in schizophrenia patients.57 P300 occurs in response to all kinds of stimuli (e.g. visual, auditory) and is thought to reflect higher level cognitive processing associated with unexpected and/or salient stimuli. Interestingly, none of these studies showed differences in the very early ERPs to facial stimuli (occurring at 50–70 ms after stimulus presentation). This would suggest that the facial affect recognition deficit in patients with schizophrenia is not due to problems in the very early stages of visual perception, but rather later, where visual information is integrated and linked with more complex cognitive processes. Johnston et al.58 has argued that a deficit at an early stage of face-specific processing (i.e., processing that occurs subsequent to early visual processing and is putatively subserved by the fusiform gyrus) accounts for impaired facial emotion recognition in schizophrenia, as opposed to the negative-emotion specific deficit model,18,19 suggesting that impaired facial information processing occurs at a later stage. Indeed, some authors59,54 have found that deficient recognition of emotions by patients with schizophrenia in comparison to controls resulted from delayed primary analysis of visual information and reduction of components in the infero-temporal and occipital cortical areas responsible for classification and decision making. Johnston et al.58 recorded ERPs for 11 patients with schizophrenia and 15 matched controls while performing a gender discrimination and facial emotion recognition task. Significant reductions of the face-specific vertex positive potential (VPP) (positive voltage over the vertex of the head), at a peak latency of 165 ms was present in the patients, while their early visual processing (positive voltage at 100 ms (P1) ) was found to be intact. The reduction in VPP was associated with a reduction in P300 amplitude and predicted accuracy during facial discrimination tasks. In a subsequent fMRI task, correlation analysis revealed that the reduction in VPP and P300 predicted blood oxygenation level-dependent activation in the fusiform, inferior frontal, middle temporal and middle occipital gyri, as well as the amygdala. It would suggest that the facial affect recognition deficits may be due to flow-on effects of a generalized deficit in early visual processing of faces. The hypothesis of a face-specific visual processing deficit in schizophrenia gained further support form a recent study by Turetsky et al.60 They recorded ERPs

168

from 16 patients and 16 healthy controls during a facial affect recognition task in which the subjects had to indicate whether faces presented were “happy”, “neutral” or “sad”, demonstrating that controls performed better than patients in recognizing facial emotions. Patients with schizophrenia showed deficits in early visual encoding of facial features that precedes the ERP response typically associated with facial affect recognition. Group differences were noted for the N170 “face processing” component (a polarity reversal of the vertex positive peak occurring 170 ms after stimulus presentation over lateral temporal scalp) that underlies the structural encoding of facial features, but not for the subsequent N250 (a negative voltage occurring 250 ms after stimulus presentation) “affect modulation” component. These results further support the view that schizophrenia is associated with disturbances in neurocognitive operations underlying visuomotor processing and facial expression analysis. Affect recognition difficulties seem to be secondary to faulty structural encoding of faces during the primary analysis of visual information. Consistent with the ERP findings of perceptual deficits in face processing, a behavioral study has demonstrated visual deficits in face but not non-face object processing in schizophrenia. Chen et al.61 examined deficits in face detection in patients with schizophrenia and normal controls by presenting them with brief line-drawings of upright or inverted faces and trees. The line-drawings were used to represent a “face-like” object and therefore contained minimal clues about identity and expressed emotion. Recognition of faces is disproportionately affected when face-stimuli are inverted (compared to non-face stimuli). A reduced effect of face inversion was present in subjects with schizophrenia, mainly attributable to their lower accuracy in detecting upright faces. fMRI and MEG studies: As previously noted, although ERP studies can provide valuable information about the timing of the neural activity, they lack the localization accuracy of other brain imaging tools (such as MEG and fMRI) and therefore are unable to show where impairments are “seated” in the brain. Recent functional imaging studies have examined whether schizophrenia patients exhibit abnormalities in areas known to be important for facial affect recognition (e.g. fusiform gyrus, amygdala) as well as areas involving the medial prefrontal cortex (modulating

R. Schoeman et al.

intense emotional responses as generated by the amygdala), the anterior cingulate cortex (involved in attention and regulation of cognitive and emotional processes) and the insula (involved in expression of internally generated feelings).62 Streit et al.15 collected magnetoencephalographic (MEG) data from a group of partially remitted schizophrenia patients and healthy volunteers. The MEG data recorded in response to facial expressions of emotion revealed that impaired behavioural performance was associated with strength of activation in the inferior prefrontal areas, the right posterior fusiform gyrus region, right anterior temporal cortex, and the right inferior parietal cortex. In an fMRI study using partially remitted patients with a first hospitalization for psychosis performing facial affect discrimination and labeling tasks,63 a significantly decreased activation of the anterior cingulate during facial affect discrimination and of the amygdala-hippocampal complex bilaterally during facial affect labeling was demonstrated. During both tasks, controls showed a significantly increased activation of the right middle frontal gyrus correlating with rising task difficulty. In addition, an increased activation of the bilateral middle frontal gyri was noted in the patients with schizophrenia, possibly reflecting a compensatory effort due to deficits in more basal limbic functions. Quintana et al.64 demonstrated failed activation of the right lateral fusiform gyrus in chronic stable schizophrenia sufferers that was present during early facial identification, as well as affect recognition tasks, irrespective of semantic processing and varying intentional demands. It has been shown that facial affect recognition can influence motor movement, especially expression of emotion. When primates passively observe others perform specific gestures or actions, premotor and motor cortical areas involved in the internal representation and actual behavioural manifestations of the actions, occurs. This mirror mechanism is important in social learning, which is deficient in schizophrenia. An interesting study by Wolf et al.65 showed that non-medicated patients with schizophrenia showed fewer joy or smile reactions compared to healthy controls with decreased previsible muscle activation of expressive muscles such as the M. orbicularis oculi and M. zygomaticus during presentation of positive emotional pictures, as measured by electromyography. This could be indicative of problems in the mirror neuron circuitry, in as much that patients with

11

Emotion Recognition Deficits as a Neurocognitive Marker of Schizophrenia Liability

schizophrenia not only have deficits in perceiving emotions, but also in expressing emotions. In support of this hypothesis, decreased activation of the mirror cortical mechanism in schizophrenia patients has been reported in a MEG study.66

Facial Affect Recognition Deficits as a Stable Trait Marker of Schizophrenia and Schizophrenia Liability If emotion recognition deficits are a trait marker of schizophrenia and contribute to vulnerability for the development thereof, it should not only be present at all stages during the course of the illness but also be specific to the illness. Researchers have conducted both longitudinal (e.g. Addington and Addington67) and cross-sectional (e.g. Gaebel and Wolwer68) studies in order to elucidate this issue and have demonstrated that these deficits tend to remain over time, despite some improvement occurring concomitant to improvement in symptoms. Addington and Addington67 assessed facial affect recognition and visual attention in a sample of 40 schizophrenic patients, evaluating their performance at baseline and again 3 months later. They performed significantly worse than individuals with bipolar disorder and healthy controls at both baseline and follow-up, despite improvement in symptoms. Edwards et al.19 confirmed small but consistent deficits in recognition of fear and sadness in first episode psychosis in comparison to patients with affective psychosis and healthy controls. A 1-year longitudinal study69 reporting on patients with a first episode of psychosis, multi-episode schizophrenia subjects and healthy controls showed that both psychotic groups were clearly impaired relative to healthy controls in terms of facial affect recognition, neurocognition, and social functioning. Gaebel and Wolwer68 also demonstrated that both patients with depression and schizophrenia express quantitative, qualitative and temporal differences in facial affect recognition deficits. Schizophrenia patients were more impaired and the deficits persisted, irrespective of medication type, dose and side-effects. In a cross-sectional study, Kurcharska-Pietura et al.70 compared 50 controls to 50 patients with earlystage and 50 with chronic schizophrenia, using the

169

Benton Facial Recognition Task (assessing facial identity recognition). Patients with chronic schizophrenia were significantly more impaired than the other two groups, which seems to reflect the progressive nature of the disease and serves as further evidence for face processing deficits as a trait feature thereof. Mueser et al.71 re-examined data on three studies of affect recognition, examining medication status and illness chronicity concluding that results across the studies suggested that poor performance on facial affect recognition tasks could be related to chronicity, but not medication status. In another interesting study72 using the Noh Mask Test 15 schizophrenia subjects and 15 healthy controls were presented with a rotated Noh Mask and an emotional label and asked to judge whether the mask expression was congruent with the emotion indicated. This test was successful in discriminating patients from controls with a discriminant ration of 99.9%. Taken together, these results suggest that facial affect recognition impairments are stable deficits (present in first episode psychosis, but also in chronic schizophrenia), specific for schizophrenia (different from affective psychosis and not present in healthy controls), and related to other neurocognitive impairments (see above). A further requirement for a trait marker is an association not just with illness presence but also illness liability, requiring the demonstration of the marker in individuals at risk for development of the illness as well as unaffected family members. Green et al.73 examined 50 healthy controls with regard to delusionproneness and correlated this with reaction time for identifying facial affect stimuli. Individuals with high delusion-prone scores displayed a significant delay in processing of angry faces in comparison to those with low delusion-proneness, evidently demonstrating an attentional bias for threat-related stimuli. Interestingly, patients with paranoid schizophrenia were more accurate than non-paranoid subtypes at recognizing and labeling negative facial affects which suggest that there may be underlying cognitive processing differences between the different subtypes.29,74 Poreh et al.75 investigated facial affect recognition in college students with high ratings on items of magical ideation, perceptual aberration and schizotypy, all of which could be suggestive of schizoptypal traits. Higher scoring individuals performed worse relative to controls on facial recognition and facial affect recognition tasks, however, on further analysis with face recognition as a

170

covariant, the differences between the groups disappeared. Similar results were found by Toomey and Schuldberg,76 suggesting that the impairment in facial affect recognition in schizotypal subjects could be related to underlying neurocognitive problems in attention and vigilance and that these deficits in emotional decoding may appear during a later stage in the development of the disorder. Both autism and schizophrenia are considered to be substantially influenced by genetic factors with endophenotypes for deficits in affect perception likely to exist in both disorders. Bolte and Poustka77 examined the capacity to recognize facial affect in subjects with autism and schizophrenia in comparison to first degree relatives (siblings and parents). Thirty-five autism plus 102 relatives and 21 schizophrenia probands plus 46 relatives, as well as an unaffected control sample were studied using a computerized facial affect recognition task. Autistic subjects were more impaired than schizophrenia probands and unaffected individuals, with probands from multiplex families (where more than one family member was diagnosed with autism) performing worse. None of the other groups were significantly impaired and as such these results therefore did not support the hypothesis that deficits in facial affect recognition are related to schizophrenia liability. One potential explanation for this negative finding could be that rather than a broad impairment in facial affect recognition, schizophrenia vulnerability may be associated with a more selective deficit with greater impairments for certain negative facial expressions, arising from functional aberrations in the amygdalacentered neural circuitry.78 To examine this possibility, Leppänen et al.78 tested healthy comparison subjects, unaffected siblings of schizophrenia patients, and schizophrenia patients with a task requiring rapid recognition of matched positive (happy), negative (angry), and neutral facial expressions. Siblings and patients demonstrated impaired recognition of negative relative to positive facial expressions whereas comparison subjects recognized negative and positive expressions at an equal level of accuracy. These results were consistent with the view that deficits in the processing of negative affect from social cues are transmitted in families and may represent a heritable component of schizophrenia and have also been supported by other studies.2,79 Abnormalities in visual scanning of faces have also been found in unaffected relatives of schizophrenia patients. In a visual scanpath study of first-degree

R. Schoeman et al.

relatives of schizophrenia probands,80,81 visual scanpaths were recorded for 65 probands, 37 first-degree relatives and 61 non-related, healthy controls. Face recognition and facial affect recognition experiments demonstrated first degree relatives to have an attenuated form of restricted scanpaths compared to probands across all facial stimuli. First-degree relatives also demonstrated avoidance of facial features compared to both probands and healthy controls, which could indicate an additional disturbance in social cognition.

Neurocognitive Endophenotypes of Schizophrenia Schizophrenia is a heterogeneous psychiatric disorder with a strong heritable component.82 It also has a polygenic basis, i.e. various highly penetrant mutations can cause a change in the neural circuitry of a patient or family, underlying a specific endophenotype, which can then be expressed in a “final common pathway” as a schizophrenia phenotype. There is mounting evidence that genetic factors not only account for 50–80% of the liability to develop the disorder but also for some of the variation in symptom expression, treatment outcome, cognitive impairment and brain morphological changes.83 Further to contributing to the phenotype expression in schizophrenia, it is also possible that these genes can contribute to an endophenotype, i.e. the underlying trait; more closely linked to the genetic abnormality than the schizophrenia phenotype as such.1 The endophenotype concept, as introduced by Gottesman and Shields84 could play a valuable role in the understanding of complex illnesses. As previously stated, the endophenotype approach seeks to identify very specific aspects/syndromes of a disorder, link them to candidate genes, and thereby explicate the more complex phenomenon by breaking it down into salient units that are amenable to rigorous scientific investigation. Endophenotypes are quantifiable traits that are conceptualized to be nearer to gene-based neuro-biological deficits than the illness itself.83 Core criteria for endophenotype markers that need to be satisfied are (1) an association with the illness, (2) heritability of the deficits, (3) stable and trait-related (i.e. the endophenotypic deficits are present during all stages of the disease and relatively state-independent), (4) co-segregation between the marker and the disorder,

11

Emotion Recognition Deficits as a Neurocognitive Marker of Schizophrenia Liability

and (5) the proband’s specific marker must occur at higher rates in first-degree relatives than in the general population.1 Suggested endophenotypes in schizophrenia can be divided into various categories. Examples include (1) Anatomical; e.g. cerebellar abnormalities,85 ventriculomegaly86 and temporal and insular grey matter abnormalities87,88 and (2) neurophysiological; e.g. deficits in inhibitory functions of sensory gating measured via suppression of the P50 evoked potential,89 sensorimotor-gating as measured by prepulse inhibition of the startle reflex90 and oculomotor measures regarding saccades and antisaccades (e.g. Ross et al.91). A further category is that of neurocognitive endophenotypes. Family, twin and adoption studies have consistently supported schizophrenia to be a highly heritable, but heterogenous disorder. Genome-wide searches have revealed possible susceptibility genes for the disorder in relatively broad regions of multiple chromosomes92 with meta-analysis of the data (see Lewis et al.93) providing some support for chromosomes 6p, 8p, 10p, 13q, 15q, 18q and 22q. Several candidate susceptibility genes have also been suggested to play a role in the pathogenesis of the schizophrenia.82 Currently some evidence exists for dysbindin (DTNBP1), neuregulin 1 (NRG1), “Disrupted in schizophrenia 1” (DISC1), D-amino-acid oxidase (DAO), D-amino-acid oxidase activator (DAOA, formerly G72) and regulator of G-protein signalling 4 (RGS4). All of these seem to only have a small or modest effect on the illness1 and gene–environment interactions still have to be clarified. Currently a number of examples of studies linking genes with neuropsychological performance can be found in the literature. Van Amelsvoort et al.94 found individuals with 22q11 deletion syndrome (velo-cardio-facial syndrome (VCFS) ) and schizophrenia to perform significantly worse in terms of spatial working memory, visual recognition and attentional tasks in comparison to individuals with VCFS without schizophrenia. Lawrie et al.95 reported an association between the COMT Val allele and increased risk for developing schizophrenia with the individuals possessing the Val allele exhibiting reduced gray matter density in the anterior cingulate and increased fMRI activation in the lateral prefrontal cortex and the anterior and posterior cingulate on a sentence task despite no increase in performance. The anterior cingulate is known to be involved in attention modulating emotional and cognitive processing (see

171

Fig. 11.1), specifically mediating emotional arousal when external information needs to be processed in the presence of conflicting internal emotional states96 and during sadness.97 Reduced gray matter density in this area could thus contribute to abnormal facial and emotional processing. As such, these deficits might be related to developmental problems in the frontal lobes and increase the risk of developing schizophrenia. Neuropsychological deficits are detectable in genetically at risk, but healthy first degree relatives of probands with schizophrenia.98 Evidence for sustained attention and vigilance, verbal declarative memory and working memory as valid endophenotypes for schizophrenia has also been demonstrated (see reviews by Aleman et al.99; Snitz et al.100). The Consortium on the Genetics of Schizophrenia (COGS) reviewed available evidence for these neurocognitive tasks as endophenotypic measures in schizophrenia genetic studies and concluded that they met the core criteria101 as described by Gottesman and Gould1 (see above). Therefore, when one takes into account the evidence as reported in this chapter, deficits in facial affect recognition would seem to satisfy endophenotype criteria. Firstly, facial affect recognition can be linked to a relatively specific neurobiological network and reliably measured. Secondly, evidence suggests significant heritability estimates for facial affect recognition.101 Thirdly, facial affect recognition deficits in schizophrenia co-segregate with illness within families. Fourthly, facial affect recognition appears to be a stateindependent trait marker of the disorder. Finally, several studies have shown that individuals with schizophrenia as well as close genetic relatives exhibit similar difficulties with the processing of this kind of social information. Studies have also started to link different measures of facial affect processing to specific genetic factors. Iidaka et al.102 studied limbic/prefrontal activation in response to a face recognition task in participants genotyped for the C178T polymorphism in the regulatory region of serotonin receptor type 3 gene (HTR3A) on chromosome 11. Individuals with C/C alleles demonstrated greater activation in the amygdala and the dorsolateral prefrontal cortex, as well as faster face recognition response times than those with the C/T alleles. Variations at the serotonin gene locus may therefore influence limbic/prefrontal brain activity and facial recognition. Hariri et al.103 demonstrated that individuals with one or two copies

172

of the short allele of the serotonin transporter (5-HTT) promoter gene (SLC6A4) had reduced 5-HTT expression and function, increased fear and anxiety-related behaviours and greater amygdala activation than their long-allele counterparts in response to fearful facial stimuli. Battaglia et al.104 also demonstrated an association between having the short allele and altered processing of angry faces, shyness and behavioural inhibition. Other studies exploring links between genes and social competence were conducted in individuals with Turner syndrome105 demonstrating a genetic locus for social-cognition expressed from the paternal X chromosome. Increased vulnerability to psychiatric disorders characterized by difficulties in social interactions, such as schizophrenia and autism, has also been reported in Klinefelter syndrome (47, XXY). Van et al.106 investigated social-emotional information processing in Klinefelter syndrome comparing 32 Klinefelter men with 26 healthy controls. Participants were evaluated on several social and emotional tasks with the Klinefelter men shown to be less accurate in facial affect recognition tasks and experiencing more difficulty in verbalizing and identifying their emotions. However, they experienced increased levels of arousal in comparison to controls. Studies have reported amygdala size to be inversely correlated with the number of X-chromosomes and that patients with Klinefelter’s have a tenfold increase in the risk to develop schizophrenia.107 The presence of abnormalities of the amygdala and facial processing in individuals at high risk for the development of schizophrenia, gives further support for the view that these behavioural and structural changes might be an endophenotype of schizophrenia that could mediate between the molecular, genetic and clinical levels of the disease.

Treatment of Facial Affect Recognition Deficits From the evidence presented it seems clear that patients with schizophrenia consistently demonstrate deficits of facial affect recognition. Studies also suggest that facial affect recognition may play a role in the development of schizophrenia and associated adjustment problems. This raises the question of whether different

R. Schoeman et al.

aspects of face processing can be changed or improved through interventions and whether such changes, in turn, could lead to therapeutic effects. To date, there is a paucity of literature regarding the effects of treatment (psychopharmacological and psychotherapeutic) on facial affect recognition abilities and consequently social outcome. However, what is available seems promising. Fakra et al.108 randomized 25 acutely ill schizophrenia patients to either haloperidol or risperidone and evaluated drug effect on facial identity discrimination and affect recognition tasks over 4 weeks. In this study, it appeared that risperidone enhanced both processing of individual facial features and affect recognition. In another study in which 14 treatment refractory schizophrenia subjects participated,109 olanzapine decreased depressive symptoms and improved the interpretation of positive prosodic affective stimuli which could enhance social adaptation. Other literature110 also mentions the possibility of using serotonergic and anxiolytic GABAergic agents to modulate the activation of the amygdala. Two studies utilizing psychotherapeutic intervention to improve facial affect recognition have been conducted. Russell et al.111 did a pilot study using a “micro-expression training tool” (METT) to improve emotion recognition. Comparing 20 patients with schizophrenia with 20 healthy matched controls they evaluated performance on an emotion-matching task pre-and post-training with the METT. The METT is a single-session computer based emotion recognition training program, aiming to improve participants’ ability to distinguish between emotions which can be confused (such as anger/disgust, fear/surprise). The participants look at 15 ms duration stimuli of microexpression (i.e. feature areas of the face such as the nose, eyes and mouth) while listening to a verbal commentary (for example “concentrate on watching how the mouth is more rounded in surprise, while in fear there is more tension and the lips are stretched horizontally”). At the end the patients are asked to identify emotions in a still image. Patients with schizophrenia improved to a level from where they could not be distinguished from pre-trained controls. Wolwer et al.81 found that “Tackling Affect Recognition” (TAR) training could be effectively used to train patients to match pre-training healthy controls. The TAR is a 12 session manual-based program with small group 45 min sessions scheduled twice-weekly. Patients progress through three different stages in which they first learn

11

Emotion Recognition Deficits as a Neurocognitive Marker of Schizophrenia Liability

to identify and discriminate prototypical facial signs of the six basic emotions and to use verbalization and self-instruction as self-aid strategies to analyze facial expressions. In the next step, patients are taught to make faster and more “holistic” decisions and also to focus on non-verbal processing and recognition of small intensity expressions. In the last part of the program, subjects are taught to process non-prototypical, ambiguous expressions (which commonly occur in real-life) and to integrate these facial expressions into social and behavioural contexts. The authors note that functionally specialized programs such as the TAR training are needed because remediation of facial affect recognition in schizophrenia patients is unlikely to be achieved with traditional cognitive rehabilitation programs.

Conclusions and Future Directions The research reviewed in this chapter provides an overview on current available evidence with regard to facial emotion recognition processing deficits in schizophrenia, as well as the cognitive mechanisms and neural networks underlying these deficits. Schizophrenia is associated with a specific impairment in facial emotion affect recognition, with some bias towards negative (fearful and angry) facial expressions. This impairment may be related to functional aberrations in emotion-related neural networks, as well as interruption of the “top-down” control on processing in these brain networks. The existing data also suggest that deficits in facial expressions meet several of the criteria for endophenotypes. Further research is required, however, to further delineate the neural bases of affect processing deficits in schizophrenia spectrum and at risk populations. Given that endophenotypes are likely to be more proximally associated with the underlying genetic mechanisms than the disease endpoint itself (i.e., diagnosed disorder), further investigation holds great potential for research on the genetics of complex psychiatric disorders. Understanding the mechanisms underlying different cognitive and social cognitive deficits is also important in order to develop targeted treatment for these deficits and associated social interaction and adjustment problems. Acknowledgements Preparation of this chapter was supported by the Academy of Finland (project #1115536; J.M.L.).

173

References 1. Gottesman II, Gould TD. The endophenotype concept in psychiatry: etymology and strategic intentions. Am J Psychiatry 2003;160: 636–645 2. Gur RE, Nimgaonkar VL, Almasy L, et al. Neurocognitive endophenotypes in a multiplex multigenerational family study of schizophrenia. Am J Psychiatry 2007;164: 813–819 3. Sharma T, Harvey P. Cognition in schizophrenia, importance and treatment strategy. New York: Oxford University Press; 2008 4. McFall RM. A review and reformulation of the concept of social skills. Behavioral Assessment 1982; 1–33 5. Mueser KT, Salyers MP, Mueser PR. A prospective analysis of work in schizophrenia. Schizophr Bull 2001;27: 281–296 6. Brothers L, Ring B, Kling A. Response of neurons in the macaque amygdala to complex social stimuli. Behav Brain Res 1990; 41: 199–213 7. Howells JG. The concept of schizophrenia: historical perspectives. Washington DC: American Psychiatric Press; 1991 8. Kee KS, Green MF, Mintz J, et al. Is emotion processing a predictor of functional outcome in schizophrenia? Schizophr Bull 2003; 29: 487–497 9. Hooker C, Park S. Emotion processing and its relationship to social functioning in schizophrenia patients. Psychiatry Res 2002; 112: 41–50 10. Brune M. Emotion recognition, ‘theory of mind,’ and social behavior in schizophrenia. Psychiatry Res 2005; 133: 135–147 11. Ekman P. Facial expression and emotion. Am Psychol 1993; 48: 384–392 12. Edwards J, Jackson HJ, Pattison PE. Emotion recognition via facial expression and affective prosody in schizophrenia: a methodological review. Clin Psychol Rev 2002; 22: 789–832 13. Morrison RL, Bellack AS, Mueser KT. Deficits in facialaffect recognition and schizophrenia. Schizophr Bull 1988; 14: 67–83 14. Bozikas VP, Kosmidis MH, Anezoulaki D, et al. Relationship of affect recognition with psychopathology and cognitive performance in schizophrenia. J Int Neuropsychol Soc 2004; 10: 549–558 15. Streit M, Ioannides A, Sinnemann T, et al. Disturbed facial affect recognition in patients with schizophrenia associated with hypoactivity in distributed brain regions: a magnetoencephalographic study. Am J Psychiatry 2001; 158: 1429–1436 16. Walker E, McGuire M, Bettes B. Recognition and identification of facial stimuli by schizophrenics and patients with affective disorders. Br J Clin Psychol 1984; 23 (Pt 1): 37–44 17. Kohler CG, Bilker W, Hagendoorn M, et al. Emotion recognition deficit in schizophrenia: association with symptomatology and cognition. Biol Psychiatry 2000; 48: 127–136 18. Borod JC, Martin CC, Alpert M, et al. Perception of facial emotion in schizophrenic and right brain-damaged patients. J Nerv Ment Dis 1993; 181: 494–502 19. Edwards J, Pattison PE, Jackson HJ, et al. Facial affect and affective prosody recognition in first-episode schizophrenia. Schizophr Res 2001; 48: 235–253 20. Leppanen JM, Niehaus DJ, Koen L, et al. Emotional face processing deficit in schizophrenia: A replication study in a

174

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34. 35.

36.

37. 38.

39.

R. Schoeman et al. South African Xhosa population. Schizophr Res 2006; 84: 323–330 Leentjens AF, Wielaert SM, van Harskamp F, et al. Disturbances of affective prosody in patients with schizophrenia; a cross sectional study. J Neurol Neurosurg Psychiatry 1998; 64: 375–378 Johnston PJ, McCabe K, Schall U. Differential susceptibility to performance degradation across categories of facial emotion–a model confirmation. Biol Psychol 2003; 63: 45–58 Johnston PJ, Devir H, Karayanidis F. Facial emotion processing in schizophrenia: no evidence for a deficit specific to negative emotions in a differential deficit design. Psychiatry Res 2006; 143: 51–61 Habel U, Krasenbrink I, Bowi U, et al. A special role of negative emotion in children and adolescents with schizophrenia and other psychoses. Psychiatry Res 2006; 145: 9–19 Suslow T, Roestel C, Arolt V. Affective priming in schizophrenia with and without affective negative symptoms. Eur Arch Psychiatry Clin Neurosci 2003; 253: 292–300 Suslow T, Roestel C, Ohrmann P, et al. Detection of facial expressions of emotions in schizophrenia. Schizophr Res 2003; 64: 137–145 Sweet LH, Primeau M, Fichtner CG, et al. Dissociation of affect recognition and mood state from blunting in patients with schizophrenia. Psychiatry Res 1998; 81: 301–308 Van’t WM, van DA, Aleman A, et al. Fearful faces in schizophrenia: the relationship between patient characteristics and facial affect recognition. J Nerv Ment Dis 2007; 195: 758–764 Kline JS, Smith JE, Ellis HC. Paranoid and nonparanoid schizophrenic processing of facially displayed affect. J Psychiatr Res 1992; 26: 169–182 Adolphs R. Recognizing emotion from facial expressions: psychological and neurological mechanisms. Behav Cogn Neurosci Rev 2002; 1: 21–62 Haxby JV, Hoffman EA, Gobbini MI. Human neural systems for face recognition and social communication. Biol Psychiatry 2002; 51: 59–67 Phillips ML, Drevets WC, Rauch SL, et al. Neurobiology of emotion perception I: The neural basis of normal emotion perception. Biol Psychiatry 2003; 54: 504–514 Hasselmo ME, Rolls ET, Baylis GC. The role of expression and identity in the face-selective responses of neurons in the temporal visual cortex of the monkey. Behav Brain Res 1989; 32: 203–218 Desimone R. Face-selective cells in the temporal cortex of monkeys. J Cogn Neurosci 1990; 3: 1–8 Kanwisher N, McDermott J, Chun MM. The fusiform face area: a module in human extrastriate cortex specialized for face perception. J Neurosci 1997; 17: 4302–4311 Allison T, Puce A, Spencer DD, et al. Electrophysiological studies of human face perception. I: Potentials generated in occipitotemporal cortex by face and non-face stimuli. Cereb Cortex 1999; 9: 415–430 Farah MJ. Is face recognition ‘special’? Evidence from neuropsychology. Behav Brain Res 1996; 76: 181–189 Marotta CR, Genovese CR, Behrmann M. A functional MRI study of face recognition in patients with prosopagnosia. Cognitive Neuroscience 2001; 12: 1–7 Narumoto J, Okada T, Sadato N, et al. Attention to emotion modulates fMRI activity in human right superior temporal sulcus. Brain Res Cogn Brain Res 2001; 12: 225–231

40. Winston JS, O’Doherty J, Dolan RJ. Common and distinct neural responses during direct and incidental processing of multiple facial emotions. Neuroimage 2003; 20: 84–97 41. Adolphs R, Tranel D, Damasio H, et al. Impaired recognition of emotion in facial expressions following bilateral damage to the human amygdala. Nature 1994; 372: 669–672 42. Adolphs R, Tranel D, Damasio AR. The human amygdala in social judgment. Nature 1998; 393: 470–474 43. Amaral DG. The amygdala, social behavior, and danger detection. Ann N Y Acad Sci 2003; 1000: 337–347 44. Aggleton JP. The Amygdala. Wiley, New York; 1993 45. Anderson AK, Phelps EA. Is the human amygdala critical for the subjective experience of emotion? Evidence of intact dispositional affect in patients with amygdala lesions. J Cogn Neurosci 2001; 14: 709–720 46. Vuilleumier P, Richardson MP, Armony JL, et al. Distant influences of amygdala lesion on visual cortical activation during emotional face processing. Nat Neurosci 2004; 7: 1271–1278 47. Rolls ET. The functions of the orbitofrontal cortex. Brain Cogn 2004; 55: 11–29 48. Yamada M, Hirao K, Namiki C, et al. Social cognition and frontal lobe pathology in schizophrenia: a voxel-based morphometric study. Neuroimage 2007; 35: 292–298 49. Gur RE, Maany V, Mozley PD, et al. Subcortical MRI volumes in neuroleptic-naive and treated patients with schizophrenia. Am J Psychiatry 1998; 155: 1711–1717 50. Lieberman J, Chakos J, Wu H. Longitudinal study of brain morphology in first episode schizophrenia. Biol Psychiatry 2001; 49: 487–499 51. Lee C, Shenton ME, Salisbury DF, et al. Fusiform gyrus volume reduction in first episode schizophrenia: a MRI study. Arch Gen Psychiatry 2002; 59: 775–781 52. Wright IC, Rabe-Hesketh S, Woodruff PW. Meta-analysis of regional brain volumes in schizophrenia. Am J Psychiatry 2000; 157: 16–25 53. Loughland CM, Williams LM, Gordon E. Schizophrenia and affective disorder show different visual scanning behavior for faces: a trait versus state-based distinction? Biol Psychiatry 2002; 52: 338–348 54. Shimizu T, Shimizu A, Yamashita K, et al. Comparison of eye-movement patterns in schizophrenic and normal adults during examination of facial affect displays. Percept Mot Skills 2000; 91: 1045–1056 55. Streit M, Wolwer W, Gaebel W. Facial-affect recognition and visual scanning behaviour in the course of schizophrenia. Schizophr Res 1997; 24: 311–317 56. Streit M, Wolwer W, Brinkmeyer J, et al. EEG-correlates of facial affect recognition and categorisation of blurred faces in schizophrenic patients and healthy volunteers. Schizophr Res 2001; 49: 145–155 57. Herrmann MJ, Reif A, Jabs BE, et al. Facial affect decoding in schizophrenic disorders: a study using event-related potentials. Psychiatry Res 2006; 141: 247–252 58. Johnston PJ, Stojanov W, Devir H, et al. Functional MRI of facial emotion recognition deficits in schizophrenia and their electrophysiological correlates. Eur J Neurosci 2005; 22: 1221–1232 59. Mikhailova ES, Tsutsul’kovskaia MI, Oleichik IV. [Neurophysiologic mechanisms of the disorders in recognition of emotions in endogenic depression]. Zh Nevrol Psikhiatr Im S S Korsakova 2000; 100: 38–43

11

Emotion Recognition Deficits as a Neurocognitive Marker of Schizophrenia Liability

60. Turetsky BI, Kohler CG, Indersmitten T, et al. Facial emotion recognition in schizophrenia: when and why does it go awry? Schizophr Res 2007; 94: 253–263 61. Chen Y, Norton D, Ongur D, et al. Ineffecient face detection in schizophrenia. Schizophr Bull 2008; 34: 367–374 62. Phan KL, Fitzgerald DA, Gao K, et al. Real-time fMRI of cortico-limbic brain activity during emotional processing. Neuroreport 2004; 15: 527–532 63. Hempel A, Hempel E, Schonknecht P, et al. Impairment in basal limbic function in schizophrenia during affect recognition. Psychiatry Res 2003; 122: 115–124 64. Quintana J, Wong T, Ortiz-Portillo E, et al. Right lateral fusiform gyrus dysfunction during facial information processing in schizophrenia. Biol Psychiatry 2003; 53: 1099–1112 65. Wolf K, Mass R, Kiefer F, et al. Characterization of the facial expression of emotions in schizophrenia patients: preliminary findings with a new electromyography method. Can J Psychiatry 2006; 51: 335–341 66. Quintana J, Davidson T, Kovalik E, et al. A compensatory mirror cortical mechanism for facial affect processing in schizophrenia. Neuropsychopharmacology 2001; 25: 915–924 67. Addington J, Addington D. Facial affect recognition and information processing in schizophrenia and bipolar disorder. Schizophr Res 1998; 32: 171–181 68. Gaebel W, Wolwer W. Facial expression and emotional face recognition in schizophrenia and depression. Eur Arch Psychiatry Clin Neurosci 1992; 242: 46–52 69. Addington J, Saeedi H, Addington D. Facial affect recognition: a mediator between cognitive and social functioning in psychosis? Schizophr Res 2006; 85: 142–150 70. Kucharska-Pietura K, David AS, Masiak M, et al. Perception of facial and vocal affect by people with schizophrenia in early and late stages of illness. Br J Psychiatry 2005; 187: 523–528 71. Mueser KT, Doonan R, Penn DL, et al. Emotion recognition and social competence in chronic schizophrenia. J Abnorm Psychol 1996; 105: 271–275 72. Minoshita S, Morita N, Yamashita T, et al. Recognition of affect in facial expression using the Noh Mask Test: comparison of individuals with schizophrenia and normal controls. Psychiatry Clin Neurosci 2005; 59: 4–10 73. Green MJ, Williams LM, Davidson DJ. Processing of threatrelated affect is delayed in delusion-prone individuals. Br J Clin Psychol 2001; 40: 157–165 74. Lewis SF, Garver DL. Treatment and diagnostic subtype in facial affect recognition in schizophrenia. J Psychiatr Res 1995; 29: 5–11 75. Poreh AM, Whitman RD, Weber M, et al. Facial recognition in hypothetically schizotypic college students. The role of generalized poor performance. J Nerv Ment Dis 1994; 182: 503–507 76. Toomey R, Schuldberg D. Recognition and judgment of facial stimuli in schizotypal subjects. J Commun Disord 1995; 28: 193–203 77. Bolte S, Poustka F. The recognition of facial affect in autistic and schizophrenic subjects and their first-degree relatives. Psychol Med 2003; 33: 907–915 78. Leppanen JM, Niehaus DJ, Koen L, et al. Deficits in facial affect recognition in unaffected siblings of Xhosa schizophrenia patients: evidence for a neurocognitive endophenotype. Schizophr Res 2008; 99: 270–273

175

79. Toomey R, Seidman LJ, Lyons MJ, et al. Poor perception of nonverbal social-emotional cues in relatives of schizophrenic patients. Schizophr Res 1999; 40: 121–130 80. Loughland CM, Williams LM, Harris AW. Visual scanpath dysfunction in first-degree relatives of schizophrenia probands: evidence for a vulnerability marker? Schizophr Res 2004; 67: 11–21 81. Wolwer W, Frommann N, Halfmann S, et al. Remediation of impairments in facial affect recognition in schizophrenia: efficacy and specificity of a new training program. Schizophr Res 2005; 80: 295–303 82. Harrison PJ, Owen MJ. Genes for schizophrenia? Recent findings and their pathophysiological implications. Lancet 2003; 361: 417–419 83. Braff D, Schork NJ, Gottesman II. Endophenotyping schizophrenia. Am J Psychiatry 2007; 164: 705–707 84. Gottesman II, Shields J. Genetic theorizing and schizophrenia. Br J Psychiatry 1973; 122: 15–30 85. Marcelis M, Suckling J, Woodruff P, et al. Searching for a structural endophenotype in psychosis using computational morphometry. Psychiatry Res 2003; 122: 153–167 86. McDonald C, Grech A, Touopoulouy T, et al. Brain volumes in familial and non-familial schizophrenic probands and their unaffected relatives. Am J Med Gen 2002; 114: 616–625 87. Dickey CC, McCarely RW, Voglmaier MM, et al. Schizotypal personality disorder and MRI abnormalities of temporal lobe gray matter. Biol Psychiatry 1999; 45: 1393–1402 88. Yoneyama E, Matsui M, Kawasaki Y, et al. Gray matter features of schizotypal disorder patients exhibiting the schizophrenia-related code types of the MMPI. Acta Psychiatrica Scand 2003; 108: 333–340 89. Freedman R, Adams CE, Adler LE, et al. Inhibitory neurophysiological deficit as a phenotype for genetic investigation of schizophrenia. Am J Med Gen 2000; 97: 58–64 90. Cadenhead KS, Swerdlow NR, Shafer KM, et al. Modulation of the startle response and startle laterality in relatives of schizophrenic patients and in subjects with schizotypal personality disorder: evidence of inhibitory deficits. Am J Psychiatry 2000; 157: 1660–1668 91. Ross RG, Meinlein S, Zerbe GO, et al. Saccadic eye movement task identifies cognitive deficits in children with schizophrenia, but not in unaffected child relatives. J Child Psychol Psychiatr 2005; 46: 1354–1362 92. Harrison PJ, Weinberger DR. Schizophrenia genes, gene expression and neuropathology: on the matter of their convergence. Mol Psychiatr 2005; 10: 40–68 93. Lewis CM, Levinson DF, Wise LH, et al. Genome scan meta-analysis of schizophrenia and bipolar disorder, part II: Schizophrenia. Am J Hum Genet 2003; 73: 34–48 94. van Amelsvoort T, Henry J, Morris R, et al. Cognitive deficits associated with schizophrenia in velo-cardio-facial syndrome. Schizophr Res 2004; 70: 223–232 95. Lawrie SM, Hall J, McIntosh AM, et al. Neuroimaging and molecular genetics of schizophrenia: pathophysiological advances and therapeutic potential. Br J Pharmacol 2008; 153(Suppl 1): S120–S124 96. Critchley H, Daly E, Phillips M, et al. Explicit and implicit neural mechanisms for processing of social information from facial expressions: a functional magnetic resonance imaging study. Hum Brain Mapp 2000; 9: 93–105

176 97. Mayberg HS. Limbic-cortical dysregulation: a proposed model of depression. J Neuropsychiatry Clin Neurosci 1997; 9: 471–481 98. Seidman LJ, Giuliano AJ, Smith CW, et al. Neuropsychological functioning in adolescents and young adults at genetic risk for schizophrenia and affective psychoses: results from the Harvard and Hillside Adolescent High Risk Studies. Schizophr Bull 2006; 32: 507–524 99. Aleman A, Hijman R, de Haan EH, et al. Memory impairment in schizophrenia: a meta-analysis. Am J Psychiatry 1999; 156: 1358–1366 100. Snitz BE, Macdonald AW, III, Carter CS. Cognitive deficits in unaffected first-degree relatives of schizophrenia patients: a meta-analytic review of putative endophenotypes. Schizophr Bull 2006; 32: 179–194 101. Gur RE, Calkins ME, Gur RC, et al. The Consortium on the Genetics of Schizophrenia: neurocognitive endophenotypes. Schizophr Bull 2007; 33: 49–68 102. Iidaka T, Ozaki N, Matsumoto A, et al. A variant C178T in the regulatory region of the serotonin receptor gene HTR3A modulates neural activation in the human amygdala. J Neurosci 2005; 25: 6460–6466 103. Hariri AR, Weinberger DR. Imaging genomics. Br Med Bull 2003; 65: 259–270 104. Battaglia M, Ogliari A, Zanoni A, et al. Influence of the serotonin transporter promoter gene and shyness on

R. Schoeman et al.

105.

106.

107.

108.

109.

110.

111.

children’s cerebral responses to facial expressions. Arch Gen Psychiatry 2005; 62: 85–94 Skuse DH, James RS, Bishop DV, et al. Evidence from Turner’s syndrome of an imprinted X-linked locus affecting cognitive function. Nature 1997; 387: 705–708 van RS, Swaab H, Aleman A, et al. X Chromosomal effects on social cognitive processing and emotion regulation: A study with Klinefelter men (47,XXY). Schizophr Res 2006; 84: 194–203 Good KP, Kopala LC, Bassett A, et al. Microsmia in postmenopausal women with genetic vulnerability to psychosis. Schizophr Res 2003; 61: 327–328 Fakra E, Salgado-Pineda P, Besnier N, et al. Risperidone versus haloperidol for facial affect recognition in schizophrenia: Findings from a randomised study. World J Biol Psychiatry 2007; 1–10 Ibarraran-Pernas GY, Guevara MA, Cerdan LF, et al. [Olanzapine effects on emotional recognition in treatment refractory schizophrenics]. Actas Esp Psiquiatr 2003; 31: 256–262 Pinkham AE, Gur RE, Gur RC. Affect recognition deficits in schizophrenia: neural substrates and psychopharmacological implications. Expert Rev Neurother 2007; 7: 807–816 Russell TA, Chu E, Phillips ML. A pilot study to investigate the effectiveness of emotion recognition remediation in schizophrenia using the micro-expression training tool. Br J Clin Psychol 2006; 45: 579–583

Chapter 12

The Use of Neurocognitive Endophenotypes in Large-Scale Family Genetic Studies of Schizophrenia William P. Horan, Tiffany A. Greenwood, David L. Braff, Raquel E. Gur, and Michael F. Green

Abstract Neurocognitive deficits are core features of schizophrenia and are among the most promising candidate endophenotypes for genetic studies of this disorder. The detection of gene–endophenotype associations requires large family cohort or case-control samples that are often only possible to collect through multisite collaborations, which presents considerable challenges for the use of endophenotypes. This chapter focuses on the rationale for using neurocognitive tasks of working memory, attention, and verbal declarative memory in large-scale, collaborative efforts to apply an endophenotype approach to schizophrenia. As an example, we describe the Consortium on the Genetics of Schizophrenia (COGS), a seven-site research network that investigates the genetic architecture of these neurocognitive and other candidate endophenotypes in families with schizophrenia. After providing a brief overview of the rigorous recruitment, data acquisition, and quality assurance procedures established by the COGS, we present recent results from this project that support utility of these procedures and the validity of neurocognitive endophenotypes. We conclude with a discussion of preliminary COGS findings that point toward specific candidate genes and future directions for this and other large-scale endophenotype studies of schizophrenia.

W. P. Horan University of California, Los Angeles T. A. Greenwood and D.L. Braff University of California, San Diego R. E. Gur University of Pennsylvania M. F. Green University of California, Los Angeles & VA Greater Los Angeles Healthcare Center

Keywords Schizophrenia • endophenotype • neurocognition Abbreviations CCC: Clinical Core Committee; CCS: Community control subjects; COGS: Consortium on the Genetics of Schizophrenia; CPT: Continuous Performance Test; CVLT-II: California Verbal Learning Test-II; DS-CPT: Degraded Stimulus-Continuous Performance Test; EGF: Epidermal growth factor; EPC: Endophenotype Committee; LNS: LetterNumber Span; NMDA: N-methyl-D-aspartate; NRG1: Neuregulin-1; Penn CNB: Penn Computerized Neurocognitive Battery; QA: Quality assurance; SNP: Single nucleotide polymorphisms; WM: Working memory; WMS-III: Wechsler Memory Scale-III

Introduction Schizophrenia is a complex neuropsychiatric disorder that is defined by diverse and fluctuating signs and symptoms. Although decades of research have clearly demonstrated that genetic factors play a key role in the pathogenesis of schizophrenia,1,2 identification of a specific susceptibility gene, or a set of genes, remains elusive. A complementary strategy that holds considerable promise for advancing our understanding the genetic architecture of this disorder is the analysis of discrete and neurobiologically relevant “endophenotypic” abnormalities.3 Endophenotypes are characteristics, usually objectively assayed in a laboratory, that reflect the actions of genes predisposing an individual to a disorder, even in the absence of any diagnosable pathology. Ideally, therefore, endophenotypes could serve as dissected components of the complex schizophrenia phenotype, reflecting fewer genes and thereby

M.S. Ritsner (ed.), The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes, © Springer Science + Business Media B.V. 2009

177

178

reducing the complexity of the genetic analyses required to identify risk genes. Among the candidate endophenotypes for use in schizophrenia, neurocognitive functions have received a great deal of attention. Neurocognitive impairment is a hallmark of schizophrenia. Affected individuals demonstrate broad compromise of neurocognitive functions, including speed of processing, attention/vigilance, working memory, verbal learning and memory, visual learning and memory, reasoning and problem solving, and verbal comprehension.4 Neurocognitive deficits appear to be core features of schizophrenia because they are: (1) often detectable before the onset of psychotic symptoms and other clinical features, (2) present in neuroleptic naïve and first episode patients without extensive exposure to antipsychotic medications, (3) present across periods of both symptom exacerbation and remission, (4) strongly related to functional outcome, and (5) detectable in attenuated form in unaffected biological relatives of schizophrenia probands (for reviews see refs.5–7). In addition, the neurophysiology of several of the cognitive functions that are affected in schizophrenia have been fairly well characterized, and various candidate genes that may impact these functions are starting to be identified. These characteristics strongly support the use of neurocognitive tests in endophenotypic studies of schizophrenia. To achieve sufficient statistical power to demonstrate gene–endophenotype associations, genetic studies require large samples that are often only possible via multi-site collaborative efforts. Multisite studies present challenges for the use of endophenotypes because differences in methodologies or test conditions across sites can introduce uncontrolled variance into the experimental measures, potentially obscuring detection of signals used to identify vulnerability genes. The COGS is the first multisite, large-scale effort to apply a comprehensive endophenotype approach to schizophrenia by investigating the genetic architecture of six carefully selected candidate neurocognitive and neurophysiological endophenotypes in families with schizophrenia and community comparison subjects. The seven participating sites in this NIMH-sponsored collaboration are Harvard University, Mount Sinai School of Medicine, University of California Los Angeles, University of California San Diego, University of Colorado, University of Pennsylvania, and University of Washington. The COGS recruitment strategy is designed to maximize the power of genetic linkage and

W.P. Horan et al.

association studies by enrolling relatively large family pedigrees that include both affected and unaffected members. To date, clinical, endophenotype, and genetic data has been collected from over 350 families and 450 community comparison subjects. To manage the complexities of reliably collecting endophenotype data in large, multisite studies, the COGS has developed rigorous data acquisition and quality assurance procedures. This chapter focuses on the candidate neurocognitive endophenotypes from the COGS project (see ref.8 for a review of the neurophysiological endopenotypes). The chapter provides an overview of (a) the rationale for the use of neurocognitive endophenotypes selected for the COGS project, (b) the recruitment strategy, data acquisition infrastructure, and quality assurance procedures established by the COGS, (c) recent findings that support the reliability and validity of these procedures, and (d) initial COGS findings pointing toward specific candidate genes. We conclude by discussing future directions for the COGS and other large scale endophenotype studies of schizophrenia.

The COGS Candidate Neurocognitive Endophenotypes The COGS endophenotype battery includes three primary measures from three neurocognitive domains plus several secondary measures. The selection of these measures was guided by the criteria that the endophenotype is: (1) associated with the illness (i.e., demonstrate large and significant separations between patients and the general population), (2) largely clinical state- and medication-independent, (3) found in unaffected relatives at a higher rate than in the general population, and (4) shows evidence of heritability.3 Long-term stability and co-segregation of endophenotype and illness within families were also considerations, although data directly relevant to these characteristics are relatively limited. Additional considerations included selection of cognitive tasks with known neurobiological circuitry (or patterns of neural activation) relevant to schizophrenia, as well as practicality of reliable task administration in a lengthy, multisite protocol. Based on these considerations and extensive literature reviews, the COGS investigators selected tasks from the domains of working memory

12

The Use of Neurocognitive Endophenotypes

(WM), attention, and verbal memory. In addition, a broad computerized neurocognitive battery was included to characterize participants and to provide additional potential endophenotypes. The following sections briefly review the rationale for selecting these domains.

Working Memory Working memory deficits have been extensively documented in schizophrenia and have been proposed to reflect fundamental disturbances that underlie many of the cognitive disturbances associated with this disorder.9 Working memory is commonly defined as a limitedcapacity storage system for the temporary maintenance and manipulation of information.10,11 Although various models of the specific component processes of WM have been proposed, schizophrenia researchers often distinguish two broad classes of WM tasks.12 The first type assesses transient on-line maintenance functions that do not involve manipulation of the stored information. Examples include spatial delayed response tasks or digit span forward tasks. The second class involves maintenance plus manipulation of information or “executive functioning WM”. Examples include N-back tasks, digit or spatial span backward tasks, or letternumber sequencing tasks, which require subjects to both categorize alternating letters and numbers into separate classes and reorder the stimuli within each task. Individuals with schizophrenia consistently show significant WM deficits across a diverse range of tasks.13 More severe impairment is often seen on tasks that involve maintenance plus complex manipulation.14 For example, in the verbal domain, patients and healthy controls show average separations of about 0.71–0.82 standard deviations on digit span forward and backward repetition tasks,15 but group differences can exceed 1.4 standard deviations on more complex letter-number sequencing tasks.12,16,17 WM deficits show minimal cross-sectional correlations with severity of delusions and hallucinations, are detectable in clinically stabilized outpatients, and are relatively stable across time and clinical status changes,18–22 suggesting that they are not merely secondary manifestations of psychotic symptoms. They are also not secondary to antipsychotic medication side effects or factors associated with chronicity, as deficits of comparable magni-

179

tude are present in neuroleptic-free patients and during the immediate post-onset period.23,24 Similar, though attenuated, WM disturbances are also present in clinically unaffected biological relatives of schizophrenia patients, who score about 0.25– 0.50 standard deviations below healthy controls across verbal and spatial WM tasks.25,26 In addition, WM ability appears to be substantially heritable, with estimates of about 0.36–0.49 found in both non-clinical samples and in family studies of schizophrenia probands.27,28 Finally, the functional neuroanatomy of WM has been fairly well characterized in animal and human studies, and abnormalities in the neural systems that are critical for intact WM performance have been extensively documented in schizophrenia. In particular, the dorsolateral prefrontal cortex and dopaminergic system, in conjunction with posterior parietal cortex, are centrally involved in WM (particularly executive) functions.29 Abnormalities in dopaminergic functioning and the structure of these cortical regions are strongly implicated in schizophrenia, and both affected individuals and their relatives show abnormal neural activation patterns while performing WM tasks.30–33 WM tasks, particularly in the verbal domain, are highly amenable to use in endophenotype batteries that are administered across multiple sites. For example, the Letter-Number Span (LNS), a subtest from the wellstandardized and normed Wechsler Memory Scale III,34 is fairly simple to administer following standardized training, requires minimal equipment, and is relatively brief (less than 15 min). Subjects are verbally presented with clusters of intermixed letters and numbers (e.g., E,6,R,2). The COGS administers the LNS under two conditions: (1) standard condition, in which the subjects reorder the letters and numbers (i.e., executivefunctioning working memory); (2) a condition in which the subjects repeat each series without reordering (i.e., transient online storage and retrieval).

Attention Deficits in attention have long been considered central features of the clinical presentation of schizophrenia35,36 and have been a consistent focus of the experimental psychopathology of schizophrenia. Attention is dysfunctional in schizophrenia in several ways, including sustained focused attention,37,38 selective attention,39

180

and cognitive control of attention.40 The deficit in sustained focused attention has garnered the most support as an attention endophenotype for schizophrenia.41–43 Continuous Performance Tests (CPTs) have become the most widely used measures of deficits in sustained focused attention in schizophrenia. CPTs typically require subjects to monitor a quickly paced series of stimuli (e.g., letters, digits) using brief stimulus durations (usually 30–100 ms) and relatively short periods of vigilance (5–15 min), and to respond each time that a target stimulus appears. The sensitivity of CPTs is often enhanced by incorporating perceptual loads (e.g., using degraded digits as stimuli) or working memory loads (e.g., using sequences of digits as target stimuli). Individuals with schizophrenia consistently show large performance deficits on CPT’s, with meta-analyses reporting a mean effect size of 1.18.44 Deficits on high information processing load CPTs are present in medication naïve and unmedicated patients45,46 and relatively persistent across time and changes in clinical status.47,48 On high processing load CPTs, impairment is also reliably detected in unaffected relatives, with medium separations (d = 0.54) between relatives and healthy subjects.26 In addition, initial evidence suggests at least moderate heritability for CPT performance; heritabilites of 0.39–0.4948 have been reported in studies of healthy families and estimates ranging from 0.48–0.7949,50 have been reported in families of schizophrenia patients. Although the neurobiological substrate of CPT performance has not been extensively studied, functional neuroimaging studies have demonstrated some meaningful patterns. Performance on tasks of sustained attention is largely subserved by an anterior network that involves coordinated activity in the prefrontal cortex, the thalamus (particularly dorsomedial nucleus), and the anterior cingulate cortex (see ref.).51 Structural abnormalities in these regions are present in schizophrenia, and both schizophrenia patients and their relatives demonstrate abnormal activation in components of this neural network while performing CPT’s.51–54 The availability of fully computerized CPT’s that include standardized stimuli, parameter settings, and administration procedures (e.g., ambient lighting requirements) facilitates multi-site assessments of attention. The COGS investigators selected two high information processing load CPTs, the Degraded Stimulus-CPT55,56 and the Identical Pairs-CPT,48,57 which each take about 15 min to administer. In the

W.P. Horan et al.

DS-CPT, single digits 0 through 9, which are made to appear highly blurred by randomizing black/white values of 40% of the pixels, are presented in quasi-random order at a 1 per second (29 ms exposures). The task is to monitor the rapid series of digits and to respond with a mouse press to each blurred 0. In the CPT-IP, subjects are asked to respond each time that the same digit string (either three- or four-digits long) occurs twice in a row in a within a quasi-random sequence.

Verbal Memory Verbal episodic or declarative memory is consistently found to be among the most impaired neurocognitive functions in schizophrenia. This dysfunction takes several forms, including deficits in acquisition/encoding, memory storage, and retrieval. These processes can be distinguished on standardized memory tasks, including list-learning tasks, such as the California Verbal Learning Test (CVLT-II58), and story recall tasks, such as the Logical Memory subtest of the Wechsler Memory Scales (WMS-III34). On these types of tasks, individuals with schizophrenia show pronounced deficits on encoding and organizing new information and retrieval of information using freerecall paradigms, but better performance on cued or recognition conditions. Well-over 100 studies document declarative memory deficits in schizophrenia patients, with effect sizes rating from 1.0–1.5.15,59,60 These deficits are present in recent-onset and chronically ill patients, present in medication naïve and unmedicated patients, and only minimally related to psychotic symptoms.61–63 Similar types of declarative memory impairment are also reliably detected among biological relatives of schizophrenia patients, with effect sizes ranging from 0.25–0.55 for relative versus control comparisons.25,26,64 In addition, declarative memory deficits are moderately heritable in healthy families (0.47–0.6365,66). The few studies of families with schizophrenia report somewhat variable estimate across different indices, with 0.21 found for list and estimates ranging from 0.49–0.66 for recognition memory.67,68 The neural correlates of declarative memory have been extensively researched, with human lesion and functional neuroimaging studies demonstrating that the medial temporal lobe (hippocampus and surrounding

12

The Use of Neurocognitive Endophenotypes

cortex) and prefrontal cortical regions are critical for intact memory formation (e.g., ref.69). Reduced hippocampal volumes are detectable in patients and their relatives,70 and both groups show abnormal patterns of frontotemporal activation during the performance of declarative memory tests.71–73 As noted above, a number of standardized verbal learning and memory tasks have been widely used in schizophrenia research. The COGS project uses the CVLT-II,58 which is normed and fairly straight-forward to administer in a standardized manner. The CVLT-II involves oral administration of a 16-item list of words from four semantic categories that is initially presented five times. Short-term retention is assessed shortly after the fifth administration trial, and then delayed retention is assessed 20 min later. The task takes approximately 20 mins to administer, excluding the delay interval, and includes a computerized scoring program that produces several indices of performance.

Penn Computerized Neurocognitive Battery In addition to the three primary candidate neurocognitive endophenotypes just described, the COGS uses a broad neuropsychological battery to more fully characterize the participants and to provide additional potential endophenotypes. This battery includes six tasks from the Penn Computerized Neurocognitive Battery (CNB) to assess the following domains: Abstraction and Mental Flexibility (Penn Conditional Exclusion Test74), Spatial Memory (Visual Object Learning Test75), Spatial Processing (Computerized Judgment of Line Orientation76), Sensorimotor Dexterity (Computerized Finger-Tapping Task and Motor Praxis test), Face Memory (Penn Face Memory Test,77 which assesses immediate and delayed recognition), and Emotion Processing (Penn Emotion Recognition Test,78 which assesses face emotion recognition). The CNB was designed for use in largescale studies; it demonstrates good psychometric properties and was validated in healthy people and individuals with schizophrenia.76,79 Administration of the COGS version of the Battery takes approximately 60 min, including standardized rest breaks. The domains selected from the Penn battery reflect a range of abilities, not all of which have been implicated

181

as candidate neurocognitive endophenotypes for schizophrenia. In particular, although the face processing abilities assessed by the Emotion Recognition and Face Memory Tasks have been studied in schizophrenia (particularly emotion recognition), they have typically not been considered in the context of research that employs the endophenotype strategy. However, emerging evidence supports their endophenotype candidacy (see ref.80). Thus, the Penn battery provides an efficient means to obtain additional information about the neurocognitive characteristics of the COGS sample and, perhaps, to identify promising additional endophenotypes.

The COGS Structure and Data Acquisition Procedures Successful multi-site collaborations that involve extensive endophenotyping require careful standardization and continuous quality assured administration of measures across sites. The COGS established an infrastructure, standardized clinical, endophenotyping, and genotyping procedures, and continuous quality assurance procedures to address the complexities of carrying out this type of research. The overall structure of the COGS is presented in Fig. 12.1 and described in the following sections.

Organizational Structure As shown in the upper right quadrant of Fig. 12.1, the COGS is organized into seven Core Units that work interactively to collect, process, and analyze data.81 In addition to annual in-person meetings at scientific conferences, the investigators and research staff are organized into four separate committees that each hold teleconferences twice per month. Under the leadership of the Director’s Unit at UCSD (David L. Braff, Director), the Principal Investigators Committee, comprised of the PI’s from each site, is the decision-making body regarding policies, procedures, and publications. At the next level, the Endophenotype Committee (EPC), comprised of the senior quality assurance (QA) scientists at each site, oversees procedures for accurate collection and data transmission of

182

W.P. Horan et al.

Fig. 12.1 Overview of COGS structure and data management procedures

the endophenotypes, and the Clinical Core Committee (CCC), comprised of one faculty-level representative from each site, oversees the recruitment and clinical assessment of research participants. Finally, the General Investigators Committee, comprised of all investigators and research staff, allows for the integration and discussion of all issues, especially those raised by the CCC and the EPC. Additional components of the COGS structure include the Bioinformatics and Data Core at UCLA, which is responsible for maintaining a secure web-based COGS data warehouse and providing statistical consultation to COGS investigators, and the Statistical Genetics Core at UCSD, which provides specialized expertise and data analysis for the statistical genetics component of the consortium.

Recruitment and Clinical Assessments The COGS recruitment procedures are designed to maximize the power of genetic linkage and association

studies. The minimal requirement for pedigree ascertainment includes a proband who meets DSM-IV criteria for schizophrenia, both parents (at least one unaffected), and at least one unaffected sibling. Probands having only one available parent but two or more available siblings (with at least one unaffected by schizophrenia) are also included, as are probands having no available parents but three or more available siblings (with at least one unaffected by schizophrenia), though these configurations are less powerful. Sampling both affected and unaffected individuals increases variation or “contrast” in the proposed endophenotypes (see refs.82,83). This approach is distinct from strategies used for studying a single qualitatively-defined disease, where affected sibling pairs and multiplex families provide the most efficient and powerful design for allele-sharing linkage analysis.84 All probands, family members, and CCS undergo standardized Axis I diagnostic85 and symptom86,87 assessments. Family members and CCS also complete an assessment of schizotypal and other axis II cluster A personality features.88 The Family Interview for

12

The Use of Neurocognitive Endophenotypes

Genetic Studies89 also completed for all families and CCS to obtain additional family history information, resulting in the creation of a composite pedigree that is informed by multiple participants. Following the interview, each subject is assigned DSM-IV best-estimate final diagnoses through a consensus diagnostic process that includes at least two faculty level clinicians. The CCS serve as a reference group against which to gauge performance of probands and family members. CCS are included if they have no personal history of psychosis or Cluster A Personality Disorder and no family history of psychosis a first- or second-degree relative. To parallel comorbidity outside of the schizophrenia spectrum in relatives of probands, non-psychotic Axis I psychopathology is accepted in approximately 30% of CCS. All participants who are endophenotyped are between the ages of 18–65 and have their blood drawn for genotyping purposes. Family members older than 65 have their blood drawn for genotyping, but do not undergo endophenotyping.

Neurocognitive Endophenotyping and Quality Assurance The full endophenotype battery (neurocognitive and neurophysiological tasks) is administered in one of two fixed orders that take a total testing time of 4–6 h. All neurocognitive tasks are administered by highly trained testers following a detailed, standardized administration manual. Each endophenotype has an assigned QA Site that reviews and scores data coming from all seven COGS sites. To ensure that each site is testing according to protocol, UCSD hosted an initial 4-day training workshop attended by testers from all seven sites along with all endophenotype faculty QA experts prior to initiation of the study. Retraining is held at UCSD on an annual basis to minimize tester drift and to train and “COGS Certify” newly hired testers.

Data Flow As shown in the lower right-hand quadrant of Fig. 12.1, blood samples for DNA analysis are sent from the local site to the Rutgers University Cell and DNA Repository. Data management depends on a constant iterative relationship between the seven sites and the

183

UCLA Data Core. Clinical assessment data and endophenotype data are uploaded separately. Once the information has been transmitted to the central Data Core, the QA scientist responsible for each endophenotype regularly downloads all new endophenotype data collected on that measure across all sites. The QA scientists review and score the endophenotype data, blind to diagnostic group, according to procedures established for each particular measure. Any data questions that arise during this QA process are reviewed with the personnel at the site at which the data were collected. Thus, the management of multisite, large-scale family studies of endophenotypes is a complex endeavor. Since its inception, the COGS has devoted considerable resources to establishing and maintaining an infrastructure to ensure high quality data collection from large samples of probands, family members, and CCS. Recent findings from this unique data set support the utility of these coordinated data acquisition procedures.

Recent Findings from the COGS Project A major goal of the first COGS funding period (“COGS 1”) has been to assess the presence of patient – control differences and heritability of the selected endophenotypes. The following sections describe findings from two recently published studies that demonstrate how the COGS data set has been used to evaluate these characteristics of the neurocognitive endophenotypes.

Working Memory: Schizophrenia Deficits and Heritability In an initial evaluation of the validity of the COGS candidate WM endophenotype, the performance of 149 probands, 337 of their first-degree relatives, and 190 community comparison subjects (CCS) on the LNS was examined.90 As shown in Table 12.1, probands had a higher proportion of males, fewer years of completed education, and higher parental education than both other groups. The relatives were significantly older than the proband and CCS groups. Probands had a typical age of onset (M = 21.4 years, SD = 6.1), were predominantly chronically ill (M = 13.5 years, SD = 10.5),

184

W.P. Horan et al. Table 12.1 Demographic and characteristics probands (n = 149), relatives (n = 337), and controls (n = 190) Probands Age (years) (SD) Sex (% male) Education (years) (SD) Parental education (years) (SD)

Relatives

a

b

34.99 (10.86) 72%a 13.52a (2.08) 15.73a (3.36)

45.41 (13.92) 42%b 15.46b (2.66) 14.60b (3.68)

Controls a

37.35 (12.10) 45.9%b 15.38b (2.81) 14.82b (2.98)

Statistic F = 44.07** X2 = 38.60** F = 31.48** F = 5.09*

Groups with different superscript are significantly different from each other. The relatives group was comprised of 118 parents and 206 siblings. * p < 0.05; **p < 0.001

Table 12.2 Letter-number sequencing performance across groups Probands

Relatives

Controls

Age

Parental education Group

(n = 149)

(n = 324)

(n = 190)

F

F

LNS Reorder 9.18 (2.55) 10.67 (2.59) 11.59 (2.55) 25.90*** 1.35 (M) (SD) LNS Forward 12.84 (2.74) 13.74 (3.02) 14.38 (2.90) 43.17*** 5.43* (M) (SD)

F

Post-hoc t-tests PvC

Effect size

PvR RvC PvC PvR RvC

33.48*** 8.10*** 5.85*** 2.52* .94

.58

.36

12.90*** 4.96*** 3.76** 1.12

.31

.22

.53

LNS = Letter-Number Sequencing. Means are based on raw scores. LNS Reordered data were missing for 2 subjects in the Relatives group (n = 322) and 1 was missing in the Proband group (n = 148). Statistical tests are based on mixed effects regression models including family as a random effect, age, sex, and parental education as covariates, and group, site, and group X site as fixed effects. Effect size estimates were calculated using the sample-size weighted pooled within-group standard deviations (LNS-Reordered = 2.56; LNS-Forward = 2.90). P v. C: a comparison of schizophrenia probands and comparison groups. P v. R: a comparison of relatives and schizophrenia probands. R v. C: a comparison of relative and comparison groups. * p < 0.05; ** p < 0.005; *** p < 0.0001

and demonstrated low to moderate levels of symptoms characteristic of an outpatient sample. The LNS data was analyzed using mixed effects regression analyses with family as a random effect to account for the nonindependence of observations among geneticallyrelated family members in the proband and relative groups, and pre-selected fixed effects (group, research site, and a group X site interaction term) and covariates (age, sex, and parental education). In the mixed effects regression models, there were no significant effects for research site, group X site, or sex. The absence of significant site-related effects bolsters confidence that WM assessments can be reliably conducted across seven geographically diverse research sites when administration of tests is rigorously standardized and continuously monitored. In this context, it is noteworthy that other COGS publications have examined the feasibility of collecting more technically challenging neurophysiological endophenotypes,

including anti-saccade and pre-pulse inhibition tasks, and also did not detect significant between-site differences.91,92 In combination, these findings support the usefulness of the rigorous COGS quality assurance procedures. The key results are summarized in Table 12.2. On the primary variable, the LNS-Reordering condition, the proband and relative groups both performed significantly worse than CCS. The magnitude of impairment was large for probands and medium for relatives, which is consistent with expectations for a valid endophenotype. A similar pattern was observed for the secondary measure, LNS-Forward condition, although only the patient versus control comparison achieved statistical significance. Age was a significant covariate in both analyses and parental education was a significant covariate in the LNS-Forward analysis. The large, well-characterized samples enabled us to meaningfully evaluate whether the relative group’s

12

The Use of Neurocognitive Endophenotypes

impairment on the LNS-Reordering condition was attributable to various psychometric or clinical confounds. First, we evaluated the possibility that the groups significantly differed on LNS-Reordered but not on LNS merely because of differences in the sensitivities of these tasks. This was done by comparing the true score variance for each task within the CCS group.93,94 True score variance was actually slightly higher in the LNS-Forward condition, which enhances confidence that the relatives’ impairment is not an artifact of the discriminating power of the tasks. Second, to determine whether these differences reflected the presence of relatives with schizophrenia spectrum disorders, the analyses were re-run after excluding relatives with any psychotic disorder or Cluster A personality disorder (which included less than 5% of the relatives). The results were nearly identical, with the relatives performing significantly worse than CCS (d = 0.33). Finally, we determined whether the impairment in the relative group persisted after accounting for the effects of lifetime mood and substance use disorder histories. Neither of these types of disorders accounted for the group differences. Thus, the impairments in the relatives were not attributable to a number of potential confounding factors. The significant executive WM impairment shown by both probands and their relatives is consistent with a number of prior family studies. It is noteworthy that the magnitude of the between-group differences is somewhat smaller than those found in previous studies using the same or comparable WM tasks.16,17,95 This could be associated with three key methodological features. First, the COGS uses a distinctive ascertainment strategy to optimize genetic analyses of endophenotypes, which involves enrolling only families with both affected and unaffected siblings to provide contrast statistical analysis. These rigorous selection criteria might have led to the inclusion of relatively intact patients and family members who were willing to complete the lengthy clinical and endophenotype assessments. Second, whereas previous studies sometimes included “super controls”,96 the COGS project includes CCS with disorders outside of the schizophrenia spectrum, which might create additional “noise” in the control sample and reduce between-group separations. Third, some previous estimates of deficits and heritability used a twin design which may lead to higher heritability estimates. This study supports the feasibility and validity of executive functioning WM as an endophenotype in

185

large-scale genetic studies that use a “contrast-based” ascertainment strategy to maximize the power of genetic analyses. Similar familiality analyses are underway for the other COGS neurocognitive endophenotypes. As an additional test of the validity of this candidate endophenotype, the COGS investigators evaluated the heritability of the WM and other neurocognitive measures in a large sample that included the participants in the WM study.

Heritability An initial heritability analysis of the primary COGS neurocognitive measures and the Penn battery measures was conducted in 638 members of 183 nuclear families with a schizophrenia proband.97 Of these families, 70% were sibships of 2, 17% were sibships of 3, 8% were sibships of 4, and 5% were sibships of 5 or more. In addition to heritability analyses, genetic and environmental correlations among the tasks were evaluated. In the first set of analyses, variance component models were used to assess heritability (h2) as implemented by the SOLAR v.2.1.2 linkage analysis package.98 Factors that that might affect the endophenotypes (e.g., age, gender, site) were screened using SOLAR for significance (p < 0.05) and included as covariates when appropriate. In addition, a correction for ascertainment bias was used as the families were ascertained through a proband with schizophrenia. All of the neurocognitive endophenotypes and the Penn CNB measures were found to be significantly heritable (p ≤ 0.005), with heritabilities ranging from 0.24 to 0.55. The observed heritabilities (standard errors in parentheses) for the neurocognitive endophenotypes were as follows: LNS-Reordered = 0.39 (0.07); DS-CPT (d’) = 0.38 (0.07), and CVLT II (Total of learning trials 1–5) = 0.25 (0.08). The observed heritabilities for the six measures from the Penn Battery (based on accuracy scores) were as follows: Abstraction and Mental Flexibility = 0.28 (0.07), Face Memory = 0.27 (0.07), Spatial Memory = 0.24 (0.08), Spatial Processing = 0.55 (0.08), Sensorimotor Dexterity = 0.39 (0.08), and Emotion Recognition = 0.32 (0.07). Age was a significant covariate in nearly all analyses and gender was a significant covariate in many analyses. Site of data collection and atypical

186

medication use were not significant covariates for any of the endophenotypes. The convergence of these findings with previously reported heritability estimates based on various types of samples also supports the validity of the COGS data. For the primary COGS variables, the heritability estimates for WM and attention are substantial and fall within the lower end of the range of values reported in non-clinical and schizophrenia.27,28,42,49,67,99,100 The heritability estimate for verbal memory is somewhat lower, and prior estimates for the CVLT were either higher in one healthy twin study (0.51; ref.27) or lower in one study of schizophrenia probands and first degree relatives (0.21; ref.67). As in the WM study, it is possible that the COGS recruitment strategy led to the preferential selection of families with less genetic loading for pathological endophenotypic values, which could result in lower estimates of heritability than other approaches (e.g., recruiting “singleton” patients). The significant heritabilities for the Penn CNB measures, including less frequently studied measures of face and emotion processing, also converge with recent findings.68 In the second phase of data analyses, bivariate genetic (ρG) and environmental (ρE) correlations were also computed using SOLAR.101,102 The genetic correlation between two endophenotypes is the component of the overall correlation that is due to pleiotropy (i.e., the influence of a gene or set of genes on both endophenotypes simultaneously), which is obtained from the kinship information in the pedigree. The environmental correlation between two endophenotypes is the component of the correlation due to environmental factors that influence both endophenotypes, which is obtained from the individual-specific error. The results of these analyses are summarized in Table 12.3, with genetic correlations presented above the diagonal and environmental correlations presented below. Significant (p < 0.05) genetic correlations were observed between most of the endophenotypes. Notably, Spatial Processing and Spatial Memory revealed significant genetic correlations with nearly all other endophenotypes. In stark contrast, Sensorimotor Dexterity was not found to be genetically correlated with any other endophenotype. Significant (p < 0.05) environmental correlations were also observed between most of the endophenotypes. Verbal Memory, Abstraction and Flexibility, and Face Memory revealed environmental correlations with nearly all other endophenotypes, and, in contrast to the genetic correlations,

W.P. Horan et al.

Sensorimotor Dexterity was found to be environmentally correlated with all other endophenotypes. Many of the genetic and environmental correlations were significant at the p < 0.001 level and remained significant even after a conservative Bonferroni correction for multiple testing. Although the significant heritability estimates in these analyses suggest that genes contribute to the endophenotypes, these genes do not appear to operate in isolation. Rather, the significant genetic correlations provide evidence that overlapping genetic architecture (pleiotropy) may underlie many of the endophenotypes. However, the lack of significant genetic correlations between several of the endophenotypes, combined with the fact that the correlated endophenotypes are not 100% co-heritable, suggests that there may be subtypes of schizophrenia with different endophenotypic characteristics. The observation of significant environmental correlations between many of the endophenotypes suggests that factors over and above genes also influence the correlations and contribute to their expression. These correlations make sense from a phenomenological and neural substrate level, as many of these measures have similar psychological, genetic, and neurobiological underpinnings. It should be noted that the observed heritabilities for these endophenotypes are lower than the highest heritability of 80% observed for schizophrenia itself.103,104 However, we expect that by assessing endophenotype heritabilities, we are parsing the entire heritability of schizophrenia into components with specific underlying neurobiology. As such, we do not necessarily expect to find a single endophenotype that is more heritable than schizophrenia but rather a series of endophenotypes that additively approach the heritability of schizophrenia and relate to specific neurobiological functions underlying schizophrenia pathogenesis. The results of this study support the fundamental COGS premise that neurocognitive endophenotypes will be important measures to consider in characterizing the genetic basis of schizophrenia. These data provide the basis for an exciting and challenging opportunity to explore the potential common underlying genetic and neurobiological substrates of these endophenotypic measures, as well as the unique genetic architecture of each measure. Using genetic data, the COGS investigators are beginning to explore the complex relationships among the endophenotypes, and between the endophenotypes and schizophrenia.

1

2

3

4

5

6

7

8

9

1. Working memory 2. Attention 3. Verbal memory

— 0.19 (0.08) 0.35 (0.15) — 0.38 (0.17)* 0.14 (0.08)

0.28 (0.08) 0.34 (0.18) —

0.42 (0.18) 0.40 (0.18) 0.34 (0.21)

0.26 (0.17) 0.32 (0.16) 0.38 (0.20)

0.38 (0.21) 0.43 (0.21) 0.68 (0.23)

0.45 (0.13)* 0.08 (0.15) .34 (.16) 0.57 (0.12)* 0.19 (0.05) .32 (.16) 0.49 (0.16) −0.23 (0.19) 0.30 (0.19)

4. Penn: Abstraction & Flexibility 5. Penn: Face memory 6. Penn: Spatial memory 7. Penn: Spatial processing 8. Penn: Sensorimotor dexterity 9. Penn: Emotion recognition

0.14 (0.08) 0.17 (0.08) 0.04 (0.09) 0.18 (0.10) 0.21 (0.09) 0.15 (0.09)

0.22 (0.08) 0.30 (0.08)* 0.17 (0.08) 0.22 (0.10) 0.33 (0.08)* 0.23 (0.08)

— 0.30 (0.08)* 0.04 (0.08) 0.22 (0.09) 0.26 (0.09) 0.23 (0.08)

0.22 (0.19) — 0.24 (0.08) 0.17 (0.10) 0.34 (0.08)* 0.35 (0.07)*

0.87 (0.27)* 0.61 (0.22) — 0.15 (0.10) 0.20 (0.09) 0.06 (0.08)

0.76 (0.16)* 0.46 (0.16) 0.49 (0.20) — 0.30 (0.11) 0.17 (0.10)

0.17 (0.08) 0.28 (0.08)* 0.08 (0.09) 0.01 (0.11) 0.31 (0.09)* 0.17 (0.09)

0.15 (0.18) 0.10 (0.17) 0.20 (0.21) 0.13 (0.15) — 0.29 (0.08)*

0.34 (0.19) 0.58 (0.14) 0.56 (0.24) 0.47 (0.15) 0.21 (0.17) —

The Use of Neurocognitive Endophenotypes in Large-Scale

Table 12.3 Genetic (ρG) and environmental (ρE) correlations between the primary endophenotypes and Penn CNB measures as assessed in the pedigrees

Correlation estimates and their standard errors (in parentheses) are indicated for each pair of endophenotypes. Genetic correlations are presented above the diagonal. Environmental correlations are presented below the diagonal. Significant (p < 0.05) correlations are indicated in bold. *Indicates correlations surviving a conservative Bonferroni correction for multiple comparisons (p < 0.001).

187

188

The Search for Candidate Genes A comprehensive review of the progress of genetic research in schizophrenia revealed a number of replicated linkages, including evidence implicating chromosomes 1q, 5q, 6p, 6q, 8p, 10p, 13q, 15q, and 22q.105 Unfortunately, none of these linkage findings has led to the identification of causative genes for schizophrenia, although several putative susceptibility genes have been identified (e.g., refs.106,107). This difficulty in mapping genetic variants predisposing to illness may be due to the modest nature of the linkage signals and the broad genetic regions they encompass, a likely result of the well-known clinical and genetic heterogeneity associated with schizophrenia. As an alternative strategy to linkage analysis in families segregating illness, the interrogation of candidate genes thought to be associated with underlying biological mechanisms may aid in the genetic dissection of complex diseases. Many genes have been investigated as candidate genes for schizophrenia with varying degrees of success. Some of the “usual suspects” include AKT1, CHRNA7, COMT, DAO, DAOA, DISC1, DTNBP1, GRM3, GSK3B, NOS1AP, NRG1, PAFAH1B1, PPP3CC, PRODH, RELN, and RGS4.106 Issues with non-replication have plagued the field,108 and the causal variants contained within even the most replicated candidate genes still remain a mystery. As is characteristic of a polygenic disease, schizophrenia likely exhibits heterogeneity, in which different genes (locus heterogeneity) or alleles within the same gene (allelic heterogeneity) influence susceptibility in different individuals, the confounding effects of which on attempts to identify disease-causing alleles are only now being fully recognized.109 It is currently thought that disturbances in several genes within the same neurobiological pathway could contribute to the pathology of the schizophrenia. The simultaneous investigation of multiple genes within a pathway of relevance, in combination with the use of phenotypes that more accurately capture the underlying biology of the disease, could therefore provide valuable insight into a polygenic disease like schizophrenia that has otherwise proved difficult to tease apart. Underlying molecular mechanisms and biological processes connect many of the already identified candidate genes for schizophrenia, and knowledge of such genetic and molecular interactions may assist in the dissection of the genetic basis of schizophrenia susceptibility.

W.P. Horan et al.

Candidate Genes Identified in the COGS Project By identifying the genetic determinants of phenotypes known or likely to represent the subclinical pathology of a disease, one can more easily identify the genes or groups of genes that impact actual manifestations of that disease. The COGS investigators have developed a custom chip containing 1,536 single nucleotide polymorphisms (SNPs) in 94 genes of relevance to schizophrenia and related neurocognitive and neurophysiological endophenotypes. These genes were chosen based on knowledge of relevant neurobiological systems, as well as an extensive review of published association, linkage, and model organism studies. In addition to schizophrenia, many of these genes have been linked to brain development and the specific neurocognitive functions assessed in the COGS project. The genes included on the COGS SNP chip cluster into several pathways, as indicated by Ingenuity Pathway Analysis (Ingenuity Systems; http://www. ingenuity.com), including cell signal transduction (12 genes), axonal guidance signaling (5 genes), amino acid metabolism (5 genes), and glutamate (16 genes), serotonin (8 genes), dopamine (9 genes), and GABA (5 genes) receptor signaling. In order to efficiently interrogate these genes, we have chosen to make extensive use of haplotype-tagging single nucleotide polymorphisms (SNPs), a small subset of SNPs that uniquely identify an inherited segment of DNA. These were derived solely from Caucasian populations, since our sample is primarily (> 70%) of Caucasian origin. Of the 1,417 tagging SNPs that were selected for 86 of the genes, 42 also had reported associations in the schizophrenia literature, and 15 were nonsynonymous SNPs, which result in a change in the amino acid sequence. For the 6 genes for which tagging SNPs were not available, 47 SNPs were chosen for even coverage via a complementary strategy. We have also included an additional 72 SNPs that were reported to be associated with schizophrenia in the literature. Many of these SNPs had been replicated by separate groups, and ten were nonsynonymous SNPs. In total, the COGS Custom SNP Chip includes 41 genes for which previously published studies indicate association with schizophrenia or related phenotypes. In a preliminary analysis that included 597 relatives from 129 families,110 variance component methods111

12

The Use of Neurocognitive Endophenotypes

were used to evaluate association between the COGS SNP chip and the same nine neurocognitive measures examined by Greenwood et al.97 and discussed in section on Heritability above. Of the 1,536 SNPs genotyped, 1,385 remained for analysis following elimination based on quality control thresholds for allele call rate and cluster separation, departures from Hardy-Weinberg Equilibrium (p < 0.00001), which can indicate genotyping error, and minor allele frequencies < 0.01 to eliminate excessively rare SNPs that might also be genotyping errors. Age and sex were included as covariates where appropriate. The results of the single-marker analyses revealed significant associations between the 9 endophenotypes and 42 of the 94 genes. There were 2 SNPs with a p < 0.0001, 30 SNPs with a p < 0.001, and 153 SNPs with a p < 0.01. These findings represent a large excess over what would be expected by chance alone and all may be of interest, considering the specific selection of these genes. Eleven genes displayed evidence for pleiotropy, revealing significant associations with three or more endophenotypes. In particular, ERBB4 and NRG1 revealed associations to eight and five endophenotypes, respectively, providing further evidence for a substantial role of these genes in mediating susceptibility to schizophrenia. NRG1 (neuregulin-1) is a trophic factor containing an epidermal growth factor (EGF)-like domain that signals through the activation of the ErbB receptor tyrosine kinases, like ErbB4 which is of particular interest to the pathophysiology of schizophrenia because of its crucial roles in neurodevelopment and in the modulation of N-methyl-Daspartate (NMDA) receptor signaling. Like NRG1 and ERBB4, many of the 42 genes found to be associated with the nine endophenotypes interact on a molecular level, and knowledge of such interactions will be utilized in future gene–gene interaction analyses. Although we are at the preliminary stages of analysis and much remains to be done (e.g., haplotype analyses, gene–gene interaction analyses), we have observed many significant associations between our endophenotypes and genes thought to be of biological relevance to schizophrenia. The observation of extensive pleiotropy for some genes (e.g., ERBB4 and NRG1) and singular associations for others in our data suggests alternative, independent pathways mediating schizophrenia pathogenesis. These results suggest that by identifying the causal variants mediating

189

the observed associations with each endophenotype, we will be in a position to dissect the polygenic basis of schizophrenia.

Conclusions and Future Directions Genetic studies of quantitative endophenotypes require the use of large samples that are often only possible to collect through multi-site collaborative efforts. To address the many logistical challenges associated with this type of research, the COGS has established an infrastructure for recruitment, clinical characterization, measurement of multiple endophenotypes, data management, rigorous quality assurance, and genotyping. Results from the COGS project demonstrate that neurocognitive tasks can be practically, efficiently, and validly employed in multisite genetic studies of schizophrenia. In addition, initial candidate gene analyses from this and other projects support the use of neurocognitive endophenotypes as a powerful approach that goes beyond the traditional focus on the complex schizophrenia phenotype. The COGS approach is to study multiple endophenotypes in schizophrenia kindreds and individuals in order to characterize the connections and likely heterogeneity among potential pathogenetic mechanisms. In so doing, COGS brings an information-intensive ‘systems biology’ approach to schizophrenia research that is highly innovative in the field of neuropsychiatry, but fully consistent with trends in other areas of modern biomedical research. The core significance of the COGS is that it is an integrated multisite network for assessing crucial functional endophenotypes in schizophrenia. While genome scans and association studies yield important information about schizophrenia, the COGS makes a unique and complementary contribution by specifying the genes associated with aberrant neurophysiological and neurocognitive measures. By discovering the “gene-to-phene” pathways – from genetic variations to brain dysfunction – COGS will provide a clear roadmap to novel interventions that will reduce or prevent the morbidity and mortality associated with schizophrenia. The COGS and other large-scale genetic studies of schizophrenia illustrate several persisting challenges for the use of neurocognitive endophenotypes.

190

For example, although the “contrast-based” ascertainment of large pedigrees in the COGS project was specifically designed to maximize the statistical power of genetic analyses, recruitment of probands and relatives from intact families may lead to a sample of relatively “healthy” participants who are not fully representative. The COGS ascertainment strategy might make positive findings more difficult to detect if, for example, the volunteer families tend to perform relatively well on neurocognitive tasks or carry less of a “paranoid” or “suspicious” genetic load. Another challenge concerns the inherent complexity of candidate neurocognitive phenotypes. Performance on these tasks can involve multiple, sometimes overlapping, cognitive operations that complicate the identification of specific disease genes,112 Although the genetic complexity of neurocognitive endophenotypes remains to be determined, the working expectation is that the relatively discrete functions that are indexed by these tasks will have greater “mapability” than the illness itself. That is, we expect that by assessing endophenotypes, we are parsing the entire heritability of schizophrenia into components with relatively specific neurobiological functions that underlie schizophrenia pathogenesis. The COGS strategy of utilizing endophenotypes and carefully selected candidate genes, in addition to more traditional linkage and whole genome association studies, offers much hope for understanding the genetics of schizophrenia. Acknowledgments The authors wish to acknowledge the Consortium on the Genetics of Schizophrenia (COGS) and the collaborative RO1 grants from the National Institute of Mental Health to the following institutions: Harvard University RO1-MH065562; Mount Sinai School of Medicine RO1-MH065554; University of California Los Angeles RO1-MH65707; University of California San Diego R01-MH065571; University of Colorado RO1-MH65588; University of Pennsylvania RO1-MH65578; University of Washington R01-MH65558.

References 1. Gottesman II, Shields J. Schizophrenia and Genetics; A Twin Study Vantage Point. New York: Academic Press; 1972. 2. Braff DL, Freedman R. Endophenotypes in studies of the genetics of schizophrenia. In: Davis KL, Charney DS, Coyle JT, Nemeroff C, eds. Neuropsychopharmacology: The Fifth Generation of Progress. Philadelphia: Lippincott Williams & Wilkens; 2002:703–716.

W.P. Horan et al. 3. Gottesman II, Gould TD. The endophenotype concept in psychiatry: Etymology and strategic intentions. American Journal of Psychiatry 2003;160(4):636–645. 4. Nuechterlein KH, Barch DM, Gold JM, Goldberg TE, Green MF, Heaton RK. Identification of separable cognitive factors in schizophrenia. Schizophrenia Research 2004;72:29–39. 5. Gold JM. Cognitive deficits as treatment targets in schizophrenia. Schizophrenia Research 2004;72:21–28. 6. Green MF, Kern, R. S., Braff, D. L., Mintz, J. Neurocognitive deficits and functional outcome in schizophrenia: Are we measuring the “right stuff”?. Schizophrenia Bulletin 2000;26(1):119–136. 7. Green MF, Kern, R.S., Heaton, R.K. Longitudinal studies of cognition and functional outcome in schizophrenia: implications for MATRICS. Schizophrenia Research 2004;72:41–51. 8. Turetsky BI, Calkins ME, Light GA, Olincy A, Radant AD, Swerdlow NR. Neurophysiological endophenotypes of schizophrenia: the viability of selected candidate measures. Schizophrenia Bulletin Jan 2007;33(1):69–94. 9. Goldman-Rakic PS. Working memory dysfunction in schizophrenia. Journal of Neuropsychiatry and Clincial Neuroscience 1994;6:348–357. 10. Miyake A, Shah P. Models of Working Memory: Mechanisms of Active Maintenance and Executive Control. New York: Cambridge University Press; 1999. 11. Baddeley AD. Working Memory. New York: Oxford University Press; 1986. 12. Perry W, Heaton RK, Potterat E, Roebuck T, Minassian A, Braff DL. Working memory in schizophrenia: Transient “online” storage versus executive functioning. Schizophrenia Bulletin 2001;27:157–176. 13. Lee J, Park S. Working memory impairments in schizophrenia: A meta-analysis. Journal of Abnormal Psychology 2005;114(4):599–611. 14. Barch DM. The cognitive neuroscience of schizophrenia. Annual Review of Clinical Psychology 2005;1(1):321–353. 15. Aleman A, Hijman R, de Haan EHF, Kahn RS. Memory impairment in schizophrenia: A meta-analysis. American Journal of Psychiatry 1999;156(9):1358–1366. 16. Gold JM, Carpenter C, Randolph C, Goldberg TE, Weinberger DR. Auditory working memory and Wisconsin Card Sorting Test performance in schizophrenia. Archives of General Psychiatry 1997;54(2):159–165. 17. Conklin HM, Curtis CE, Calkins ME, Iacono WG. Working memory functioning in schizophrenia patients and their first-degree relatives: cognitive functioning shedding light on etiology. Neuropsychologia 2005;43(6):930–942. 18. Heaton RK, Gladsjo JA, Palmer BW, Kuck J, Marcotte TD, Jeste DV. Stability and course of neuropsychological deficits in schizophrenia. Archives of General Psychiatry 2001;58(1):24–32. 19. Hill SK, Schuepbach D, Herbener ES, Keshavan MS, Sweeney JA. Pretreatment and longitudinal studies of neuropsychological deficits in antipsychotic-naïve patients with schizophrenia. Schizophrenia Research 2004;68(1):49–63. 20. Park S, Püschel J, Sauter BH, Rentsch M, Hell D. Spatial working memory deficits and clinical symptoms in schizophrenia: A 4-month follow-up study. Biological Psychiatry 1999;46(3):392–400.

12

The Use of Neurocognitive Endophenotypes

21. Park S, Püschel J, Sauter BH, Rentsch M, Hell D. Spatial selective attention and inhibition in schizophrenia patients during acute psychosis and at 4-month follow-up. Biological Psychiatry 2002;51(6):498–506. 22. Tyson PJ, Laws KR, Roberts KH, Mortimer AM. A longitudinal analysis of memory in patients with schizophrenia. Journal of Clinical and Experimental Neuropsychology 2005;27(6):718–734. 23. Barch DM, Carter CS, Braver TS, Sabb FW, MacDonald A, III, Noll DC, Cohen JD. Selective deficits in prefrontal cortex function in medication-naive patients with schizophrenia. Archives of General Psychiatry 2001;58(3):280–288. 24. Carter C, Robertson L, Nordahl T, Chaderjian M. Spatial working memory deficits and their relationship to negative symptoms in unmedicated schizophrenia patients. Biological Psychiatry 1996;40(9):930–932. 25. Trandafir A, Meary A, Schurhoff F, Leboyer M, Szoke A. Memory tests in first-degree adult relatives of schizophrenic patients: a meta-analysis. Schizophrenia Research Jan 31 2006;81(2–3):217–226. 26. Snitz BE, MacDonald AW, Carter CS. Cognitive deficits in unaffected first-degree relatives of schizophrenia patients: a meta-analytic review of putative endophenotypes. Schizophrenia Bulletin 2006;32:179–194. 27. Ando J, Ono Y, Wright MJ. Genetic structure of spatial and verbal working memory. Behavioral Genetics 2001;31:615–624. 28. Hansell NK, Wright MJ, Luciano M, Geffen GM, Geffen LB, Martin NG. Genetic covariation between event-related potential (ERP) and behavioral non-ERP measures of working-memory, processing speed, and IQ. Behavioral Genetics 2005 ;5(6):695–706. 29. Wager TD, Smith EE. Neuroimaging studies of working memory: A meta-analysis. Cognitive, Affective and Behavioral Neuroscience 2003;3(4):255–274. 30. Cannon TD, Glahn DC, Kim J, et al. Dorsolateral prefrontal cortex activity during maintenance and manipulation of information in working memory in patients with schizophrenia. Archives of General Psychiatry 2005;62(10):1071–1080. 31. Seidman LJ, Thermenos HW, Poldrack RA, Peace NK, Koch JK, Faraone SV, Tsuang MT. Altered brain activation in dorsolateral prefrontal cortex in adolescents and young adults at genetic risk for schizophrenia: an fMRI study of working memory. Schizophernia Research Jul 2006;85(1–3):58–72. 32. Thermenos HW, Goldstein JM, Buka SL, Poldrack RA, Koch JK, Tsuang MT, Seidman LJ. The effect of working memory performance on functional MRI in schizophrenia. Schizophernia Research May 1 2005;74(2–3):179–194. 33. Karlsgodt KH, van Erp TG, Poldrack RA, Bearden CE, Nuechterlein KH, Cannon TD. Diffusion tensor imaging of the superior longitudinal fasciculus and working memory in recent-onset schizophrenia. Biological Psychiatry Mar 1 2008;63(5):512–518. 34. Wechsler D. Wechsler Memory Scale - Third Edition. San Antonio, Texas: The Psychological Corporation/ Harcourt Brace & Company; 1997. 35. Bleuler E. Dementia Praecox or the Group of Schizophrenias. New York: International Universities Press; 1950. 36. Kraepelin E. Dementia Praecox and Paraphrenia. Edinburgh: E. & S. Livingston; 1919.

191 37. Cornblatt BA, Keilp JG. Impaired attention, genetics, and the pathophysiology of schizophrenia. Schizophrenia Bulletin 1994;20:31–46. 38. Nuechterlein KH. Vigilance in schizophrenia and related disorders. In: Steinhauer SR, Gruzelier JH, Zubin J, eds. Handbook of Schizophrenia, Neuropsychology, Psychophysiology and Information Processing. Vol 5. Amsterdam: Elsevier Science; 1991:397–433. 39. Nestor PG, Han SD, Niznikiewicz M, Salisbury D, Spencer K, Shenton ME, McCarley RW. Semantic disturbance in schizophrenia and its relationship to the cognitive neuroscience of attention. Biol Psychol Jul–Aug 2001;57(1–3):23–46. 40. Cohen JD, Braver TS, O’Reilly RC. A computational approach to prefrontal cortex, cognitive control and schizophrenia: recent developments and current challenges. Philosophical Transactions of the Royal Society of LondonSeries B Oct 29 1996;351(1346):1515–1527. 41. Chen WJ, Faraone SV. Sustained attention deficits as markers of genetic susceptibility to schizophrenia. American Journal of Medical Genetics Spring 2000;97(1):52–57. 42. Cornblatt BA, Malhotra AK. Impaired attention as an endophenotype for molecular genetic studies of schizophrenia. American Journal Medical Genetics Jan 8 2001;105(1):11–15. 43. Nuechterlein KH, Asarnow RF, Subotnik KL, Fogelson DL, Ventura J, Torquato R, et al. Neurocognitive vulnerability factors for schizophrenia: Convergence across genetic risk studies and longitudinal trait/state studies. In: Dworkin R, ed. Origins and Development of Schizophrenia: Advances in Experimental Psychopathology. Washington, D.C: American Psychological Association; 1998:299–327. 44. Heinrichs RW, Zakzanis KK. Neurocognitive deficit in schizophrenia: A quantitative review of the evidence. Neuropsychology 1998;12:426–445. 45. Finkelstein JRJ, Cannon TD, Gur RE, Gur RC, Moberg P. Attentional dysfunctions in neuroleptic-naive and neuroleptic-withdrawn schizophrenic patients and their siblings. Journal of Abnormal Psychology 1997;106:203–212. 46. Wohlberg GW, Kornetsky C. Sustained attention in remitted schizophrenics. Archives of General Psychiatry 1973;28:533–537. 47. Nuechterlein KH, Dawson ME, Ventura J, Yee-Bradbury C. Longitudinal stability of vigilance and span of apprehension deficits in the early phase of schizophrenia. Sixth Annual Meeting of the Society for Research in Psychopathology. Cambridge, MA; 1991. 48. Cornblatt BA, Risch NJ, Faris G, Friedman D, ErlenmeyerKimling L. The Continuous Performance Test, Identical Pairs Version (CPT-IP): I. New findings about sustained attention in normal families. Psychiatry Research 1988;26:223–238. 49. Chen WJ, Liu SK, Chang CJ, Lien YJ, Chang YH, Hwu HG. Sustained attention deficit and schizotypal personality features in nonpsychotic relatives of schizophrenic patients. American Journal of Psychiatry Sept 1998; 155(9):1214–1220. 50. Grove WM, Lebow BS, Clementz BA, Cerri A, Medus C, Iacono WG. Familial prevalence and coaggregation of schizotypy indicators: a multitrait family study. Journal of Abnormal Psychology May 1991;100(2):115–121.

192 51. Seidman LJ, Thermenos HW, Koch JK, et al. Auditory verbal working memory load and thalamic activation in nonpsychotic relatives of persons with schizophrenia: an fMRI replication. Neuropsychology Sept 2007;21(5):599–610. 52. Thermenos HW, Seidman LJ, Breiter H, Goldstein JM, Goodman JM, Poldrack R, Faraone SV, Tsuang MT. Functional MRI during auditory verbal working memory in non-psychotic relatives of persons with schizophrenia: A pilot study. Biological Psychiatry 2004;55:490–500. 53. Delawalla Z, Csernansky JG, Barch DM. Prefrontal cortex function in nonpsychotic siblings of individuals with schizophrenia. Biological Psychiatry Mar 1 2008;63(5):490–497. 54. Seidman LJ, Breiter HC, Goodman JM, et al. A functional magnetic resonance imaging study of auditory vigilance with low and high information processing demands. Neuropsychology Oct 1998;12(4):505–518. 55. Nuechterlein KH. Signal detection in vigilance tasks and behavioral attributes among offspring of schizophrenic mothers and among hyperactive children. Journal of Abnormal Psychology 1983;92:4–28. 56. Nuechterlein KH, Parasuraman R, Jiang Q. Visual sustained attention: Image degradation produces rapid sensitivity decrement over time. Science 1983;220:327–329. 57. Cornblatt BA, Lenzenweger MF, Erlenmeyer-Kimling L. The continuous performance test, identical pairs version: II. Contrasting attentional profiles in schizophrenic and depressed patients. Psychiatry Research Jul 1989;29(1):65–85. 58. Delis D, Kramer J, Kaplan E, Ober B. California Verbal Learning Test: Second Edition. Adult version. Manual. New York: The Psychological Corporation; 2000. 59. Cirillo M, Seidman LJ. A review of verbal declarative memory function in schizophrenia: From clinical assessment to genetics and brain mechanisms. Neuropsychol Review 2003;13:43–77. 60. Heinrichs RW, Zakzanis KK. Neurocognitive deficits in schizophrenia: A quantitative review of the evidence. Neuropsychology 1998;12(3):426–445. 61. Harvey PD, Palmer BW, Heaton RK, Mohamed S, Kennedy J, Brickman A. Stability of cognitive performance in older patients with schizophrenia: an 8-week test-retest study. American Journal of Psychiatry 2005;162:110–117. 62. Joyce E. Origins of cognitive dysfunction in schizophrenia: clues from age at onset. British Journal of Psychiatry 2005;186:93–95. 63. Saykin AJ, Shtasel DL, Gur RE, Kester DB, Mozley LH, Stafiniak P, Gur RC. Neuropsychological deficits in neuroleptic naive patients with first-episode schizophrenia. Archives of General Psychiatry 1994;51:124–131. 64. Sitskoorn MM, Aleman A, Ebisch SJ, Appels MC, Kahn RS. Cognitive deficits in relatives of patients with schizophrenia: a meta-analysis. Schizophrenia Research Dec 1 2004;71(2–3):285–295. 65. Bouchard T, Jr. Genetic and environmental influences on adult intelligence and special mental abilities. Human Biology 1988;70:257–279. 66. Lee JH, Flaquer A, Stern Y, Tycko B, Mayeux R. Genetic influences on memory performance in familial Alzheimer disease. Neurology 2004;62:414–421. 67. Tuulio-Henriksson A, Haukka J, Partonen T, et al. Heritability and number of quantitative trait loci of neurocognitive func-

W.P. Horan et al.

68.

69.

70.

71.

72.

73.

74.

75.

76.

77.

78.

79.

80.

81.

tions in families with schizophrenia American Journal of Medical Genetics 2002;114:483–490. Gur RE, Nimgaonkar VL, Almasy L, et al. Neurocognitive endophenotypes in a multiplex multigenerational family study of schizophrenia. American Journal of Psychiatry May 2007;164(5):813–819. Ranganath C, Minzenberg MJ, Ragland JD. The cognitive neuroscience of memory function and dysfunction in schizophrenia. Biological Psychiatry Jul 1 2008;64(1):18–25. Boos HB, Aleman A, Cahn W, Pol HH, Kahn RS. Brain volumes in relatives of patients with schizophrenia: a meta-analysis. Archives of General Psychiatry Mar 2007;64(3):297–304. Achim AM, Bertrand MC, Sutton H, et al. Selective abnormal modulation of hippocampal activity during memory formation in first-episode psychosis. Archives of General Psychiatry Sep 2007;64(9):999–1014. Thermenos HW, Seidman LJ, Poldrack RA, Peace NK, Koch JK, Faraone SV, Tsuang MT. Elaborative verbal encoding and altered anterior parahippocampal activation in adolescents and young adults at genetic risk for schizophrenia using FMRI. Biological Psychiatry Feb 15 2007;61(4):564–574. Ragland JD, Gur RC, Valdez J, et al. Event-related fMRI of frontotemporal activity during word encoding and recognition in schizophrenia. American Journal of Psychiatry Jun 2004;161(6):1004–1015. Kurtz MM, Ragland JD, Moberg PJ, Gur RC. The Penn Conditional Exclusion Test: A new measure of executivefunction with alternate forms for repeat administration. Archives of Clinical Neuropsychology 2004;19:191–201. Glahn DC, Gur RC, Ragland JD, Gur RE. Reliability, performance characteristics, and construct validity and initial application of the visual object learning test (VOLT). Neuropsychology 1997;11:602–612. Gur RC, Ragland JD, Moberg PJ, Turner TH, Bilker WB, Kohler C, Siegel SJ, Gur RE. Computerized neurocognitive scanning: I. Methodology and validation in healthy people. Neuropsychopharmacology 2001;25(5):766–788. Gur RC, Jaggi JL, Ragland JD, Resnick SM, Shtasel DL, Muenz L, Gur RE. Effects of memory processing on regional brain activation: cerebral blood flow in normal subjects. International Journal of Neuroscience 1993;72:31–44. Kohler CG, Turner TH, Bilker WB, Brensinger CM, Siegel SJ, Kanes SJ, Gur RE, Gur RC. Facial emotion recognition in schizophrenia: intensity effects and error pattern. American Journal of Psychiatry 2003;160:1768–1774. Gur RC, Ragland JD, Moberg PJ, Bilker WB, Kohler C, Siegel SJ, Gur RE. Computerized neurocognitive scanning: II. The profile of schizophrenia. Neuropsychopharmacology 2001;25(5):777–788. Gur RE, Calkins ME, Gur RC, Horan WP, Nuechterlein KH, Seidman LJ, Stone WS. The Consortium on the Genetics of Schizophrenia (COGS): Neurocognitive Endophenotypes. Schizophrenia Bulletin 2007;33:49–68. Calkins ME, Dobie DJ, Cadenhead KS, et al. The Consortium on the Genetics of Endophenotypes in Schizophrenia: model recruitment, assessment, and endophenotyping methods for a multisite collaboration. Schizophrenia Bulletin Jan 2007;33(1):33–48.

12

The Use of Neurocognitive Endophenotypes

82. Schork NJ, Greenwood TA, Braff DL. Statistical genetics concepts and approaches in schizophrenia and related neuropsychiatric research. Schizophrenia Bulletin Jan 2007;33(1):95–104. 83. Braff DL, Freedman R, Schork NJ, Gottesman, II. Deconstructing schizophrenia: an overview of the use of endophenotypes in order to understand a complex disorder. Schizophrenia Bulletin Jan 2007;33(1):21–32. 84. Lander ES, Schork NJ. The genetic dissection of complex traits. Science 1994;265:2037–2048. 85. Nurnberger JI, Jr., Blehar MC, Kaufmann CA, et al. Diagnostic interview for genetic studies. Rationale, unique features, and training. NIMH Genetics Initiative. Archives of General Psychiatry Nov 1994;51(11):849–859; discussion 863–844. 86. Andreasen NC. The Scale for the Assessment of Negative Symptoms (SANS). Iowa CIty, Iowa: The University of Iowa; 1983. 87. Andreasen NC. The Scale for the Assessment of Positive Symptoms. Iowa City, Iowa: The University of Iowa; 1984. 88. Kendler KS, Lieberman, J.A., Walsh, D. The structured interview for schizotypy (SIS): a preliminary report. Schizophrenia Bulletin 1989;15:559–571. 89. NIMH. Genetics Initiative: Family Interview for Genetic Studies (FIGS). Rockville, Md: National Institute of Mental Health; 1992. 90. Horan WP, Braff DL, Nuechterlein KH, et al. Verbal working memory impairments in individuals with schizophrenia and their first-degree relatives: Findings from the Consortium on the Genetics of Schizophrenia. Schizophrenia Research Apr 10, 2008. 91. Radant AD, Dobie DJ, Calkins ME, et al. Successful multi-site measurement of antisaccade performance deficits in schizophrenia. Schizophrenia Research Jan 2007;89(1–3):320–329. 92. Swerdlow NR, Sprock J, Light GA, et al. Multi-site studies of acoustic startle and prepulse inhibition in humans: initial experience and methodological considerations based on studies by the Consortium on the Genetics of Schizophrenia. Schizophrenia Research May 2007;92(1–3):237–251. 93. Chapman LJ, Chapman, J. P. Problems in the measurement of cognitive deficits. Psychological Bulletin 1973;79:380–385. 94. Chapman LJ, Chapman, J. P. The measurement of differential deficit. Journal of Psychiatric Research 1978;14(1, Suppl. 4):303–311. 95. Perry W, Heaton, R. K., Potterat, E., Roebuck, T., Minassian, A., Braff, D. L. Working memory in schizophrenia: Transient “online” storage versus executive functioning. Schizophrenia Bulletin 2001;27(1):157–176. 96. Kendler KS. The super-normal control group in psychiatric genetics: possible artifactual evidence for coaggregation. Psychiatric Genetics 1990;1:45–53. 97. Greenwood TA, Braff DL, Light GA, et al. Initial heritability analyses of endophenotypic measures for schizophrenia:

193 the consortium on the genetics of schizophrenia. Archives of General Psychiatry Nov 2007;64(11):1242–1250. 98. Almasy L, Blangero J. Multipoint quantitative-trait linkage analysis in general pedigrees. American Journal of Human Genetics 1998;62:1198–1211. 99. Toulopoulou T, Picchioni M, Rijsdijk F, Hua-Hall M, Ettinger U, Sham P, Murray R. Substantial genetic overlap between neurocognition and schizophrenia: genetic modeling in twin samples. Archives of General Psychiatry Dec 2007;64(12):1348–1355. 100. Tuulio-Henriksson A, Arajarvi R, Partonen T, Haukka J, Varilo T, Schreck M. Familial loading associates with impairment in visual span among healthy siblings of schizophrenia patients. Biological Psychiatry 2003;54(623–628). 101. Hopper JL, Mathews JD. Extensiosn to multivariate normal models for pedigree analysis. Annals of Human Genetics 1982;46:373–383. 102. Almasy L, Dyer TD, Blangero J. Bivariate quantitative trait linkage analysis: pleiotropy versus co-incident linkages. Genetic Epidemiology 1997;14:953–958. 103. Owen MJ, O’Donovan MC, Gottesman II. Schizophrenia. In: McGuffin P, Owen MJ, Gottesman II, eds. Psychiatric Genetics & Genomics. Oxford: Oxford University Press; 2002:247–266. 104. Sullivan PF. The Genetics of Schizophrenia. PLoS Med. Vol 2; 2005. 105. Baron M. Genetics of schizophrenia and the new millennium: progress and pitfalls. American Journal of Human Genetics 2001;68:299–312. 106. Harrison PJ, Weinberger DR. Schizophrenia genes, gene expression, and neuropathology: on the matter of their convergence. Molecular Psychiatry Jan 2005;10(1):40–68; image 45. 107. Gogos JA, Gerber DJ. Schizophrenia susceptibility genes: emergence of positional candidates and future directions. Trends in Pharmacological Science Apr 2006;27(4):226–233. 108. Risch N, Botstein D. A manic depressive history. Nature Genetics Apr 1996;12(4):351–353. 109. Mutsuddi M, Morris DW, Waggoner SG, Daly MJ, Scolnick EM, Sklar P. Analysis of high-resolution HapMap of DTNBP1 (Dysbindin) suggests no consistency between reported common variant associations and schizophrenia. American Journal of Human Genetics 2006;79:903–909. 110. Greenwood TA, Light GA, Cadenhead KS, et al. Initial analyses of 94 candidate genes and twelve endophenotypes for schizophrenia from the Consortium on the Genetics of Schizophrenia. Biological Psychiatry, 64:82S, 63rd Annual Meeting of the Society for Biological Psychiatry 2008. 111. Abecasis GR, Cherny SS, Cookson WO, Cardon LR. Merlin–rapid analysis of dense genetic maps using sparse gene flow trees. Nature Genetics Jan 2002;30(1):97–101. 112. Flint J, Munafo MR. The endophenotype concept in psychiatric genetics. Psychological Medicine Feb 2007;37(2):163–180.

Chapter 13

Neurocognitive Endophenotypes for Bipolar Disorder: Evidence from Case-Control, Family and Twin Studies Eugenia Kravariti, Fergus Kane, and Robin M. Murray

Abstract Endophenotypes are observable and quantifiable traits, which are considered to be more immediate sequelae of a complex disease genotype than the clinical syndrome. Their presumed greater genetic simplicity and biological proximity to the genetic substrates, has sparked enthusiasm about a potential role of endophenotypes in the genetic ‘dissection’ of psychiatric disorders. Bipolar illness, and several of its reliable cognitive correlates, have substantial heritabilities, inviting the hypothesis that the two phenotypes tend to co-occur because they share susceptibility genes. We critically discuss the notion that cognitive dysfunction is integral to the bipolar diathesis by drawing evidence from population based studies of children, adolescents and young adults who develop bipolar illness in later life, studies of euthymic bipolar patients, and investigations of non-bipolar first-degree relatives of affected probands. The balance of the evidence suggests that immediate recall of word lists, and learning that occurs over repeated presentations of list items, deserve consideration as putative endophenotypes for bipolar illness, as do delayed verbal recall, and circumscribed aspects of selective attention, response inhibition and resistance to interference. However, the most consistent evidence from family and twin studies to date involves negative rather than positive findings. In addition, studies with genetically informative designs are scarce, while the assumptions underlying the endophenotypic approach need to undergo rigorous empirical investigation.

E. Kravariti, F. Kane, and R. M. Murray NIHR Biomedical Research Centre for Mental Health, South London and Maudsley NHS Foundation Trust and Institute of Psychiatry, King’s College London

Keywords Bipolar disorder • endophenotype • cognitive deficit • family study • twin study • review Abbreviations B-P: Brown-Peterson; BDNF: Brain derived neurotrophic factor; BFLT-E: Biber Figure Learning Test-Extended; CAMCOR: Cambridge Cognitive Examination – Revised; CANTAB: Cambridge Neuropsychological Test Automated Battery; COMT: Catechol-O-methyl-transferase; CPT: Continuous Performance Test; CVLT: California Verbal Learning Test; DISC1: Disrupted-in-schizophrenia-1; DZ: Dizygotic; E-RBMT: Extended Rivermead Behavioral Memory Test; ED: Extra-dimensional; HSCT: Hayling Sentence Completion Test; ID: Intra-dimensional; IES: Intradimensional Extra-dimensional Shift; IQ: Intelligence Quotient; MZ: Monozygotic; NART: National Adult Reading Test; RAVLT: Rey Auditory Verbal Learning Test; ROCFT: Rey-Osterrieth Complex Figure Test; SOC: Stockings of Cambridge; TOL: Tower of London; WAIS-III: Wechsler Adult Intelligence Scale-III; WAIS-R: Wechsler Adult Intelligence Scale-Revised; WASI: Wechsler Abbreviated Scale of Intelligence; WCST: Wisconsin Card Sorting Test; WMS/-R: Wechsler Memory Scale/-Revised

Introduction Since Emil Kraepelin made the distinction between dementia praecox (schizophrenia) and manic depressive insanity (bipolar disorder), these disorders have been the focus of much research. Until recently, schizophrenia received most of the attention, but there is now an increased focus on bipolar disorder. Bipolar disorder is characterized by episodes of clinically significant mood dysregulation, usually alternating

M.S. Ritsner (ed.), The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes, © Springer Science + Business Media B.V. 2009

195

196

with periods of recovery or sub-threshold mood symptomatology. The last few decades have seen a departure from the traditional view that the majority of bipolar patients have complete remission between episodes. Recent evidence suggests that less than 50% of patients reach full functional recovery after their first hospitalisation1 and that cognitive dysfunction is a stable companion of the disorder regardless of symptom state.2–6 Parallel developments in genetically informative research have established substantial heritabilities for both bipolar disorder (80–93%)7,8 and several of the neuropsychological functions that may be altered in the syndrome (31–76%).9,10 The coincidence of considerable genetic aetiology in both bipolar disorder and cognitive dysfunction, and the tendency of the two phenotypes to co-occur, has spurred the notion that their genetic determinants may be shared to some extent. The momentum of this view parallels the increasing popularity of the concept of ‘endophenotypes’. These are observable and quantifiable traits, which are earlier – more proximal and direct – sequelae of a disease genotype than the temporally remote, phenotypically complex, clinical syndrome.11 Such traits are assumed to be simpler from a genetic point of view, and likely to be associated with fewer genetic loci, than the disease entity. Research focused on endophenotypes is a potential step towards resolving the genetic complexity of multifactorial diseases, such as bipolar disorder.11,12 Despite many positive findings and the availability of large family pedigrees in several molecular genetic studies, no gene has yet been conclusively identified for bipolar illness.13 The lack of agreement between studies emphasizes that susceptibility alleles are likely to be modest in effect size, requiring even larger samples for detection, and that phenotypic heterogeneity, unusual mutation mechanisms, interactions between genes, or between genes and the environment, and false-positive or -negative linkage findings are likely to have hampered consistency across studies.13–16 Rather than looking for genes that code a complex disorder, endophenotypic research searches for genes underlying simpler, ideally monogenic, traits, which accompany the illness and are likely to contribute to its pathophysiology.12 To enhance the likelihood that a phenotype is indeed intermediate between disease genotype and clinical presentation, it must be associated with the illness in the population, be heritable and state independent (i.e. present during remission), co-segregate with the disease within affected families, and have higher prevalence in unaffected first-degree relatives of affected probands than in the general population.11

E. Kravariti et al.

In the present chapter, we critically discuss the notion that cognitive dysfunction is integral to the bipolar diathesis by drawing evidence from four lines of research: Whether cognitive deficits predate the onset of bipolar disorder (and hence are present throughout the lifespan) is addressed in a brief overview of population based studies of children, adolescents and young adults who develop bipolar illness in later life. As these studies are scarce, and have used a limited number of neuropsychological measures, a possible dissociation of cognitive impairment from symptom states is further addressed through evidence from studies of euthymic bipolar patients. The latter type of research is confounded by ‘noise’ generated by disease-related factors which may not reflect genetic influences (e.g. possible cognitive dysfunction due to medication). Hence, we further draw upon studies of nonbipolar first-degree relatives of affected probands.1 The strength of this approach is that such subjects provide a more decisive control for effects associated with the disease; the caveat is a presumed rather than established susceptibility for bipolar illness. Finally, we briefly evaluate examples of recent research efforts at the interface of ‘cognitive genetics’ and ‘psychiatric genetics’. The majority of studies reviewed in the following sections have included patients, or relatives of patients, with bipolar I disorder.2 However, (usually a minority of) bipolar II patients3 have been included in some reports.

Premorbid Cognitive Function in Bipolar Disorder A small number of studies have examined cognitive function in large, representative, i.e. epidemiologically derived, samples of children, adolescents and young adults who later develop bipolar illness. Three major studies of unselected samples have shown intact, or even superior, performance in a variety of intelligence, language and motor development tasks. This pattern contrasts 1 Only samples of first-degree relatives with a mean age in the adult range have been reviewed in this chapter. For studies of child and adolescent offspring of bipolar patients, see a recent review by Balanza-Martinez et al.17 2 Bipolar I disorder is characterised by the occurrence of one or more manic or mixed episodes (commonly, patients also experience episodes of depression).22 3 Bipolar II disorder is characterised by the occurrence of one or more episodes of hypomania and one or more episodes of depression.22

13 Neurocognitive Endophenotypes for Bipolar Disorder

with reports of abnormal cognitive and developmental precursors in heterogeneous samples with mixed affective disorders in earlier population-based studies.18–20 The Dunedin Multidisciplinary Health and Development Study21 followed 1,037 individuals from birth to adulthood, 20 of whom received a DSM-IV22 diagnosis of mania at the age of 26. The pre-manic cases had shown no evidence of impaired premorbid IQ, or receptive and expressive language development throughout childhood, but had shown better motor development compared to healthy controls. In contrast, children who later developed schizophreniform disorders were impaired in most cognitive and developmental milestones. These findings were corroborated by two subsequent reports: Firstly, Zammit et al.23 examined diagnostic outcome in 50,053 male Swedish conscripts, 108 of whom developed bipolar disorder during the 27-year follow-up. At baseline, the sample had been assessed on four subtests of intelligence (general knowledge, verbal IQ, visuospatial ability, and mechanical ability), which gave rise to an aggregate IQ score. None of the intelligence measures predicted the development of bipolar disorder, although they did predict schizophrenia, severe depression and non-affective psychoses. Secondly, Reichenberg et al.24 merged the Israeli Draft Board and National Psychiatric Hospitalisation Case Registries, identifying 68 16- to 17-year-olds with no evidence of mental illness, who were later hospitalised for non-psychotic bipolar disorder. In contrast to individuals with schizoaffective or schizophrenia outcomes, pre-bipolar adolescents showed no deficits in eight measures of intelligence and language skills compared to controls with no psychiatric hospitalisations. Conversely, two population-based studies found evidence of circumscribed cognitive deficits in pre-bipolar disorder. Tiihonen et al.25 administered verbal, arithmetic, and visuospatial reasoning subtests to 195,019 apparently healthy males conscripted into the Finnish Defence Forces. Through linkage with the Finnish Hospital Discharge Register, the authors identified 100 individuals who later developed bipolar disorder. Poor performance on the visuospatial reasoning test, and better arithmetic reasoning, were both associated with increased risk for bipolar disorder (in the case of schizophrenia and other psychoses, it was only impaired visuospatial reasoning which increased risk). Most recently, Osler et al.26 examined psychiatric outcomes in early to middle adulthood in a birth cohort of 6,923 Danish men, who had completed assessments of cognitive performance (measured by a total test score that

197

combined verbal, arithmetic and spatial functions) at ages 12 and 18. Using linkage with the national Psychiatric Register, the authors identified 16 men who were diagnosed as having a bipolar affective disorder. Cognitive function measured at 12 and 18 was inversely associated with risk for bipolar disorders. However, the authors commented that the association was imprecisely estimated due to small numbers. It is of interest to note that, while the above studies did not produce a consensus of premorbid cognitive deficits in bipolar disorder, they all reported premorbid cognitive deficits in schizophrenia. This is likely to reflect the presence in schizophrenia of deficits which are stronger, more consistent, and easier to detect reliably, than those in bipolar disorder.

Cognitive Function in Euthymic Patients with Bipolar Disorder To date, more than 70 studies have investigated the presence of cognitive deficits in euthymic bipolar disorder. The majority have examined small sample sizes (often including fewer than 30 patients), a caveat which restricts both their power to detect effects, and their generalisability to the wider patient population. Three recent meta-analyses2,6,5 have attempted to overcome this problem. Robinson et al.2 pooled and quantified neurocognitive evidence from 26 independent studies of euthymic bipolar patients and controls published between 1980 and August 2005. Sixteen cognitive variables (each used in at least four studies), as well as years of education, were included in the meta-analysis. With the exception of IQ, neuropsychological measures were divided into three domains: executive function, memory, and attention/psychomotor speed. The number of patients for any single neuropsychological variable ranged from 97 to 418, and of controls from 102 to 368. Overall, bipolar probands performed significantly worse than controls on all variables except IQ (and years of education). Effect sizes (categorised according to Cohen’s convention27) varied from small (0.2–0.49) to medium (0.5–0.79) to large (0.8+). In the executive domain, category fluency and digit span backward (mental manipulation/ working memory) showed large effect sizes, stroop performance (response inhibition and susceptibility to interference) and Wisconsin Card Sorting Test (WCST) (set-shifting and abstraction) gave rise to medium

198

E. Kravariti et al.

effects, while verbal fluency had a small effect. In the memory domain, a large effect was seen for immediate verbal recall/learning, medium effects for short- and long-delay free verbal recall, and a small effect for digit span forward. Finally, in the attention/psychomotor speed domain, medium effect sizes were seen for three variables (response latency in sustained attention tasks, digit symbol substitution and trails A), and a small effect for a fourth variable – sustained attention sensitivity. Robinson et al.2 found no evidence of publication bias in the available literature. Two later meta-analyses by Arts et al.6 and Torres et al.5 largely confirmed the findings by Robinson et al.2 Table 13.1 summarises the effect sizes estimated by the three meta-analyses for various components of

intelligence, verbal memory/learning, attention/psychomotor speed, and executive function (Arts et al.6 further estimated effect sizes for first-degree relatives of bipolar patients – also included in Table 13.1). It should be noted that, as the three systematic reviews inevitably included many of the same studies, they should be considered as confirmatory analyses rather than as replication studies. Subsyndromal mood symptoms, past illness burden, alcohol abuse, and medication, are potentially serious confounds in the existing literature on euthymic bipolar patients. Subtle mood abnormalities, which do not meet full criteria for a manic, hypomanic, depressive or mixed episode, often persist during periods of remission, and may go undetected in phases of

Table 13.1 Meta-analyses of cognitive function in euthymic bipolar patients and their non-bipolar first-degree relatives Effect sizes (Cohen’s d* (# studies) ) Relatives

Euthmic patients Torres et al.5 IQ Measures IQ Reading Vocabulary Verbal Learning/Memory Immediate Recall/Learning Short Delay Long Delay Delay (unspecified) Recognition Hits Digit Span Forward Attention/Psychomotor Speed Trails A Sustained Attention (CPT) Hits Sensitivity Reaction Time Digit Symbol Substitution Test Executive Function Digit Span Backward Trails B Wisconsin Card Sorting Test Categories Perseverative Errors Category Fluency Letter Fluency (FAS) Stroop All Measures Correct Time ROCFT Figure Recall ROCFT Figure Copy

0.04 0.08

(19) (10)

0.81 0.74 0.72

(12) (10) (12)

0.43

(10)

Robinson et al.2

Arts et al.6

Arts et al.6

0.19

(12)

0.16

(8)

0.19

(5)

0.90 0.73 0.71

(10) (10) (11)

0.82

(12)

0.42

(4)

0.85

(10)

0.56

(4)

0.47

(5)

0.37

(6)

0.04

(4)

0.52

(11)

0.71

(10)

0.13

(7)

0.60

(10)

0.74

(8) (4) (7) (9)

(4)

(10) (8)

0.48 0.60 0.59

0.58

0.62 0.79

0.84

(7)

0.14

(4)

0.54 0.55

(8) (11)

0.98 0.78

(5) (12)

1.02 0.99

(6) (10)

0.18 0.37

(5) (7)

0.69

(8)

0.52 0.88 0.87 0.59

(10) (10) (7) (12)

(4) (6)

(11)

(7) (7) (4) (8)

0.04 0.17

0.47

0.62 0.76 1.09 0.34

0.27

(4)

0.71

(13)

0.63

(11)

0.49

(4)

0.65 0.73 0.62 0.22

(8) (6) (4) (4)

*Effect sizes > 0.8 are considered large; effect sizes 0.5–0.79 are considered medium; effect sizes 0.2–0.49 are considered small. ROCFT: Rey-Osterrieth Complex Figure Test.

13 Neurocognitive Endophenotypes for Bipolar Disorder

‘euthymia’. A recent study by Henry et al.28 has shown that euthymic patients have higher affective lability and more intense emotions than controls.28 Although a number of studies29–31 have attempted to ‘control’ statistically (i.e. co-vary) for subthreshold mood symptoms, despite its common use in the psychological literature, the validity of this statistic approach is under debate.4 In theory, a very strict definition of euthymia that excludes subsyndromal mood symptoms would overcome this caveat, but is logistically difficult. Another problem is that studies of bipolar illness are rarely representative of all patients. Those who recover and return to normal community life are less frequently included in studies than those who are regularly seen by hospital psychiatrists. One of the most consistent findings in the relevant literature to date is an association of cognitive performance with past illness burden (e.g. number of episodes or hospitalisations, overall illness duration, etc.), which raises the possibility that persistent cognitive dysfunction in asymptomatic states reflects neurodegenerative processes in bipolar illness.33 However, an alternative explanation is that patients with more severe forms of the disorder are more cognitively impaired, experience more episodes or hospitalisations, and are treated with higher doses of (cognition altering) medication33 (see below). The extent to which cognitive deficits during euthymia reflect alcohol or iatrogenic effects is unclear. In their recent review, Savitz et al.33 emphasise the frequency of alcohol abuse in bipolar disorder, and consider the failure of most existing studies to account for past alcohol exposures a weak point in the literature. More studies have investigated (or tried to account for) the effects of mood stabilisers and antipsychotic medication on cognition. Joffe et al.34 did not find significant differences between medicated and unmedicated patients, but also failed to detect differences between patients and controls. Gualtieri and Johnson35 4

While adjusting for the effect of covariates is a valid and useful way to specify a model and remove noise from data, it can only be done given certain conditions. If groups are significantly different on a variable due to their group membership, it is not valid to adjust for that variable. This is because, in such cases, it is not possible to partition the variance due to the group membership and that due to the covariate. This is exactly the case when comparing patients with subsyndromal symptoms to healthy controls. For a more detailed discussion of this issue, please see ‘Misunderstanding analysis of covariance’ by Miller and Chapman.32

199

estimated (in a patient-only sample) the relative impact of different anticonvulsants and lithium on cognition. The authors concluded that each treatment was associated with different degrees of ‘cognitive toxicity’, with lamotrigine and oxcarbazepine having the least, lithium intermediate, and topiramate, valproic acid and carbamazepine, the most impact on cognition. Considering the possibility that cognitive dysfunction in bipolar disorder is iatrogenic, Savitz et al.33 reviewed several studies with mixed findings. Many investigations showed adverse effects of mood stabilisers and neuroleptic medication on memory, attention, executive processes, and psychomotor functioning, while others indicated that lithium and valproate, rather than impacting negatively on cognition, exert a neuroprotective effect. This possibility was offered as a tentative explanation for the worse verbal memory of unmedicated first-degree relatives of bipolar probands compared to the index cases in Frantom et al.36 However, Savitz et al.33 emphasized that patients are often treated with complex combinations of mood stabilisers, antidepressants, antipsychotics and anxiolytics, whose combined effects are a matter of speculation.

Cognitive Function in First-Degree Relatives of Bipolar Disorder Patients The multiple confounds of subsyndromal mood symptomatology, past illness burden, and medication can be addressed to some extent by examining non-bipolar first-degree relatives (including co-twins) of affected probands. Healthy family members of bipolar patients are less likely to be taking psychotropic medication or manifest mood symptoms. In practice, excluding the possibility of subthreshold affective symptoms in this population is less straightforward, as unipolar depression and other psychopathology is observed at a higher rate in the non-bipolar first-degree relatives of affected probands than in the general population.17,37 Another limitation of most family studies is their inability to partition familial resemblance into covariance resulting from shared environmental influences (e.g. diet, or social class) and covariance due to shared genetic factors.38 The classical twin design overcomes this limitation by combining information from monozygotic (MZ) and dizygotic (DZ) twin pairs to decompose

200

individual differences in a trait into genetic and environmental sources of variance.38 Such decomposition is computationally feasible, because MZ and DZ twins are assumed to have the same degrees of intra-pair correlation for environmental influences, but different degrees of correlation for genetic influences. Therefore, it becomes possible to estimate the heritability of a particular phenotype (e.g. Verbal Intelligence Quotient) by quantifying the relative contribution of genetic effects to the total phenotypic variance. More importantly, modern techniques of biometric genetic analysis, such as structural equation model fitting,

E. Kravariti et al.

can make use of twin data to partition the phenotypic correlation between two traits (e.g. bipolar illness and verbal learning) into covariance determined by common genes and covariance determined by common environmental effects. In other words, suitable manipulation of zygosity and diagnostic status in twin studies enables an estimation of genetic correlation between bipolar illness and various indexes of neuropsychological function. Table 13.2 lists 18 neurocognitive studies of nonbipolar first-degree relatives (including co-twins) of affected probands, divided on the basis of their positive

Table 13. 2 Family and twin studies of neurocognitive function in bipolar disorder Relatives = Controlsa Intelligence – General Cognitive Function NART IQ Fer04, McI05 Fra05, Gou99, McI05, Zal04 Wechslerc Full Scale IQ McI05, Tou06 Wechslerc Verbal IQ (VIQ) Tou06 Wechslerc Performance IQ (PIQ) Wechslerc VIQ - PIQ Ant07, Kie05 Wechslerc Vocabulary CAMCOR Score Verbal Learning/Memory/Recognition Ant07, Cla05b, Fer04, Sob03 Word Liste Immediate Recall/Learning Word Liste Semantic Clustering Ant07 Ant07 Word Liste Learning Slope Gou99 Word Liste Primacy Recall Ant07, Cla05b, Fra08, Gou99, Ker01 Word Liste Short Delay Free Recall Ant07, Gou99 Word List e Short Delay Cued Recall Ant07, Cla05b, Fra08, Sob03 Word Liste Long Delay Free Recall Word Liste Long Delay Cued Recall Ant07, Kie05 Ant07, Gou99 Word Liste Intrusions (for Free Recall) B-P Intrusions Ant07, Cla05b, Fer04, Fra08, Gou99, Ker01 Word Liste/B-P Recognition Hits Kie05 Word Liste Recognition Discriminability Gou99 Word Liste Response Bias E-RBMT Score WMS Memory Quotient WMS/-R Story Recall-Immediate Gou99, Kie05, Kre98 WMS/-R Story/Text Recall-Delayed Gou99, Kie05, Kre98 Visual (-Spatial) Learning/Memory/Recognition Visual Backward Masking Ker01 WMS/-R Visual Reproduction-Immediate Gou99, Kie05, Kre98 WMS/-R Visual Reproduction-Delayed Kie05, Kre98 BFLT-E Immediate Recall/Learning BFLT-E Short Delay Recall Fra08 BFLT-E Long Delay Recall ROCFT Short Delay Figure Recall ROCFT Long Delay Figure Recall Fra08, Gou99 WMS-III Faces I/Faces II Pennsylvania Facial Recognition Kie05 CANTAB Spatial/Pattern Recognition Working Memory Digit Span Forward Ant07, Fer04, Ker01, Pir05 Digit Span Backward Ant07, Gou99, Ker01, Pir05 Word/Alphabet span Ker01

Relatives ≠ Controlsb

Fra08, Tou06 McI05d Tou06 Tou06 Chr06 Fra08, Gou99, Kie05f

Gou99, Kie05f Gou99 Gou99 Gou99, Sob03 McI05 Gou99 Ker01

Fra08 Fra08 Fra08 Fra08 Fer04/Fer04g

Fer04 (continued)

13 Neurocognitive Endophenotypes for Bipolar Disorder

201

Table 13. 2 (continued) Relatives = Controlsa B-P Total Correct WMS-R Visual Span Forward/Backward Spatial Span Spatial Working Memory Attention – Executive Function – Psychomotor Speed WMS Mental Control CPT Reaction Time/Latency CPT Omissions/Commissions CPT Hits/Sensitivity/Other Measures Trails A (Time) Trails B (Time) Trails B (Correct) Trails B - Trails A (Time) Stroop Correct Stroop Interference HSCT Errors Dichotic Listening (A’) WCST Categories WCST Perseverative Errors Letter Fluency Category Fluency SOC/TOL Planning Accuracy IES Completion Rate IES Total/Reversal/ID-Shifting Errors IES ED Shifting Errors Digit Symbol (Substitution Test) Go/No Go Reaction Time/Number Correct Purdue Grooved Pegboard Choice Reaction Time Visuospatial/Constructional/Motor Function ROCFT Figure Copy Benton Judgement of Line Orientation Benton Facial Recognition Wechsler Block Design Simple Reaction Time Motor Speed Grip Strength

Relatives ≠ Controlsb Gou99

Ant07, Pir05 Fra08, Ker01 Fer04, Tri08

Fer04

Gou99 Cla05a, Fer04, Kie05, Tri08 Fer04 Cla05a, Fra08, Gou99, Kre98 Ant07, Fer04, Fra08, Gou99, Szo06, Zal04 Ant07, Fer04, Fra08, Gou99, Zal04 Ant07, Szo06, Zal04 Chr06, Fer04 Sob03 McI05 Kre98 Fra05, Fra08, Gou99, Ker01, Zal04, Fra05, Gou99, Ker01, Kre98, Szo06, Zal04 Fer04, Fra08, Gou99, Ker01, McI05, Sob03, Zal04 Fra08 Sob03

Tri08

Szo06 Ant07

Chr06, Zal04 Fra05 Sob03 Kre98, Tri08 Fra08, Tri08

McI05 Cla05b

Cla05b Fer04, Fra08, Kie05, McI05 Sob03 Fra08 McI05, Sob03 Fra08 Gou99 Fra08, Gou99 Tou06 McI05

Cla05b Ant07

Fra08 Fra08 Sob03 Fra08

Term Abbreviations: B-P: Brown-Peterson; BFLT-E: Biber Figure Learning Test-Extended; CAMCOR: Cambridge Cognitive Examination – Revised; CANTAB: Cambridge Neuropsychological Test Automated Battery; CPT: Continuous Performance Test; ED: Extra-dimensional; E-RBMT: Extended Rivermead Behavioral Memory Test; HSCT: Hayling Sentence Completion Test; ID: Intra-dimensional; IES: (CANTAB) Intra-dimensional Extra-dimensional Shift; NART: National Adult Reading Test; ROCFT: Rey-Osterrieth Complex Figure Test; SOC: Stockings of Cambridge; TOL: Tower of London; WCST: Wisconsin Card Sorting Test; WMS/-R: Wechsler Memory Scale/-Revised. Study Abbreviations: Ant07: Antila et al.45; Chr06: Christensen et al.53; Cla05a: Clark et al.41; Cla05b: Clark et al.52; Fer04: Ferrier et al.51; Fra05: Frangou et al.42; Fra08: Frantom et al.36; Gou99: Gourovitch et al.39; Ker01: Keri et al.49; Kie05: Kieseppa et al.7; Kre98: Kremen et al.48; McI05: McIntosh et al.43; Pir05: Pirkola et al.44; Sob03: Sobczak et al.50; Szo06: Szoke et al.46; Tou06: Toulopoulou et al.47; Tri08: Trivedi et al.54; Zal04: Zalla et al.40 a First-degree relatives (including MZ and DZ co-twins) show no statistically significant differences compared to controls. b Statistically significant difference denotes impairment or atypical pattern (e.g. pronounced discrepancy between Verbal IQ and Performance IQ) relative to controls. c WAIS-R (Wechsler Adult Intelligence Scale-Revised) or WAIS-III (Wechsler Adult Intelligence Scale-III) or WASI (Wechsler Abbreviated Scale of Intelligence). d Sub-sample of unaffected relatives from ‘mixed’ families (affected with both bipolar disorder and schizophrenia). e Word list tasks include the California Verbal Learning Test (CVLT), Rey Auditory Verbal Learning Test (RAVLT), Verbal Visual Learning, and their procedural analogues. f Female sub-sample only. g In sub-sample with no Axis I diagnoses.

202

or negative findings in relation to a putative association between different aspects of neurocognition, and presumed genetic liability for bipolar disorder. The numbers of first-degree relatives in the various investigations have ranged from 7 to 64, and of controls from 8 to 114. Eleven of the studies performed direct comparisons between bipolar probands, their non-bipolar first-degree relatives, and controls,7,36,39–42,43–47 and seven studies compared non-bipolar first-degree relatives with controls.48–54 The latter group included participants with no personal psychiatric history, or less frequently, nonbipolar individuals with no personal or family history of psychotic and bipolar disorders. In this literature, a failure to support an association between neurocognitive function and the genetic determinants of bipolar illness was inferred from a lack of statistically significant differences in neuropsychological performance between first-degree relatives and controls. In contrast, a deficit (or atypical pattern) in relatives compared to controls was taken to support an association with the bipolar diathesis. Finally, a graded deficit in bipolar twins compared to identical co-twins, and in the latter compared to controls (bipolar twins < identical co-twins < controls) was thought to result from the combined effects of genetic vulnerability and the manifest disease on neurocognitive function. Of over 70 neuropsychological dimensions investigated across the 18 studies (spanning six broad neurocognitive domains: see Table 13.2), about half did not emerge as putative endophenotypes in any investigation, 31 received support in a single study each, and fewer than 10 were supported from more than one different studies. For the majority of neuropsychological measures which emerged as potential endophenotypes in at least two publications, for every supporting finding, there were two or more negative ones. Within each composite domain, at least three processes received support as tentative markers. However, some domains have been better represented (more broadly investigated) in the existing literature than others (Table 13.2).

Intellectual Function None of the 18 studies showed deficits in pre-morbid or Verbal IQ in first-degree relatives compared to controls.39,40,42,43,51 However, contrary to a report of higher general intellectual function in female relatives

E. Kravariti et al.

compared to controls in the investigation by Kremen and colleagues,48 two studies reported deficits in FullScale IQ47,36 and another in Performance IQ43 in their respective samples. In addition, Toulopoulou et al.47 found a significantly higher discrepancy between Verbal IQ (96) and Performance IQ (108) in a highrisk cohort compared to controls (12 as opposed to 5 points). It is of interest to note that two of the three studies that reported deficits in current IQ, or a pronounced discrepancy between verbal and practical intelligence, had selected their relatives for a family history of a ‘familial’ form of bipolar disorder (at least two family members affected with bipolar disorder or bipolar disorder and other psychosis),43,47 while the third one reported a relatively high proportion (58%) of ‘familial’ probands.36

Verbal Memory, Learning and Recognition The best supported candidate neurocognitive endophenotypes in the family literature to date are immediate verbal recall/learning7,36,39 and long delay verbal recall,7,39,49 as each process received support from three independent studies (Table 13.2). However, both aspects also failed to emerge as genetically mediated risk indicators of bipolar illness in an even larger number of investigations,7,39,45,48,50–52 Other candidate markers are two composite indexes of memory function,39,43 as well as recognition discriminability (i.e. the ability to differentiate between targets and distracters during delayed recognition of word list items). However, a recent study by Frantom et al.36 failed to detect differences in verbal learning and memory among bipolar probands, their unaffected relatives and healthy controls, despite demonstrating group differences in several other neurocognitive domains.36 In the same study, verbal learning and memory was the only composite function not to show an intermediate pattern of performance in the unaffected relatives compared to patients and normal controls (i.e. patients < relatives < controls). The findings of this investigation36 are in line with those reported in three other studies: Kremen et al.48 proposed that verbal memory may be an indicator of risk specifically for schizophrenia as opposed to bipolar disorder, Ferrier et al.51 concluded that memory dysfunction

13 Neurocognitive Endophenotypes for Bipolar Disorder

in bipolar illness may be limited to the visuospatial domain, and Antila et al.45 suggested that verbal memory impairment may be related to the fully developed disease.

Visual (-spatial) Learning, Memory and Recognition The non-verbal modality of memory has shown a less mixed pattern of findings in the family literature, but this is likely to have resulted from the small number of studies exploring processes within this neuropsychological domain (Table 13.2). Using the Biber Figure Learning Test-Extended (BFLT-E), described as the visual analogue of the California Verbal Learning Test (CVLT) (the most widely used list learning task), Frantom et al.36 identified immediate visual recall/ learning and long delay visual recall as potential endophenotypes for bipolar illness. This finding raises the possibility that learning which occurs over repeated (3–5) presentations of test items, and recall of the same material following a delay of 20–30 min (as opposed to 1–3 min), may be promising endophenotypes regardless of modality, as the same processes have received the most robust support in the search of memory markers within the verbal modality. However, this possibility requires further investigation, as only one family study has used the BFLT-E to date,36 while an alternative test, the Rey-Osterrieth Complex Figure Test, gave rise to the reverse pattern: It was short- rather than long delay figure recall that emerged as a putative endophenotype.36 Other candidate markers (each receiving support from a single study), are pattern and spatial recognition,51 as well as processes indexed by the immediate and delayed conditions of the Wechsler Memory Scale-III (WMS-III) Faces subtest.36

Attention, Executive Function and Psychomotor Speed Mental control, a WMS index of attention/concentration based on the speeded rehearsal of familiar material,39 correct responses in trails B (alphanumeric sequencing and set shifting),45 completion rate in Intradimensional/Extra-dimensional Shift (a computerised

203

analogue of the Wisconsin Card Sorting Test),52 and extra-dimensional shifting errors52 have each received tentative support as endophenotypes in a single study (while no studies have produced negative findings in relation to these measures so far). On the other hand, aspects of selective attention, response inhibition (dichotic listening, stroop interference, Hayling Sentence Completion Test errors) and planning accuracy have received mixed reports in the few existing studies (Table 13.2).

Working Memory and Visual-Spatial, Constructional and Motor Function Working memory, visual-spatial and constructional abilities, and motor function have been the focus of relatively few family studies (Table 13.2). Preliminary support for the endophenotypic status of digit span backward,51 correct responses in a verbal working memory task with an interference component,39 spatial span,51 judgement of line orientation,36 Wechsler Block Design,36 motor speed50 and grip strength36 has come from no more than a single study for each component. To date, the most consistent evidence across family and twin studies involves negative rather than positive findings. Indeed, short-delay recall of word-list items, recognition hits in verbal learning tasks, letter fluency, and trails A (alphabetical sequencing, attention and psychomotor speed) have consistently failed to receive support as putative endophenotypes in several studies (Table 13.2). Verbal and visual span tasks (tapping working memory) have given rise to negative findings in five reports,39,44,45,51,49 combined with limited support for digit span backward in one study.51 Various components of sustained attention, as measured by continuous performance tasks, have shown intact performance in seven studies of first-degree relatives,7,36,39,41,48,51,54 with the exception of omission and commission errors, which emerged as potential endophenotypes in a recent study.54 Trails B (time) or trails B–trails A (time) have received no support as risk indicators in six family and twin studies,39,40,45,46,36,51 and contrasting evidence (for trails B accuracy) in one investigation.46 Each component of the Wisconsin Card Sorting Test (number of categories and perseverative errors) has elicited negative findings in at least five studies, and supporting evidence in just 2 (Table 13.2). Finally, Digit Symbol, a

204

measure of processing psychomotor speed, has shown intact performance in four cohorts of first-degree relatives,7,36,43,51 and deficits in a population based sample selected for a family history of ‘familial’ bipolar disorder.45

Cognitive Function in Co-Twins of Bipolar Disorder Patients Despite the promise of the twin method as a useful tool for partitioning genetic and environmental sources of familial resemblance, only four of the 18 studies reviewed in the preceding section have used twin designs. Furthermore, none of the four studies has assessed sufficiently large numbers of MZ and DZ twin pairs in genetically informative combinations to allow an estimation of genetic correlations with bipolar illness of different neuropsychological measures. Reduced power to compare MZ and DZ discordant pairs has subjected the existing twin studies to the limitations of other family designs in partitioning genetic and environmental sources of familial resemblance. The four existing twin studies are briefly outlined below. Gourovitch et al.39 administered a comprehensive neuropsychological battery to seven monozygotic twin pairs discordant for bipolar illness and seven normal control MZ pairs. At the time of testing, three of the affected twins were euthymic, two had depressive symptoms, and two were in the manic state, while cotwins were free from psychiatric pathology. Delayed verbal recall, recognition discriminability, a composite index of general memory function (WMS memory quotient), intrusions during verbal working memory, and attention/concentration (WMS mental control) were impaired in both high-risk groups compared to controls, but failed to differentiate between the two discordant groups. Therefore, the respective functions emerged as genetically mediated indicators of risk for bipolar illness, with no evidence of modifying effects of the fully developed illness. In contrast, immediate verbal recall/learning, and total number of words recalled in a verbal working memory task with an interference component (Brown-Peterson task) appeared to be under the influence of both the bipolar diathesis and the manifest disease, as they elicited impairments in the affected twins compared to co-twins, and in both highrisk groups compared to normal controls.

E. Kravariti et al.

Kieseppa et al.7 investigated general intellectual functioning, information-processing speed, verbal memory and learning, and non-verbal memory in a populationbased Finnish sample of 26 euthymic twins with bipolar I disorder, 19 non-bipolar twins from discordant pairs (largely free from affective disorders), and 114 controls with no history of psychotic disorders. The high-risk sample comprised concordant and discordant pairs, as well as 13 affected or unaffected individuals with no participating co-twins. More than 68% of the high-risk participants were dizygotic, and the sample included only two monozygotic discordant pairs. The bipolar twins showed impairments on all processing speed, memory and learning variables compared to controls, but no deficit was detected in the total sample of unaffected co-twins. However, female co-twins showed deficits in immediate verbal recall/learning and delayed verbal recall, partially replicating the findings by Gourovitch et al.39 Pirkola et al.44 examined verbal and spatial working memory in a nationwide Finnish sample of 16 twin pairs discordant, and 3 pairs concordant, for bipolar I disorder, compared with 100 control twins with no history of psychotic or current affective disorders, and 78 control twins at established or presumed genetic risk for schizophrenia. The sample overlapped with the cohort studied by Kieseppa et al.,7 and included only three MZ pairs discordant for bipolar disorder. In contrast to individuals at high risk for schizophrenia, neither bipolar probands nor their co-twins showed deficits in verbal or visual span tasks compared to normal controls. In addition, bipolar patients performed worse than their co-twins, but the latter group outperformed the normal controls, in the forward condition of digit span. Lastly, Christensen et al.53 studied a nationwide Danish sample of affectively non-disordered co-twins of patients with unipolar or bipolar affective disorders, compared to affectively non-disordered co-twins of individuals with no hospital contacts for affective disorders. Statistical comparisons were performed between high- and low-risk participants, as well as between subgroups defined by zygosity and co-twin diagnosis (unipolar, bipolar, none). The high-risk subgroup with bipolar co-twins comprised 7 MZ and 14 DZ participants. Compared to zygosity-matched controls, MZ co-twins of bipolar probands showed a significant deficit on a composite score of cognitive function (CAMCOR), and DZ co-twins on a stroop measure of susceptibility to interference.

13 Neurocognitive Endophenotypes for Bipolar Disorder

205

Methodological Considerations in Neurocognitive Studies of First-Degree Relatives

A recent such effort by Arts et al.6 (which included most studies of first-degree relatives – including cotwins - reviewed in this chapter, but excluded Clark et al.,41 Antila et al.,45 Frantom et al.,36 and Trivedi et al.54) found a near-significant moderate effect for CVLT delayed verbal recall (0.56), and a statistically significant small-to-medium effect for CVLT immediate verbal recall/learning (0.42) (Table 13.1). Both measures (and their procedural analogues) were identified as the best supported candidate endophenotypes in the present review. Interestingly, stroop performance, for which we found inconsistent evidence across studies, and trails B, which elicited negative findings in nearly all relevant investigations,36,39,40,45,51 gave rise to the second (0.49) and fourth (0.37) largest effect sizes, respectively, in this meta-analysis.6 The finding attests to the confounding effect of limited statistical power in the existing family studies. Other measures (WCST categories, digit span forward, trails A, digit symbol substitution, WCST perseverative errors, digit span backward, IQ and letter fluency), yielded both a preponderance of negative findings in the present review, and small effect sizes (0.04–0.27) in Arts et al.6 (Table 13.1), proving less promising as candidate endophenotypes.

The picture which emerges from the family studies reviewed is one of statistically non-significant differences between first-degree relatives and controls for numerous neuropsychological measures, of preliminary, isolated and non-replicated support for many candidate markers, and of mixed evidence for several other cognitive processes, usually tipping the balance towards negative findings. This obscure pattern is compounded by a number of important methodological aspects, which have varied across family studies. The resulting heterogeneity has determined the extent to which evidence is comparable across studies, and may account for some of the divergent findings for overlapping measures.

Number of Studies, Statistical Power, and Measure Overlap Only a small number of neurocognitive investigations of first degree relatives have been published to date, and these overlap to a limited extent with regard to the measures employed. Fragmenting the neuropsychological focus into different functional subcomponents across studies has resulted in unavoidable trade-offs between breadth, diversity and replicability of findings, increasing the number of working hypotheses, but limiting conclusions. In many investigations, the groups of relatives have performed, on average, slightly below the level of controls, against a background of statistically non-significant differences. Clark et al.41 commented that in order to confirm a statistically significant difference at a small-tomedium effect of 0.38, case control studies should aim to recruit 115 subjects per group. However, over 60% of family studies to date have included fewer than 25 relatives and 50 controls in their designs, strongly raising the possibility that some investigations may have lacked statistical power to detect small or small-to-medium effects. Methodological obstacles such as the dearth of relevant studies, and their small sample sizes can be overcome to some extent in meta-analytic studies.

Degree of Genetic Susceptibility in First-Degree Relatives The degree of genetic susceptibility among firstdegree relatives can be presumed to vary across studies based on the selection criteria and mean age of participants. A small number of studies have purposefully selected first-degree relatives from families with at least two affected members (‘familial’ bipolar disorder),43,45,47 while Trivedi et al.54 specifically tested unaffected siblings of patients with sporadic bipolar disorder. In addition, the mean age of relatives in the various investigations has ranged from 27 to 51 years. Therefore, studies of younger adults may have included more individuals likely to develop bipolar disorder or psychosis in the future, while studies of older unaffected participants may have inadvertently included some people with ‘resilience’ to affective disorders due to unknown ‘protective’ factors (which may or may not be related to the putative marker under investigation).

206

Symmetry of Exclusion Criteria for First-Degree Relatives and Controls Another potential confounder, which has received less mention in existing studies, is the degree of symmetry in the exclusion criteria applied to first-degree relatives and controls. Some studies have excluded controls with personal (and frequently family) histories of psychiatric disorders, whilst including relatives with lifetime diagnoses of unipolar affective or other Axis I disorders.41,42,46,51,52 Other studies have used largely symmetric, relatively strict, exclusion criteria for relatives and controls (excluding participants with histories of affective or other Axis I disorders).36,43,49,50,54 In contrast, a small number of studies have used largely symmetric, but relatively ‘relaxed’, criteria for relatives and controls, ‘tolerating’ unipolar affective or other Axis I disorders in their samples.45,47,48 Arguments for including participants with unipolar depression (and sometimes past substance or alcohol abuse) histories are: (i) Such psychopathology is relatively common in the general population55,56; (ii) major depressive disorder is the most prevalent mood disorder among relatives of bipolar probands57; and (iii) the genes that predispose to bipolar disorder are likely to have pleiotropic effects, increasing the risk of a range of psychopathology in the relatives of affective probands.33 Therefore, reducing such psychopathology to nearzero rates may lead to selection of atypical samples of first degree relatives, with artificially ‘deflated’ degrees of genetic susceptibility, and ‘super-normal’ controls. On the other hand, the presumed genetic continuity between bipolar disorder and unipolar depression8 implies that including controls with histories of depressive symptomatology may reduce both genetic distinctiveness from high-risk samples and the likelihood to detect small or small-to-medium effects.

Genetic Heterogeneity of Bipolar Disorder? MacQueen et al.12 have commented that the relatively slow progress of gene-mapping efforts is partly due to the ‘complex’ nature of bipolar disorder, which most likely is a heterogeneous entity. Their recent review suggests that more homogeneous groups of bipolar

E. Kravariti et al.

patients could be defined using a number of criteria, including age of onset, presence or absence of psychotic symptoms/comorbid conditions, familial patterns of illness, suicide attempts, treatment response, gender of transmitting parent, and clinical course of illness.12 Since most family studies to date have not attempted to define homogeneous forms of bipolar illness, it is possible that the effects of the bipolar diathesis on neuropsychological function have been concealed by the disparate associations with cognition of different genetic forms of the disorder.

Bridging ‘Cognitive’ Genetics with ‘Psychiatric’ Genetics There have been strenuous efforts to identify the genes that underpin cognitive dysfunction and/or vulnerability to bipolar disorder. The point where psychiatric genetics and cognitive genetics intersect, gives a clue to the plausibility of cognitive dysfunction as a key to unlocking the genetic complexity of bipolar illness. To mention a few examples of relevant research, Savitz et al.58 have recently evaluated evidence favouring the hypothesis that COMT (catechol-O-methyl-transferase) and BDNF (brain derived neurotrophic factor), two of the most studied genes for schizophrenia and bipolar disorder, influence cognitive function. In a study of 723 members of 179 Finnish families affected with bipolar disorder, Palo et al.59 found that haplotypes at the 30 end of DISC1 (disrupted-in-schizophrenia-1) were associated with both bipolar spectrum disorders and several cognitive traits, with the most robust signal detected for rs821616 and verbal fluency, and rs980989 and psychomotor processing speed. These results are corroborated by additional credible evidence, both genetic and biological, for a dual role of DISC1 in determining susceptibility to psychiatric illness and in biochemical processes that are directly linked to learning, memory, and mood.60–62 Finally, Savitz et al.63 assessed 190 individuals of the reproductively isolated Afrikaner population with a battery of neuropsychological tests, and found that bipolar disorder patients performed significantly worse than their unaffected relatives on memory tasks. Using this memory-related putative endophenotype, they carried out a focused linkage and family-based association study, and detected evidence for linkage on chromosome 22q11, a region implicated in bipolar disorder.

13 Neurocognitive Endophenotypes for Bipolar Disorder

Evidence that allelic variation in some genes is likely to be involved in both cognitive function and the aetiology of bipolar disorder reinforces the notion that endophenotypic approaches may prove a fruitful strategy in psychiatric genetics. Savitz et al.58 concluded that ‘far from being an esoteric backwater of modern psychiatry, cognitive genetics is probably indispensable to a comprehensive account of both the aetiology and pathophysiology of psychiatric illness’. A less optimistic view has been put forward by Flint and Munafo.64 These authors examined the assumptions that endophenotypes have a simpler genetic basis, and bear a closer relationship to the biological determinants of psychiatric illness, than diagnostic categories. Applying meta-analytic techniques, they showed that the number of cases (and an equal number of controls) that would be required to detect a genuine association between COMT and schizophrenia, and between COMT and two promising neurocognitive endophenotypes (WCST perseverative errors and N-Back performance) would be in excess of 900 and 1,700, respectively. The authors found no evidence that the genetic effects of COMT on neurocognitive performance are any larger than on schizophrenia. They further commented that there are very few robust estimates of effect sizes for susceptibility loci in either psychiatric disease or endophenotypes. Discussing examples of traits in model organisms, the authors concluded that endophenotypes are not likely to be any easier to dissect at a genetic level than the disorders to which they are related. Lending more weight to the above reservations, Savitz et al.58 observed that with approximately half of the 32,000 Homo sapiens genes expressed in the brain, and a significant portion expressed in the prefrontal cortex, narrowing down the list of candidates for involvement in cognition becomes a speculative exercise. In addition, the gene-cognition correlations reported in the literature usually rest on the assumption that the effect of a genetic variant is specific to the cognitive process under investigation – as opposed to general intelligence, an assumption that may not survive scrutiny.58 Finally, Craddock et al.65 commented that despite compelling physiological relevance to brain pathology in psychosis, and a role in executive function that is hard to dispute, genetic variation at COMT has proved particularly resistant to the substantial research efforts aimed at demonstrating a clear relationship with psychiatric phenotypes. They predicted

207

that the experiences with this gene will provide a foretaste of the complexity of genotype–phenotype relationships in psychiatric genetics.65

Conclusions and Future Directions The concept of endophenotypes for complex genetic diseases has sparked increased optimism regarding genetic research into the causes of bipolar disorder. The presence of neuropsychological deficits in individuals with a diagnosis of bipolar illness is a matter of little controversy or debate, as is the genetic determinism of much of the vulnerability for both phenotypes. There is mounting and converging evidence from neurocognitive investigations of euthymic and at-riskstates of bipolar illness that there is more to cognitive dysfunction than the adverse effects of affective symptoms, medication, past illness burden and the detrimental recurrence of mood episodes. However, counter evidence also compels attention: The focus on the cognitive course of pre-bipolar individuals is yet to produce evidence of an inherent neuropsychological deficit, while some of the few family studies may have compared relatives with typical rates of psychopathology with controls with an atypical lack of it. Equally, while large twin studies of combined monozygotic and dizygotic twin samples hold the key to confirming the existence of true endophenotypes, to date, they have not been represented in the literature. The balance of evidence suggests that immediate recall of word lists, and learning that occurs over repeated presentations of list items deserve further investigation as putative endophenotypes of bipolar illness. The same observation can be extended to the delayed recall of verbal material, as well as circumscribed aspects of selective attention, response inhibition and resistance to interference that can be measured using the stroop neuropsychological paradigm. We think that caution should be exercised with generalisations that ‘verbal memory’ or ‘executive function’ are endophenotypes for bipolar illness. Both are composite neurocognitive domains, comprising multiple subcomponent processes, most of which have not received substantial support as endophenotypes in the existing literature. The probable genetic heterogeneity of bipolar illness, combined with the limited number, statistical power

208

and neuropsychological overlap of family studies to date, may have prevented discernible patterns of neurocognitive ‘weaknesses’ or ‘strengths’ from characterising the bipolar diathesis. Future studies should aim to identify neuropsychological endophenotypes in genetic sub-forms of the disorder, informing their selection criteria from current advances in molecular genetics. Given the relatively low prevalence of bipolar disorder, and the equally low prevalence of the most informative section of the population (twins), international collaborations may be the way forward for attaining large and homogeneous study samples. Identifying neurocognitive endophenotypes in bipolar illness is just a first step. Whether their utility in deciphering the genetic complexity of bipolar disorder will stand the test of evidence is difficult to answer. Some of the assumptions about the genetic simplicity of endophenotypes and their proximity to the genetic substrates of psychiatric illness have been recently challenged.64 Despite the plausibility of some biomarkers as indicators of ‘core’ pathophysiological processes in psychiatric illness, the sensitivity and specificity with which they can detect fundamental features of bipolar illness are not established. In addition, none of the putative neurocognitive markers identified to date can be claimed to be specific for bipolar disorder compared with schizophrenia.40 Finally, although neuropsychological endophenotypes are certainly not ‘symptom dimensions’, we cannot exclude the possibility that some of the variance in cognitive performance can be accounted by residual, or sub-threshold affective symptoms. Therefore, at least some of the criteria for useful diagnostic biomarkers (see Chapter 1 in this book by Ritsner and Gottesman) have not been convincingly met in relation to bipolar illness. If we should project to the future, we would guess that Craddock et al.’s65 caution about the lessons learned from the COMT gene are more realistic than pessimistic. On the other hand, neuropsychological endophenotypes can be measured repeatedly over time. Given their relative temporal stability and (at least partial) dissociation from symptom state, combined with the satisfactory psychometric properties of several neurocognitive instruments, neuropsychological markers are reproducible. The extensive evidence base associated with many neuropsychological paradigms has established them as valid and reliable tools. Most are relatively in-expensive, easy-to-perform procedures,

E. Kravariti et al.

and can flexibly be administered in diverse settings, including research centres, outpatient facilities and the examinee’s own home. Provided that the recommended physical conditions for testing (e.g. lighting, ventilation, lack of distractions etc.) are provided, neurocognitive markers should be reliably measured in all these testing environments. In addition, neuropsychological tasks carry low risk for harm to the individual being assessed, and being non-invasive, they may be perceived as more appealing than certain medical procedures or biological sampling techniques. These properties satisfy several of the criteria for useful diagnostic biomarkers discussed by Ritsner and Gottesman in Chapter 1 of this book. Therefore, we will conclude with an optimistic message: Even if neurocognitive endophenotypes are not easier to dissect genetically than complex disease phenotypes, being quantitative and reliable, they may still prove more suitable for collecting the large samples needed for genetic analysis than full-blown psychiatric illness.64

References 1. Tohen M, Zarate CA, Hennen J et al. The McLean-Harvard First-Episode Mania Study: prediction of recovery and first recurrence. Am J Psychiatry 2003;160:2099–2107 2. Robinson LJ, Thompson JM, Gallagher P et al. A metaanalysis of cognitive deficits in euthymic patients with bipolar disorder. J Affect Disorder 2006;93:105–115 3. Bearden CE, Hoffman KM, Cannon TD. The neuropsychology and neuroanatomy of bipolar affective disorder: a critical review. Bipolar Disorders 2001;3:106–150 4. Krabbendam L, Arts B, van Os J et al. Cognitive functioning in patients with schizophrenia and bipolar disorder: a quantitative review. Schizophr Res 2005;80:137–149 5. Torres IJ, Boudreau VG, Yatham LN. Neuropsychological functioning in euthymic bipolar disorder: a meta-analysis. Acta Psychiatrica Scandinavica 2007;116:17–26 6. Arts B, Jabben N, Krabbendam L et al. Meta-analyses of cognitive functioning in euthymic bipolar patients and their first-degree relatives. Psychol Med 2008;38:771–785 7. Kieseppa T, Tuulio-Henriksson A, Haukka J et al. Memory and verbal learning functions in twins with bipolar-I disorder, and the role of information-processing speed. Psychol Med 2005;35:205–215 8. McGuffin P, Rijsdijk F, Andrew M et al. The heritability of bipolar affective disorder and the genetic relationship to unipolar depression. Arch Gen Psychiatry 2003;60:497–502 9. Tuulio-Henriksson A, Haukka J, Partonen T et al. Heritability and number of quantitative trait loci of neurocognitive functions in families with schizophrenia. Am J Med Genet 2002;114:483–490

13 Neurocognitive Endophenotypes for Bipolar Disorder 10. Singer JJ, MacGregor AJ, Cherkas LF et al. Genetic influences on cognitive function using The Cambridge Neuropsychological Test Automated Battery. Intelligence 2006;34:421–428 11. Gottesman II and Gould TD. The endophenotype concept in psychiatry: etymology and strategic intentions. Am J Psychiatry 2003;160:636–645 12. MacQueen GM, Hajek T, Alda M. The phenotypes of bipolar disorder: relevance for genetic investigations. Mol Psychiatry 2005;10:811–826 13. Zuchner S, Roberts ST, Speer MC et al. Update on psychiatric genetics. Genet Med 2007;9:332–340 14. Segurado R, tera-Wadleigh SD, Levinson DF et al. Genome scan meta-analysis of schizophrenia and bipolar disorder, part III: Bipolar disorder. Am J Hum Genet 2003;73:49–62 15. Farmer A, Elkin A, McGuffin P. The genetics of bipolar affective disorder. Curr Opin Psychiatry 2007;20:8–12 16. Sklar P, Smoller JW, Fan J et al. Whole-genome association study of bipolar disorder. Mol Psychiatry 2008;13:558–569 17. Balanza-Martinez V, Rubio C, Selva-Vera G et al. Neurocognitive endophenotypes (Endophenocognitypes) from studies of relatives of bipolar disorder subjects: A systematic review. Neurosci Biobehav Rev 2008 [Epub ahead of print]. Available at http://tinyurl.com/6kfjjj. Accessed Jun 1, 2008 18. Crow TJ, Done DJ, Sacker A. Childhood precursors of psychosis as clues to its evolutionary origins. Eur Arch Psychiatry Clin Neurosci 1995;245:61–69 19. van Os J, Jones P, Lewis G et al. Developmental precursors of affective illness in a general population birth cohort. Arch Gen Psychiatry 1997;54:625–631 20. Jone PB, Done DJ. From birth to onset: a developmental perspective of schizophrenia in two national birth cohorts. In: Murray RM and Keshavan MS, eds. Cambridge University Press; 1997;119–136 21. Cannon M, Caspi A, Moffitt TE et al. Evidence for earlychildhood, pan-developmental impairment specific to schizophreniform disorder: results from a longitudinal birth cohort. Arch Gen Psychiatry 2002;59:449–456 22. American Psychiatric Association. Diagnostic and statistical manual of mental disorders: DSM-IV-TR, 4th ed., text revision. Washington, DC, American Psychiatric Association; 2000 23. Zammit S, Allebeck P, David AS et al. A longitudinal study of premorbid IQ Score and risk of developing schizophrenia, bipolar disorder, severe depression, and other nonaffective psychoses. Arch Gen Psychiatry 2004;61:354–360 24. Reichenberg A, Weiser M, Rabinowitz J et al. A population-based cohort study of premorbid intellectual, language, and behavioral functioning in patients with schizophrenia, schizoaffective disorder, and nonpsychotic bipolar disorder. Am J Psychiatry 2002;159:2027–2035 25. Tiihonen J, Haukka J, Henriksson M et al. Premorbid intellectual functioning in bipolar disorder and schizophrenia: results from a cohort study of male conscripts. Am J Psychiatry 2005;162:1904–1910 26. Osler M, Lawlor DA, Nordentoft M. Cognitive function in childhood and early adulthood and hospital admission for schizophrenia and bipolar disorders in Danish men born in 1953. Schizophr Res 2007;92:132–141

209 27. Cohen J. Statistical power analysis for the behavioral sciences, 2nd ed. Hillsdale, NJ, Erlbaum; 1988 28. Henry C, Van den Bulke D, Bellivier F et al. Affective lability and affect intensity as core dimensions of bipolar disorders during euthymic period. Psychiatry Res 2008;159:1–6 29. Ferrier IN, Stanton BR, Kelly TP et al. Neuropsychological function in euthymic patients with bipolar disorder. Br J Psychiatry 1999;175:246–251 30. Clark L, Iversen SD, Goodwin GM. Sustained attention deficit in bipolar disorder. Br J Psychiatry 2002;180:313–319 31. Thompson JM, Gallagher P, Hughes JH et al. Neurocognitive impairment in euthymic patients with bipolar affective disorder. Br J Psychiatry 2005;186:32–40 32. Miller GA and Chapman JP. Misunderstanding analysis of covariance. J Abnorm Psychol 2001;110:40–48 33. Savitz J, Solms M, Ramesar R. Neuropsychological dysfunction in bipolar affective disorder: a critical opinion. Bipolar Disord 2005;7:216–235 34. Joffe RT, MacDonald C, Kutcher SP. Lack of differential cognitive effects of lithium and carbamazepine in bipolar affective disorder. J Clin Psychopharmacol 1988;8:425–428 35. Gualtieri CT and Johnson LG. Comparative neurocognitive effects of 5 psychotropic anticonvulsants and lithium. MedGenMed 2006;8:46–46 36. Frantom LV, Allen DN, Cross CL. Neurocognitive endophenotypes for bipolar disorder. Bipolar Disord 2008;10: 387–399 37. Tsuang MT, Farone SV. The genetics of mood disorders, Baltimore, MD, Johns Hopkins University Press; 1990 38. Rijsdijk FV and Sham PC. Analytic approaches to twin data using structural equation models. Brief Bioinform 2002;3:119–133 39. Gourovitch ML, Torrey EF, Gold JM et al. Neuropsychological performance of monozygotic twins discordant for bipolar disorder. Biol Psychiatry 1999;45:639–646 40. Zalla T, Joyce C, Szoke A et al. Executive dysfunctions as potential markers of familial vulnerability to bipolar disorder and schizophrenia. Psychiatry Res 2004;121:207–217 41. Clark L, Kempton MJ, Scarn A, et al. Sustained attentiondeficit confirmed in euthymic bipolar disorder but not in first-degree relatives of bipolar patients or euthymic unipolar depression. Biol Psychiatry 2005;57:183–187 42. Frangou S, Haldane M, Roddy D et al. Evidence for deficit in tasks of ventral, but not dorsal, prefrontal executive function as an endophenotypic marker for bipolar disorder. Biol Psychiatry 2005;58:838–839 43. McIntosh AM, Harrison LK, Forrester K et al. Neuropsychological impairments in people with schizophrenia or bipolar disorder and their unaffected relatives. Br J Psychiatry 2005;186:378–385 44. Pirkola T, Tuulio-Henriksson A, Glahn D et al. Spatial working memory function in twins with schizophrenia and bipolar disorder. Biol Psychiatry 2005;58:930–936 45. Antila M, Tuulio-Henriksson A, Kieseppa T et al. Cognitive functioning in patients with familial bipolar I disorder and their unaffected relatives. Psychol Med 2007;37:679–687 46. Szoke A, Schurhoff F, Golmard JL et al. Familial resemblance for executive functions in families of schizophrenic and bipolar patients. Psychiatry Res 2006;144:131–138

210 47. Toulopoulou T, Quraishi S, McDonald C et al. The Maudsley Family Study: premorbid and current general intellectual function levels in familial bipolar I disorder and schizophrenia. J Clin Exp Neuropsychol 2006;28:243–259 48. Kremen WS, Faraone SV, Seidman LJ et al. Neuropsychological risk indicators for schizophrenia: a preliminary study of female relatives of schizophrenic and bipolar probands. Psychiatry Res 1998;79:227–240 49. Keri S, Kelemen O, Benedek G et al. Different trait markers for schizophrenia and bipolar disorder: a neurocognitive approach. Psychol Med 2001;31:915–922 50. Sobczak S, Honig A, Schmitt JAJ et al. Pronounced cognitive deficits following an intravenous L-tryptophan challenge in first-degree relatives of bipolar patients compared to healthy controls. Neuropsychopharmacology 2003;28:711–719 51. Ferrier IN, Chowdhury R, Thompson JM et al. Neurocognitive function in unaffected first-degree relatives of patients with bipolar disorder: a preliminary report. Bipolar Disord 2004;6:319–322 52. Clark L, Sarna A, Goodwin GM. Impairment of executive function but not memory in first-degree relatives of patients with bipolar I disorder and in euthymic patients with unipolar depression. Am J Psychiatry 2005;162:1980–1982 53. Christensen MV, Kyvik KO, Kessing LV. Cognitive function in unaffected twins discordant for affective disorder. Psychol Med 2006;36:1119–1129 54. Trivedi JK, Goel D, Dhyani M et al. Neurocognition in firstdegree healthy relatives (siblings) of bipolar affective disorder patients. Psychiatry Clin Neurosci 2008;62:190–196 55. Chisholm D, Sanderson K, yuso-Mateos JL et al. Reducing the global burden of depression: Population-level analysis of intervention cost-effectiveness in 14 world regions. Brit J Psychiatry 2004;184:393–403

E. Kravariti et al. 56. Galea S, Nandi A, Vlahov D. The Social Epidemiology of Substance Use. Epidemiol Rev 2004;26:36–52 57. Benazzi F. Mood patterns and classification in bipolar disorder. Curr Opin Psychiatry 2006;19:1–8 58. Savitz J, Solms M, Ramesar R. The molecular genetics of cognition: dopamine, COMT and BDNF. Genes Brain Behav 2006;5:311–328 59. Palo OM, Antila M, Silander K et al. Association of distinct allelic haplotypes of DISC1 with psychotic and bipolar spectrum disorders and with underlying cognitive impairments. Hum Mol Genet 2007;16:2517–2528 60. Ishizuka K, Paek M, Kamiya A et al. A review of DisruptedIn-Schizophrenia-1 (DISC1): neurodevelopment, cognition, and mental conditions. Biol Psychiatry 2006;59: 1189–1197 61. Arnsten AFT. Catecholamine and second messenger influences on prefrontal cortical networks of “representational knowledge”: a rational bridge between genetics and the symptoms of mental illness. Cereb Cortex 2007;17(Suppl 1):6–15 62. Porteous DJ, Thomson P, Brandon NJ et al. The genetics and biology of DISC1–an emerging role in psychosis and cognition. Biol Psychiatry 2006;60:123–131 63. Savitz J, van der Merwe L, Solms M et al. A linkage and family-based association analysis of a potential neurocognitive endophenotype of bipolar disorder. Neuromolecular Med 2007;9:101–116 64. Flint J and Munafo MR. The endophenotype concept in psychiatric genetics. Psychol Med 2007;37:163–180 65. Craddock N, Owen MJ, O’Donovan MC. The catechol-Omethyl transferase (COMT) gene as a candidate for psychiatric phenotypes: evidence and lessons. Mol Psychiatry 2006;11:446–458

Chapter 14

Trait and State Markers of Schizophrenia in Visual Processing Yue Chen, Daniel Norton, and Ryan McBain

Abstract A shift in recent neuropsychiatric research has led to the identification and characterization of trait and state markers which are intrinsic to the inner workings of the brain, and suitable to serve as targets for detection, treatment, and prevention of psychiatric disorders. A trait marker represents the properties of the biological and behavioral processes that play an antecedent, and possibly causal, role in the pathophysiology of the disorder, whereas a state marker reflects the status of clinical manifestations in patients. The established connections among visual processing, brain function, and the abnormal behaviors in patients suggest that certain visual responses are useful candidates for such markers. A series of visual processing studies has begun to address the question of what types of visual functions can serve as trait or state markers. In the case of schizophrenia, evaluating clinically unaffected relatives of schizophrenia patients and patients with bipolar disorder can provide information on the relationship between a schizophrenic disposition and visual response traits. It has been found that motion integration is dysfunctional in schizophrenia patients but not in their relatives or in bipolar patients, whereas motion discrimination is dysfunctional in schizophrenia patients and their relatives, but not in bipolar patients. By synthesizing these findings, this chapter suggests that visual processing trait and state markers of schizophrenia can be distinguished, as illustrated in the examples of motion discrimination vs. motion integration. Identifying and distinguishing these markers is useful for developing

Y. Chen McLean Hospital, Department of Psychiatry, Harvard Medical School D. Norton and R. McBain McLean Hospital

pharmacological and behavioral interventions for patients with schizophrenia. Keywords Schizophrenic • visual functions • endophenotype • genetics • neurophysiology • behavior

Introduction The brain receives a great proportion of information about the world through vision.1 The processing of visual information provides a basis for forming not only what we see, but also what we think and how we act. Therefore, when the brain does not work properly, as in the case of schizophrenia, the question arises as to whether visual information processing is normal. Many studies have shown that visual processing is in fact altered in the disease. Over the past two decades, researchers have sought to understand whether (or which) visual impairments are at the core of schizophrenia. Since the modern characterization of schizophrenia2,3 the etiology of this psychiatric disorder has been a topic of intense inquiry. Psychosis, the clinical hallmark of schizophrenia, has traditionally been a major target of investigation for understanding both the predisposition to, and symptoms of, the disorder. This symptom-based approach, while useful in understanding and treating less complex disorders, has not yet led us to a comprehensive picture of the pathophysiological processes underlying schizophrenia. One difficulty with this approach is that psychosis may be merely one adverse effect of the illness, rather than a root problem. If so, focusing on psychosis would be a less productive approach for understanding the biological underpinnings of schizophrenia. Moreover, the clinical complexity

M.S. Ritsner (ed.), The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes, © Springer Science + Business Media B.V. 2009

211

212

associated with schizophrenia warrants a novel approach to unravel its altered biological processes. In recent years, a translational approach, which examines genetic, neurobiological, and behavioral responses concomitantly, has been applied, and has yielded significant progress. One important conceptual and empirical advance for schizophrenia research has been the distinction of trait vs. state markers. A trait is a behavioral characteristic brought about primarily by the expression of a gene or many genes. Trait markers for schizophrenia refer to the properties of the behavioral and biological processes that play an antecedent and potentially causal role in the etiology of the psychiatric disorder, whereas state markers refer to the current status of clinical manifestations in patients. Typically, though not necessarily, a trait characteristic is enduring and a state characteristic is transient. In the genetics of schizophrenia, another term, endophenotype, is often used to describe a trait that is associated with the disorder, but is not immediately visible within the clinical domain.4 The trait markers for schizophrenia we refer to here reflect a broader definition, which includes the manifestation of altered behavioral and biological processes that are linked to functional abnormalities at the core of the disorder, but not necessarily linked to psychosis. Trait markers are most useful when they are present in clinically unaffected relatives of schizophrenia patients (co-familial traits), since those individuals are spared from suffering psychosis. However, trait markers are not limited to those that co-segregate with psychosis. One challenge when applying the translational approach to the study of schizophrenia is identifying and characterizing particular altered responses that can serve as trait markers. Studying trait markers, as opposed to state markers, makes the complex disorder more tractable at genetic, neurobiological and psychological levels.5–7 This chapter describes current progress in distinguishing trait markers from state markers in schizophrenia with respect to visual processing. We highlight visual processes in particular because of their established connection with the brain and with abnormal behaviors in patients. Our selection for discussion is not intended to provide a general review of visual processing in schizophrenia. Instead, this chapter is meant to offer current answers to several specific questions: (1) How is visual processing altered in schizophrenia (integrity)? (2) Is altered processing associated with psychosis (psychosis-dependent or independent deficit)? (3) Is the altered processing observed in patients also present in

Y. Chen et al.

clinically-unaffected relatives (co-familiality)? (4) Does altered processing implicate a specific neural pathway in schizophrenia (pathophysiology)? This chapter discusses the associations between select visual processes and a predisposition to (as opposed to the effect of) schizophrenia. We will not describe in detail the experimental paradigms that produced the results that elucidate this distinction, as these methods are well outlined in the original research articles. Rather, we will discuss the genetic, neurophysiological and psychological basis of altered visual and visuomotor processes (focusing on eye tracking). We will then use visual motion processing as a model to describe the way in which trait vs. state markers can be distinguished at a perceptual level.

Visual and Visuomotor Processes Whether and how basic sensory processes are involved in schizophrenia is an active question in current research. Exploration in the past decades has shown that several vision-related processes are altered in this disorder (Table 14.1). A prominent example is dysfunction in eye tracking, an oculomotor response used to follow a moving object. Extensive studies have shown that eye tracking dysfunction occurs not only in schizophrenia patients but also in their biological relatives.8–13 It has also been demonstrated that eye tracking dysfunction is present in bipolar patients,14–16 but to a lesser extent, if at all, in relatives of bipolar patients.17 The combination of these results provides strong evidence that eye tracking dysfunction is a trait characteristic in schizophrenia.10,18,19 Genetic linkage studies suggest that eye tracking dysfunction is a susceptibility marker for schizophrenia.20,21 One challenge in further applying knowledge from eye tracking to schizophrenia research is that the biological basis of the trait is quite complex. In order to follow a moving target with the eyes, the brain systems involved need to first encode the information of visual motion, which occurs in the sensory cortex.22 The encoded motion signals are then projected to the motor and the cognitive cortical systems to generate decision signals for initiating and maintaining eye movements.23 As essential as visual motion signals are for initiating eye tracking, non-visual signals that temporally represent target movement are required in order to maintain eye tracking. The reason for storing the visual

14 Trait and State Markers of Schizophrenia in Visual Processing

213

Table 14.1 Visual and visuomotor processes in schizophrenia

Relatives of Schizophrenia Schizophrenia Patients Patients

Bipolar Patients

Relatives of Bipolar Patients

Eye Tracking

Motion Discrimination

Motion Integration

?

Visual Backward Masking

?

Contrast Detection

?

? Impaired

Spatial Interaction

?

?

??

Possibly Impaired Unimpaired

Temporal Interaction

?

information is that the external inputs of the motion signal become unavailable when eyes start to move and follow a moving target. Physiological and neurological studies have shown the involvement of multiple neural mechanisms in different aspects of eye tracking; disruption to any part of the cortical network causes eye tracking dysfunction.24–26 In schizophrenia, both visual and non-visual signals have been shown to play a role in deficient eye tracking.12,27 Multiple cortical areas have been implicated in eye tracking dysfunction in schizophrenia.28–31 The extensive exploration of eye tracking dysfunction as a trait marker for schizophrenia can serve as a useful exemplar for studying other visual processes. Another visual process that has been extensively studied is visual backward masking. A backward masking paradigm measures how detection of one target is influenced by another target that is presented shortly

?

??

?

Uninvestigated

after. Patients with schizophrenia and bipolar disorder show altered behavioral responses on this measure.32 This altered visual masking is independent of medication,33 and is accompanied by altered electrophysiological responses in schizophrenia patients.34 The altered behavioral responses also occur in clinically unaffected relatives of schizophrenia patients.35,36 These results suggest that altered visual backward masking is a trait characteristic for schizophrenia. However, visual backward masking may not be a schizophrenia-specific trait since bipolar patients are similarly affected. Another visual response that has been examined is contrast detection. Patients with schizophrenia have shown reduced sensitivities in some studies,37 but not in others.38 Several factors may contribute to these diverse results. For instance, differences in the spatial and

214

Y. Chen et al.

temporal patterns of the experimental stimuli have been shown to modulate visual contrast sensitivity.39,40 In addition, various pharmacological treatments have been shown to affect contrast detection performance. For example, one study reported significantly different contrast detection abilities for schizophrenia patients taking typical antipsychotics, atypical antipsychotics, and no antipsychotic medication.38 Understanding the effects of pharmacology on visual contrast detection will be important in determining whether contrast detection is a state or a trait marker for schizophrenia. That particular study also addressed the state vs. trait question by comparing performance of schizophrenia patients and their relatives on contrast detection, and reported that contrast detection in the relatives of schizophrenia patients was normal.38 However, the question of a contrast detection deficit as a trait or state marker for schizophrenia merits further investigation. With continued examination of these responses in relatives of schizophrenia patients and other psychiatric populations,41,42 the value of these visual functions as trait or state markers for schizophrenia will become increasingly evident. One key premise in studying visual and visuomotor functions is that certain behavioral responses could be more reflective of their underlying genetic and neurophysiological processes than other responses. In the case of schizophrenia, validating and specifying such behav-

Fig. 14.1 Visual trait versus state markers in relation to the biology of schizophrenia

ioral responses are essential for understanding alterations at the genetic and neurophysiological levels. Patients’ behavioral responses may ultimately be the test for animal-driven various pathophysiological models of schizophrenia, since psychosis, the clinical hallmark of the disorder, is extremely difficult to assess in animals. One such behavioral response that we have been studying is motion perception. There are several advantages to studying visual motion processing in relation to schizophrenia. Broadly speaking, motion perception has a well-defined neural basis within the sensory cortex. Additionally, compared with other visual processes, motion perception is more vulnerable to abnormalities in the brain.43 Thirdly, motion perception normally involves a straightforward behavioral task that does not depend substantially on cognitive processing. Thus, performance on motion perception is less contaminated by the various cognitive deficits associated with schizophrenia. Lastly, motion perception plays an important role in eye tracking, which is known to be impaired in schizophrenia. Studying motion perception may thus provide a sensitive measurement for assessing pathophysiological processes in schizophrenia. In the follow section, we will focus on using this visual response as a means for exploring trait vs. state markers for schizophrenia, after outlining its underlying genetic and neurophysiological mechanisms (Fig. 14.1).

Trait-like

State-like

Motion Integration

Motion Discrimination

Genetics

Brain Physiology

Visual Processing

Psychosis

14 Trait and State Markers of Schizophrenia in Visual Processing

Neurophysiology and Genetics of Visual Motion Processing Genetic factors play an essential role in determining the structure and function of the visual system. Before endeavoring to link visual responses and genetic factors in schizophrenia research, it is important to ask whether it is possible to study the genetics of motion processing by measuring behavioral responses. The genes essential for the development of neural pathways involved in visual processing have been screened and elicited in animal studies.44,45 Similar genetic studies have been conducted on motion processing.46 Abnormal visual behaviors have been shown in patients with heritable vision-related diseases such as visual color deficiency.47 In healthy individuals, even simple visual processes such as contrast detection appear to possess a modest heritability component.48 These studies from various visual domains suggest that examining the relationships between certain visual behaviors and their genetic origins is becoming increasingly feasible. At the neurochemical level, several types of neurotransmitters are involved in forming and shaping neural responses to visual signals. Dopamine, for example, acts to reduce responsiveness of light-sensitive neurons in the retina, and to uncouple the electric junction between these neurons, which controls neural amplification under different light conditions.49 Glutamate, through NMDA and AMPA receptors, connects neighboring neurons in the retina and cortex to provide the basis of direction selectivity, a special neural property that is fundamental for visual motion processing.50 GABA, at a later stage of the visual system, selectively suppresses some neural responses but not others, allowing more sophisticated non-linear neural computation for motion processing.51 Further research efforts are needed to understand how diverse types of neurotransmitters regulate vision-related behaviors.52 At the neurophysiological level, specific pathways for visual motion processing have been identified.53 Specifically, neural processing of motion signals begins in subcortical areas such as the retina and lateral geniculate nucleus of the thalamus and in the striate cortex. The motion signals are then transmitted to the middle temporal area (MT), an extrastriate cortical area essential for motion-specific processing. Other areas in the occipital, parietal and temporal cortices receive projections from the MT for motion-related cognitive and motor processing. These biological and behavioral results

215

suggest that applying motion perception to the study of neural functions in schizophrenia is plausible. There are two pressing challenges which must be contended with when applying motion perception to the characterization of trait vs. state markers in schizophrenia, and when studying abnormal brain functioning and genetics of the disorder. The first is to identify those visual motion functions that are altered and to examine the relationships between these functions and schizophrenic psychosis.54 The second is to characterize those altered motion responses via a behavioral-genetic analysis approach.55,56 Recent studies on motion perception in schizophrenia represent an effort in this direction.

Visual Motion Perception Studies on visual motion processing hoping to understand the nature of eye tracking dysfunction have shown that motion discrimination (Fig. 14.2) is impaired in schizophrenia patients27,57,58 and in clinically unaffected relatives,27 whereas motion integration is impaired in schizophrenia patients59–62 but not in relatives of schizophrenia patients or patients with bipolar disorder.61 Since motion discrimination and motion integration are mediated by different cortical processes, this pattern of results has several implications for understanding schizophrenia. First, the presence of altered motion discrimination in both schizophrenia patients and their relatives, but not in bipolar patients, suggests that neither psychosis nor mood disturbances significantly affect the visual process. Second, the motion discrimination deficit may be a trait characteristic that is specific to schizophrenia. Third, motion integration seems to be specifically associated with clinical schizophrenia, and thus may serve as a state marker for this disorder. In combination with neurophysiological knowledge of motion processing,63 these results suggest that motion discrimination and motion integration are indicators of different stages of the disorder – predisposition and psychosis, respectively.

Visual Responses as Trait or State Markers In addition to comparing visual responses in different populations associated with schizophrenia (e.g., patients and relatives), another method for identifying altered

216

Y. Chen et al.

Fig. 14.2 Schematic illustration of experimental paradigms used in studying visual motion processing in schizophrenia

visual responses as trait markers is to track their development and progression throughout the course of illness. With respect to development, children at high risk for schizophrenia show vision problems well before they develop any psychotic symptoms. As illustrated in an epidemiological study,64 the prevalence of visual difficulties is significantly higher in children who have a family history of schizophrenia. Furthermore, visual difficulties are present more frequently than other sensory problems (such as auditory difficulties). These epidemiological results suggest that certain deficits in visual processing are more closely linked to a schizophrenic predisposition than schizophrenic psychosis. Data on motion discrimination in children at high risk for schizophrenia are not available at present. However,

data on motion discrimination in schizophrenia patients of different ages suggest that this visual process is altered regardless of the age or stage of illness. A recent study showed that performance in motion discrimination in schizophrenia patients, while dysfunctional, is independent of age in the range of 18–55 years, whereas performance in normal controls gets worse starting at 45–55 years.65 Interestingly, an additional report found that the rate of age-related decline in other visual responses was more rapid in relatives of schizophrenia patients than in normal controls.66 This pattern of results suggests that altered motion discrimination is a stable feature over the course of the illness, furthering the case that this visual response may be used as a trait marker.

14 Trait and State Markers of Schizophrenia in Visual Processing

One relevant question concerns whether schizophrenia-related markers are modifiable by antipsychotic drugs. The distinction between trait and state markers implies that as the medications alleviate psychotic symptoms they may preferentially alter responses associated with state markers (rather than trait markers). Pharmacological agents such as antipsychotic drugs could in principle modulate any behavioral response, including those associated with trait markers of schizophrenia. Modulation of trait markers by the drugs, however, could be independent of changes in psychotic symptoms or associated state markers. For example, antipsychotic medications are known to have differential effects on abnormal behaviors of schizophrenia patients – they are more effective in treating positive symptoms than cognitive dysfunction or negative symptoms.67 How antipsychotic medications affect visual responses (such as trait markers) remains a largely unexplored question.38 One useful strategy in future studies may be to compare performance on a certain visual task before and after patients are treated by antipsychotic medications. This strategy, however, is complicated by the fact that a majority of patients are now treated with antipsychotics. However, many antipsychotic drugs, especially those of the new generation (such as clozapine) have a relatively short occupancy period. Such temporal dynamics suggests windows of opportunity during which the effects of antipsychotic drugs may be minimal or maximal – e.g., immediately before or after dose administration. Comparisons on performance before and after the drugs are administered would provide a practical way to assess the drug effect in medicated patients. Another strategy would be to examine performance on a visual task in both patient and relative groups. If a trait marker manifests in both groups, comparing associated visual performance of the patients who are medicated with the relatives who are not would provide a useful index on the effects of the drugs. Another important pharmacological issue is how the relationships among visual responses and neurochemical underpinnings inform neurochemical theories of schizophrenia. While current theories implicate several types of neurotransmitters such as dopamine, glutamate and GABA, empirical evidence is largely generated from neurochemical studies using either postmortem schizophrenic brains or animal models. As discussed earlier, these very same neurotransmitters also modulate visual responses. This heuristic link provides a basis for using visual perception as a means

217

to test these theories directly in schizophrenia patients. One practical example of this is found in recent studies which have shown hypersensitivity on visual contrast detection for unmedicated schizophrenia patients, and normal-to-hyposensitivity for patients treated with antipsychotic drugs (D2 receptor blockade).38,68 This relation is consistent with the theory that abnormally high dopamine activity in the subcortical system may be associated with schizophrenia, and that antipsychotic drugs act to reduce this activity.69 Another example is the abnormal performance of patients during the visual tasks mediated by center-surround interaction,70–72 a ubiquitous neural mechanism that is modulated by GABAergic activity. Much progress has been made towards understanding schizophrenia at a biological level via postmortem73 and genetic studies.56 It is important to synthesize such research with behavioral studies. With advancements in both basic and clinical research, the application of visual responses (e.g., motion discrimination and integration, which have been extensively investigated psychophysically and physiologically) in schizophrenia research is becoming an increasingly powerful approach for probing into the underlying pathophysiological processes in this disorder. Using the methods described above, systematic exploration of visual responses will also be useful for understanding other sensory74–76 and related cognitive processes77–80 implicated in schizophrenia. Identification and characterization of deficient visual responses and their neural correlates in schizophrenia may ultimately provide important clues for developing intervention strategies used to help patients. Understanding how neuromodulation affects deficient visual processing and its neural correlates in schizophrenia will be useful for identifying new and better targets for pharmacological treatment. Measuring pharmacological effects on behavioral performance is already being done with sensory gating, another behavioral process that is compromised in schizophrenia. Extensive research into the neuromodulation of the sensory gating deficit has informed the search for pharmacological interventions that target the alpha-7 nicotinic receptor system (see Martin et al.81 for a review). In addition, characterization of visual responses and their neural correlations may be useful in another, more novel way: Basic research has shown that perceptual and cognitive capacities are plastic and adaptive not only during child development but also throughout adulthood.82–84

218

In healthy people, visual function and cortical modulation can be enhanced through perceptual learning.84 In light of such knowledge, the identification of visionrelated trait markers could provide concrete targets for special behavioral training, analogous to cognitive rehabilitation, to improve patients’ performance starting at the sensory level.

Conclusions and Future Directions It is being increasingly recognized that schizophrenia is a pleiotropic disorder. The likely involvement of many genes and environmental factors makes it difficult to draw an outright picture of the disorder’s underlying biological and behavioral mechanisms. With advances in basic research and the implementation of a new translational research strategy, we can move towards a genetically- and neurobiologically-based approach that focuses on those traits influenced by well-understood brain physiology and accessible genetic pathways. One important step in taking this novel approach is to differentiate and characterize trait and state markers of schizophrenia. Psychotic symptoms alone are not adequate to determine the genetic and neural basis of schizophrenia. Study of selected visual processes provides an opportunity to identify genetic variations within a range of visual phenotypes that are specific to schizophrenia. The results so far have pointed towards motion discrimination as a trait characteristic that may be useful for screening schizophrenia-related genes and for exploring neural mechanisms underlying perceptual deficits associated with schizophrenia. Of foremost significance is the role that motion discrimination may play in linking genetics, neurophysiology and perceptual performance in schizophrenia. Future studies should underscore whether and to what extent the enduring trait markers of schizophrenia are modifiable and how they are related to state markers of schizophrenia. This information will help define phenotypes more precisely for use in genetic, physiological and psychological studies. Knowledge acquired from these studies will eventually be applied to unravel the pathophysiology at the core of schizophrenia, and to inform prevention and intervention strategies. Acknowledgement The work was supported in part by UHS Grant MH 61824 and a faculty pilot research award (Harvard University).

Y. Chen et al.

References 1. Van Essen D, Drury H. Structural and functional analyses of human cerebral cortex using a surface-based atlas. J Neurosci 1997;17:7079–7102 2. Bleuler E. Dementia Praecox or the Group of Schizophrenias. New York: International Universities Press; 1911 3. Kraepelin E. Dementia Praecox and Paraphrenia. Chicago, IL: Medical Books; 1919 4. Gottesman I, Shields J. Genetic theorizing and schizophrenia. Br J Psychiatry 1973;122:15–30 5. Holzman PS. Behavioral markers of schizophrenia useful for genetic studies. J Psychiatr Res 1992;26:427–454 6. Tamminga CA, Holcomb HH. Phenotype of schizophrenia: a review and formulation. Mol Psychiatry 2005 Jan;10:27–39 7. Saccuzzo DP, Braff DL. Information-processing abnormalities: trait- and state-dependent components. Schizophr Bull 1986;12:447–459 8. Holzman PS, Proctor LR, Hughes DW. Eye-tracking patterns in schizophrenia. Science 1973;181:179–181 9. Levin S, Luebke A, Zee DS, et al. Smooth pursuit eye movements in schizophrenics: quantitative measurements with the search-coil technique. J Psychiatr Res 1988;22:195–206 10. Clementz BA, Grove WM, Iacono WG, et al. Smooth-pursuit eye movement dysfunction and liability for schizophrenia: implications for genetic modeling. J Abnorm Psychol 1992;101:117–129 11. Sweeney JA, Clementz BA, Escobar MD, et al. Mixture analysis of pursuit eye-tracking dysf unction in schizophrenia. Biol Psychiatry 1993;34:331–340 12. Thaker GK, Ross DE, Cassady SL, et al. Smooth pursuit eye movements to extraretinal motion signals: deficits in relatives of patients with schizophrenia. Arch Gen Psychiatry 1998;55:830–836 13. Holzman PS, Proctor LR, Levy DL, et al. Eye-tracking dysfunctions in schizophrenic patients and their relatives. Arch Gen Psychiatry 1974;31:143–151 14. Shagass C, Amadeo M, Overton DA. Eye-tracking performance in psychiatric patients. Biol Psychiatry 1974;9:245–260 15. Levin S, Holzman PS, Rothenberg SJ, et al. Saccadic eye movements in psychotic patients. Psychiatry Res 1981;5:47–58 16. Iacono WG, Peloquin LJ, Lumry AE, et al. Eye tracking in patients with unipolar and bipolar affective disorders in remission. J Abnorm Psychol 1982;91:35–44 17. Levy DL, Yasillo NJ, Dorus E, et al. Relatives of unipolar and bipolar patients have normal pursuit. Psychiatry Res 1983;10:285–293 18. Holzman P, Kringlen E, Matthysse S, et al. A single dominant gene can account for eye tracking dysfunctions and schizophrenia in offspring of discordant twins. Arch Gen Psychiatry 1988;45:641–647 19. Thaker GK, Avila MT, Hong EL, et al. A model of smooth pursuit eye movement deficit associated with the schizophrenia phenotype. Psychophysiology 2003;40:277–284 20. Arolt V, Lencer R, Nolte A, et al. Eye tracking dysfunction is a putative phenotypic susceptibility marker of schizophrenia and maps to a locus on chromosome 6p in families with multiple occurrence of the disease. Am J Med Genet 1996;67:564–579

14 Trait and State Markers of Schizophrenia in Visual Processing 21. Matthysse S, Holzman PS, Gusella JF, et al. Linkage of eye movement dysfunction to chromosome 6p in schizophrenia: additional evidence. Am J Med Genet B Neuropsychiatr Genet 2004;128:30–36 22. Newsome WT, Wurtz RH, Komatsu H. Relation of cortical areas MT and MST to pursuit eye movements. II. Differentiation of retinal from extraretinal inputs. J Neurophysiol 1988;60:604–620 23. Lisberger SG, Morris EJ, Tychsen L. Visual motion processing and sensory-motor integration for smooth pursuit eye movements. Annu Rev Neurosci 1987;10:97–129 24. Heide W, Kurzidim K, Kompf D. Deficits of smooth pursuit eye movements after frontal and parietal lesions. Brain 1996;119:1951–1969 25. Dursteler MR, Wurtz RH. Pursuit and optokinetic deficits following chemical lesions of cortical areas MT and MST. J Neurophysiol 1988;60:940–965 26. Morrow MJ, Sharpe JA. Retinotopic and directional deficits of smooth pursuit initiation after posterior cerebral hemispheric lesions. Neurology 1993;43:595–603 27. Chen Y, Nakayama K, Levy DL, et al. Psychophysical isolation of a motion-processing deficit in schizophrenics and their relatives and its association with impaired smooth pursuit. Proc Natl Acad Sci USA 1999;96:4724–4729 28. O’Driscoll GA, Benkelfat C, Florencio PS, et al. Neural correlates of eye tracking deficits in first-degree relatives of schizophrenic patients: a positron emission tomography study. Arch Gen Psychiatry 1999;56:1127–1134 29. Tregellas JR, Tanabe JL, Miller DE, et al. Neurobiology of smooth pursuit eye movement deficits in schizophrenia: an fMRI study. Am J Psychiatry 2004;161:315–321 30. Hong LE, Tagamets M, Avila M, et al. Specific motion processing pathway deficit during eye tracking in schizophrenia: a performance-matched functional magnetic resonance imaging study. Biol Psychiatry 2005;57:726–732 31. Lencer R, Nagel M, Sprenger A, et al. Reduced neuronal activity in the V5 complex underlies smooth-pursuit deficit in schizophrenia: evidence from an fMRI study. Neuroimage 2005;24:1256–1259 32. Green MF, Nuechterlein KH, Mintz J. Backward masking in schizophrenia and mania. I. Specifying a mechanism. Arch Gen Psychiatry 1994;51:939–944 33. Butler PD, Harkavy-Friedman JM, Amador XF, et al. Backward masking in schizophrenia: relationship to medication status, neuropsychological functioning, and dopamine metabolism. Biol Psychiatry 1996;40:295–298 34. Schechter I, Butler PD, Silipo G, et al. Magnocellular and parvocellular contributions to backward masking dysfunction in schizophrenia. Schizophr Res 2003;64:91–101 35. Green MF, Nuechterlein KH, Breitmeyer B, Mintz J. Forward and backward visual masking in unaffected siblings of schizophrenic patients. Biol Psychiatry 2006; 59:446–451 36. Green MF, Nuechterlein KH, Breitmeyer B. Backward masking performance in unaffected siblings of schizophrenic patients. Evidence for a vulnerability indicator. Arch Gen Psychiatry 1997;54:465–472 37. Slaghuis WL. Contrast sensitivity for stationary and drifting spatial frequency gratings in positive- and negative-symptom schizophrenia. J Abnorm Psychol 1998; 107:49–62

219 38. Chen Y, Levy DL, Sheremata S, et al. Effects of typical, atypical, and no antipsychotic drugs on visual contrast detection in schizophrenia. Am J Psychiatry 2003; 160:1795–801 39. Kelly DH. Motion and vision. II. Stabilized spatio-temporal threshold surface. J Opt Soc Am 1979;69:1340–1349 40. Watson AB, Barlow HB, Robson JG. What does the eye see best? Nature 1983;302:419–422 41. Keri S, Kelemen O, Benedek G, et al. Vernier threshold in patients with schizophrenia and in their unaffected siblings. Neuropsychology 2004;18:537–542 42. Keri S, Kelemen O, Benedek G, et al. Different trait markers for schizophrenia and bipolar disorder: a neurocognitive approach. Psychol Med 2001;31:915–922 43. Braddick O, Atkinson J, Wattam-Bell J. Normal and anomalous development of visual motion processing: motion coherence and ‘dorsal-stream vulnerability’. Neuropsychologia 2003;41:1769–1784 44. Li L, Dowling JE. A dominant form of inherited retinal degeneration caused by a non-photoreceptor cell-specific mutation. Proc Natl Acad Sci USA 1997;94:11645–11650 45. Neuhauss SC. Behavioral genetic approaches to visual system development and function in zebrafish. J Neurobiol 2003;54:148–160 46. Baier H. Zebrafish on the move: towards a behavior-genetic analysis of vertebrate vision. Curr Opin Neurobiol 2000;10:451–455 47. Tovee MJ. The molecular genetics and evolution of primate colour vision. Trends Neurosci1 994;17:30–37 48. Cronin-Golomb A, Panizzon MS, Lyons MJ, et al. Genetic influence on contrast sensitivity in middle-aged male twins. Vision Res 2007;47:2179–2186 49. Dowling JE. Functional and pharmacological organization of the retina: dopamine, interplexiform cells, and neuromodulation. Res Publ Assoc Res Nerv Ment Dis 1990; 67:1–18 50. Rivadulla C, Sharma J, Sur M. Specific roles of NMDA and AMPA receptors in direction-selective and spatial phaseselective responses in visual cortex. J Neurosci 2001; 21:1710–1719 51. Egelhaaf M, Borst A, Pilz B. The role of GABA in detecting visual motion. Brain Res 1990;509:156–160 52. Carter OL, Pettigrew JD, Hasler F, et al. Modulating the rate and rhythmicity of perceptual rivalry alternations with the mixed 5-HT2A and 5-HT1A agonist psilocybin. Neuropsychopharmacology 2005;30:1154–1162 53. Maunsell JHR, Newsome WT. Visual processing in monkey extrastriate cortex. Ann Rev Neurosci 1987;10:363–401 54. Andreasen NC. Linking mind and brain in the study of mental illnesses: a project for a scientific psychopathology. Science 1997;275:1586–1593 55. Gottesman I, Shields J. Schizophrenia: The Eigenetic Pzzle. Cambridge: Cambridge University Press; 1982 56. Owen MJ, Williams NM, O’Donovan MC. The molecular genetics of schizophrenia: new findings promise new insights. Mol Psychiatry 2004;9:14–27 57. Kim D, Wylie G, Pasternak R, et al. Magnocellular contributions to impaired motion processing in schizophrenia. Schizophr Res 2006;82:1–8 58. Clementz BA, McDowell JE, Dobkins KR. Compromised speed discrimination among schizophrenia patients when

220 viewing smooth pursuit targets. Schizophr Res 2007;95:61–44 59. Stuve T, Friedman L, Jesberger JA, et al. The relationship between smooth pursuit performance, motion perception and sustained visual attention in patients with schizophrenia and normal controls. Psychol Med 1997;27:143–152 60. Li CS. Impaired detection of visual motion in schizophrenia patients. Prog Neuropsychopharmacol and Biol Psychiatry 2002;26:929–934 61. Chen Y, Bidwell LC, Holzman PS. Visual motion integration in schizophrenia patients, their first-degree relatives, and patients with bipolar disorder. Schizophr Res 2005; 74:271–281 62. Chen Y, Nakayama K, Levy DL, et al. Processing of global, but not local, motion direction is deficient in schizophrenia. Schizophr Res 2003;61:215–227 63. Wurtz R, Yamasaki D, Duffy C, et al. Functional specialization for visual motion processing in primate cerebral cortex. Cold Spring Harbor Symposia on Quantum Biology 1990;55:717–27 64. Henriksson KM, McNeil TF. Health and development in the first 4 years of life in offspring of women with schizophrenia and affective psychoses: Well-Baby Clinic information. Schizophr Res 2004;70:39–48 65. Bidwell LC, Holzman PS, Chen Y. Aging and visual motion discrimination in normal adults and schizophrenia patients. Psychiatry Res 2006;145:1–8 66. Bedwell JS, Esposito S, Miller LS. Accelerated age-related decline of visual information processing in first-degree relatives of persons with schizophrenia. Psychiatry Res 2004;125:225–235 67. Kopala LC, Fredrikson D, Good KP, et al. Symptoms in neuroleptic-naive, first-episode schizophrenia: response to risperidone. Biol Psychiatry 1996;39:296–298 68. Keri S, Antal A, Szekeres G, et al. Transient visual channel functions in schizophrenia. Int J Psychophysiol 1998;30:170 69. Davis KL, Kahn RS, Ko G, et al. Dopamine in schizophrenia: a review and reconceptualization. Am J Psychiatry 1991;148:1474–1486 70. Tadin D, Kim J, Doop M, et al. Weakened center-surround interactions in visual motion processing in schizophrenia. J Neurosci 2006;26:11403–11412

Y. Chen et al. 71. Dakin S, Carlin P, Hemsley D. Weak suppression of visual context in chronic schizophrenia. Curr Biol 2005; 15:R822–824 72. Chen Y, Norton D, Ongur D. Altered center-surround motion inhibition in schizophrenia. Biol Psychiatry 2008;17:17 73. Benes FM. Emerging principles of altered neural circuitry in schizophrenia. Brain Res Brain Res Rev 2000;31:251–269 74. Saccuzzo DP, Braff DL. Early information processing deficit in schizophrenia. New findings using schizophrenic subgroups and manic control subjects. Arch Gen Psychiatry 1981;38:175–179 75. Freedman R, Adler LE, Gerhardt GA, et al. Neurobiological studies of sensory gating in schizophrenia. Schizophr Bull 1987;13:669–678 76. Javitt DC, Shelley AM, Silipo G, Lieberman JA. Deficits in auditory and visual context-dependent processing in schizophrenia: defining the pattern. Arch Gen Psychiatry 2000;57:1131–1137 77. Park S, Holzman PS, Goldman-Rakic PS. Spatial working memory deficits in the relatives of schizophrenic patients. Arch Gen Psychiatry 1995;52:821–828 78. Silverstein SM, Kovacs, I., Corry, et al. Perceptual organization, the disorganization syndrome, and context processing in chronic schizophrenia. Schizophr Res 2000;43:11–20 79. Kim J, Doop ML, Blake R, et al. Impaired visual recognition of biological motion in schizophrenia. Schizophr Res 2005;77:299–300 80. Kelemen O, Erdelyi R, Pataki I, et al. Theory of mind and motion perception in schizophrenia. Neuropsychology 2005;19:494–500 81. Martin LF, Kem WR, Freedman R. Alpha-7 nicotinic receptor agonists: potential new candidates for the treatment of schizophrenia. Psychopharmacology 2004;174:54–64 82. Levine SM. Adult visual system plasticity. J Am Optom Assoc 988;59:135–139 83. Gilbert CD. Plasticity in visual perception and physiology. Current Opinion in Neurobiology 1996;6:269–274 84. Schwartz S, Maquet P, Frith C. Neural correlates of perceptual learning: a functional MRI study of visual texture discrimination. Proc Natl Acad Sci USA 2002;99:17137–17142

Chapter 15

Visual Scanning Abnormalities as Biomarker for Schizophrenia Patricia E.G. Bestelmeyer

Abstract Visual scan paths are a pattern of foveal fixations and voluntary saccades produced when an individual views an image. These scan paths have repeatedly been shown to be abnormal in patients with schizophrenia. Most clinical scan path studies have focused on the investigation of eye movements to images of social scenes, or most commonly, of faces. While healthy participants scan face images extensively, particularly the eyes and mouth, patients with schizophrenia demonstrate restricted scanning of these images. In other words, patients with schizophrenia exhibit fewer fixations, longer fixation duration, longer saccade duration and peak velocity, and smaller saccade amplitude compared with healthy controls. However, recent evidence has shown that this abnormality is not specific to faces but can be observed to other image types. In addition, it has been shown that the spatial distribution of fixation locations is different in patients with schizophrenia compared to healthy controls and that this abnormal scanning behaviour extends to patients with bipolar disorder. A small study on healthy twins has provided preliminary evidence that the distribution of fixations may be genetically determined and may serve as a biomarker for schizophrenia and bipolar disorder. Keywords Visual scan paths • trait marker • eye movements • schizophrenia • bipolar disorder

P. E. G. Bestelmeyer () Department of Psychology, Centre for Cognitive Neuroimaging, University of Glasgow

Eye Movement Recording Background to Eye Movement Recording Ganglion cells and photoreceptors in the eye are most densely packed at the centre of the retina, the fovea. These cells allow for high visual acuity and colour vision, both of which decrease rapidly and continuously with visual eccentricity. A visual fixation is a focusing of the fovea at the same position, generally for 100 ms or more,1 and provides the visual system with detailed information about an inspected stimulus. Even though the general meaning of a scene can be quickly identified within a single fixation,2 the acquisition of more detailed information depends on the serial movement of the eyes by means of saccades.3 Saccades direct the fovea from one region of interest to the next but the processing of visual information is suppressed during these movements.4 Three to four times per second a viewer selects and fixates a particular region for detailed visual processing.5,6 The pattern of foveal fixations and voluntary saccades produced during the exploration of a scene make up the visual scan path.7 Attention is usually paid to the area of fixation due to highest visual acuity at this site. It is generally assumed in the scan path literature that scan paths reflect initial sensory input and directed visual attention.8 Scan path parameters are most commonly measured using video-oculography. This non-invasive technique records eye movements by means of an infrared light source, which is reflected from the pupil, and the video recording of these reflections. Generally, scan path variables of interest can be split into two main categories, temporal and spatial variables. Temporal variables, for instance, are the number of fixations, fixation duration, number of saccades, saccade duration and

M.S. Ritsner (ed.), The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes, © Springer Science + Business Media B.V. 2009

221

222

saccade peak velocity. Spatial variables are the fixation distributions in a particular image and provide information about which regions of an image the individual finds interesting and therefore selects for fixation. Differences between two or more scan path distributions can be assessed to determine whether a particular group significantly differs from another.

Scan Path Paradigms and Analysis Typically, scan path experiments on clinical populations involve the presentation of images (e.g. pictures of faces or social scenes) for a few seconds. During every trial the scan paths for each stimulus are recorded. Measured variables are then averaged for every particular stimulus category and compared to scan paths of healthy control participants. Most studies have concentrated on the analysis of temporal variables. The analysis of spatial scan path parameters (i.e. fixation locations) has proven difficult. More recent studies, however, tend to also examine the difference between fixation locations of clinical samples compared to healthy viewers.9 In this analysis the images, typically images of faces, are divided up into predefined regions of interest such as feature areas (e.g. eyes, mouth, nose) and non-feature areas (e.g. hair, cheeks). The number of times a fixation falls into one of these predefined regions is then counted and compared to that of other regions and other experimental groups.10 This method is usually limited to the analysis of images of objects that consist of the same features (e.g. the eyes, nose and mouth in images of faces) and the definition of these regions of interest varies between studies. Studies investigating the scanning behaviour to other types of images (e.g. geometric figures) have concentrated on variables that describe the temporal characteristics of scan paths rather than their spatial extent. A more flexible analysis tool for fixation location and measure to determine differences between scan paths at a group level has been lacking. The analysis of dissimilar images, such as landscape pictures each of which contains several different (subjective) regions of interest (e.g. a picture of mountains another image of a forest or seaside), has only been attempted recently.11,12

P.E.G. Bestelmeyer

Other current approaches (used in scan path studies on non-clinical populations) for assessing the spatial similarity between scan paths take the form of nearest saccade metrics, often based upon the sum of squared differences between fixations (e.g. refs.13,14). The problem with this approach is that there is no logical reason to suggest that fixations 20° apart are twice as different as those 10° apart (or four times if using a squared difference). The technique developed by Tatler et al.15 avoids this problem and simply classes fixations within the radius of the fovea as the same, and those beyond the radius of the fovea as different. Tatler et al.’s15 technique involves estimating spatial probability distributions of the fixation locations and uses a technique known as the Kullback-Leibler divergence to estimate the difference between these distributions (for a full discussion of the benefits of this metric see ref.15).

Genetic Influence On Eye Movements There are several criteria a trait marker needs to fulfil.16 First, a trait marker must be state independent and it must be related to the transmission of the illness (i.e. the characteristic must be found in probands as well as their unaffected relatives at a higher rate than in the general population). Further, a trait marker must be under genetic control and it must be associated with an increased likelihood of the illness so that the abnormal behaviour is more often present in the affected individuals than the healthy population. Streit et al.17 tested whether the visual scanning behaviour is a state or trait abnormality. First, the researchers tested patients with acute schizophrenia showing abnormal visual scanning of faces compared to healthy controls. Four weeks later the patients were assessed again and found to be in a partially remitted state. Repeated testing after this period of time revealed the same eye movement abnormalities. Since this disturbance seems to be stable over time, Streit et al.17 suggested that abnormal visual scanning of face images represents a trait rather than a state marker of the illness. Loughland et al.18 analysed the scan paths of healthy control participants, schizophrenia patients and their first degree relatives to assess whether scan paths may represent a genetic trait. Generally, the scan path parameters of relatives resembled those of healthy controls

15

Visual Scanning Abnormalities as Biomarker for Schizophrenia

more than those of patients with schizophrenia so that scan paths of relatives were generally not significantly different from those of healthy controls but were significantly less restricted than those of patients with schizophrenia. However, the median fixation duration and raw scan path length of the relatives fell in between those of patients with schizophrenia and healthy controls. Loughland et al.18 therefore published the first study which hinted that relatives of patients with schizophrenia show some degree of abnormal scanning behaviour and concluded that scan path behaviour may be a vulnerability marker for schizophrenia. Bestelmeyer et al.12 recently investigated the genetic influence on visual scanning behaviour by comparing the temporal and spatial aspects of scan paths made by monozygotic twins, dizygotic twins and unrelated control pairs when viewing a variety of complex scenes (landscapes, everyday scenes, fractals and pink noise patterns). Bestelmeyer et al.12 examined the similarity in temporal scanning parameters as well as the participant pairs’ similarity in the spatial distributions of fixations when viewing the scenes using an information theoretic measure based on the Kullback-Leibler divergence.15 The data suggest that certain aspects of eye movement behaviour when viewing complex scenes may be influenced by the underlying genetics of the observer. Specifically, Bestelmeyer et al.12 found that the locations chosen for fixation in the scenes were more similar for monozygotic twins than for dizygotic twins or unrelated controls. No previous studies have indicated a genetic basis for the selection of spatial locations for visual fixation. The results of this study point toward the suitability of eye movement behaviour as a trait marker for psychiatric disorders.

Visual Scan Paths in Schizophrenia To date, clinical scan path studies have focused on investigations of eye movements to images of social scenes or, most commonly, of faces. Typically, healthy participants tend to fixate salient features of a face, such as the eyes, nose or mouth.19 Gordon et al.20 were the first to investigate the scan paths of schizophrenia patients. The patients were found to make fewer visual fixations and tended to focus less on salient facial features compared to healthy controls. Restricted visual scan paths to faces and social scenes have been con-

223

firmed repeatedly9,21–25 and have been used to explain, at least in part, the social disturbances most individuals with schizophrenia experience.25,26 Williams et al.26 investigated visual scan paths of schizophrenia patients to four stimuli of faces; one stimulus was an image of a non-degraded face (i.e. recognisable) and the other one was an image of a degraded face (i.e. facial identity was not recognisable due to the use of a mosaic filter). Each non-degraded and degraded face was presented twice. Participants were asked to view these two different types of face stimuli to demonstrate distinct scan path aberrations to the non-degraded faces but not to the non-recognisable, degraded faces. Following each presentation, the ability to recognise the identity of the displayed face was assessed in two different multiple choice conditions. For the first exposure participants had to choose the correct face from among seven similar faces and for the second exposure participants were asked to pick the correct face from an array of three faces. Williams et al.26 found that patients with schizophrenia as well as healthy control participants could not recognise degraded faces but schizophrenia patients were significantly less accurate than healthy controls in recognising non-degraded faces in the seven response condition. However, the performance of schizophrenia patients significantly improved for the three option condition. Replicating Gordon et al.,20 Williams et al.26 found that schizophrenia patients fixated equally as much on nonfeature (e.g. cheeks) as on feature (e.g. eyes) areas which may have prevented the processing of fine details accounting for the difficulty in recognising non-degraded faces in the difficult condition. Yet, this limited processing strategy may suffice for a multiple choice task with fewer options.26 A problem with the design of Williams et al.’s26 study is that stimuli but not task conditions were randomly administered so that the easier task was conducted after the more difficult one with the same pictures being displayed in both tasks. The design of carrying out the easier task after the more difficult one may have led to a practice effect in this finding. The scan paths of schizophrenia patients and healthy control participants were found to be similar for degraded faces but not for non-degraded faces. This result led Williams et al.26 to argue that schizophrenia patients have a specific deficit in the scanning of faces which underlies their social difficulties. Since the non-degraded face stimulus looked unnatural and

224

because this non-degraded face stimulus was still recognisable as a face Williams et al.’s26 study does not confirm a face specific scanning abnormality in schizophrenia. In contrast to Williams et al.,26 Phillips and colleagues9,24 have demonstrated that schizophrenia patients are able to recognise neutral faces in a forced-choice recognition task. However, patients with schizophrenia may be unable to correctly recognise emotional expressions. In other words, schizophrenia patients’ abnormal information gathering techniques might be sufficient to recognise faces but their techniques might not allow them to successfully delineate more complex and detailed facial information like emotional expressions. Studies have examined whether visual scanning of images without any social content is impaired in schizophrenia. The results are equivocal. For example, Kojima et al.27 explored scan path behaviour to geometric figures and found that patients with schizophrenia as compared to healthy controls exhibited fewer visual fixations (see also Ryu et al.28 for a similar study and Green et al.29 for a comprehensive review of studies using mainly geometric figures). Similarly, Minassian et al.30 demonstrated that patients with schizophrenia displayed fewer fixations of longer duration to abstract Rorschach stimuli compared to healthy controls. Minassian et al.30 suggested that patients with schizophrenia have a general impairment with regards to visual scanning. However, in contrast to images of faces, Rorschach stimuli and line drawings seem neither aesthetically interesting nor biologically important. Therefore increased staring behaviour to these abstract pictures may have occurred due to a lack of motivation or disinterest of the patients in the displayed images rather than an actual visual scanning impairment. A selection of image categories of equivalent complexity and/or interest to that of a face may be crucial to determine whether these scanning impairments are global or confined to socially salient stimuli. Only few studies have investigated the specificity of restricted scan paths in patients with schizophrenia to socially salient stimuli. Manor et al.31 directly compared the scan paths of schizophrenia patients and healthy control participants to an image of a face and an image of a line-drawing (the Rey complex figure), i.e. a non-socially salient stimulus. Manor et al.31 concluded that in comparison to controls, individuals with schizophrenia have dysfunctional scanning of faces, leading to social deficits. Inspection of their results

P.E.G. Bestelmeyer

reveals that overall length of the scan path in response to face images was significantly shorter for patients with schizophrenia as compared to healthy controls. However, there was no statistically significant difference in the number of fixations or fixation duration between the two groups for faces or the Rey figure. In addition, qualitative differences between the group of patients with schizophrenia and the control group were evident to the Rey figure as well as the face image. Other explanations for the overall shorter scan path length to the face image are possible. The novelty of the Rey figure and the fact that this figure was always presented first may have caused patients with schizophrenia to scan more extensively than they would usually do if several of these two categories of stimuli had been presented in a random order. A more serious confound is that the Rey figure subtended a larger visual angle than the face stimulus, which may well explain the greater scan path length for the Rey figure compared to the face stimulus. These methodological confounds cast doubts on Manor et al.’s31 conclusion and the accepted view in the literature that restricted visual scanning in schizophrenia is specific to faces, which has been used to partly explain patients’ difficulties in social functioning.23 A thorough investigation of the nature and extent of the scanning impairment in schizophrenia in response to complex social and non-social stimuli is therefore warranted. For scan paths to serve as a useful diagnostic tool for schizophrenia, scanning abnormalities should be present in most, if not all affected individuals. In addition, this abnormality should be specific to schizophrenia and should not be observed in patient groups suffering from other kinds of functional psychoses. An interesting study by Loughland et al.23 has addressed the issue of specificity of visual scanning impairments in schizophrenia by comparing scan paths of patients with schizophrenia to those of a mixed group of patients with affective disorder (i.e. patients with unipolar depression and bipolar disorder). Loughland et al.23 found that the temporal variables of affective disorder patients fell between those of healthy controls and patients with schizophrenia and were significantly different only from those of the participants with schizophrenia but not healthy controls. Loughland et al.23 concluded that the scan path aberrations of affective disorder patients were relatively minor and that more serious scan path abnormalities are specific to schizophrenia.

15

Visual Scanning Abnormalities as Biomarker for Schizophrenia

There seems consensus that schizophrenia patients display aberrant scan path behaviour and reduced attention to salient facial features leading to poor emotion recognition performance.17,22 Most researchers have suggested that this dysfunction in processing of detailed information (fewer fixations and therefore less attention to salient facial features) underlies schizophrenia patients’ disturbed social interactions with others. Unclear in this particular literature is whether impaired viewing behaviour in schizophrenia is specific to images of socially salient stimuli (e.g. faces) or whether this scanning abnormality can be measured to other types of images (e.g. other biologically relevant stimuli). Further, it is uncertain whether this abnormal scanning behaviour is specific to schizophrenia or can be observed in patients with other types of functional psychosis. A recent study addressed both aforementioned issues by investigating the specificity of the visual scanning abnormality to schizophrenia and the speci-

225

ficity of this abnormality to socially salient stimuli.11 Bestelmeyer et al.11 compared the scan path parameters of patients with schizophrenia to patients with bipolar disorder. Scan paths were recorded in response to faces as well as other biologically or aesthetically interesting images (e.g. landscapes and fractal patterns). Patients with schizophrenia and bipolar disorder showed restricted scan paths to all types of images suggesting that this scanning abnormality is neither specific to schizophrenia nor is it specific to stimuli with social content (see Fig. 15.1 for an illustration). Temporal scan path variables were therefore not suitable to distinguish between patients with schizophrenia and bipolar disorder. However, the analysis of which regions patients with schizophrenia and bipolar disorder chose for fixation differed between the two illnesses (and healthy controls). Given that fixation location seems to be genetically influenced fixation location may be a trait marker specific to schizophrenia.

Healthy control

Patient with schizophrenia

Patient with bipolar disorder

Fig. 15.1 Illustrations of representative scan paths of a healthy control, patient with schizophrenia and bipolar disorder. Circles represent fixations; dashed lines represent saccades. The size of the circles is directly proportional to the duration of the fixation (example data plotted from Bestelmeyer et al.11).

226

Future Directions Although it has been known for several decades that patients with schizophrenia display abnormal scanning of images it is unclear what causes this restricted scan path behaviour. It is possible that patients with functional psychosis perseverate on a small area of an image. Thus restricted scanning could be linked with abnormal frontal function and could be tested by correlating performance scores with fixation duration or saccade amplitude. A recent study has shown that patients with psychosis can improve on facial emotion recognition.32 Here patients were actively trained to explore the image and pay attention to certain feature areas of faces. As a next step it would be interesting to investigate how long this knowledge is retained and whether patients perceive this knowledge as helpful when interacting with others.

References 1. Manor BR, Gordon E. Defining the temporal threshold for ocular fixation in free- viewing visuocognitive tasks. J Neurosc Methods 2003;128:85–93. 2. Schyns PG, Olivia A. From blobs to boundary edges: Evidence for time– and spatial-scale-dependent scene recognition. Psychol Sci 1994;5:195–200. 3. Hollingworth A, Henderson JM. Accurate visual memory for previously attended objects and natural scenes. Journal of Experimental Psychology: Human Perception and Performance 2002;28:113–136. 4. Matin E. Saccadic suppression: A review and an analysis. Psychol Bull 1974;81:899–917. 5. Buswell GT. How people look at pictures: A study of the psychology of perception in art. Chicago: University of Chicago Press, 1935. 6. Yarbus AL. Eye movements and vision. New York: Plenum Press, 1967. 7. Noton D, Stark I. Eye movements and visual perception. Sci Am 1971;224:35–43. 8. Phillips ML, David AS. Understanding the Symptoms of Schizophrenia Using Visual Scan Paths. Br J Psychiatry 1994;165:673–675. 9. Phillips ML, David AS. Abnormal visual scan paths: a psychophysiological marker of delusions in schizophrenia. Schizophr Res 1998;29:235–245. 10. Latimer CR. Eye–movement data: Cumulative fixation time and cluster analysis. Behav Res Methods Instrum Comput 1988;20:437–430. 11. Bestelmeyer PEG, Tatler, B.W., Phillips, L.H., Fraser, G., Benson, P., & St.Clair, D. Global visual scanning abnormalities in schizophrenia and bipolar disorder. Schizophr Res 2006;87:212–222. 12. Bestelmeyer PEG, Tatler, B.W., Phillips, L.H., Benson, P., & St.Clair, D. Are temporal characteristics of visual scan paths heritable? Perception 2006;35:136.

P.E.G. Bestelmeyer 13. Mannan SK, Ruddock KH, Wooding DS. Automatic control of saccadic eye movements made in visual inspection of briefly presented 2–D images. Spat Vis 1995;9:363–386. 14. Mannan SK, Ruddock KH, Wooding DS. The relationship between the locations of spatial features and those of fixations made during visual examination of briefly presented images. Spat Vis 1996;10:165–188. 15. Tatler BW, Baddeley RJ, Gilchrist ID. Visual correlates of fixation selection: effects of scale and time. Vis Res 2005;45:643–659. 16. Gould TD, Gottesman II. Psychiatric endophenotypes and the development of valid animal models. Genes Brain Behav 2006;5:113–119. 17. Streit M, Wolwer W, Gaebel W. Facial–affect recognition and visual scanning behaviour in the course of schizophrenia. Schizophr Res 1997;24:311–317. 18. Loughland CM, Williams LM, Harris AW. Visual scanpath dysfunction in first–degree relatives of schizophrenia probands: evidence for a vulnerability marker? Schizophr Res 2004;67:11–21. 19. Walker–Smith GJ, Gale AG, Findlay JM. Eye movement strategies involved in face perception. Perception 1977;6:313–326. 20. Gordon E, Coyle S, Anderson J, et al. Eye movement response to a facial stimulus in schizophrenia. Biol Psychiatry 1992;31:626–629. 21. Gaebel W, Ulrich G, Frick K. Visuomotor performance of schizophrenic patients and normal controls in a picture viewing task. Biol Psychiatry 1987;22:1227–1237. 22. Loughland CM, Williams LM, Gordon E. Visual scanpaths to positive and negative facial emotions in an outpatient schizophrenia sample. Schizophr Res 2002;55:159–170. 23. Loughland CM, Williams LM, Gordon E. Schizophrenia and affective disorder show different visual scanning behavior for faces: A trait versus state–based distinction? Biol Psychiatry 2002;52:338–348. 24. Phillips ML, David AS. Visual scan paths are abnormal in deluded schizophrenics. Neuropsychologia 1997;35:99–105. 25. Williams LM, Loughland CM, Gordon E, Davidson D. Visual scanpaths in schizophrenia: is there a deficit in face recognition? Schizophr Res 1999;40:189–199. 26. Williams LM, Loughland CM, Green MJ, Harris AWF, Gordon E. Emotion perception in schizophrenia: an eye movement study comparing the effectiveness of risperidone vs. haloperidol. Psychiatry Res 2003;120:13–27. 27. Kojima T, Matsushima E, Ando K, et al. Exploratory Eye– Movements and Neuropsychological Tests in Schizophrenic– Patients. Schizophr Bull 1992;18:85–94. 28. Ryu H, Morita K, Shoji Y, Waseda Y, Maeda H. Abnormal exploratory eye movements in schizophrenic patients vs healthy subjects. Acta Neurol Scand 2001;104:369–376. 29. Green MJ, Williams LM, Hemsley DR. Cognitive theories of delusion formation: The contribution of visual scanpath research. Cogn Neuropsychiatry 2000;5:63–74. 30. Minassian A, Granholm E, Verney S, Perry W. Visual scanning deficits in schizophrenia and their relationship to executive functioning impairment. Schizophr Res 2005;74:69–79. 31. Manor BR, Gordon E, Williams LM, et al. Eye movements reflect impaired face processing in patients with schizophrenia. Biol Psychiatry 1999;46:963–969. 32. Russel TA, Green MJ, Simpson I, Coltheart M. Remediation of facial emotion perception in schizophrenia: Concomitant changes in visual attention. Schizophr Res 2008;103:248–256.

Chapter16

Biomarkers and Endophenotypes in Eating Disorders Carolina Lopez, Marion Roberts, and Janet Treasure

Abstract The aim of this chapter is to discuss the concept of endophenotypes within the eating disorder (ED) population. As is common in most forms of psychiatric illness, people with ED have additional psychological problems, as do their family members. Affective disorders, addictions, and autistic spectrum disorders commonly occur in association with the individual and/or in the families of those with ED. This chapter will explore the literature related to underlying biological markers that potentially underpin all of these disorders. Some of the biomarkers fulfil the criteria for an endophenotype, although this is an area that is poorly developed so far in the eating disorder field. It is possible that core inherited vulnerability traits such as anxiety, rigidity, weak coherence, poor social cognition, compulsivity and impulsivity are some of the features held in common with other disorders. Keywords Biomarker • endophenotype • anorexia nervosa • bulimia nervosa • set-shifting • weak central coherence • 5-HT • dopamine • anxiety • OCPD • social cognition

Abbreviations ADHD: Attention-deficit hyperactivity disorder; AN: Anorexia nervosa; ASD: Autistic spectrum disorders; BN: Bulimia nervosa BED: Binge eating disorder DA:Dopamine; ED: Eating disorders; EDNOS: Eating disorder not otherwise specified; OA: Object assembly task; OCPD: Obsessive-compulsive personality disorder; OCPT: Obsessive-compulsive

C. Lopez, M. Roberts, and J. Treasure Institute of Psychiatry, Section of Eating Disorders, Division of Psychological Medicine & Psychiatry, King’s College London

personality traits; 5-HT: Serotonin; WCST: Wisconsin Card Sorting Test

Introduction Eating disorders (ED) are complex psychiatric illnesses that have severe consequences for the sufferer.1,2 The ED phenotype involves a range of syndromes characterised by a disturbance of body (or self) image, associated with over or under control of eating and extreme behaviours to control weight (e.g. food restriction, vomiting, laxative abuse etc.). These factors are associated with significant impairment in psychosocial and physical health.3,4 There are still uncertainties with regard to the causal and maintaining factors of ED. Recently, the concept of endophenotypes and biological markers have entered the discussion surrounding the aetiology of ED, highlighting the increasing recognition of the importance of biology in the development and maintenance of these disorders. In this chapter, the main findings with regards to biological markers in ED will be outlined, with a focus on the main diagnostic categories of ED: Anorexia Nervosa (AN) and Bulimia Nervosa (BN).

Eating Disorders: An Overview What are the Eating Disorders? The essential feature of AN is a low body weight attained and sustained by insufficient food intake. Often there is an increase in energy expenditure by

M.S. Ritsner (ed.), The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes, © Springer Science + Business Media B.V. 2009

227

228

compulsive activity. A subgroup use vomiting, laxatives and/or diet pills to control their energy intake (purge subtype). Continued poor nutritional intake is commonly explained by fears that the individuals’ weight or shape will become enormous if they start to eat and they will never regain a normal appetite. Other explanations for not eating include fears of bloating and abdominal discomfort or that food will be toxic in some way. The essential feature of BN is recurrent binge eating. A binge is defined as an episode of eating which is excessive for the context and is accompanied by a subjective sense of loss of control. Craving and overeating usually develop after a period of strict selfimposed dietary rules such as skipping meals or avoidance of carbohydrate, and is often triggered by stress, food cues or time alone. Thinness is over valued. Self-acceptance and regard is dominated by weight and shape esteem. The tension between concern about weight gain and the urges to overeat leads to the use of a variety of behaviours employed to either reduce intake such as vomiting or purging, or increase expenditure such as over exercise, metabolic stimulants or in the case of diabetes mellitus omitting insulin. The essential features of a binge eating disorder (BED) are episodes of overeating. In contrast to BN there is less investment in weight and shape for the construction of self-esteem. Less extreme methods are used to make reparation for the over eating and there are fewer impulsive, disinhibited, behaviours. BED merges with both the non-purging subtype of BN and obesity. BED is commonly associated with bipolar disorder and type 2 diabetes. Eating disorder not other wise specified (EDNOS) is a residual category including people who do not fully meet the criteria of one of the other eating disorders. Examples of EDNOS include: someone of normal weight who engages in purging such as vomiting or laxative use in the absence of binging episodes (‘purging disorder’), or someone with anorexia nervosa whose weight is not low enough to satisfy the criteria for anorexia nervosa, or whose periods have not stopped. This difficulty with classification is especially obvious in the child and adolescent population. Night eating syndrome (NES) is one category within this group. In this condition the greater part of calorie consumption occurs in the second half of the day.

C. Lopez et al.

Diagnostic Categories and Comorbidity The diagnostic categories within the ED spectrum are in a state of flux. Restricting anorexia (RAN) has a long medical record and was described in the mid nineteenth century, whereas the binge eating group of disorders were first recognized in cohorts born after 1950. Their prevalence continues to increase and as many as 10% of women may have had the problem within their lifetime. In 19795 Russell described BN as an “ominous variant” of AN marked by a switch in food related behaviour from extreme restriction to intense cravings for food with irresistible urges to eat large amounts of food. BN was the tip of the iceberg, with a variety of disorders including binge eating/ purging now identified, including the aforementioned EDNOS, “subtypes” BED, NES, and purging disorder. These variants of EDNOS have not yet been accepted into the American Psychiatric Association’s diagnostic and statistical manual, however BED was acknowledged and marked for further study. Despite EDNOS being a heterogeneous residual category, it can result in severe impairment for sufferers that present for treatment but do not meet the full criteria for the prototypical disorders. The diagnostic boundaries within the eating disorder spectrum are fuzzy with transitions over time. Also there is overlap at the edges with developmental disorders, personality disorders, obesity and other axis I disorders. This is shown in Fig. 16.1 together with the relationship with potential endophenotypes. At the AN end of the spectrum, 30–50% of persistent cases develop binge eating within seven years of onset. People with AN often show obsessive-compulsive traits of rigidity, attention to detail and perfectionism in childhood.6 Anxiety disorders such as separation anxiety and obsessive-compulsive disorder (OCD) are also common. Approximately 20% of cases have features consistent with autistic spectrum disorder (ASD).7 Some of these disorders precede or follow the onset of the ED. Depression tends to occur in parallel with starvation. At the binge eating side of the spectrum, other forms of disinhibited and impulsive behaviours are common. Comorbidity with alcohol/substance abuse and antisocial behaviours such as lying and stealing can occur (see Fig. 16.1). People with attention deficit hyperactivity disorder (ADHD) in childhood are at an increased risk of BN. There can be comorbidity with the affective spectrum of disorders. A very small

16 Biomarkers and Endophenotypes in Eating Disorders

229

Fig. 16.1 Diagnostic overlap between the subtypes of ED and other psychiatric disorders and the possible relationship with potential endophenotypes

proportion may develop the binge purge subtype of AN (AN-BN) but this is rarely sustained.

The Health and Social Costs of Eating Disorders Anorexia and bulimia nervosa affect mainly, but not exclusively, young females. Both are usually chronic and life-threatening psychiatric illnesses, with one of the highest risks of premature death of all psychiatric disorders.8 Lifetime prevalence rates for AN are, at the most conservative, 1.6% in women and 0.3% in men,9 with subclinical disordered eating even more common.10 The median duration of AN is 6 years.11,12 Lifetime prevalence for BN is even higher with different studies reporting rates that vary from 1–4.6% in females2,13–15 and between 0.1 to 0.4% in males.2,16 These illnesses strike early in the course of development, profoundly disrupting the quality of life (for a systematic review see ref.17). The burden of disease is high18 with severe physical and psychological comorbidity1,11,19 and disruption to vocational20 and social development21 for the individual, in addition to high costs of care for the family 22,23 and society.24–26

Despite this high level of disability and increasing interest among clinicians and scientists to find answers to these illness, there is a profound lack of clarity as to how ED develop. There is sufficient evidence to suggest that ED encompass psychological, physical and environmental elements in their aetiology, development and maintenance.27 However, the role of biology in ED had been somewhat neglected. This led to the drafting of a position statement from the Academy of Eating Disorders (AED), the largest international organisation of eating disorder professionals, which states “Current science affirms that eating disorders are influenced by neurobiological alterations of the brain; significantly impair cognitive function, judgment, and emotional stability; and limit the life activities of the person with the illness (page 2).”28 The search for suitable biological markers in people with ED is complex as starvation and impaired nutrition, core elements of the disorder, have a profoundly disrupting effect on both psychology and physiology. Thus a proliferation of biological abnormalities is present in the acute state. One way of trying to distinguish between the consequences of starvation and features that suggest an underlying biological vulnerability is to examine people after weight recovery. This in itself is not without problems; for example, how does one define recovery?; is it important to have a minimal duration of

230

recovery?; is the abnormality a risk factor or a scar from the illness? Some risk factors may only occur in the context of environmental or developmental risk factors and may remain hidden. Despite these difficulties several groups have undertaken studies of biomarkers in patients after recovery. Fewer studies have taken the next step which is to examine whether the biomarkers represent a familial and genetic vulnerability. Recent reviews have included the concept of endophenotypes in the discussion of the aetiology of ED.29–33

Endophenotypes and Biological Markers in Eating Disorders Gottesman and Gould34 have defined specific endophenotype criteria for psychiatric disorders which includes (1) association with the illness in the population, (2) heritability, (3) state-independence, (4) co-segregation with the illness in families, and (5) presence in unaffected relatives at higher level than in the general population. Although the endophenotype and biological marker concepts have been used interchangeably in the literature, Gottesman and Gould use the former when there are some signs of heritability, and biomarkers when the trait does not fulfil the criteria of genetic underpinnings. In this chapter we follow this convention.

C. Lopez et al.

bias,43 (d) difficulties in set-shifting,44,45 and (e) weak central coherence.46 Of the evidence-based biomarkers mentioned above, only a few have evidence that would support the proposition that they are potential endophenotypes. Thus, only two published studies have identified potential endophenotype candidates for the ED population which fulfil the criteria at least to be a biomarker for ED, and also be present in healthy relatives of affected individuals: impaired set-shifting for AN47 and reduced brain serotonin (5-HT) transporter activity in BN.48 More recently, weak central coherence has become a promising candidate as it was found to be present in people with current and past ED46,49 as well as in unaffected family members. We will focus the following section on those evidence-based potential biomarkers that have shown to be present in people recovered from ED in order to distinguish biomarkers from correlates of starvation. However, this has its own difficulties as traits found in a recovered state can represent scars from the illness, or a biological vulnerability that only emerges in a specific context.

Evidence-Based Biomarkers Potential Biomarkers for Eating Disorders

Neurotransmitters Biomarkers

Biomarkers that are of interest include those derived from conceptual and theoretical models of the illness, and those that result from strong empirical grounding. Those that arise from conceptual and theoretical derived models of the illness include (a) neurodevelopmental-chronic stress hypothesis associated with hyperactivity of the hypothalamus-pituitary-adrenal axis (HPA axis) associated which chronic interpersonal difficulties and stress,35 (b) fear conditioning,36 (c) deficits in habit (implicit) learning,37 and (d) shared features with ASD and OCD with obsessive-compulsive personality traits (OCPT) such as rigidity and emotional avoidance.30,38 Evidence-based relevant biomarkers include neurotransmitters biomarkers such as (a) altered serotonin function39–41 and (b) dopamine disturbance42; neuropsychological biomarkers such as (c) attentional

Altered serotonin (5-HT) and dopamine (DA) systems have been described as possible biomarkers in the ED literature (see Table 16.1). These are proposed to have a key role in the appetite control systems as shown in Fig. 16.2.

Table 16.1 Neurotransmitters abnormalities in eating disorders AN AN-BN BN 5-HT2A 5-HT1A 5-HTT DA D2/D3

↓ = ↑ ↑

↓ ↑ ↓

= ↑

AN (Anorexia Nervosa); AN-BN (Binge Eating Purging Anorexia Nervosa); BN (Bulimia Nervosa) 5-HT (Serotonin); DA (Dopamine D2/D3 receptor); ↑ (Increased); ↓ (Decreased); = (Normal)

16 Biomarkers and Endophenotypes in Eating Disorders

231

Fig. 16.2 Central control of appetite. This diagram illustrates the two aspects of the appetite control system. The nutrostat which is a homeostatic system to ensure nutritional balance, and the drive system that determines behavioural choice. It is possible that endophenotypes relating to either or both of these networks are linked to the vulnerability to develop an eating disorder. GI: Gastrointestinal; DA: Dopamine; 5-HT: Serotonin

Altered Serotonin Function as a Biomarker of ED A number of anomalies in the serotonin (5-HT) system have been described as enduring features in ED. The main findings are described below. 5-HT 2A There is wide evidence that altered serotonin systems are involved in the aetiology of ED.50 Decreased 5-HT2A binding has been found in both women with acute RAN and AN-BN, and also in those recovered from the illness.40,41,51–53 No differences compared to healthy controls were evident in a group of BN patients.52 However, Kaye and collaborators54 found that women who had recovered from BN had a reduction of medial orbital frontal cortex 5-HT2A binding. 5-HT1A 5-HT1A activity has been found to be increased in current AN, current BN, and recovered AN-BN groups.55–57 5-HT1A binding is higher in BN patients than in control subjects in diverse brain regions with the most robust differences in the angular gyrus, the medial prefrontal cortex, and the posterior cingulate cortex.57

5-HT Transporter Activity Reduced 5-HT transporter availability was found in the hypothalamus and thalamus in BN especially in the group with a longer duration of illness.58 Also, increased 5-HTT activity was found in recovered RAN and BN groups in the antero-ventral striatum.51 Steiger et al.48 investigated levels of platelet 5-HT reuptake activity compared to healthy controls in current BN patients, healthy and affected relatives. Results indicated that healthy mothers and sisters of women suffering from BN obtained lower values of paroxetinebinding density indicating decreased central 5-HT reuptake activity. These results are similar to what has been described previously in people with current BN. Replications of this study with larger sample sizes are required.

Altered Dopamine Systems Frank et al.42 found higher dopamine D2/D3 receptor binding in women recovered from AN in the anteroventral striatum. This finding suggests that there is either a decreased volume of intrasynaptic dopamine or increased D2/D3 receptors density or affinity in AN. Interestingly, this finding might be associated to the

232

altered response to reward seen in AN (difficulties to differentiate between positive and negative feedback) which has been also reported in fully recovered AN individuals.59

Neuropsychological Biomarkers

C. Lopez et al.

Set-Shifting in Recovery Women recovered from AN also demonstrate impairment in set-shifting ability but with reduced effect sizes.44,45 Small to moderate effect sizes have been found for the TMT, Brixton Task and CatBat Task, with a large effect size reported for the Haptic Illusion task.

Attentional Biases An attentional bias towards food and body-related stimuli has been reported in both currently ill AN and BN, with a medium effect size.60,61 In recovery, women with past AN but not those with past BN are still more distracted by shape concerns.62 This report is consistent with the finding of abnormally higher activation in the medial prefrontal and anterior cingulate cortices in response to food stimuli using fMRI in both currently ill patients with AN and long-term recovered women,43 suggesting that it might be a biological marker for AN.

Set-Shifting Set-shifting is one of the main aspects of executive functioning and refers to the ability to be flexible with one’s mindset in adapting to new task demands or changes in situations.63

Set-Shifting in the Acute Illness State A series of studies using a variety of neuropsychological tasks to measure this concept found that people with AN have difficulties in set-shifting.64–66 These difficulties have also been described in people with current BN.67 A meta-analytic systematic review of the literature confirmed this impairment in cognitive flexibility across diagnostic groups and illness state.45 People with ED were found to perform generally poorer than the healthy controls in measures of setshifting with pooled effect sizes (Cohen’s d) in metaanalyses varying from small in the Trail Making Test-B, medium in the Wisconsin Card Sort Test (WCST) and CatBat Task, to large in Haptic illusion Test. Conflicting results have been reported using the Brixton task (small to large effects in AN and negligible in BN).45

Set-Shifting in First Degree Relatives First-degree relatives of women with ED have also been found to have poor set-shifting.47 Specifically, sisters of women with AN took longer to shift set in the CatBat Task (d = 0.7) and Brixton Test (d = 0.5) and showed significantly greater perceptual rigidity on the Haptic Illusion Task (d = 0.7). A replication study including 30 AN and 20 BN discordant sister pairs found that healthy sisters made more perseverative errors on the WCST (d = 0.66; d = 0.36, respectively), and on the Brixton Task for healthy BN sisters only (d = 0.56) on. Replication of these studies with larger samples is required as well as genetic examination to complete the criteria for an endophenotype.

Central Coherence Central coherence was originally defined by Frith in 198968 as the natural cognitive style of most people in adult life to integrate large amounts of incoming information into context, gestalt and meaning. The opposite tendency, weak central coherence, is therefore the cognitive style characterised by a tendency to process information in parts (detail) rather than as a whole, with relative difficulty in global or integrative processing.69 Weak central coherence is a key component of the cognitive style of those with autistic spectrum disorders (ASD)69,70 and has also been found in first-degree relatives in this population.71–73

Weak Central Coherence in the Acute Illness State Initial findings regarding this trait in the field of ED are attributed to Gillberg and collaborators’ longitudinal study which found that people with early onset of AN were persistently impaired over time on only one task from WAIS was the Object Assembly subtest (OA)

16 Biomarkers and Endophenotypes in Eating Disorders

which requires the ability to integrate pieces of information into a whole to construct a familiar object.74,75 They also found that a sub-group of people with AN who met criteria for ASD also had a general cognitive profile resembling the profile found in people with ASD. Although the authors initially hypothesised that the poor performance in the OA was due to poor abstract thinking74 they later associated their findings with weak central coherence.75 Our group has furthered the investigation of the weak central coherence hypothesis in people with ED with a systematic and hypothesis driven approach. We have found that weak central coherence may be a feature of AN and BN (see Table 16.2) as there is superiority in tasks that require detail processing with moderate effect sizes (e.g. Group/Embedded Figures and Matching Familiar Figures) and a weakness in tests that require a global strategy such as the Sentences Completion and Homograph Reading Tasks, the ReyOsterrieth Complex Figure and OA with moderate to large effect sizes (see Table 16.2).49,76,77 A recent systematic review of the literature found some evidence supporting this account particularly for decreased global processing.77

Weak Central Coherence in Recovery This trait has been reported as present in women in recovered state of ED, with medium effect sizes for poor visual global integration and large for detail focused processing. In the verbal domain, recovered women also performed poorer than healthy controls with medium to large effect sizes46 (see Table 16.2).

Weak Central Coherence in First-Degree Relatives A recent study from our group, has found that weak central coherence is also present in the healthy sisters of those with AN and BN. Using the Group Embedded Figure Test and the Rey-Osterreith Complex Figure copy trial (central coherence index), it was found that like their AN sisters, 30 healthy sisters of those with AN displayed a more detail focussed processing style on both tasks (d = −0.2; d = 1.16, respectively). Likewise, 20 healthy sisters of BN patients displayed weak coherence on these tasks compared to healthy controls (d = −0.4; d = 0.94).

233 Table 16.2 Effect sizes (Cohen’s d)* for neuropsychological tasks relating to set-shifting and weak central coherence in ED Task AN.C AN.R BN Time Repeat Set-shifting TMT-B WCST Brixton CatBat Haptic

0.8 0.6 0.2 0.6 1.2

Central Coherence D- EFT 0.5 D- GEFT 0.8 D- BD 0.0 D- MFFT 0.9 G- Rey CC 0.9 G- OA 0.6 G- Frag. Pic 1.7 G- SCT 0.7 G- HRT 0.0

0.5 0.4 0.3 0.4 0.9

0.2

1.0

0.7

0.2 0.6 0.6

0.3 0.5 0.6 0.4

1.2 0.4

1.1 0.5

0.1 0.9 1.1

5 10 5 3 5

Yes-2 No Yes-2 No No

10 15 10 10 5 15 10 10 10

Yes-2 ? ? No Yes-2 No Yes-2 No No

TMT-B (Trail Making Task, Trial B); WCST (Wisconsin Card Sorting Test); EFT (Embedded Figures Task); BD (Block Design); MFFT (Efficiency Matching Familiar Figures Test); Rey CC (Rey Central Coherence Index); OA (Object Assembly); Frag. Pic. (Fragmented Pictures Task); SCT (Total Score in Sentence Completion Task); HRT (Total Score in Homograph Reading Task) AN.C (Current AN); AN.R (Recovered AN); BN (Current Bulimia Nervosa) Time (Time taken in minutes to administer each task) Repeat (Can the task be repeated?); Yes, Yes-2 (have 2 versions of the task), No, ? (unsure if task can be repeated). The effect size in the table represent the mean of the published studies * Cohen’s d effect sizes are understood as follows: 0 to < 0.15 negligible; > 0.15–0.40 - small; > 0.40–0.75 - medium; > 0.75– 1.10 - large; > 1.10 - huge.

As yet we do not know if weak central coherence and set-shifting fulfil the criteria for an endophenotype for ED and whether is linked to OCPD traits in ED. Table 16.2 illustrates an updated summary of the promising measures used in the examination of both set-shifting and weak central coherence in ED.

Potential Biomarkers Associated with Emotional Processing in ED Emotional disturbances in patients with ED have been largely reported (e.g. anxiety, alexithymia, impaired social cognition). However, evidence of this in the recovered state is extremely limited.

234

C. Lopez et al.

Table 16.3 Effect sizes (Cohen’s d) for neuropsychological tasks relating social cognition in AN AN.C Social Cognition RMI Eyes RMI Voice RMI Film E-Stroop

1.1 0.8 0.8 1.6

RMI (Reading the Mind in the Eyes; Voice and Films); E-Stroop (Emotional Stroop Task); AN.C (Current AN)

We have found that in the acute illness state, women with AN have an attentional bias to social stimuli using the E-Stroop (Emotional Stroop) paradigm, where those with higher levels of depressive symptoms also display a bias to angry over neutral faces. Additionally, women with acute AN appear to be impaired in visual and verbal recognition of emotions.78–81 This contrasts with the finding of normal self-reported levels of empathic response.82 A recent study from our group, that administered tasks developed in the field of ASD to women with acute AN (e.g. Reading the Mind in the Voice, Films, Eyes) found large effects in all of the task domains suggesting greater difficulties in the acute state of the illness relative to a healthy comparison group (Table 16.3). High levels of anxiety have been found in both AN and BN, and seem to persist in recovery although with reduced effect sizes. Self-report anxiety shows a very large effect size in the AN and BN population with measures such as the Hospital Anxiety and Depression Scale (Cohen’s d effect size = 2.07 for AN49; and 1.72 for BN76) and the State Trait Anxiety Inventory (State anxiety d = 1.06 for AN and 1.7 for BN; Trait anxiety d = 1.44 for AN and 2.4 for BN83). Recovered people maintain high levels of anxiety with large effect size (d = 1.2646). Also, higher rates of lifetime anxiety disorders such as generalised anxiety, obsessive-compulsive disorder, separation anxiety, social phobia, and panic disorder have also been found in first-degree relatives of people with AN which suggest a possible heritability for anxiety and fear propensity.84

Implications for Treatment The search for potential biomarkers and endophenotypes in ED has revealed a number of features that are present in the acute state of the illness that also persist after recovery, suggesting specific vulnerabilities in

the development and maintenance of these disorders. This evidence can be of use to inform treatment strategies that may reduce the duration of the illness and improve outcome. Our group have developed treatments aimed at either remediating or moderating processing biases such as set-shifting and weak central coherence. Tchanturia and collaborators have introduced cognitive remediation therapy in which exercises to strengthen cognitive flexibility and global integration are used in people with severe AN as a pre-therapy intervention for in-patients.85,86 In addition, a pre-therapy module of motivational enhancement based on neuropsychological traits with individualised feedback, formulation and goal setting has been developed as part of outpatient therapy.87,88 Individuals with extreme difficulty processing the whole and/or with cognitive inflexibility and/or high anxiety are given personalised feedback on their neuropsychological assessment in visual format. Bar charts for each task allow patients to compare their performance with a clinical (ED) group and a healthy comparison group. Feedback on neuropsychological assessment is provided in such a way that the therapist and patient collaboratively discuss how such information processing biases can be a strength or weakness, depending on the circumstances. This feedback intervention follows the principles of motivational interviewing, using ambivalence about change in the “how” of thinking to produce an enhanced meta-cognitive ability. Patients are encouraged to reflect on their own cognitive functioning and to undertake behavioural experiments to adjust their response when their natural tendency may not be advantageous to them. While this intervention is still new, it has received positive feedback from patients and therapists and has been described as a safe, nonthreatening introduction to therapy. Work is needed to explore these interesting findings further. This will involve testing in a randomised controlled trial whether the addition of a module focused on these core processes improves outcome and whether change is mediated by the behavioural representations of these traits. Finally, emotional disturbances have been highlighted as a core target in the treatment of ED due to the key role in the maintenance of these disorders. For example, a reward-emotion module to attain balance in the approach-avoidance systems have been included in the new Maudsley Cognitive Motivational Therapy for AN.

16 Biomarkers and Endophenotypes in Eating Disorders

Conclusions and Future Directions Eating Disorders have a complex aetiology and a clear picture of what makes people vulnerable to develop these disorders is as yet unknown. It is hoped that a better understanding of the causal and maintaining factors may translate into more sophisticated and successful treatments and a reduced stigma for the sufferers. The introduction of the research into endophenotypes for ED may be a pathway to contribute to this goal. The progress towards revealing potential biomarkers in ED has considerably advanced in the last decade. Main findings in this direction highlight the involvement of altered neurotransmitters systems such as 5-HT and DA as well as cognitive information processing anomalies such as weak central coherence and set-shifting in both AN and BN. Although these advances in research have contributed to the understanding and recognition of some of the cognitive and neurobiological basis of ED, more research is needed to clarify the real effect of these findings in the long-term course of the illness and also in other salient aspects of the illness that remains relatively unexplored such as social cognition and emotional processing. We proposed that specifically tailored interventions which address these enduring vulnerability features may help to improve the outcome of ED. It is hoped that such a deeper understanding of ED based on biomarkers and the search for endophenotypes will improve awareness, early intervention and make for better prognosis in the long term. Acknowledgements Carolina Lopez and Marion Roberts are Ph.D. students funded by Nina Jackson Foundation Research Into Eating Disorders (RIED).

References 1. Hudson JI, Hiripi E, Pope HG, et al. The prevalence and correlates of eating disorders in the national comorbidity survey replication. Biol Psychiatry 2007; 2007: 348–358 2. van Hoeken D, Seidell J, Wijbrand H. Epidemiology. In Treasure J, Schmidt U and van Furth E, eds. Handbook of Eating Disorders, 2nd ed. Chichester: Wiley; 2003: 11–34 3. Fairburn C, Harrison P. Eating disorders. Lancet 2003; 361: 407–416 4. Striegel-Moore R, Bulik C. Risk factors for eating disorders. Am Psychol 2007; 62: 181–198 5. Russell G. Bulimia nervosa: An ominous variant of anorexia nervosa. Psychol Med 1979; 9: 429–448

235 6. Anderluh M, Tchanturia K, Rabe-Hesketh S, et al. Childhood obsessive-compulsive personality traits in adult women with eating disorders: Defining a broader eating disorder phenotype. Am J Psychiatry 2003; 160: 242–247 7. Gillberg IC, Rastam M, Gillberg C. Anorexia nervosa outcome: Six-year controlled longitudinal study of 51 cases including a population cohort. J Am Acad Child Adolesc Psychiatry 1994; 33: 729–739 8. Harris E, Barraclough B. Excess mortality of mental disorder. Br J Psychiatry 1998; 173: 11–53 9. Bulik C, Sullivan P, Tozzi F, et al. Prevalence, heritability, and prospective risk factors for anorexia nervosa. Arch Gen Psychiatry 2006; 63: 305–312 10. Ackard D, Fulkerson J, Neumark-Sztainer D. Prevalence and utility of DSM-IV eating disorder diagnostic criteria among youth. Int J Eat Disord 2007; 40: 409–417 11. Herzog W, Deter H, Fiehn W, et al. Medical findings and predictors of long-term physical outcome in anorexia nervosa: A prospective, 12-year follow-up study. Psychol Med 1997; 27: 269–279 12. Fichter M, Quadflieg N. Long-term stability of eating disorder diagnoses. Int J Eat Disord 2007; 40: S61–S66 13. Favaro A, Ferrara S, Santonastaso P. The spectrum of eating disorders in young women: A prevalence study in a general population sample. Psychosom Med 2003; 65: 701–708 14. Garfinkel P, Lin E, Goering P, et al. Bulimia nervosa in a Canadian community sample: Prevalence and comparison of subgroups. Am J Psychiatry 1995; 152: 1052–1058 15. Lewinsohn P, Striegel-Moore R, Seeley J. Epidemiology and natural course of eating disorders in young women from adolescence to young adulthood. J Am Acad Child Adolesc Psychiatry 2000; 39: 1284–1292 16. Kjelsas E, Bjornstrom C, Gotestam K. Prevalence of eating disorders in female and male adolescents (14–15 years). Eat Behav 2004; 5: 13–25 17. Mond J, Hay P, Rodgers B, et al. Assessing quality of life in eating disorder patients. Qual Life Res 2005; 14: 171–178 18. Mathers C, Vos E, Stevenson C, et al. The Australian burden of disease study: Measuring the loss of health from diseases, injuries and risk factors. Med J Aust 2000; 172: 592–596 19. O’Brien K, Vincent N. Psychiatric comorbidity in anorexia and bulimia nervosa: Nature, prevalence, and causal relationships. Clin Psychol Rev 2003; 23: 57–74 20. Hjern A, Lindberg L, Lindblad F. Outcome and prognostic factors for adolescent female in-patients with anorexia nervosa: 9- to 14-year follow-up. Br J Psychiatry 2006; 189: 428–432 21. Tiller J, Sloane G, Schmidt U, et al. Social support in patients with anorexia nervosa and bulimia nervosa. Int J Eat Disord 1997; 21: 31–38 22. Treasure J, Murphy T, Szmukler G, et al. The experience of caregiving for severe mental illness: A comparison between anorexia nervosa and psychosis. Soc Psychiatry Psychiatr Epidemiol 2001; 36: 343–347 23. Whitney J, Haigh R, Weinman J, et al. Caring for people with eating disorders: Factors associated with psychological distress and negative caregiving appraisals in carers of people with eating disorders. Br J Clin Psychol 2007; 46: 413–428 24. Striegel-Moore R, Leslie D, Petrill S, et al. One-year use and cost of inpatient and outpatient services among female and male patients with an eating disorder: Evidence from a

236

25.

26. 27. 28.

29.

30.

31.

32.

33.

34.

35. 36.

37.

38.

39.

40.

41.

42.

43.

C. Lopez et al. national database of health insurance claims. Int J Eat Disord 2000; 27: 381–389 Byford S, Barrett B, Roberts C, et al. Economic evaluation of a randomised controlled trial for anorexia nervosa in adolescents. Br J Psychiatry 2007; 191: 436–440 Striegel-Moore R, DeBar L, Wilson G, et al. Health services use in eating disorders. Psychol Med 2007; 2: 1–10 Collier DA, Treasure J. The aetiology of eating disorders. Br J Psychiatry 2004; Nov: 363–365 Klump K, Bulik C, Kaye W, et al. Eating disorders are serious mental illnesses. Int J Eat Disord 2008 (in press). doi 10.1002/eat.20589 Bulik CM, Hebebrand J, Keski-Rahkonen A, et al. Genetic epidemiology, endophenotypes, and eating disorder classification. Int J Eat Disord 2007; 40: S52–S60 Treasure J, Lopez C, Roberts M. Endophenotypes in eating disorders: Moving toward etiologically based diagnosis and treatment focused on pathophysiology. Pediatric Health 2007; 1: 171–181 Treasure J. Getting beneath the phenotype of anorexia nervosa: The search for viable endophenotypes and genotypes. Can J Psychiatry 2007; 52: 212–209 Steiger H, Bruce KR. Phenotypes, endophenotypes, and genotypes in bulimia spectrum eating disorders. Can J Psychiatry 2007; 52: 220–227 Hudson J, Poper H. Genetic epidemiology of eating disorders and co-occurring conditions: The role of endophenotypes. Int J Eat Disord 2007; 40: S76–S78 Gottesman I, Gould T. The endophenotype concept in psychiatry: Etymology and strategic intentions. Am J Psychiatry 2003; 160: 636–645 Connan F, Campbell I, Katzman M, et al. A neurodevelopmental model for anorexia nervosa. Physiol Behav 2003; 79: 13–24 Strober M. Pathologic fear conditioning and anorexia nervosa: On the search for novel paradigms. Int J Eat Disord 2004; 35: 504–508 Steinglass J, Walsh B. Habit learning and anorexia nervosa: A cognitive neuroscience hypothesis. Int J Eat Disord 2006; 39: 267–275 Wade T, Tiggemann M, Bulik C, et al. Shared temperament risk factors for anorexia nervosa: A twin study. Psychol Med 2008; 70: 239–244 Kaye WH, Gwirtsman HE, George DT, et al. Altered serotonin activity in anorexia nervosa after long-term weight restoration. Does elevated cerebrospinal fluid 5-hydroxyindoleacetic acid level correlate with rigid and obsessive behavior? Arch Gen Psychiatry 1991; 48: 556–562 Frank GK, Kaye WH, Meltzer CC, et al. Reduced 5-HT2A receptor binding after recovery from anorexia nervosa. Biol Psychiatry 2002; 52: 896–906 Bailer U, Price J, Meltzer C, et al. Altered 5-HT(2A) receptor binding after recovery from bulimia-type anorexia nervosa: Relationships to harm avoidance and drive for thinness. Neuropsychopharmacology 2004; 29: 1143–1155 Frank G, Bailer U, Henry S, et al. Increased dopamine D2/ D3 receptor binding after recovery from anorexia nervosa measured by positron emission tomography and [11c]raclopride. Biol Psychiatry 2005; 58: 908–912 Uher R, Brammer MJ, Murphy T, et al. Recovery and chronicity in anorexia nervosa: Brain activity associated with differential outcomes. Biol Psychiatry 2003; 54: 934–942

44. Tchanturia K, Morris R, Anderluh M, et al. Set shifting in anorexia nervosa: An examination before and after weight gain, in full recovery and relationship to childhood and adult OCPD traits. J Psychiatr Res2004; 38: 545–552 45. Roberts ME, Tchanturia K, Stahl D, et al. A systematic review and meta-analysis of set shifting ability in eating disorders. Psychol Med 2007; 37: 1075–1084 46. Lopez C, Tchanturia K, Stahl D, et al. Weak central coherence in eating disorders: A step towards looking for an endophenotype of eating disorders. J Clin Exp Neuropsychol 2008 (in press). doi 10.1080/13803390802036092 47. Holliday J, Tchanturia K, Landau S, et al. Is impaired setshifting an endophenotype of anorexia nervosa? Am J Psychiatry 2005; 162: 2269–2275. 48. Steiger H, Gauvin L, Joober R, et al. Intrafamilial correspondences on platelet [3H-]paroxetine-binding indices in bulimic probands and their unaffected first-degree relatives. Neuropsychopharmacology 2006; 31: 1785–1792 49. Lopez C, Tchanturia K, Stahl D, et al. An examination of the concept of central coherence in women with anorexia nervosa. Int J Eat Disord 2008; 41: 143–152 50. Kaye W. Neurobiology of anorexia and bulimia nervosa. Physiol Behav 2008; 94: 121–135 51. Bailer UF, Frank GK, Henry SE, et al. Serotonin transporter binding after recovery from eating disorders. Psychopharmacology (Berl) 2007; 195: 315–324 52. Goethals I, Vervaet M, Audenaert K, et al. Comparison of cortical 5-HT2A receptor binding in bulimia nervosa patients and healthy volunteers. Am J Psychiatry 2004; 161: 1916–1918 53. Audenaert K, Van Laere K, Dumont F, et al. Decreased 5-HT2A receptor binding in patients with anorexia nervosa. J Nucl Med 2003; 44: 163–169 54. Kaye W, Frank G, Meltzer C, et al. Altered serotonin 2A receptor activity in women who have recovered from bulimia nervosa. Am J Psychiatry 2001; 158: 1152–1155 55. Bailer U, Frank G, Henry S, et al. Exaggerated 5-HT1A but normal 5-HT2A receptor activity in individuals ill with anorexia nervosa. Biol Psychiatry 2007; 61: 1090–1099 56. Bailer U, Frank G, Henry S, et al. Altered brain serotonin 5-HT1A receptor binding after recovery from anorexia nervosa measured by positron emission tomography and [carbonyl11c]way-100635. Arch Gen Psychiatry 2005; 62: 1032–1041 57. Tiihonen J, Keski-Rahkonen A, Muhonen M, et al. Brain serotonin 1A receptor binding in bulimia nervosa. Biol Psychiatry 2004; 55: 871–873 58. Tauscher J, Pirker W, Willeit M, et al. [123i] beta-CIT and single photon emission computed tomography reveal reduced brain serotonin transporter availability in bulimia nervosa. Biol Psychiatry 2001; 49: 326–332 59. Wagner A, Aizenstein H, Venkatraman VK, et al. Altered reward processing in women recovered from anorexia nervosa. Am J Psychiatry 2007; 164: 1842–1849 60. Johansson L, Ghaderi A, Anderson G. Stroop interference for food - and body - related words: A meta-analysis. Eat Behav 2005; 6: 271–281 61. Dobson KS, Dozoi DJ. Attentional biases in eating disorders: A meta-analytic review of stroop performance. Clin Psychol Rev 2004; 23: 1001–1022

16 Biomarkers and Endophenotypes in Eating Disorders 62. Lovell D, Williams J, Hill A. Selective processing of shaperelated words in women with eating disorders, and those who have recovered. Br J Clin Psychol 1997; 36: 421–432 63. Miyake A, Freidman NP, Emerson MJ, et al. The unity and diversity of executive functions and their contributions to complex “frontal lobe” tasks: A latent variable analysis. Cognit Psychol 2000; 41: 49–100 64. Tchanturia K, Morris R, Surguladze S, et al. An examination of perceptual and cognitive set shifting tasks in acute anorexia nervosa and following recovery. Eating Weight Disord 2002; 7: 312–315 65. Tchanturia K, Serpell L, Troop N, et al. Perceptual illusions in eating disorders: Rigid and fluctuating styles. J Behav Ther Exp Psychiatry 2001; 32: 107–115 66. Steinglass JE, Walsh T, Stern Y. Set shifting deficit in anorexia nervosa. Int J Eat Disord 2006; 12: 431–435 67. Tchanturia K, Anderluh M, Morris R, et al. Cognitive flexibility in anorexia nervosa and bulimia nervosa. J Int Neuropsychol Soc 2004; 10: 513–520 68. Frith U. Autism: Explaining the enigma. Oxford: Basil Blackwell; 1989 69. Happé F, Frith U. The weak coherence account: Detailfocused cognitive style in autism spectrum disorders. J Autism Dev Disord 2006; 36: 5–25 70. Happé F, Booth R. The power of the positive: Revisiting weak coherence in autism spectrum disorders. Q J Exp Psychol (Colchester) 2008; 61: 50–63 71. Smalley S, Asarnov R. Brief report: Cogntive subclinical markers in autism. J Autism Dev Disord 1990; 20: 271–278 72. Happé F, Briskman J, Frith U. Exploring the cognitive phenotype of autism: Weak “Central coherence” In parents and siblings of children with autism: I.Experimental tests. J Child Psychol Psychiatry 2001; 42: 299–307 73. Baron-Cohen S, Hammer J. Parents of children with asperger syndrome: What is the cognitive phenotype? J Cogn Neurosci 1997; 9: 548–554 74. Gillberg I, Gillberg C, Rastam M, et al. The cognitive profile of anorexia nervosa: A comparative study including a community-based sample. Compr Psychiatry 1996; 37: 23–30 75. Gillberg I, Rastam M, Wentz E, et al. Cognitive and executive functions in anorexia nervosa ten years after onset of eating disorder. J Clin Exp Neuropsychol 2007; 29: 170–178 76. Lopez C, Tchanturia K, Stahl D, et al. Central coherence in women with bulimia nervosa. Int J Eat Disord 2008; 41: 340–347

237 77. Lopez C, Tchanturia K, Stahl D, et al. Central coherence in eating disorders: A systematic review. Psychol Med 2008; 38: 1393–1404 78. Zonnevylle-Bender M, van Goozen S, Cohen-Kettenis P, et al. Emotional functioning in anorexia nervosa patients: Adolescents compared to adults. Depress Anxiety 2004; 19: 35–42 79. Zonnevylle-Bender M, van Goozen S, Cohen-Kettenis P, et al. Emotional functioning in adolescent anorexia nervosa patients: A controlled study. Eur Child Adolesc Psychiatry 2004; 13: 28–34 80. Zonnevijlle-Bender M, van Goozen S, Cohen-Kettenis P, et al. Do adolescent anorexia nervosa patients have deficits in emotional functioning? Eur Child Adolecent Psychiatry 2002; 11: 38–42 81. Kucharska-Pietura K, Nikolaou V, Masiak M, et al. The recognition of emotion in the faces and voice of anorexia nervosa. Int J Eat Disord 2004; 35: 42–47 82. Hambrook D, Tchanturia K, Schmidt U, et al. Empathy, systemazing, and autistic traits in anorexia nervosa: A pilot study. Br J Clin Psychol 2008; 47: 335–339 83. Wagner A, Barbarich-Marsteller NC, Frank GK, et al. Personality traits after recovery from eating disorders: Do subtypes differ? Int J Eat Disord 2006; 39: 276–284 84. Strober M, Freeman R, Lampert C, et al. The association of anxiety disorders and obsessive compulsive personality disorder with anorexia nervosa: Evidence from a family study with discussion of nosological and neurodevelopmental implications. Int J Eat Disord 2007; 40: S46–S51 85. Tchanturia K, Whitney J, Treasure J. Can cognitive exercises help treat anorexia nervosa? Eating Weight Disord 2006; 11: e112–116 86. Tchanturia K, Davies H, Lopez C, et al. Neuropsychological task performance before and after cognitive remediation in anorexia nervosa: A pilot case series. Psychol Med 2008; 38: 1371–1373 87. Lopez C, Roberts ME, Tchanturia K, et al. Using neuropsychological feedback therapeutically in treatment for anorexia nervosa: Two illustrative case reports. Eur Eat Disord Rev 2008; 16: 411–420 88. Lopez CA, Roberts ME, Sepulveda AR, et al. Using an evidence based approach to develop new tailored treatment for anorexia nervosa. In Buckroyd J and Rother S, eds. Psychological Responses to Eating Disorders. Chichester: Wiley; 2008: 39–54

Chapter 17

Movement Abnormalities: A Putative Biomarker of Risk for Psychosis Vijay A. Mittal and Elaine F. Walker

Abstract The link between movement abnormalities and psychotic disorders is presumed to reflect common neural mechanisms that influence both motor functions and vulnerability to psychosis. The high-risk or prodromal period preceding psychotic disorders represents both a viable point for intervention and a developmental period with high potential to shed light on etiology. In this chapter, we provide a background for understanding subtypes of movement abnormalities and underlying neurological mechanisms within the context of a neurodevelopmental conceptualization of psychosis and schizophrenia. We use these sections to provide a framework for reviewing and integrating studies of movement abnormalities in high-risk populations. These data are then evaluated to determine the viability of movement abnormalities as a biomarker. Potential directions for movement and biomarker research are evaluated and discussed. Keywords Movement abnormalities • biomarker • hyperkinetic • psychosis • high-risk • schizophrenia Abbreviation ANCOVA: Analysis of covariance: CNS: Central nervous system: CSTC: Corticos triatothalamocotical: CI: Confidence interval: DSM: Diagnostic and Statistical Manual: DA: Dopamine: DRD2: Dopamine receptor D2: DRD3: Dopamine receptor D3: GPe: External pallidal segment,: fMRI: Functional magnetic resonance imaging: GPi: Globus pallidus: HR: High-risk: HPA: Hypothalamic-pituitary-adrenal: MRI:

V. A. Mittal Department of Psychology, University of California, Los Angeles, USA E. F. Walker Department of Psychology, Emory University, USA

Magnetic resonance imaging: MPA: Minor physical anomaly: NAPLS: North American Prodromal Longitudinal Study: NC: Normal control: OPD: Other personality disorder: PPN Pedunculopontine nucleus: PET: Positron emission tomography: SPD: Schizotypal personality disorder: SNr: Substantia nigra: SCID: Structured clinical interview for axis I DSM-IV disorders: SCD: Schizoid: VA/VL: Ventrolanterior/ventrolatereal thalamus: WBIM: Whole-body involuntary movement

Introduction Obscured by drug-induced side-effects common during the early neuroleptic era, spontaneous movement abnormalities have only recently been understood as a medication-independent phenomena, occurring across the life span in patients with schizophrenia and other psychotic disorders.1–3 This shift began with conceptualizations that schizophrenia involves a developmentally moderated neuropathological process,4–5 and continued as scientists integrated this new paradigm into their interpretations of movement research.4,6,7 One of the most exciting products to come from the recent acceleration in research is support for a link between movement abnormalities and psychotic disorders, presumed to reflect common neural mechanisms that influence both motor functions and vulnerability to psychotic symptoms.4,8–13 Another very important development in the field of schizophrenia research and intervention involves prospective longitudinal studies designed to following individuals at high-risk for schizophrenia (i.e. individuals with a schizophrenic parent; individuals with schizotypal personality disorder (SPD); individuals with a prodromal syndrome). These studies are designed to

M.S. Ritsner (ed.), The Handbook of Neuropsychiatric Biomarkers, Endophenotypes and Genes, © Springer Science + Business Media B.V. 2009

239

240

identify and treat individuals with the prodrome, follow the early course of illness, and potentially ameliorate or prevent the course of illness.14–19 Recent evidence has indicated that as many as 35% of clinical high-risk individuals (i e., individuals identified as prodromal base on subclinical symptoms) will develop a psychotic disorder, such as schizophrenia, within a two-year period of index.14,20,21 Given the possibility that interventions applied during this period may ameliorate or even prevent the onset of illness,17 attention to understanding and identifying this early period could advance our understanding of the prevention of psychotic illnesses. However, it is also important to consider that less than half of the individuals who are identified as prodromal actually go on to develop schizophrenia or a spectrum disorder.14,20,21 We must, therefore, achieve greater power in the prediction of psychosis before we can conduct large scale trials of the preventive use of antipsychotics, given their considerable potential side-effects (e.g., weight gain, diabetes). Thus, recognition of putative markers that can serve to refine identification and enhance predication has become a priority for the field. In the present chapter we examine the potential of movement abnormalities as one such marker. As noted, an accumulating body of evidence suggests an overlap between circuits that give rise to movement abnormalities, and brain regions commonly implicated in the pathogenesis of schizophrenia. More specifically, it has been proposed that both hyperkinesias and psychotic symptoms are due to overactivation of compromised dopamine pathways, specifically, the striatal pathway mediated by the D2 receptor subtype.4,8,10,22–24 Further, there is evidence that movement abnormalities manifest a developmental trajectory that converges with changes posited by neurodevelopmental models of schizophrenia.5,25–27 In the present context, this is particularly important because (1) movements occur before the onset of symptoms4,6,8,28–30 and could therefore serve as potential markers, and (2) movements are likely to change as a function of normative and abnormal neurodevelopmental processes commonly indicated in schizophrenia, highlighting a potential of movement to serve as a changing “barometer,” reflective of progressing neuropathology. Taken together, this evidence suggests the potential for movement abnormalities to serve as a viable biological marker. In the following sections, we provide a background for understanding subtypes of movement abnormali-

V.A. Mittal and E.F. Walker

ties, underlying neurological mechanisms, and movement abnormalities within the context of a neurodevelopmental conceptualization of schizophrenia. We use these sections to provide a framework for reviewing and integrating studies of movement abnormalities in high-risk populations. These data are then evaluated to determine the practicality and viability of movement abnormalities as a potentially viable biomarker. We conclude the chapter by discussing new areas of emphasis for future biomarker research.

History and Typology Abnormalities in movement function have been observed in schizophrenia patients since the inception of the diagnosis.31 However, when compared to other lines of investigations (e.g. cognitive, and socioemotional functioning), there have relatively few studies of movement deficits associated with schizophrenia and spectrum disorders. This relative neglect is in large part the result of confusion surrounding the etiology of movement abnormalities. Historically, abnormalities in movement have been the concern of neurology, whereas abnormalities in socioemotional behavior have been the focus of psychiatry and clinical psychology.4 While the medical field has long held the conception that movement disorders have an etiology based in nervous system dysfunction, acceptance of the link between CNS dysfunction and disturbances in mood and thought has only recently began to receive support.5 In addition to a stifling effect resulting from rigid intradisciplinary boundaries and erroneous historically rooted etiological assumptions, the advent of antipsychotic medication in the 1950s contributed further to the confusion about movement signs in psychiatric populations. Because neuroleptic medication produced characteristic movement disorder in some patients (i.e. extrapyramidal side-effects known as tardive dyskinesia), drug induced movement disturbances became a primary focus of research attention. This narrow focus led scientist to relatively ignore the spontaneous movements that occur in schizophrenia patients regardless of medication history.1,7,32 The following sections aim to clarify ambiguity around terminology and classification, and provide a framework with which to evaluate literature presented in the remainder of the chapter.

17

Movement Abnormalities: A Putative Biomarker of Risk for Psychosis

241

Table 17.1 Movement abnormalities Appearance

Representative Disease

Hypokinesia (associated with decreased DA activity in striatal pathways) Parkinson’s Disease Akinesia – poverty of movement and movement initiation. Bradykinesia – slowness of movement. Rigidity – uniform increase in resistance to passive movements about individual joints. Tremor – rhythmic, involuntary muscular contraction characterized by oscillations of a part of the body. Hyperkinesia (associated with increased DA activity in striatal pathways) Huntington’s disease Choreoathetoid – brief, irregular contractions that are not repetitive or rhythmic, but appear to flow from one muscle to the next. Ballistic movements – violent or flinging movement. Dystonia – sustained muscle contractions cause twisting and repetitive movements or abnormal postures. Stereotypy – Movements of frequent repetition that serve no obvious purpose.

Movement Disorders: A Typology Disorders of movement can be categorized into two broad categories, deficits in motor proficiency and movement abnormalities. There are two distinct types of movement abnormality commonly contained under the umbrella term dyskinesia: hypokinesias and hyperkinesias. Hypokinesias comprise akinesia (i.e., poverty of movement and movement initiation), bradykinesias (i.e., slowness of movement), rigidity (i.e., uniform increase in resistance to passive movements about individual joints), and tremor (rhythmic, involuntary muscular contraction characterized by oscillations of a part of the body). These movements are associated with dopamine depletion disorders (i.e., Parkinson’s disease), but can also been seen in schizophrenia patients with long-term use of traditional antipsychotic medications (see the below section for a discussion). In contrast, hyperkinetic movements include chorea (brief, irregular contractions that are not repetitive or rhythmic, but appear to flow from one muscle to the next), athetoid movements (writhing movements of the limbs, fingers or face), ballismus (violent or flinging movement), dystonia (sustained muscle contractions causing twisting and repetitive movements or abnormal postures) and stereotypy (movements of frequent repetition that appear to serve no purpose). The term ‘choreoathetoid’ is used to refer to combined chorea and athetoid movements, as they often co-occur. Hyperkinetic movements are associated with increased DA activity and are commonly observed in patients receiving DA agonists

(i.e., l-dopa, stimulants) and in diseases presumed to be linked with heightened striatal DA (e.g., Huntington’s Chorea, schizophrenia). Please see Table 17.1 for description and definitions of the noted movement abnormalities. The movement abnormalities that have been observed in untreated patients with psychosis have been referred to as “spontaneous dyskinesias”, to distinguish them from the movement side effects attributable to antipsychotics, and are typically of the hyperkinetic variety described above (e g., choreoathetoid, repetitive and ballistic movements). As noted these movements are the product of the overactivation of DA neurocircuitry.33 In contrast, the movement abnormalities induced by DA antagonists often take the form of hypokinesias and bradykinesias, and parallel those observed in DA depletion disorders.34 In tardive dyskinesia, the most common syndrome induced by antipsychotics, the abnormalities are most common in the oral region, and less often in the extremities.1 It is important to note that the picture is not always as clear-cut, particularly so when neuroleptic medications are introduced. It is not uncommon for antipsychotic “rebound” effects to result temporary heightened DA activity and subsequent fleeting hyperkinetic movements (e.g., if antipsychotic treatment is suddenly suspended); this phenomena points to the importance of longitudinal movement observation. Further, animal models have shown that DA blockade in different neural regions can result in concomitant hyperkinetic and hypokinetic movements.35

242

This chapter is primarily focused on the hyperkinesia subtype of movement abnormality. First, hyperkinesias are of particular interest because they are relatively unique to schizophrenia and spectrum disorders.36,37 Second, hyperkinetic abnormalities are characterized by developmental changes that converge with neurodevelopmental models of schizophrenia.5,27 This allows the potential for a dynamic biological marker, which precedes symptom onset, and then increases throughout the progression of illness. Third, it has been proposed that both hyperkinesias and psychotic symptoms are due to overactivation of compromised dopamine pathways, specifically, the striatal pathway mediated by the D2 receptor subtype.4,23–24 Finally, as a practical consideration, hyperkinetic movement abnormalities are relatively easy to measure, and thus hold the potential to serve as risk indicators that, in combination with other behavioral and biological measures, may enhance prediction of psychiatric outcome.38

Neurological Mechanisms Neural Substrates Implicated in Movement Abnormalities Because movement abnormalities have historically been studied in clinical disorders within the boundaries of neurology (e.g., Parkinson’s and Huntington’s disease), general understanding of the neural processes underlying movement is well developed. In fact, the most researched cortico-subcortical circuit is the “motor circuit”.22 Although a comprehensive survey of the neurophysiology underlying movement disorders is outside the scope of the present chapter, a brief description is provided as familiarity with a basic model will be useful when considering the implications of an overlapping mechanism with psychosis. Although some degree of uncertainty remains to as if the degree of segregation between circuits or if they are truly close or partially open, it is generally accepted that the basal ganglia comprise a series of segregated circuits that tie into the cerebral cortex and the thalamus.22 Each circuit originates in specific cortical areas, passes though portions of the basal ganglia and thalamus and then projects back into the frontal cortical area of origin (i.e., in closed loop); for each of these

V.A. Mittal and E.F. Walker

segregated circuits, the striatum serves as the input stage and the internal segment of the globus pallidus (GPi) and the parts reticulata of the substantia nigra (SNr) serves as the output station to the thalamus and brainstem.22–24 Among this series of circuits (e.g., oculomotor, limbic, associative) the motor circuit is highly relevant to movement disorders. At the cortical level, it comprises pre and post central sensorimotor areas.22 At the subcortical level, it includes sensorimotor areas in both the basal ganglia and ventroanterior/ventrolateral (VA/ VL) thalamus.22 The cortical projections terminate largely in the putamen where output is directed toward the globus pallidus (GPi) and the substantia nigra (SNr) via two distinct pathways: a direct monosynaptic inhibitory pathway and a net excitatory polysynaptic indirect polysynaptic pathway via the external pallidal (GPe) segment and STN.22 GPe output reaches the GPi/SNr both directly and via the subthalamic nucleus. Basal ganglia output is then directed at the thalamus.22 GPi efferents reach the nucleus ventralis oralis anterioris and SNr output reaches the magna cellular portion of the nucleus anterioris.22 Collaterals of basal ganglia projections are also directed to the pedunculopontine nucleus (PPN).22 As noted, the present chapter focuses on hyperkinetic movement abnormalities, which are characterized by reduced basal ganglia output to the thalamus, leading to disinhibition of the thalamocortical neurons. This disinhibition in turn leads to the development of involuntary movements. For example, during the early stages of Huntington’s disease striatal output projections to the GPe are affected; this can lead to reduced inhibition of neurons in the GPe, and subsequent increased inhibition of the STN neurons.2,39 The resulting decreased GPi output underlies hyperkinetic movements.2,39

The Role of Dopamine in a Movement Circuit It is important to note that dopamine in the striatum is assumed to be responsible for balancing the activity of both the noted direct and indirect pathways which comprise the motor circuit. The dopaminergic input to the striatum arises in the substantia nigra pars compacta (SNc). The release of DA from terminals in this area facilitates transmission over the direct pathway (via D1

17

Movement Abnormalities: A Putative Biomarker of Risk for Psychosis

receptors) and inhibits transmission over the indirect pathway (via D2 receptors). Therefore, the overall effect of striatal DA release is a reduction in basal ganglia output from the GPi/SNr; this in turn leads to increased activity of the thalamocortical projection neurons.22–33 As noted above, this reduction in basal ganglia output is assumed to be responsible for hyperkinetic movement. Therefore, increases of striatal dopamine are directly associated with increases in hyperkinetic movement. In contrast, disorders related to dampened levels of dopamine (e.g., Parkinson’s disease) show an increase in the basal ganglia output to the thalamus; this results in a decrease in the activity of thalamocortical projection neurons and subsequently the hypokinetic movement (i.e., bradykinesias, rigidity, tremor).22,40 Thus, schizophrenia, a disease proposed to be associated with heightened striatal dopamine, manifests heightened rates of hyperkinesias.33 Irregularities of corticostriatothalamocortical (CSTC) circuits, basal ganglia, and dopaminergic activity are also responsible for abnormal movements seen in other disorders, such as the tics associated with Tourette’s syndrome; in congruence with the theory that these brain regions are responsible for involuntary movements, syndromes like Tourette’s are also treated with neuroleptic drugs designed to dampen dopamine transmission.41 The relationship between striatal DA, movement, and psychotic symptoms can be demonstrated by studies examining compounds, such as levodopa, that are used to treat Parkinson’s disease by increasing DA transmission. As noted, movement abnormalities associated with Parkinson’s disease (i.e., hypokinesias; slow jerking movements, rigidity) are related to low levels of dopamine characteristic of the disease. However, researchers have observed that patients with Parkinson’s disease, who are being treated with dopamine agonists (i.e., levodopa induced elevated striatal DA) show drug-induced dyskinesias,34 and in some cases, psychotic symptomatology.42 In a similar vein, researchers have observed that other amphetamines, such cocaine, increase dopamine activity, and can cause both hyperkinesias and psychotic symptoms.43 The dopamine irregularity is also supported by drug challenge studies that examine dopamine activity in areas hypothesized to be associated with dyskinesias and prodromal or high-risk symptoms. For example, researchers using PET (positron emission tomography) noted that in response to an amphetamine challenge, high-risk participants showed abnormalities in

243

binding to D2 receptors in the striatum when compared to matched non-psychiatric controls.44 This indicates that like schizophrenia, high-risk periods preceding the illness are associated with DA abnormality in the striatum. In a related study, researchers found that in response to dextroamphetamine challenge to the dopaminergic system, non-psychiatric participants showed an increase in movement abnormalities.12

Potential Overlapping Neural Substrates for Movement and Psychosis As discussed above, hyperkinetic movements are hypothesized to be due to overactivation of compromised dopamine pathways, specifically, the striatal pathway mediated by the D2 receptor subtype.23–24,39 It is also possible that excess activation of DA neurons projecting from the ventral tegmental area, to the accumbens and caudate area of the striatum result in the psychotic symptoms observed in patients with schizophrenia.45,46 Recently, researchers have began to suspect that cortico-striato-pallido-thalamic circuit malfunction, moderated by dopaminergic function, is responsible for schizophrenia symptoms including motivation and cognitive functioning,47 as well as both positive and negative psychotic symptoms.4,8,13 Walker(1994)4 suggests that the abnormalities in the basal ganglia may be involved in the generation of dyskinesias and schizophrenia symptoms. The basal ganglia are linked with forebrain structures that are involved in planning, goal-directed behavior, and monitoring intentions. Irregular neuroactivation, and/or compromise structural/neural circuitry in the basal ganglia could lead to schizophrenia symptomatology. For example, if the striatum plays a role as a goal selector for potential adaptive processes,48 then a disturbance in this subcortical region might result in a “disconnect” between the goals represented in the prefrontal cortex and the basal ganglia’s selection of a particular response. In other words, an afflicted individual might experience this disconnection as a feeling that they are not in control of their own behavior or motivations.38 Consistent with this hypothesis, Moller and Husby49 characterized the initial prodrome of psychosis in first episode schizophrenia as self-disturbances that were related to losing control of cognitive and affective experiences. Another possibility is that

244

because cortico-striato-pallio-thalamic pathways play a role in linking motor and emotional pathways,50 motor functioning is potentially linked with symptoms relating to emotion and insight.51

Development and Movement Abnormalities Biological Risk Markers: A Neural Diathesis Stress Perspective The neurodevelopmental view of schizophrenia, within a diathesis-stress framework, underlies high-risk research, an area concerned with identifying “biobehavioral markers” of the predisposition to schizophrenia (such as movement abnormalities) and then using these markers to intervene and, it is hoped, prevent the expression or ameliorate the course of the schizophrenia illness. The starting point of the model is with a genetic vulnerability, some type of environmental insult, or a combination of both, that occurs very early in development and leads to a biological susceptibility or vulnerability to schizophrenia.5,25,27,52 A critical aspect of this model, following Weinberger5 and others, is that while the underlying pathophysiology causing the susceptibility (sometimes referred to as the “lesion”) occurs very early in development, behavior does not become affected until much later. The lesion remains dormant until triggered by some developmental event, at which point the clinical symptoms start to emerge. It is assumed that susceptibility takes the form of one or basic brain abnormalities. For example, demonstrated structural abnormalities include enlarged ventricles or reduced temporal lobes; abnormal DA activity is a well-established biochemical abnormality.45–46 There is also considerable evidence from PET and fMRI studies suggesting functional impairments of the brain in schizophrenia. Hypofrontality (i.e., under activity) of the frontal lobes, especially in the dorsolateral prefrontal cortex, for example, has been reported by Weinberger and his colleagues.53 The brain abnormalities are, in turn, thought to be associated with a variety of neurocognitive deficits. In this model, these deficits are called “biobehavioral markers”. It is important to note that by definition, a biobehavioral marker does not require information about

V.A. Mittal and E.F. Walker

the location of a disease gene(s). Instead, it indicates the presence of the biological susceptibility for schizophrenia, and provides information about the pathophysiology of the disorder.54 This is not to say that an endophenotype cannot also be used in the function of a biobehavioral marker. As discussed below, movement abnormalities might indeed be both indicative of a disease gene, and also serve as marker of susceptibility.8 The neurocognitive deficits are labeled “biobehavioral” because they are considered to be intermediate between basic brain functions (the “bio”) and more complex clinical features of the illness (the behaviors). They consist of such variables as the ability to process information in the environment (i.e., attention) or to move the eyes smoothly when tracking a moving target. Since they are abnormalities measured on the behavioral level, they are phenotypic (as opposed to genetic) markers. The “markers” part of the label refers to the notion that, if valid, the deficits signal or “mark” the presence of the biological susceptibility to schizophrenia. The model also stresses another critical neurodevelopmental assumption: the marker deficits can be detected before clinical symptoms begin to emerge. These markers can therefore serve as predictors of future illness, paving the way for future intervention programs. That is, once markers have been validated, they can be used to identify individuals with a true susceptibility to illness. This could provide a way to screen at-risk and other populations for individuals most in need of intervention. In addition, because the marker deficits are thought to reflect the pathophysiology of schizophrenia, they may point to the most effective types of interventions. The goal of such programs is to interrupt the developmental pathway and to prevent the clinical expression of illness, or, at minimum, to lessen its severity.25 It is clear that in any model, the developmental pattern most characteristic of schizophrenia must be explained; that is, why the onset of schizophrenia typically occurs in late adolescence/early adulthood. There are numerous developmental changes in the CNS system during puberty and early adulthood and our recent understanding of the interplay between neuroendocrine function and neurotransmission has significantly contributed to integrating developmental complexities into the present model. As noted, there is evidence that some cases of psychotic disorders involve an abnormality in striatal DA

17

Movement Abnormalities: A Putative Biomarker of Risk for Psychosis

receptors, either in the regional distribution, density, affinity, subtype ration, or an interaction that results in a heightened sensitivity to DA.4,26–27,55 The Hypothalamic-Pituitary-Adrenal axis (HPA) governs a hormonal cascade that has important effects on brain function and facilitates both physiological and behavioral responses to threats. Walker and colleagues26,27 have proposed that this biological diathesis can be moderated by the HPA axis due to the augmenting effects of cortisol on DA activity. From a neural developmental framework, this is particularly relevant; the gradual normative increase in HPA activity throughout adolescence56,57 may trigger this cycle in individuals with a pre-existing constitutional vulnerability (i.e., hippocampal abnormality, striatal abnormality, innervated DA pathways). In other words, during adolescence, normatively elevated cortisol might interact with a constitutional vulnerability (the result of prenatal insult and/or genetic factors), as well as normative and abnormal neural maturation (e.g., synaptic pruning) thus augmenting DA (expressing itself as behavior problems, prodromal symptoms or eventually Axis I psychotic disorder). This augmented DA reciprocally affects cortisol levels, thus heightening stress responsivity. These biological mechanisms are hypothesized to collectively result in a stress cascade and in the clinical expression of the individual’s constitutional vulnerability to psychosis.26,27 This phenomenon, in part, may explain the escalating behavior problems observed in preschizophrenic participants and the modal age at onset of the prodrome in late adolescents.58 Another normative neuromaturational process is likely to co-occur and potentially interact. In healthy development, the peak of synaptic density occurs at 2 years of age and is followed by a slow drop during childhood followed by a steep decline in adolescence.59,60 Early in development many synaptic connections are superfluous and the increase in synaptic pruning occurring during adolescence, particularly in the prefrontal brain regions, is thought to be related to efficiency and adult reasoning capabilities.59,61 McGlashan and colleagues62 proposed that an abnormally aggressive synaptic pruning process, which would lead to a reduction in synaptic connectivity, may be a contributing factor to psychosis. This idea is supported by postmortem studies, which have shown reduced neuropil, representing a loss of connections between neurons, without neural loss.63 In continuing

245

within the framework of this model, early lesions in a particular neural structure (e.g., the hippocampus) would, in effect, reduce the amount of synaptic pruning necessary to pass a threshold of synaptic loss, resulting in psychosis.64 At the outcome side of the model, it can be seen that the diathesis can have a number of alternate expressions, ranging from the full-blown illness (sometimes called “core” schizophrenia) through a number of less severe variants of it (e.g., personality disorders or schizophrenia-related features) to no behavioral expression. It has been proposed by several researchers that the diathesis is most likely to be expressed clinically in the form of a nonpsychotic, schizotypallike personality disorder, with some type of environmental stressor required to trigger psychosis.65,67 It is further assumed that a variety of additional factors (environmental, biological, cognitive, personality) can serve as moderators, either exacerbating or reducing the severity of the expressed illness.

Movement Abnormalities and the Lifetime Course of Schizophrenia Fish62 proposed that “pandysmaturation”, disorganized motor and sensory-motor development, reflects this defect during the first 2 years of life and can, therefore, serve as an early marker. In her long-term prospective study of a small group of high-risk offspring,6,68 she found pandysmaturation in infancy and early childhood to be highly predictive of subsequent schizophrenia spectrum disorders. Further support for a relationship between early neurological abnormalities and schizophrenia has been provided by both Walker and associates69 and Jones and collegues.70 In both studies, motor dysfunctions and delays in early childhood (up to approximately 2–3 years of age) were found to characterize children who later developed schizophrenia. In their study, Walker69 used home movies taken during childhood to compare preschizophrenic children (i.e., those who became schizophrenic as adults) with their healthy siblings, preaffective disorder subjects, the healthy siblings of patients with affective disorder, and subjects from families with no mental illness. Results of the study indicated that preschizophrenic children were significantly differentiated from controls on a number of motor dysfunctions, including

246

a variety of dyskinetic signs (e g., motor overflow; “associated” involuntary movements), abnormal hand posture, choreoathetoid movements, and abnormal tonicity). Similarly, data from the Northern Finland 1966 Birth Cohort showed that compared to the rest of the cohort, children who later developed schizophrenia were delayed in the attainment of motor milestones at age one.71 Because infancy is a period in which motor skills are in ascendance, individual differences are more easily detected during this period. However, the significant differences appear to fade during early childhood.4 This may be because a potential ceiling effect for mastery of motor skills during middle childhood renders the observable signs of motor dysfunction less pronounced.69 It is important to note that, although motor dysfunction may be less apparent with development, prospective research shows that preschizophrenic children continue to manifest motor abnormalities,30 and as many of 75% of high-risk offspring do so.28 It has been further hypothesized that during adolescence, neuromaturational processes trigger the behavioral manifestation of latent vulnerability, and movement abnormalities may again become more pronounced.38 In one study, researchers observed that adolescents with SPD exhibit elevated rates of movement abnormalities in comparison to normal and psychiatric controls.29 Schiffman and colleagues72 examined videotaped social interactions of 11- to 13-year-old children who later developed schizophrenia. The authors found that a high occurrence of movement abnormalities distinguished the pre-schizophrenia children from matched controls.72 As the present topic surrounds biological markers of risk during this period, a more detailed account of recent developments in the high-risk literature will be addressed in a following section. Movement abnormalities continue to manifest past the prodromal period, and are associated with psychiatric disorder in adulthood. Both clinical exams and laboratory measures of motor speed, proficiency, steadiness, rhythmicity, and involuntary movement yield evidence of motor dysfunction in patients of all ages. Retrospective research (using pre-neurolepticera archives) and some recent studies of nonmedicated subjects indicate that movement abnormalities, in particular dyskinesias, often accompany schizophrenia.73 For example, investigators have found that adult

V.A. Mittal and E.F. Walker

patients with schizophrenia show more movement abnormalities than healthy matched controls74; this does not appear to be a function of neuroleptic treatment. A prospective study of first-onset patients with schizophrenia compared treatment naïve and neuroleptic treated patients and found that both groups exhibited a high level of dyskinesias.1 Similarly, Gervin and colleagues75 also observed spontaneous dyskinesias and involuntary orofacial movements in first episode schizophrenia patients who were neuroleptic naïve, or had received treatment for less than 1 month. Screening of medical records revealed that about 23% of nonmedicated schizophrenia patients manifested readily observable dyskinesia.7 A similar study of records from the United Kingdom observed the prevalence of movement abnormalities as high as 33%.76 Given these and other findings, Khot and Wyatt2 argue that some of the movement disorder attributed to neuroleptics in medicated patients may have been preexisting. Consistent with the above findings, research on both child77 and adult subjects with SPD12,78 also reveals an elevated rate of motor dysfunction. Researchers have observed that the prevalence of movement abnormalities in patients with schizophrenia dramatically increases with advanced age. In a cross-sectional study of individuals with chronic schizophrenia, researchers found that as age increased, the prevalence of involuntary movements approaches 100%. Research have found that the prevalence across age groups was 63% (< 65 years), 80% (65–75 years, and 93% (> 75 years).79 However, it should be noted that samples including elderly individuals with schizophrenia untreated with first generation antipsychotics are very scarce, and the association between continued exacerbation of spontaneous movement abnormalities as a function of advanced age is not entirely understood at the present time.80

Movement Abnormalities Within the Neural Diathesis-Stress Framework As noted, this longitudinal course is of particular interest because it mirrors what would be expected given a neural diathesis-stress formulation of schizophrenia and psychotic disorders.27 Investigators have examined movement abnormalities in a neurodevelopmental framework, under the assumption that schizophrenia

17

Movement Abnormalities: A Putative Biomarker of Risk for Psychosis

and other psychotic disorders involve a developmentally moderated neuropathological process.5,27 Genetic and prenatal complications occurring in the prenatal and perinatal periods interact resulting in one or more organic or constitutional vulnerabilities.52 With the case of movement abnormalities, evidence of one such constitutional vulnerability (striatal abnormalities) is evident early in infancy.4,6 However, these movement abnormalities become less noticeable as a milestone ceiling is reached in early childhood4 and may become more pronounced in adolescence, when a series of interacting normative and abnormal maturational processes occur.8 Walker4 proposes that the functionally segregated neural circuits that include the striatum mature at different rates and as a consequence, the behavioral effects of striatal dopamine abnormalities will change with development. This notion may provide a basis for explaining the presence of premorbid abnormalities, as well as the modal age at onset in late adolescence.4 As noted above, normative processes such as synaptic pruning and elevated cortisol interact with early lesions and may augment preexisting DA abnormalities.8,27,62 Integration of movement into this model is directly supported by recent research focusing on prodromal adolescents, that has noted that movement abnormalities are associated with other risk indicators, including MPAs (a proxy for early lesion) and elevated salivary cortisol (indicative of heightened HPA activity).10 The resulting increases in striatal DA activity may first present as increased movement abnormalities and prodromal symptomatology, and eventually, as psychotic symptomatology.9,23–24 As the disease progresses with time, the movement abnormalities continue and eventually increase with advanced age.79

Movement Abnormalities and High Risk Populations A Brief History of High-Risk and Early Intervention Research A cursory understanding of the history of high-risk research is important for framing the role of movement abnormalities as a biological marker of risk. Because

247

the risk for schizophrenia in the general population is about 1%, if of 1,000 youngsters were randomly selected and then closely followed for, say, 25 years, only about ten subjects could be expected to develop schizophrenia. Consequently, researchers turned to a population with higher than average risk for developing schizophrenia: the offspring of schizophrenic parents, who are about 12 times as likely to become ill as members of the general population. Pearson and Kley81 were the first to recognize the advantages of a prospective design that involved the offspring of schizophrenic parents. Their proposal was followed by the pioneering work of Mednick, who, in 1962, initiated the first offspring high-risk study on a large scale,82 and then by several additional high-risk studies that were begun between the late 1960s and early 1970s. These have been referred to as the “first generation” of high-risk studies. Many of the first-generation high-risk (HR) projects focused on detecting subtle neurocognitive deficits in the young at-risk offspring under study.83 Observations that the majority of individuals who succumb to schizophrenia and other psychotic disorders manifest prodromal signs of behavioral disturbance served as a foundation for a new generation of high-risk studies.84 These signs usually begin in early adolescence and become progressively worse as the individual approaches young adulthood.85 It is estimated that at least 70% of preschizophrenic people manifest behavioral dysfunction during adolescence,58,86 with many showing schizotypal signs, such as social withdrawal and thought abnormalities,87 deficits in memory and executive function,88 and neurological soft signs.38 Many early studies in this second generation of high-risk research followed individuals with SPD; this proved a productive strategy as a fair portion of adolescents with SPD eventually developed schizophrenia,48,89 researchers could maintain international continuity when discussing research, and finally, in the early stages of this field, as DSM-IV category was largely a requisite for funding. As the field continued to develop, the high-risk concept was empirically refined to increase prediction and allow for a broader and more accurate conceptualization of symptoms.90 For example, research from a large collaboration of American prodromal research groups (North American Prodromal Longitudinal Study, NAPLS)14 recently indicated that 35% of individuals who meet criteria for a prodromal syndrome (exhibiting a particular

248

threshold of severity and frequency of pre-psychotic symptoms/ showing a large drop in global functioning accompanied by genetic risk or SPD diagnosis90) have been shown to develop a psychotic disorder within 2 years of index.14 The high-risk period (comprising the late adolescent years immediately preceding modal onset age) represents both a viable point for intervention, and a developmental period with high potential to shed light on the etiology of schizophrenia.91 While the idea was first introduced by Sullivan,92 factors such as a negative prognostic attitude towards schizophrenia, the prominent role of the institution (prior to deinstitutionalization during the late 1980s), and lack of a working neurobiological model contributed to a relatively late development of this concept.93 This changed with the advent of first-episode research during the 1980s that highlighted a better prognosis,94–97 a neural-diathesis stress conceptualization of illness,27 and the introduction of atypicals, which served as evidence for an improved life course. Beginning with a handful of landmark studies,98–101 this field has shown tremendous growth and promise, and now includes approximately 200-research and treatment centers worldwide.102 Ongoing longitudinal studies hold significant promise for elucidating etiology, increasing the complexity of our developmental understanding, and highlighting viable risk-markers that may potentially enable preventive intervention. Within this context, the use of biological markers is rapidly become an international research priority.

Movement Abnormalities in High-Risk Populations A growing body of literature suggests that, independent of psychotropic medications, high-risk individuals with demonstrate regionally specific abnormalities that are consistent with those observed in schizophrenic populations. As noted above, neurodevelopmental view of schizophrenia is the foundation of high-risk research and a major focus of risk researchers is on the identification of developmental abnormalities that act as warning signals of future illness and therefore make intervention possible. The studies reviewed below represent available empirical investigations

V.A. Mittal and E.F. Walker

of movement abnormalities in high-risk populations, and are presented in this framework.

Group Differences in Movement Walker and colleagues29 coded movement from videotaped structured interviews of adolescents with SPD (n = 15), age/sex-matched controls with other personality disorders (n = 16), and an additional demographically matched group with no personality disorder (n = 18). Trained raters (alpha ≥ 0.80) coded four 6-min segments for a total of 24 min of the Structured Clinical Interview for Axis I DSM-IV Disorders (SCID)103 (i.e., the first 6 min, 15–20, 30–36, and the last 6 min). The coding scheme was devised to obtain measures of abnormal movement in seven body regions including (e.g., head, trunk, limbs). Raters coded movements as involuntary if they occurred with no apparent intent or goal (e.g., writhing and alternating extensions and flexions of the fingers or wrist, facial grimacing, foot tapping). Voluntary movements were defined as any movement intended to convey meaning (e.g., nodding, gesturing) or achieve a purpose (e.g. reaching for an object, scratching rearranging clothing, repositioning). Consistent with observations of schizophrenia patients, in reviewing the entire body, the researchers observed that the high-risk group showed significantly higher frequencies of hyperkinesias, specific to the facial and upper-body regions.29 The authors also found that there were no diagnostic group differences in the rate of voluntary movements. This second finding is particularly important because it demonstrates that the involuntary movements seen in the high-risk group were not simply a function of global movement abnormality. In support of this finding, Mittal and colleagues104 observed a different group of adolescents with SPD (a second cohort), and noted that that while adolescents exhibited significantly fewer gestures that accompany speech, this was not the case for overall movements when compared to normal and psychiatric controls. Taken together, these results suggest that the observed movement abnormalities represent a specific subtype rather than a global deficit in movement. One important factor shared by both studies was that psychotropic medication was limited or entirely absent. During the period these studies were conducted, psychotropic medications were considerably less common in the

17

Movement Abnormalities: A Putative Biomarker of Risk for Psychosis

treatment of youngsters. As a result, these studies provide a rare and likely unrepeatable view into movement behavior patterns in prodromal populations. In a related investigation, Cassady and colleagues12 compared rates of dyskinetic movements in adults with SPD (n = 17) and schizoid personality disorder (SCD) (n = 15) and normal controls (n = 22). As in the Walker29 study, trained raters, blind to diagnostic status (alpha ≥ 0.80), coded video taped structured examinations for involuntary movements. The raters coded movements with the Maryland Psychiatric Research Center Involuntary Movement Scale, a measure originally developed for tardive dyskinesia. Researchers observed that SCD patients manifested significantly more dyskinetic movements than controls with SCD as well as controls with no personality disorder.12 Twenty-four percent of the SPD group showed dyskinesias, most frequently, side-to-side and undulating tongue, lip puckering, facial contraction, lateral jaw deviation, and twitches in the arm or shoulder. Dyskinetic movement scores were not associated with age, gender, and history of major depression. However, one significant limitation to this early research was that the study did not control for psychotropic medication. Region specificity. One consistent finding among this series of studies is that movement abnormalities appear more common in the facial and upper-body regions. It is possible that the current methodology, where individuals were seated during the assessment of movement abnormalities, may be confounding assessment by limiting movements in the lower extremities. Indeed, the noted earlier study27 also used a coding procedure with seated individuals. However, other reports examining movement abnormalities have also noted a trend for more pronounced movement abnormalities in facial and upper-body regions. For example, in a report examining movement abnormalities in adults with psychosis, utilizing a coding method in which participants were not seated, researchers observed abnormalities in the facial region roughly three-times more often and abnormalities in upper-body roughly twice as often than movements in the lower extremities.1 One potential explanation for the region differences lies in the neurological basis of movement abnormalities; because there are topographically organized motor sub-circuits for different body regions, it is possible for movement abnormalities to be limited to one area. Circuits underlying the facial and upper-body regions are in the ventral medial area of the putamen whereas the

249

circuits for the lower-body region are in the dorsal lateral putamen.4 Although this is highly speculative, it is of interest that the regions of the most salient movement abnormalities exhibited by the SPD are regulated by neural circuits adjacent to the hippocampus whereas the dorsal lateral putamen (underlying lower-body movement) is the furthest away.4 As discussed in the proceeding sections, this structure is hypothesized to plays a key developmental role in vulnerability for psychosis. The relationship between movement circuits and planning centers in the striatum and other brain structures is also a noteworthy consideration; motor circuits underlying facial and upper-body movement are tied into other neurological regions implicated in the etiology of psychosis. More specifically, nonverbal behaviors such as gesturing and facial expression involve a complex coordination between motor circuits looping through the limbic system, and communication and language centers such as Broca’s center.104 As noted, abnormalities in expressive nonverbal communication are highly prevalent in psychosis.104 This relationship may help to explain the null findings with regard to the lower body region, which is not tied to communication or language centers. Because the systems underlying facial and upper-body regions involve a greater degree of complexity and coordination between several structures also implicated as dysfunctional in psychosis, these circuits may be more impervious to dysfunction. Overall, this pattern of findings suggests that the facial and upper-body movement abnormalities may be indicative of the resulting interaction between several dysfunctional neural circuits. However, future research is necessary to elucidate the direction of interactions between these hypothesized regions of dysfunction in relation to both movements and deficits in expressive nonverbal behavior. In the following section, we discuss data from a longitudinal study that indicate lowerbody movement abnormalities emerge during the later progression of the prodromal period.

The Progression of Movement Abnormalities Over Time In a series of recent studies,8–11 researchers examined the incidence of movement abnormalities among adolescents during a 3–4 year period; at baseline 42 met criteria for SPD, 30 met criteria for an Axis II, Cluster

250

‘B’ disorder (other personality disorder – OPD group), and 49 did not met criteria for any DSM IV disorder (normal comparison – NC group). Following earlier studies,27 trained raters blind to diagnostic status coded structured interviews of participants, without audio track. In addition, medication status was included as a covariate and each analysis was also conducted on the subsection of participants not using medication. In the initial study, we utilized a mixed ANCOVA design to examine differences between groups over three annual time points (baseline, follow-up 1, and follow-up 2). Findings of baseline group differences were highly consistent with early literature27; individuals with SPD uniquely exhibited elevated movement abnormalities in facial and upper-body regions but not lower body regions. Analyses of the facial and upperbody region revealed that the SPD group remained significantly elevated in comparison to both controls for each of the three time points and for this group alone the changes in movement score between the three time points were highly significant. Movement abnormalities remained relatively constant for the NC and OPD group across time periods for each of the regions. Analyses of the lower-body region revealed that the SPD group was not different from controls at baseline, but became significantly elevated in comparison to both comparison groups at follow-up. Walker4 suggested that one interpretation of the overall pattern of findings is that vulnerability for psychosis involves striatal abnormalities that can be initially manifested as movement abnormalities during childhood and adolescence, then, as frontal-striatal circuitry matures through late adolescence, striatal dysfunction is manifested as prodromal and eventually psychotic symptoms. The later onset of movement abnormalities in the lower extremities may reflect that dysfunction in the dorsal putamen, the region responsible for hyperkinetic movement in the lower limbs, may occur later in the course of illness. Medication covariates. It is also noteworthy to mention stimulants were a significant covariate in this analysis. For the entire sample, individuals being treated with stimulant medication developed significantly more upper-body movement abnormalities overtime. This fining may reflect the relationship between stimulant medication and dopamine activity. Because the movement abnormalities in question are moderated by DA activity,8,22 it is possible the increase in movement observed in the present study reflect a stimulant

V.A. Mittal and E.F. Walker

induced increase in DA. Because DA activity in the striatal region is also closely tied with psychotic symptomatology,13,45 this trend is a serious concern. The psychotropic drugs for which increased prescriptions to pediatrics have been most clearly documented are stimulants, but to date, there has been little published material on the side effects of this class of psychotropics on the developing brain. One potential mechanism that may explain the current pattern of findings is the subsequent increase in DA that has been shown to follow the use of stimulants.105 Because involuntary movements are also the result of irregular DA activity,4,22 the moderating relationship can be understood under a paradigm where elevated movement abnormalities serve as an external marker of ongoing neural events.8 There has been a recent acceleration in the research literature expanding on our understanding of the relationship between stimulants and psychosis. Most notably, stimulants have been found to precipitate psychotic symptoms106 and the onset of bipolar disorder107 in some youth, particularly those with an elevated risk for developing Axis I disorder. Consistent with these findings, Mittal and colleagues (2008)8 report that stimulant medication was linked with sustained higher levels of positive and negative symptoms over the course of a year and these results held even when controlling for the initial level of symptoms.

Movement Abnormalities Associated with Symptoms Until recently, it was unclear whether movement abnormalities are related to positive, negative, or total psychotic symptomatology. In a study of adolescents with SPD, other Axis II disorders, and no disorder, Neumann and Walker38 found that movement abnormality was related to negative but not to positive symptoms. Conversely, Caligiuri and Lohr108 and Cassady and colleagues12 found that only positive symptoms were related to movement abnormalities. In another study, authors found that both positive and negative symptoms were associated with movement abnormalities.29 Because each of the studies that found a positive symptom association were conducted on adults, whereas the study that found support for only negative symptoms38 was conducted on children, the results

17

Movement Abnormalities: A Putative Biomarker of Risk for Psychosis

may be a result of the developmental course of schizophrenia; positive symptoms usually emerge later in young adulthood.109 It is also possible that because positive are less pronounced in SPD, limited variability weakens statistical power necessary to detect relations with neuromotor functioning. The noted series of studies, examining the developmental trajectory of movement abnormalities and symptomatology in the prodromal years immediately preceding onset, have significantly helped to clarify this issue; because a longitudinal study was utilized, potential differences of symptom associations due to course of illness were teased out for the first time.8,11 Movement abnormalities were correlated with symptomatology at three annual time points and further, the magnitude of change over time was examined. Because symptoms were assessed in each participant on a continuous scale, it was possible for individuals in the NC and OPD groups to endorse some items, yet still remain at sub-threshold for prodromal or SPD status. Movement abnormalities in the facial region were positively associated with total prodromal symptomatology at baseline, but the strength of this relationship appeared to diminish as a function of time. Correlations were significant for the upper-body for each of the three assessments, and increased in magnitude at subsequent time points. Finally, associations for the lowerbody region were not significant for baseline, but the magnitude of the correlations subsequently increased, and were significant at follow-up 1 and follow-up 2. The division of movement into regions provides some explanation of previously conflicting evidence (that had lumped all regions of movement abnormalities together). Fisher-Z transformations were used to test the hypothesis that the strength of the relation between movement abnormalities and symptoms would increase over time. For the facial region, the correlations between movement abnormalities and positive, negative, and total symptoms did not significantly change in magnitude over time. For the upper-body region, the magnitude of the relationship with increased in magnitude for both positive and total symptoms, but only showed a small positive trend for negative symptoms. ( p =0.04). Consistent with the longitudinal progression in-group differences noted above, lower-body movements, there were significant changes in magnitude over time for each symptom domain. These relationships provide support for the notion of overlapping cortico-striato-pallido-thalamic circuit mal-

251

function underlying both hyperkinetic movement and psychosis. While it is not possible to say for certain of dopamine was the moderating factor here, the longitudinal progression seen in the magnitude of movement abnormality symptom associations is highly suggestive that neuromaturational processes are at play.

Movement Abnormalities Predict Progression of Symptoms When our group saw early indication of these longitudinal trends, including both the group differences and the changes in magnitude over time, it was hypothesized that movement abnormalities might be able to predict future symptomatology, and soon turned to test this potential. Using the noted videotape coding strategy, we analyzed data from participants who had completed baseline and one follow-up assessment: (SPD; n = 42) and both psychiatric (OPD; n = 30) and nonpsychiatric (NC; n = 49).10 Utilizing a hierarchal regression strategy, symptoms at time one were entered in the first block, dummy coded medication status was entered in the second black, and finally movement abnormalities were entered in the third block; symptoms at time 2 (1-year later were the criterion variable. This model was run for the combined population as well as each individual group. Consistent with our early predictions, it was found that within the SPD group baseline movement abnormalities predicted symptom severity 1-year later. In predicting global symptomatology, there was a significant increment in R2 for Block III in the upper-body region (Beta = 0.38, p =.04) and for the facial region (Beta = 0.39, p =0.03). There were no significant findings for positive symptomatology. For negative symptomatology, there was a significant increment in the R2 for Block III in the facial region (Beta = 0.34, p = 0.03). Also consistent with early findings, there were no significant findings for lower-body. In these analyses, the control for psychotropic medication (in Block II) indicated that stimulant medication was uniquely significant. While it is important to consider that individuals prescribed stimulants are likely to have qualitatively different clinical presentations than individuals prescribed other medications classes or no medication (and the present findings should be interpreted with caution), the study does indicate that stimulant

252

medication use was associated with an increase in positive prodromal symptoms one year after index 11(Beta = 0.16, p ≤ 0.01).10

V.A. Mittal and E.F. Walker

the end of the study period (OR 2.98, CI: 1.23,7.19). At the present time, these findings provide the strongest evidence that individuals with elevated movement abnormalities may represent a subgroup of prodromal adolescents who are at the highest risk for conversion.

Movement Abnormalities and Conversion This progression of studies led us to formally test the possibility that movement abnormalities could serve as a biological marker of risk.9 The noted coding strategy was employed and movement abnormalities were coded from videotapes of 40 participants. However, to mirror advances in the prodromal literature (noted above), in this study participating adolescents met criteria for a prodromal syndrome (Attenuated Positive Prodromal Syndrome, designated on the Structured Interview for Prodromal Symptoms; SIPS.90 Following initial assessment, participants were evaluated for diagnostic status at four annual time points. A total of ten participants converted to an Axis I psychosis (schizophrenia, and affective disorders with psychotic features) over the 4-year period. This 25% rate is consistent with what is to be expected of highrisk populations.14 It should be noted that in an effort to provide consistency with genetic evidence of shared etiological factors among DSM psychotic disorders,110,111 prodromal researchers have been including both schizophrenia and affective disorder with psychotic features under the common term “psychosis” as the outcome variable. Comparisons with converted and non-converted participants at baseline indicated that the groups did not differ on demographic characteristics or levels of prodromal symptomatology. The participants in the converted group exhibited significantly more baseline movement abnormalities- Facial region: Non-Converted (Mean = 0.65, SD = 1.20) and the Converted group (mean = 2.10, SD = 1.52); Upper-body region: NonConverted (mean = 2.10, SD = 2.17), and the Converted group (mean = 5.10, SD = 3.10); Lower-body region: Non-Converted (mean = 1.46, SD = 1.31), Converted (mean = 1.60, SD = 1.26). While movement abnormalities and prodromal symptoms were significantly associated, logistic regression indicated that abnormalities in facial and upper-body regions were most predictive of conversion. Movement abnormalities in the facial region during the baseline period were found to be indicative of threefold greater odds of conversion at

Summary Consistent with previous predictions4,6,112 early movement abnormalities appear to be associated with a latent constitutional vulnerability. Subsequently, the literature has developed to suggest that hyperkinetic movement abnormalities are indicative of dysfunction in DA striatal circuits. Through neuromaturation and biological reactions to chronic and acute environmental stressors, increases in DA activity in this region may contribute to the onset of prodromal and eventually psychotic symptoms. As movement abnormalities precede the onset of the clinical prodrome, and then continue to increase in severity throughout the early progression of subtle prodromal symptoms, they have potential as a viable marker of risk for psychosis. Nonetheless, though promising, our current understanding of the underlying neural mechanisms is limited. In the following section, we suggest future research directions, and speculate on emerging technologies that will facilitate understanding of these intricate processes.

Future Research Directions Family Studies While we have a solid body of evidence indicating that abnormal striatal DA activity can result in hyperkinetic movement, there is little available data on whether such abnormalities are the product of genetic factors, environmental insult, or an interaction between the two. As noted, the ability to point to a specific causal agent or neuropathophsyiology is not a formal requirement for movement abnormalities to serve as a viable biomarker. Nonetheless, it behooves us study potential genetic and environmental contributions. As research has not revealed any single genes of major, or even moderate effects, it is generally assumed

17

Movement Abnormalities: A Putative Biomarker of Risk for Psychosis

that multiple genes affect the vulnerability to schizophrenia.113 Given that the diagnosis itself is not highly associated with any specific genetic etiologies, a promising strategy lies in identifying endophenotypes (supplemental phenotypes that are associated more with a specific genetic liabilities than the diagnosis of schizophrenia is in itself).114 Among several neurocognitive markers, Gottesman and Gould)115 formally identified movement abnormalities as one such potential supplemental phenotype. To date, research findings have been quite mixed. While some authors have noted an increase in hyperkinetic movement in the relatives of individuals with schizophrenia (in comparison to controls),74,116–118 other studies have found no such relationship.74,119–121 Tarboux and Pogue-Geile119 note that methodological discrepancies are likely to have played a role in these disparate findings, as both exclusion criteria and demographic matching quality for control groups have varied widely. Also, most studies have not used scales designed to specifically detect hyperkinetic movements (which is not necessarily a problem as most scales include hyperkinetic movements but can certainly affect consistency across studies), and blind raters have only been utilized in a proportion of the studies. In addition, a large prospective birth-cohort study observed that unusual movements during childhood were indeed predictive of adulthood schizophrenia onset, but the authors also noted that the childhood movements did not predict unaffected sibling status; this unique perspective indicates that movement abnormalities may be specific to high-risk individuals who develop the actual clinical phenotype.30 Nonetheless, at least one study utilizing the Abnormal Involuntary Movement Scale, and conducted on medication naive schizophrenia patients and their relatives in rural India, shows good promise for this area of inquiry, as authors observed the presence of hyperkinetic movements in at least 25 out of 181 (14%) of the first-degree relatives of individuals with schizophrenia.116 For the continued development of movement abnormality research within the field of psychosis, it will be particularly important to determine if relatives of genetic high-risk individuals also exhibit these movements. To date, this study has not been conducted among prodromal populations. In addition to this approach, we stress the importance of exploring specific candidate genes in the process. To date, this represents a relatively unexplored

253

area of inquiry. However, there is a body of literature examining DRD2 polymorphisms in tardive dyskinesia.122 It is somewhat ironic that genetic research on the drug-induced motor abnormalities that impeded research on spontaneous movement abnormalities might now serve as a springboard for genetic studies of spontaneous movement abnormalities in psychosis. Based on this literature, we note that the Ser9Gly polymorphism of the DRD3 gene, and both C957T and the adjacent C939T polymorphism (spanning the DRD2 gene) appear prominent in DA function in the nigrostriatal system believed to underlie movement and psychosis.122

Neuroimaging and Emerging Technology Although early studies have also been successful in examining childhood movement abnormalities, and relating them to MRI scans of individuals who later developed schizophrenia in adulthood,123 to date, there is little published research utilizing imaging or other recent technological advances to understand movement disorders in clinical populations. However, some preliminary studies have used fMRI to examine the neural basis for motor sequencing in healthy subjects,124,125 and positron emission tomography (PET) has been used to trace the anatomy of feedback loops during particular movements, such as reaching126 and one recent study employed an ultrasonic movement analysis system in cognitive movement performance among individuals with schizophrenia.127 In an early study, Caligiuri and Lohr108 attempted to distinguish spontaneous from neuroleptic induced movement abnormalities by utilizing spectral analyses of hand force instability94; this procedures continues to appear promising in determining whether a particular series of movements is the result of mediation side-effects or a natural occurring “spontaneous” abnormality. Research methods for examining whole-body involuntary movement (WBIM) in disorders such as Parkinson’s128 may also contribute to the study of movement abnormalities in psychosis. Finally, advances in in-vivo investigation of the dopaminergic system129 may be applied to dyskinesias in clinical populations. Taken together, the advent of these new technologies is promising for the continued development and understanding of movement abnormalities and schizophrenia.

254

Combining Markers The overall goal of prodromal research is to determine if we can use signs and signals to improve our ability to predict conversion beyond using subthreshold symptoms alone. Evidence presented in this chapter suggests that movement abnormalities may serve as a potentially useful marker of risk, indicating who, among clinical high-risk individuals are at the greatest risk for schizophrenia. The next step in this overall area of inquiry (prodromal research) will be to utilize longitudinal prospective research designs, following high-risk individuals from adolescence through early adulthood, and utilizing this “course of illness” data to determine if combining biological markers can serve to enhance our ability to identify high risk individuals or predict risk. As noted, knowledge from this type of research will serve to elucidate etiological processes, as well as helping to identify those individuals who are most likely to benefit from preventive interventions. To date, one significant limitation hampering this promising research is sample size. Identifying highrisk individuals, encouraging them to participate, and then retaining them for a number of years is an enormously costly and complex endeavor. As a result, few existing studies have yielded sample sizes large enough to provide sufficient power to testing the predictive validity of multiple markers of risk.

The Practical Detection and Treatment of High-Risk Individuals Interest in the possibility of preventing the onset of psychotic disorders has burgeoned in the past decade. There is early evidence that this may be feasible; preliminary findings suggest that antipsychotic drugs may delay or reduce the likelihood of transition to psychosis in vulnerable young adults who show prodromal or schizotypal signs.90,130 The critical mechanisms of action responsible for such an effect are not understood. Although a reduction in dopaminergic activity figures prominently in theories about antipsychotic efficacy, these drugs have a myriad of other neurobiological

V.A. Mittal and E.F. Walker

effects, including a significant reduction in stress hormone levels.27 The latter effect is of particular interest because it has potential implications for both neuromaturational processes and preventive mechanisms. This is further illustrated by a body of research showing that shorter duration of untreated psychotic illness is significantly related to a better course of illness.131 While research is beginning to support the notion that pharmacological intervention, the side-effects of available medications (e.g., weight gain, diabetes) make providing blanket treatment to all prodromal individuals an inappropriate strategy.132,133 As current strategies are effective in identifying less than half of the individuals who will go on to develop psychosis, factors serving to strengthen the identification of those 35% that can most benefit from preventive treatment is rapidly becoming a priority. The current literature suggests that, for some individuals at risk, movement processes may be intricately tied with the development of psychosis. Recent studies have indicated that movement abnormalities occur long before the onset of symptoms, and then gradually increase as the symptoms increase in severity and frequency.8 As the strength of the association between movement abnormalities and symptoms has been shown to increase in magnitude with illness progression, and this phenomenon appears specific to individuals with schizophrenia and other psychoses, there is the potential of a clinically useful application. It is important to emphasize that no single risk indicator, including movement abnormalities, is likely to be sufficient in the prediction of conversion to psychosis. Instead, combinations of behavioral and biological measures are expected to hold the greatest potential for predicting imminent risk for psychosis. As the health model continues to turn towards early intervention and preventive strategies, the investigation of biomarkers will represents an increasingly critical step in this process. As this research demands the integrating empirical data with theory to inform practice, it is very likely to meet the challenge, setting a standard for other fields to follow. As researchers continue to incorporate innovations in technology, and set the standards for interdisciplinary science, biological marker research will contribute significantly to understanding and treating psychosis.

17

Movement Abnormalities: A Putative Biomarker of Risk for Psychosis

References 1. Puri BK, Barnes TR, Chapman MJ, Hutton SB, Joyce EM. Spontaneous dyskinesia in first episode schizophrenia. J Neurol Neurosurg Psychiatry 1999;66(1):76–78. 2. Khot V, Wyatt RJ. Not all that moves is tardive dyskinesia. Am J Psychiatry, 1991;148:661–666. 3. Yarden PE, Discipio W. Abnormal movements and prognosis in schizophrenia. Am J Psychiatry, 1971;128:317–323. 4. Walker EF. Developmentally Moderated Expressions of the Neuropathology Underlying Schizophrenia. Schizophr Bull 1994;20:453–480. 5. Weinberger D. Implications of normal brain development for pathogenesis of schizophrenia. Arch Gen Psychiatry 1987;44:660–669. 6. Fish B, Marcus J, Hans SL, Auerbach JG, Purdue S. Infants at risk for schizophrenia: Sequelae of a genetic neurointegrative defect. Arch Gen Psychiatry 1992;49:221–235. 7. Fenton WS, Wyatt RJ, McGlashan TH. Risk factors for spontaneous dyskinesia in schizophrenia. Arch Gen Psychiatry 1994;51(8):643–50. 8. Mittal VA, Neumann C, Saczawa M, Walker EF. The longitudinal progression of movement abnormalities and psychotic symptoms in adolescents at high-risk for psychosis. Arch Gen Psychiatry 2008;65(2):165–170. 9. Mittal VA, Walker EF. Movement abnormalities predict conversion to Axis I psychosis among prodromal adolescents. J Abn Psychol 2007;116(4):796–803. 10. Mittal VA, Tessner KD, Trottman HD, Dhruv S, Esterberg M, Simeonova D, McMillan AL, Murphy E, Saczawa M, Walker EF. Movement abnormalities and the progression of prodromal symptomatology in adolescents at risk for psychotic disorders. J Abn Psychol 2007;116(2):260–267. 11. Mittal VA, Dhruv S, Tessner KD, Walder DJ, Walker EF. The relations among putative bio risk markers in schizotypal adolescents: minor physical anomalies, movement abnormalities and salivary cortisol. Biol Psychiatry 2007;61(10):1179–1186. 12. Cassady SL, Adami H, Moran M, Kunkel R, Thaker GK. Spontaneous dyskinesia in subjects with schizophrenia spectrum personality. Am J Psychiatry 1998;155(1):70–5. 13. Graybiel AM. The basal ganglia and cognitive pattern generators. Schizophr Bull 1997;23(3):459–469. 14. Cannon TD, Cadenhead K, Cornblatt B et al., Prediction of psychosis in youth at high clinical risk: a multisite longitudinal study in North America. Arch Gen Psychiatry 2008;65(1):28–37. 15. Klosterkotter J, Hellmich M, Steinmeyer EM, SchultzeLutter F: Diagnosing schizophrenia in the initial prodromal phase. Arch Gen Psychiatry 2001;58:158–164. 16. Yung AR, McGorry PD. The initial prodrome in psychosis: Descriptive and qualitative aspects. Aust NZ J Psychiatry 1996;30:587–599. 17. Haroun N, Dunn L, Haroun A, Cadenhead KS. Risk and protection in prodromal schizophrenia: ethical implications for clinical practice and future research. Schizophrenia Bull 2006;32(1):166–178. 18. Falloon IRH. Early detection and intervention through integration with primary health services: The Buckingham Project. In: Miller T, Mednick SA, McGlashan TH,

19.

20.

21.

22. 23.

24.

25.

26.

27. 28.

29.

30.

31. 32.

33.

34.

35.

255

Libiger J, Johannessen JO, (Eds.), Early Intervention in Psychotic Disorders. Dordrecht, The Netherlands: Kluwer; 2001;123–133. Preda A, Miller TJ, Rosen JL, Somjee L, McGlashan TH, Woods SW. Treatment histories of patients with a syndrome putatively prodromal to schizophrenia. Psychiatr Serv 2002;53:342–344. Yung AR, Phillips LJ, McGorry PD, Hallgren MA, McFarlane CA, Jackson HJ, Francey S, Patton GC. Can we predict the onset of first-episode psychosis in a high-risk group? Int Clin Psychopharm 1998;13(Suppl 1):S23–S30. Yung AR, McGorry PD. The initial prodrome in psychosis: descriptive and qualitative aspects. Aust New Zealand J Psychiatry 1996;30:587–599. Delong MR, Wichman T. Circuits and circuit disorders of the basal ganglia. Arch Neurol 2007;64(1):20–24. Smith Y, Bevan MD, Shink E, Bolam P. Microcicuitry of the direct and indirect pathways of the basal ganglia. Neurosci 1998;86:353–387. Alexander GE, Crutcher MD, DeLong MR. Basal ganglia -thalamocortical circuits: Parallel substrates for motor, oculomotor, “prefrontal” and “limbic” functions. Prog Brain Res 1990;85:119–145. Walker EF, Mittal VA, Tessner KD. HPA activity and the developmental course of schizophrenia. Annual Rev Clin Psychol 2008;4:5.1–5.28. Corcoran C, Walker EF, Huot R, Mittal VA, Tessner K, Kestler K, Malaspina D. The stress cascade and schizophrenia: etiology and onset. Schizophr Bull 2003;29(4): 671–692. Walker EF, Diforio D. Schizophrenia: A neural diathesisstress model. Psychol Rev 1997;104:667–685. Erlenmeyer-Kimling L, Rock E, Roberts SA, Janal M, Kestenbaum C, Cornblatt B, et al. Attention, memory, and motor skills as childhood predictors of schizophrenia related psychoses: the New York Hih-Risk Project. Am J Psychiatry 2000;157(9):1416–1422. Walker E., Lewis N, Loewy R., Palyo S. Motor dysfunction and risk for schizophrenia. Dev Psychopathol 1999;11: 509–523. Rosso IM, Bearden CE, Hollister M, Gasperoni TM, Sanchez LE, Hadley T, et al. Childhood neuromotor dysfunction in schizophrenia patients and their unaffected siblings: a perspective cohort study. Schizophr Bull 2000;26(2): 367–378. Reiter PJ. Extrapyramidal motor-disturbances in dementia praecox. Acta Psychiatr Neurol Scand 1926;1:287–309. Honer WG, Kopala LC, Rabinowitz J. Extrapyramidal symptoms and signs in first episode, antipsychotic exposed and non-exposed patients with schizophrenia or related psychotic illness. Psychopharmacol 2005;18:277–228. Mehler-Wex C, Riederer P, Gerlach M. Dopaminergic dysbalance in distinct basal ganglia neurocircuits: implications for the pathophysiology of Parkinson’s disease, schizophrenia and attention deficit hyperactivity disorder. Neurotoxicity Res 2006;10(3–4):167–179. Hoff JI, van den Plas AA, Wagemans EA, van Hilten JJ. Accelerometric assessment of levodopa-induced dyskinesias in Parkinson’s disease. Mov Dis 2001;16(1):58–61. Gerlach J. Pathophysiological mechanisms underlying tardive dyskinesia. Psychopharmacol Suppl 1985;2:98–103.

256 36. Woods BT, Kinnery DK, Yurgelun-Todd D. Neurological abnormalities in schizophrenic patients and their families: I. Comparison of schizophrenic, bipolar, and substance abuse controls. Arch Gen Psychiatry 1986;43:657–663. 37. Boks M. P., Liddle P. F., Burgerhof J. G., Knegtering, R. van den Bosch R. J. Neurological soft signs discriminating mood disorders from first episode schizophrenia. Act Psych Scand 2004;110(1):29–35. 38. Neumann, CS, Walker EF. Childhood neuromotor softsigns, behavior problems and adult psychopathology. In: Ollendick T, Prinz R (Eds.), Advances in Clinical Child Psychology. New York: Plenum Press, 1996. 39. Roitberg BZ, Emborg ME, Sramek JG, Palfi S, Kordower JH. Behavioral and morphological comparison of two nonhuman primate models of Huntington’s disease. Neurosurgery 2002:50(1);137–145. 40. Stahl SM, Davis KL, Berger PA. The neuropharmacology of tardive dyskinesia, spontaneous dyskinesia, and other dystonias. J Clin Psychopharm 1982;2(5):321–328. 41. McMahon WM, Filloux FM, Ashworth JC, Jensen J. Movement disorders in children and adolescents. Neurologic Clin 2002;20(4):1101–1124. 42. Papapetropoulos S, Mash DC. Psychotic symptoms in Parkinson’s disease: from description to etiology. J Neurol 2005;252(7):753–764. 43. Weiner WJ, Rabinstein A, Levin B, Weiner, C, Shulman, L. Cocaine-induced persistent dyskinesias. Neurol 2001;56(7):964–965. 44. Abi-Dargham A, Kegeles LS, Zea-Ponce Y, et al. Striatal amphetamine-induced dopamine increase in patients with schizotypal personality disorder studied with single photon emission computed tomography and [123l]lodobenzamide. Biol Psychiatry 2004;55:1001–1006. 45. Kestler LA, Walker EF, Vega EM. Dopamine receptors in the brains of schizophrenia patients: A meta-analysis of the findings. Behav Pharm 2001;12(5):355–371. 46. Swerdlow NR, Koob GF. Dopamine, schizophrenia, mania, and depression: Toward a unified hypothesis of cortico-striato-pallido-thalamic function. Behav Brain Sci 1987;10:197–245. 47. Gray JA, Veena K, Lawrence N, Young AM. Functions of the dopaminergic innervation of the nucleus accumbens. Psychobiol 1999;27(2):225–235. 48. Schulz PM, Schulz C, Goldberg SC, Ettigi P, Resnick RJ, Friedel RO. Diagnoses of the relatives of schizotypal outpatients. J Nerv Ment Dis 1986;174(8):457–463. 49. Moller P, Husby R. The initial prodrome in schizophrenia: Searching for naturalistic core dimensions of experience and behavior. Schizophr Bull 2000;26(1):217–232. 50. Lichter DG, Cummings JL. Frontal-Subcortical Circuits in Psychiatric and Neurological Disorders. New York: Guilford Press; 2001. 51. Ritsner MS, Blumenkrantz H. Predicting domain-specific insight of schizophrenia patients from symptomatology, multiple neurocognitive functions, and personality related traits. Psychiatry Res 2007;149(1–3):59–69. 52. Mednick SA, Machon RA, Huttunen MO, Bonet D. Adult schizophrenia following prenatal exposure to an influenza epidemic. Arch Gen Psychiatry 1988;45:189–192. 53. Weinberger DR, Berman KF. Speculation on the meaning of cerebral metabolic hypofrontality in schizophrenia. Schizophr Bull 1988;14:157–168.

V.A. Mittal and E.F. Walker 54. Cornblatt BA, Walker EF, Green MF, Mittal VA. Schizophrenia: Etiology and Neurocognition. In Millon T, Blaney PH, Davis RD (Eds.), Oxford Textbook of Psychopathology (2nd edition). Oxford University Press: New York; 2008. 55. Weinstock M. The potential influence of maternal stress hormones on development and mental health of the offspring. Brain Behav Immun 2005;19(4):296–308. 56. Kiess W, Meidert A, Dressendorfer RA, et al. Salivary cortisol levels throughout childhood and adolescence: Relation with age, pubertal stage and weight. Ped Res 1995;37:502–506. 57. Walker EF. Adolescent neurodevelopment and psychopathology. Current Directions in Psychol Sci 2002;11(1):24–28. 58. Neumann C, Grimes K, Walker E, Baum K. Developmental pathways to schizophrenia: Behavioral subtypes. J Abn Psychol 1995;104(4):558–566. 59. Spear LP. Neurodevelopment during adolescence. In: Cicchetti D., Walker E. (Eds.), Neurodevelopmental mechanisms in psychopathology. New York: Cambridge University Press; 2003:62–83. 60. Feinberg I. Schizophrenia: caused by a fault in programmed synaptic elimination during adolescence? J Psychiatr Res 1982;17(4):319–334. 61. Huttenlocher PR. Synaptic density in human frontal cortex: Developmental changes and effects of again. Brain Res 1979;163(2):195–205. 62. McGlashan TH, Hoffman RE. Schizophrenia as a disorder of developmentally reduced synaptic connectivity. Arch Gen Psychiatry 2000;57:637–647. 63. Glantz LA, Lewis DA. Decreased dendritic spine density on prefrontal cortical pyramidal neurons in schizophrenia. Arch Gen Psychiatry 2000;57:65–73. 64. Mittal VA, Ellman LM, Cannon TD. Gene-environment interaction and covariation in schizophrenia: the role in obstetric complications. Schizophr Bull 2008;34:1083–1094. 65. Cannon TD, Mednick SA. The schizophrenia high-risk project in Copenhagen: Three decades of progress. Act Psychiatr Scand 1993;370(Suppl):33–47. 66. Cannon TD, Mednick SA, Parnas J, Schulsinger F, Praestholm J, Vestergaard, A. Developmental brain abnormalities in the offspring of schizophrenic mothers: II. Structural brain characteristics of schizophrenia and schizotypal personality disorder. Arch Gen Psychiatry 1994; 51:955–962. 67. Siever LJ, Silverman JM, Horvath TB, Howard K, Coccaro E, Keefe RSE. Increased morbid risk for schizophreniarelated disorders in relatives of schizotypal personality disordered patients. Arch Gen Psychiatry 1990;47:634–640. 68. Fish B. Infant predictors of the longitudinal course of schizophrenic development. Schizophr Bull 1987;13(3):395–409. 69. Walker EF, Savoie T, Davis D. Neuromotor precursors of schizophrenia. Schizophr Bull 1994;20(3):441–451. 70. Jones P, Rodgers B, Murray R, Marmot M. Child development risk factors for adult schizoprhiena in the British in 1946 birth cohort. Lancet 1994;344(8934):1398–1402. 71. Isohanni M, Jones P, Kemppainen L, Croudace T, Isohanni I, Veijola J, Rasanen S, Wahlberg KE, Tienari P, Rantakallio P. Childhood and adolescent predictors of schizophrenia in the Northern Finland 1966 birth cohort– a descriptive life-span model. Eur Arch Psychiatry Clin Neurosci 2003;250(6):311–319. 72. Schiffman J, Walker E, Ekstrom M, Schulsinger F, Sorensen H, Mednick S. Childhood videotaped social and neuromotor

17

73.

74.

75.

76. 77. 78.

79.

80.

81.

82.

83. 84.

85.

86.

87.

88.

89.

90.

Movement Abnormalities: A Putative Biomarker of Risk for Psychosis precursors of schizophrenia: A prospective investigation. Am J Psychiatry 2004; 161:2021–2027. Caligiuri MP, Lohr JB. A disturbance in the control of muscle force in neuroleptic-naive schizophrenic patients. Biol Psychiatry 1994;35(2):104–111. McCreadie RG, Thara R, Kamath S, et al. Abnormal movements in never-medicated Indian patients with schizophrenia. Brit J Psychiatry 1996;168:221–226. Gervin M, Browne S, Lane,A, Clarke M, Waddington JL, Larkin C, O’Callaghan E. Spontaneous abnormal involuntary movements in first-episode schizophrenia and schizophreniform disorder: baseline rate in a group of patients from an Irish catchment area. Am J Psychiatry 1998; 155(9):1202–1206. Turner T. Rich and mad in Victorian England. Psychol Med 1989;19:29–44. Nagy J, Szatmari P. A chart review of schizotypal personality disorders in children. J Autism Dev Dis 1986;16:351–367. Webb CT, Levinson DF. Schizotypal and paranoid personality disorder in the relatives of patients with schizophrenia and affective disorders: A review. Schizophr Res 1993; 11:81–92. Quinn J, Meagher D, Murphy P, Kinsella A, Mullaney J, Waddington JL. Vulnerability to involuntary movements over a lifetime trajectory of schizophrenia approaches 100%, in association with executive (frontal) dysfunction. Schizophr Res 2001;49(1–2):79–87. Jeste D, Twamley E, Zorrilla L, Golshan S, Patterson T, Palmer B. Aging and outcome in schizophrenia. Acta Psychiatr Scand 2003;107(5):336–343. Pearson JS, Kley IB. On the application of genetic expectancies as age specific base rates in the study of human behavior disorders. Psychol Bull 1957;54:406–420. Mednick SA, Schulsinger F. Some premorbid characteristics related to breakdown in children with schizophrenic mothers. J Psychiatry Res 1968;6:267–291. Cornblatt, B., Obuchowski, M. Update of high-risk research: 1987–1997. Int Rev Psychiatry 1997;9:437–447. Larsen TK. McGlashan TH, Moe LC. First-episode schizophrenia: I. Early course parameters. Schizophr Bull 1996; 22(2):241–256. Cornblatt B, Lencz T, Obuchowski M. The schizophrenia prodrome: treatment and high-risk perspectives. Schizophr Res 2002;54:177–186. Neumann CS, Grimes, K., Walker, E F, Baum K. Developmental pathways to schizophrenia: Behavioral subtypes. J Abn Psychol 1995;104;1–9. Walker E, Baum K, Diforio D. Developmental changes in the behavioral expression of vulnerability for schizophrenia. In Lenzenweger M., B. Dworkin (Eds.), Experimental psychopathology and the pathogenesis of schizophrenia. Washington, DC: American Psychological Association;1998. Silverstein ML, Mavrolefteros G, Turnbull A. Premorbid factors in relation to motor, memory, and executive functions deficits in adult schizophrenia. Schizophr Res 2003;61(2–3):271–280. Wolff S. ‘Schizoid’ personality in childhood and adult life I: The vagaries of diagnostic labeling. Brit J Psychiatry 1991;159:615–620. Miller TJ, McGlashan TH, Rosen JL, Somjee L, Markovich PJ, Stein K, Woods SW. Prospective diagnosis of the initial prodrome for schizophrenia based on the Structured

257

Interview for Prodromal Syndromes: preliminary evidence of interrater reliability and predictive validity. Am J Psychiatry 2002;159(5):863–865. 91. McGorry PD, Bell RC, Dudgeon PL, Jackson HJ. The dimensional structure of first episode psychosis: an exploratory factor analysis. Psychol Med 1998;28(4):935–947. 92. Sullivan, H. The Onset of Schizophrenia. Am J Psychiatry 1927;84:105–134. 93. Sanbrook M, Harris A. Evidence for early intervention in firstepisode psychosis. Can J Psychiatry 2004;49(4):280–281. 94. Owens DG, Johnstone EC, Crow TJ, Frith CD, Jagoe JR, Kreel L. Lateral ventricular size in schizophrenia: relationship to the disease process and its clinical manifestations. Psychol Med 1985;15(1):27–41. 95. Szymanski S, Kane JM, Lieberman JA. A selective review of biological markers in schizophrenia. Schizophr Bull 1991;17(1):99–111. 96. Szymanski S, Lieberman JA, Alvir JM, et al. Gender differences in onset of illness, treatment response, course, and biologic indexes in first-episode schizophrenic patients. Am J Psychiatry 1995;152(5):698–703. 97. Dawson ME, Nuechterlein KH, Schell AM, Mintz J. Concurrent and predictive electrodermal correlates of symptomatology in recent-onset schizophrenic patients. J Abn Psychol 1992;101(1):153–164. 98. Falloon IR. Early intervention for first episodes of schizophrenia: a preliminary exploration. Psychiatry 1992; 55(1):4–15. 99. Loebel AD, Lieberman JA, Alvir JM, Mayerhoff DI, Geisler SH, Szymanski SR. Duration of psychosis and outcome in first-episode schizophrenia. Am J Psychiatry 1992;149(9):1183–1188. 100. McGlashan TH. Early detection and intervention in schizophrenia: research. Schizophr Bull 1996;22(2):327–345. 101. Yung AR, Phillips LJ, Yuen HP, Francey SM, McFarlane CA, Hallgren M, McGorry PD: Psychosis prediction: 12-month follow up of a high-risk (“prodromal”) group. Schizophr Res 2003; 60:21–32. 102. Early International Psychosis Association (IEPA). Available at: http://www.iepa.org.au/services/listing.cfm. Accessed June 15, 2008. 103. First M, Spitzer RL, Gibbon M, Williams JB. Structured Clinical Interview for the DSM-IV Axis I Disorders (SCID-I), Patient Edition. Washington, DC: American Psychiatric Press;1995. 104. Mittal VA, Tessner K, Delawalla Z, McMillan AK, Trottman H, Walker, EF. Gesture behavior in unmedicated schizotypal adolescents. J Abn Psychology 2006;115(2):351–358. 105. Grady TA, Broocks A, Canter SK, Pigott TA, Dubbert B, Hill JL, Murphy DL. Biological and behavioral responses to D-amphetamine, alone and in combination with serotonin3 receptor antagonist ondasterone, in healthy volunteers. Psychiatry Res 1996;64:1–10. 106. Cherland E, Fitzpatrick R. Psychotic side effects of psychostimulants: a 5-year review. Can J Psychiatry 1999;44(8): 811–813. 107. Del Bello MP, Soutullo CA, Hendricks W, Niemeier RT, McElroy SL, Strakowski SM. Prior stimulant treatment in adolescents with bipolar disorder: association with age at onset. Bipol Dis 2001;3(2):53–57. 108. Caligiuri MP, Lohr JB. A disturbance in the control of muscle force in neuroleptic-naïve schizophrenic patients. Biol Psychiatry 1994;47:434–441. 109. Walker E, Savoie T, Davis, D. Neuromotor precursors of schizophrenia. Schizophr Bull 1994;20:453–480.

258 110. Cardno AG, Rijsdijk FV, Sham PC, Murray RM, McGuffin, P. A twin stud of genetic relationships between psychotic symptoms. Am J Psychiatry 2002;159(4):539–545. 111. Riley B, Kendler KS. Molecular genetic studies of schizophrenia. Eur J Hum Gen 2006;14(6):669–680. 112. Fish B. Neurobiologic antecedents of schizophrenia in children: Evidence for an inherited, congenital neurointegrative defect. Arch Gen Psychiatry 1977;34(11):1297–313. 113. Pogue-Geile MF, Gottesman I. Schizophrenia: Study of a genetically complex phenotype. In: Jones BC, Mormede P (Eds.), Neurobehavioral genetics: Methods and Applications. New York:CRC Press. 114. Gottesman II, Shields J. Genetic theorizing and Schizophrenia. Br J Psychiatry 1973;122:15–30. 115. Gottesman II, Gould TD. The endophenotype concept in psychiatry: etymology and strategic intentions. Am J Psychiatry 2003;160:636–645. 116. McCreadie, R. G., Thara, R., Srinivasan, T. N., Padmavathi, R. Spontaneous dyskinesia in first-degree relatives of chronically ill, never-treated people with schizophrenia. Brit J Psychiatry 2003;183:45–49. 117. Egan MF, Hyde TM, Bonomo JB, Mattay VS, et al. Relative risk of neurological signs in siblings of patients with schizophrenia. Am J Psychiatry 2001;158:1827–1834. 118. Flyckt L, Sydow O, Bjerkenstedt L, Edman G, Gydin E, Wiesel F. Neurological signs and psychomotor performance in patients with schizophrenia their relatives, and healthy controls. Psychiatry Res 1999;86:113–129. 119. Tarbox SI, Pogue-Geile MF. Spontaneous dyskinesia and familial liability to schizophrenia. Schizophr Res 2006;81(2–3):125–123. 120. Chen R, Chen E, Mak F. Soft neurological signs in schizophrenic patients and their nonpsychotic siblings. J Nerv Ment Disease 2000;188:84–89. 121. Kinney DK, Yurgelun-Todd DA, Woods BT. Hard neurologic signs and psychopathology in relatives of schizophrenic patients. Psychiatry Res 1991;39:45–53. 122. Zai CC, Hwang RW, De Luca V et al. Association study of tardive dyskinesia and twelve DRD2 polymorphisms in schizophrenia patients. Intl J Neuropsychopharmacol 2007;10(5):639–651.

V.A. Mittal and E.F. Walker 123. Walker EF, Lewine RR, Neumann C. Childhood behavioral characteristics and adult brain morphology in schizophrenia. Schizophr Res 1996;22(2):93–101. 124. Smith JF, Chen K, Johnson S, Morrone-Strupinsky J. Network analysis of single-subject fMRI during a finger opposition task. Neuroimage 2006;32(1):325–332. 125. Chan RC, Rao H, Chen EE, Ye B, Zhang C. The neural basis of motor sequencing: an fMRI study of healthy subjects. Neurosci Letters 2006;398(3):189–194. 126. Desmurget M, Grea H, Grethe JS, Prablanc C, Alexander GE, Grafton ST. Functional anatomy of nonvisual feedback loops during reaching: a positron emission tomography study. J Neurosci 2001;21(8):2919–2928. 127. Putzhammer A, Klein HE. Quantitative analysis of motor disturbances in schizophrenic patients. Dialogues Clin Neurosci 2006;8(1):123–130. 128. Fenney A, Jog MS, Duval C. Short-term variability in amplitude and motor topography of whole-body involuntary movements in Parkinson’s disease dyskinesias and in Huntington’s chorea. Clin Neurol Neurosur 2008;110(2):160–167. 129. Nikolaus S, Antke C, Kley K, Poeppel TD, Hautzel H, Schmidt D, Muller HW. Investigating the dopaminergic synapse in vivo. I. Molecular imaging studies in humans. Rev Neurosci 2007;18:439–472. 130. McGorry PD, Yung AR, Phillips LJ, et al. Randomized controlled trial of interventions designed to reduce the risk of progression to first-episode psychosis in a clinical sample with subthreshold symptoms. Arch Gen Psychiatry 2002;59(10):921–928. 131. Harrigan S, McGorry PD, Krstev H. Does treatment delay in first-episode psychosis really matter? Psychol Med 2003;33:97–110. 132. Gerbino-Rosen G, Roofeh D, Tompkins DA, et al., Hernatological adverse events in clozapine-treated children and adolescents. J Am Acad Child Adoles Psychiatry 2005;44:1024–1031. 133. Sporn A, Greenstein D, Gogtay N, Sailer F, Hommer DW, Rawlings R. Childhood-onset schizophrenia: smooth pursuit eye-tracking dysfunction in family members. Schizophr Res 2005;73:243–252.

Contents to Volumes 2, 3, and 4

Volume 2 Part III

Neuroanatomical and Neuroimaging Findings

18

Neuroimaging Biomarkers in Alzheimer’s Disease Chong Mei Sian and Lim Wee Shiong

19

Role of Imaging Techniques in Discerning Neurobehavioral Changes in Ischemic, Neurodegenerative and Demyelinating Disorders Turi O. Dalaker, Mona K. Beyer, Milena Stosic, and Robert Zivadinov

20

Towards a Functional Neuroanatomy of Symptoms and Cognitive Deficits of Schizophrenia David Linden

21

Functional and Structural Endophenotypes in Schizophrenia Stephan Bender, Matthias Weisbrod, and Franz Resch

22

Neuromorphometric Measures as Endophenotypes of Schizophrenia Spectrum Disorders Daniel Mamah, Deanna M. Barch, and John G. Csernansky

23

Magnetic Resonance Imaging Biomarkers in Schizophrenia Research Heike Tost, Shabnam Hakimi, and Andreas Meyer-Lindenberg

24

Neurostructural Endophenotypes In Autism Spectrum Disorder Armin Raznahan, Jay N. Giedd, and Patrick F. Bolton

25

Neuroimaging Biomarkers for Bipolar Disorder Across the Lifespan Nick C. Patel, Michael A. Cerullo, David E. Fleck, Jayasree J. Nandagopal, Caleb M. Adler, Stephen M. Strakowski, and Melissa P. DelBello

26

Neuroimaging Studies of Pediatric Obsessive-Compulsive Disorder: Special Emphasis on Genetics and Biomarkers Frank P. MacMaster and David R. Rosenberg

259

260

27

Contents to Volumes 2, 3, and 4

Structural Brain Alterations in Cannabis Users: Association with Cognitive Deficits and Psychiatric Symptoms Nadia Solowij, Murat Yücel, Valentina Lorenzetti, and Dan I. Lubman

Contents to Volumes 1, 3, and 4 Contributors to Volumes 1, 3, and 4 Index

Volume 3 Part IV

Possible Metabolic and Peripheral Biomarkers

28

Peripheral Biomarkers in Dementia and Alzheimer’s Disease Christian Humpel and Josef Marksteiner

29

S100B as a Potential Neurochemical Biomarker in a Variety of Neurological, Neuropsychiatric and Neurosurgical Disorders Patrick Wainwright, Jon Sen, and Antonio Belli

30

Can the Cortisol to DHEA Molar Ratio be Used as a Peripheral Biomarker for Schizophrenia and Mood Disorders? Peter Gallagher and Michael S. Ritsner

31

Neuroactive Steroid Biomarkers of Alcohol Sensitivity and Alcoholism Risk A. Leslie Morrow and Patrizia Porcu

32

Neuroendocrine Markers of Psychopathy Andrea L. Glenn

33

Mitochondrial Complex I as a Possible Novel Peripheral Biomarker for Schizophrenia Dorit Ben-Shachar

34

Peripheral Biomarkers of Excitotoxicity in Neurological Diseases Lucio Tremolizzo, Gessica Sala, and Carlo Ferrarese

35

Melatonin as a Biological Marker in Schizophrenia Armando L. Morera, Pedro Abreu-Gonzalez, and Manuel Henry

36

Peripheral Biological Markers for Mood Disorders Ghanshyam N. Pandey and Yogesh Dwivedi

37

The Diagnosis of Alcoholism Through the Identification of Biochemical Markers in Hair Nadia De Giovanni

Contents to Volumes 2, 3, and 4

261

38

Retinoic Acid Signalling in Neuropsychiatric Disease: Possible Markers and Treatment Agents Sarah J. Bailey and Peter J. McCaffery

39

Abnormalities of Inositol Metabolism in Lymphocytes as Biomarkers for Bipolar Disorder Galila Agam, Yuly Bersudsky, and Robert H. Belmaker

Contents to Volumes 1, 2, and 4 Contributors to Volumes 1, 2, and 4 Index

Volume 4 Part V

Molecular Genetic and Genomic Markers

40

Pharmacogenomic Biomarkers in Neuropsychiatry: The Path to Personalized Medicine in Mental Disorders Ramón Cacabelos

41

Pharmacogenetics in Neurological Diseases Chantal Depondt

42

Gene Expression Changes and Potential Impact of Endophenotypes in Major Psychiatric Disorders Gursharan Chana, Janet Kwok, Stephen J. Glatt, Ian P. Everall, and Ming T. Tsuang

43

Molecular Genetics of Schizophrenia: Focus on Symptom Dimensions Michael S. Ritsner and Ehud Susser

44

Genetics of Mood Disorders Laura Mandelli, Alessandra Nivoli, and Alessandro Serretti

45

Endophenotype Strategy in Epilepsy Genetics Dalila Pinto, Dorothée Kasteleijn-Nolst Trenité, and Dick Lindhout

46

Epilepsy, Biomarkers, and Genes Danielle M. Andrade

47

Molecular and Imaging Genetic Markers in Panic Disorder Katharina Domschke and Jürgen Deckert

262

Contents to Volumes 2, 3, and 4

48

The Genetics of Obsessive–Compulsive Disorder Marco A. Grados and Rebecca Dang

49

Development of Biomarkers for Alcoholism and PolysubstanceAbuse Hiroki Ishiguro, Minori Koga, Yasue Horiuchi, Emmanuel S. Onaivi, and Susumu Higuchi

50

Role of Members of the Nur (NR4A) Transcription Factors in Dopamine-Related Neurodegenerative and Neuropsychiatric Disorders Daniel Lévesque and Claude Rouillard

Contents to Volumes 1, 2, and 3 Contributors to Volumes 1, 2, and 3 Index

Contributors to Volumes 2, 3, and 4

Volume 2 Caleb M. Adler, M.D., Associate Professor of Psychiatry, Co-Director, Division of Bipolar Disorders Research, Department of Psychiatry, University of Cincinnati College of Medicine, Cincinnati, OH, USA E-mail: [email protected] Deanna M. Barch, Ph.D., Professor, Departments of Psychology, Psychiatry and Radiology, Washington University, St. Louis, MO, USA E-mail: [email protected] Stephan Bender Senior scientist and commissionary Head of the joint Neurophysiological Laboratory of the Psychiatric, Psychosomatic and Child and Adolescent Psychiatric Hospital of the University of Heidelberg, Germany E-mail: [email protected] Mona K. Beyer, M.D., Ph.D., Department of Radiology, Stavanger University Hospital, Stavanger, Norway; The Norwegian Centre for Movement Disorders, Stavanger University Hospital, Stavanger, Norway Michael A. Cerullo, M.D., Assistant Professor of Psychiatry,Division of Bipolar Disorders Research, Department of Psychiatry,University of Cincinnati College of Medicine, Cincinnati, OH, USA E-mail: [email protected] John G. Csernansky, M.D., Lizzie Gilman Professor and Chairman, Department of Psychiatry and Behavioral Sciences, Northwestern University Feinberg School of Medicine, Chicago, IL, USA E-mail: [email protected] Turi O. Dalaker, M.D., Buffalo Neuroimaging Analysis Center, Department of Neurology, State University of New York at Buffalo, Buffalo, NY, USA; Department of Radiology, Stavanger University Hospital, Stavanger, Norway; The Norwegian Centre for Movement Disorders, Stavanger University Hospital, Stavanger, Norway E-mail: [email protected]

263

264

Contributors to Volumes 2, 3, and 4

Melissa P. DelBello, M.D., M.S., Vice-Chair for Clinical Research, Department of Psychiatry; Associate Professor of Psychiatry and Pediatrics, Division of Bipolar Disorders Research, University of Cincinnati College of Medicine, Cincinnati, OH, USA E-mail: [email protected] David E. Fleck, Ph.D., Assistant Professor of Psychiatry, Division of Bipolar Disorders Research, Department of Psychiatry,University of Cincinnati College of Medicine, Cincinnati, OH, USA E-mail: [email protected] Shabnam Hakimi, B.A., Clinical Brain Disorders Branch, Genes, Cognition, and Psychosis Program, National Institute of Mental Health, National Institutes of Health, Bethesda, MD, USA David Linden Professor of Biological Psychiatry, Wales Institute of Cognitive Neuroscience and North Wales Clinical School, School of Psychology, University of Wales Bangor, Bangor, UK E-mail: [email protected] Valentina Lorenzetti Ph.D. candidate, Melbourne Neuropsychiatry Centre, Department of Psychiatry, The University of Melbourne and Melbourne Health, Australia E-mail: [email protected] Dan I. Lubman, Ph.D., FRANZCP, FAChAM; Associate Professor, ORYGEN Research Centre, Department of Psychiatry, University of Melbourne, Victoria, Australia E-mail: [email protected] Frank P. MacMaster, Ph.D., Psychiatry and Behavioral Neurosciences, Wayne State University School of Medicine Detroit, MI, USA E-mail: [email protected] Daniel Mamah, M.D., M.P.E., Instructor, Department of psychiatry, Washington University School of Medicine St. Louis; President, Eastern Missouri Psychiatric Society, USA E-mail: [email protected] Chong Mei Sian Consultant, Department of Geriatric Medicine, Tan Tock Seng Hospital, 11 Jalan Tan Tock Seng, Singapore E-mail: [email protected] Andreas Meyer-Lindenberg, M.D., Ph.D., Director of the Central Institute of Mental Health, Professor of Psychiatry and Psychotherapy, Faculty of Clinical Medicine Mannheim, University of Heidelberg, Germany E-mail: [email protected] Jayasree J. Nandagopal, MD, Assistant Professor of Psychiatry, Division of Bipolar Disorders Research, Department of Psychiatry, University of Cincinnati College of Medicine, Cincinnati, OH, USA E-mail: [email protected]

Contributors to Volumes 2, 3, and 4

265

Nick C. Patel, Pharm.D., Ph.D., Clinical Pharmacist, Lifesynch; and Clinical Assistant Professor & Health Behavior, Medical College of Georgia; USA E-mail: [email protected] Armin Raznahan, MBBS, MRCPCH, MRCPsych., Medical Research Council Clinical Research Training Fellow, Institute of Psychiatry, King’s College London, UK E-mail: [email protected] Franz Resch Professor, Director of the Child and Adolescent Psychiatric Hospital of the University of Heidelberg, Germany E-mail: [email protected] David R. Rosenberg, M.D., Psychiatry and Behavioral Neurosciences, Wayne State University School of Medicine, Children’s Hospital of Michigan, Detroit, MI, USA E-mail: [email protected] Lim Wee Shiong Consultant, Department of Geriatric Medicine, Tan Tock Seng Hospital, 11 Jalan Tan Tock Seng, Singapore Nadia Solowij, Ph.D., Senior Lecturer, School of Psychology and Illawarra Institute for Mental Health, University of Wollongong, Australia, Affiliated Scientist, Schizophrenia Research Institute, Sydney, Australia E-mail: [email protected] Milena Stosic, M.D., Buffalo Neuroimaging Analysis Center, Department of Neurology, State University of New York at Buffalo, Buffalo, NY, USA Stephen M. Strakowski, MD, The Stanley and Mickey Kaplan Professor and Chair of Psychiatry Professor of Psychology and Biomedical Engineering Director, Center for Imaging Research University of Cincinnati College of Medicine, Cincinnati, OH, USA E-mail: [email protected] Heike Tost, M.D., Ph.D., Post-Doctoral Research Fellow, Clinical Brain Disorders Branch, Genes, Cognition, and Psychosis Program, National Institute of Mental Health, National Institutes of Health, Bethesda, MD, USA E-mail: [email protected] Matthias Weisbrod, Professor, Director of the SRH Psychiatric Hospital Karlsbad-Langensteinbach; Head of the Section for Experimental Psychopathology of the University of Heidelberg, Germany E-mail: [email protected] Murat Yücel, Ph.D., MAPS; Senior Lecturer and Clinical Neuropsychologist, Melbourne Neuropsychiatry Centre and ORYGEN Research Centre, Department of Psychiatry, University of Melbourne and Melbourne Health, National Neuroscience Facility, Melbourne, Australia E-mail: [email protected] Robert Zivadinov, M.D., Ph.D., Buffalo Neuroimaging Analysis Center, Department of Neurology, State University of New York at Buffalo, Buffalo, NY, USA E-mail: [email protected]

266

Contributors to Volumes 2, 3, and 4

Volume 3 Pedro Abreu-Gonzalez Professor of Biochemistry, Department of Physiology, School of Medicine, University of La Laguna, La Laguna, Santa Cruz de Tenerife, Canary Islands, Spain Galila Agam, Ph.D., Associate Professor, Psychiatry Research Unit and Department of Clinical Biochemistry, Faculty of Medicine, Ben Gurion University, Israel Sarah J. Bailey Lecturer, Department of Pharmacy and Pharmacology, University of Bath, Claverton Down, UK E-mail: [email protected] Dorit Ben-Shachar, Ph.D., Head of Lab, Laboratory of Psychobiology, Department of psychiatry, B. Rappaport Faculty of Medicine, Rambam Medical Center, Technion IIT, Haifa, Israel E-mail: [email protected] Yuly Bersudsky, M.D., Ph.D., Senior Lecturer, Faculty of Medicine, Ben Gurion University, Beersheva Mental Health Center, Beersheva, Israel Yogesh Dwivedi, Ph.D., Associate Professor, University of Illinois at Chicago, Department of Psychiatry, Chicago, IL, USA E-mail: [email protected] Carlo Ferrarese, M.D., Ph.D., Professor of Neurology, Director of the Department of Neurology and of the Neurology Residency School, University of Milano-Bicocca, Ospedale San Gerardo, Monza, Italy E-mail: [email protected] Peter Gallagher Research Associate in Psychiatry, School of Neurology, Neurobiology and Psychiatry, Newcastle University, Leazes Wing (Psychiatry), Newcastle upon Tyne, UK E-mail: [email protected] Nadia De Giovanni Istituto Medicina Legale, Università Cattolica S. Cuore, Roma, Italy E-mail: [email protected] Andrea L. Glenn, M.A., Doctoral Student, University of Pennsylvania, Philadelphia, PA, USA E-mail: [email protected] Manuel Henry Professor of Psychiatry, Department of Internal Medicine, Dermatology and Psychiatry, School of Medicine, University of La Laguna, La Laguna, Santa Cruz de Tenerife, Canary Islands, Spain Christian Humpel Associate Professor Dr., Laboratory of Psychiatry and Exp. Alzheimer’s Research, Department of General Psychiatry, Innsbruck Medical University, Innsbruck, Austria E-mail: [email protected]

Contributors to Volumes 2, 3, and 4

267

Josef Marksteiner, M.D., Associate Professor Dr., Laboratory of Psychiatry and Exp. Alzheimer’s Research, Department of General Psychiatry, Innsbruck Medical University, Innsbruck, Austria E-mail: [email protected] Peter McCaffery Professor, University of Aberdeen, Institute of Medical Sciences, Foresterhill, Aberdeen, UK E-mail: [email protected] Armando L. Morera Professor of Psychiatry, Department of Internal Medicine, Dermatology and Psychiatry, School of Medicine, University of La Laguna, La Laguna, Santa Cruz de Tenerife, Canary Islands, Spain E-mail: [email protected] A. Leslie Morrow, Ph.D., Professor of Psychiatry and Pharmacology, Associate Director, Bowles Center for Alcohol Studies, University of North Carolina School of Medicine, USA E-mail: [email protected] Ghanshyam N. Pandey, Ph.D., Professor, University of Illinois at Chicago, Department of Psychiatry, Chicago, IL, USA E-mail: [email protected] Patrizia Porcu Assistant Professor of Psychiatry, Bowles Center for Alcohol Studies, University of North Carolina School of Medicine, USA E-mail: [email protected] Michael S. Ritsner, M.D., Ph.D., Associate Professor of Psychiatry and Head of Cognitive and Psychobiology Research Laboratory, The Rappaport Faculty of Medicine, Technion – Israel Institute of Technology, Haifa and Chair, Acute Department, Sha’ar Menashe Mental Health Center, Hadera, Israel E-mail: [email protected] Gessica Sala, Ph.D., Post-doctoral Research Associate, Department of Neuroscience and Biomedical Technologies, University of Milano-Bicocca, Italy E-mail: [email protected] Jon Sen Specialty Registrar in Neurosurgery; Wessex Neurological Centre, Southampton University Hospitals, UK E-mail: [email protected] Lucio Tremolizzo, M.D., Ph.D., Neurologist and Post-doctoral Research Associate, University of Milano-Bicocca; Ospedale San Gerardo, Monza, Italy E-mail: [email protected]

Volume 4 Danielle M. Andrade, M.D., M.Sc., Assistant Professor, Department of Medicine, University of Toronto, Division of Neurology – Epilepsy Program UHN – Toronto, Western Hospital, Toronto, Canada E-mail: [email protected]

268

Contributors to Volumes 2, 3, and 4

Ramón Cacabelos Professor and Chairman EuroEspes Biomedical Research Center, Institute for CNS Disorders and Genomic Medicin, EuroEspes Chair of Biotechnology and Genomics, Camilo José Cela University, Coruña, Spain E-mail: [email protected] Gursharan Chana, Ph.D., Department of Psychiatry, University of California, San Diego, La Jolla, CA, 92093-0603, USA Rebecca Dang Johns Hopkins University, Baltimore, MD, USA Juergen Deckert, M.D., Full Professor and Chairman, Department of Psychiatry, University of Wuerzburg, Germany E-mail: [email protected] Chantal Depondt, M.D., Ph.D., Department of Neurology, Hôpital Erasme, Université Libre de Bruxelles, Belgium E-mail: [email protected] Katharina Domschke, M.D., M.A., Head of Group “Genetics of Affective Disorders”, Department of Psychiatry, University of Muenster, Germany E-mail: [email protected] Ian P. Everall, M.D., Ph.D, FRCPsych., FRCPath., Department Of Psychiatry, University of California, San Diego, CA, USA Stephen J. Glatt, Ph.D., Department of Psychiatry and Behavioral Sciences, State University of New York, Syracuse, NY 13210, USA Marco A. Grados, M.D., M.P.H., Assistant Professor, Johns Hopkins University School of Medicine, Baltimore, MD, USA E-mail: [email protected] Susumu Higuchi, M.D., Ph.D., National Hospital Organization, Kurihama Alcoholism Center, Kanagawa, Japan E-mail: [email protected] Yasue Horiuchi, Ph.D., Department of Medical Genetics, Graduate School of Comprehensive Human Sciences, University of Tsukuba, Ibaraki, Japan E-mail: [email protected] Hiroki Ishiguro, M.D., Ph.D., Assistant Professor of Department of Medical Genetics, Graduate School of Comprehensive Human Sciences, University of Tsukuba, Ibaraki, Japan E-mail: [email protected] Dorothée Kasteleijn-Nolst Trenité Professor Medical Genetics, University of Utrecht; Department of Medical Genetics, University Medical Centre, Utrecht, the Netherlands; Professor Neuroscience, University of Rome “Sapienza”, Department of Neuroscience, Rome, Italy E-mail: [email protected]; [email protected] Minori Koga, Ph.D., Department of Medical Genetics, Graduate School of Comprehensive Human Sciences, University of Tsukuba, Ibaraki, Japan E-mail: [email protected]

Contributors to Volumes 2, 3, and 4

269

Janet Kwok, B.Sc., Department Of Psychiatry, University of California, San Diego, La Jolla, CA, 92093-0603, USA Daniel Lévesque Associate Professor, Senior Scientist, Faculty of Pharmacy University of Montreal, Canada E-mail: [email protected] Dick Lindhout Department of Medical Genetics, University Medical Centre Utrecht, Utrecht, The Netherlands Laura Mandelli, Psy.D., Assistant Professor of Psychiatry, Institute of Psychiatry, University of Bologna, Italy E-mail: [email protected] Alessandra Nivoli, M.D., Associate Professor of Psychiatry, Institute of Psychiatry, University of Sassari, Italy E-mail: [email protected] Emmanuel S. Onaivi, Ph.D., Associate Professor of Department of Biology, William Paterson University, Wayne, NJ, USA E-mail: [email protected] Dalila Pinto, Ph.D., Research Fellow. Department of Medical Genetics. University Medical Center Utrecht, Utrecht, the Netherlands; and Genetics and Genome Biology, The Center for Applied Genetics, The Hospital of Sick Children, Toronto, Canada. E-mail: [email protected] Michael S. Ritsner, M.D., Ph.D., Associate Professor of Psychiatry and Head of Cognitive and Psychobiology Research Laboratory, The Rappaport Faculty of Medicine, Technion – Israel Institute of Technology, Haifa and Chair, Acute Department, Sha’ar Menashe Mental Health Center, Hadera, Israel E-mail: [email protected] Claude Rouillard, Ph.D., Professor, Department of Medicine, Faculty of Medicine, Laval University and Neuroscience Research Centre, Laval University Hospital Research Centre, Québec City, Québec, Canada E-mail: [email protected] Alessandro Serretti, M.D., Associate Professor of Psychiatry, Institute of Psychiatry, University of Sassari, Via Luna e Sole 55, 07100, Sassari, Italy E-mail: [email protected] Ehud Susser, M.D., Senior Psychiatrist, Sha’ar Menashe Mental Health Center, Hadera, and the Rappaport Faculty of Medicine, Technion – Israel Institute of Technology, Haifa, Israel E-mail: [email protected] Ming T. Tsuang, M.D., Ph.D., D.Sc., Behavioral Genomics Endowed Chair and University Professor, University of California; Distinguished Professor of Psychiatry and Director, Center for Behavioral Genomics, Department of Psychiatry, University of California, San Diego, CA; Director, Harvard Institute of Psychiatric Epidemiology and Genetics, Harvard Medical School and Harvard School of Public Health, USA E-mail: [email protected]

Index

A Addiction, 74 Alcohol dependence, 8 Alzheimer’s disease, 3–6, 14, 24, 30–35, 45, 59, 74, 76, 89, 98, 99, 103, 104, 106, 107, 109, 147, 259 Amygdala-Hippocampal Complex (AHC), 168 Antipsychotic agents, 72 Anorexia nervosa, 227–235 Antecedent markers, 6 Antisaccade task, 12, 48, 50, 51, 118 Antipsychotics, 25, 47, 49, 75, 109, 199, 214, 217, 240, 241, 246 Antisocial, 228 Anxiety, 104, 160, 161, 166, 172, 227, 228, 233, 234 Apolipoprotein E (APOE), 23, 24, 33, 34, 98, 107 Attention, 11, 12, 24, 28, 31, 42, 44–48, 51, 53, 58, 59, 66–68, 70, 71, 73, 74, 77, 78, 91, 98, 117–120, 126, 127, 129, 131–135, 144, 155, 157, 158, 160, 165, 166, 168–171, 177–180, 186, 187, 195, 197–199, 201, 203, 204, 207, 221, 225–228, 240, 244, 259 Attentional Deficit Hyperactive Disorder (ADHD), 14, 98, 227, 228 Autism, 8, 32, 39, 42, 76, 108, 170, 172

B Basal Ganglia, 166, 242, 243 Behavioral markers, 3, 244 Biomarker, 3–15, 23–38, 41, 53, 57–63, 65–68, 70–74, 76, 79, 89, 103–109, 208, 221–226, 227–235, 239–254, 259–261 Biomarker signatures, 6 Bipolar disorder, 6, 8, 11, 13, 14, 42, 72, 75, 87–89, 96, 98, 99, 108, 114, 125–127, 131, 132, 134–136, 160, 161, 169, 195–208, 211, 213, 215, 221, 224, 225, 228, 250, 259–261 Brain, 4, 8, 10–13, 25, 34, 41, 43, 45, 47, 48, 51, 53, 54, 59–71, 74, 75, 77, 87, 89–92, 95–98, 105–107, 114–118, 135, 136, 141, 142, 144, 146–148, 151, 152, 165–168, 170, 171, 173, 188, 189, 195, 206, 207, 211, 212, 214, 215, 218, 229–231, 240, 243–245, 249, 250, 259, 260 Brain Derived Neurotrophic Factor (BDNF), 12, 13, 34, 98, 107, 195, 206 Bulimia nervosa, 227, 229, 230, 233 Brief Psychiatric Rating Scale (BPRS), 74

C Calcineurin (PPP3CC), 11, 188 California Verbal Learning Test (CVLT), 46, 50, 125, 129, 132–134, 177, 180, 181, 185, 186, 195, 201, 203, 205

Cambridge Neuropsychological Test Automated Battery (CANTAB), 195, 200, 201 Capillary electrophoresis (CE), 103, 105 Catechol-O-methyltransferase (COMT), 11, 23, 27, 37, 46, 49, 50, 57, 69, 74, 79, 96–98, 114, 117, 125, 132, 135, 136, 146, 151, 155, 156, 161, 171, 188, 195, 206–208 Central nervous system (CNS), 43, 63, 103, 106, 239, 240, 244 Cerebral spinal fluid (CSF), 3, 6, 10, 13, 14, 32, 34, 103–105, 107, 108 Cognition, 41, 46, 53, 57–60, 68, 70, 79, 89, 90, 96, 97, 113, 125–127, 143, 155, 158, 160, 164, 166, 170, 172, 199, 206, 207, 227, 233–235 Cognitive endophenotypes, 44, 117, 119, 135, 261 Corticotrophin-releasing hormone (CRH), 11, 13 Computer Tomography (CT), 88, 91 Comorbidity, 24, 30, 35, 183, 228, 229 Continuous Performance Test (CPT), 46, 129, 177, 180, 185, 195, 198, 201 Corpus Callosum (CC), 117 Cortisol, 12, 13, 31, 245, 247

D D-amino-acid oxidase activator (G72 or DAOA), 11, 125, 132, 135, 136, 146, 171, 188 Degraded Stimulus-Continuous Performance Test (DS-CPT), 46, 177, 180, 185 Dehydroepiandrosterone (DHEA), 12, 13 Dementia, 30, 34, 44, 51, 59, 74, 106, 107, 157, 195 Depression, 3, 6, 8, 11, 13, 14, 27, 31, 78, 79, 89, 98, 106, 127, 160, 169, 196, 197, 199, 206, 224, 228, 234, 249 Dexamethasone suppression test (DST), 13, 23, 24, 31 Diagnostic markers, 6, 36, 108 Disrupted in Schizophrenia, 1 (DISC1), 46, 48, 49, 96, 97, 114, 117, 118, 125, 131–136, 155, 171, 188, 195, 206 Dizygotic twins (DZ), 43, 125, 126, 132, 133, 155, 158, 159, 195, 199–201, 204, 223 DNA-Deoxyribonucleic acid, 5, 14, 87, 88, 96, 98, 114, 116, 130, 135, 150, 183, 188, 261 DNA markers, 5, 11 Dopamine (DA), 49, 57, 60, 68–70, 79, 91, 96, 98, 117, 135, 145, 149, 151, 188, 215, 217, 227, 230, 231, 235, 239–245, 247, 250–253 Dopamine receptor D2 (DRD2), 49, 57, 69, 74, 98, 146, 217, 230, 231, 239, 240, 242, 243, 253 Dopamine receptor D3 (DRD3), 74, 230, 231, 239 Dorsolateral Prefrontal Cortex (DLPFC), 3, 14, 46, 47, 50, 73, 79, 96, 97, 98, 133, 135, 165, 244 Drug response, 6, 36 Dysbindin (DTNBP1), 118, 132, 155, 161, 171, 188 271

Index

272 E Eating Disorders, 227–235 Electroencephalography, 3, 11, 119, 163 Emotion recognition deficit, 163–173 Endophenotypes, 3–15, 25, 30, 36, 41–49, 51–54, 71, 87, 89, 90, 93, 96, 99, 113–121, 125–136, 141, 142, 146, 147, 155–161, 163, 164, 170–173, 177–190, 195–208, 211, 212, 227–235, 244, 253 Endophenotype strategy, 3, 5, 7, 15, 125, 129, 131, 134, 136, 156, 181, 259, 261 Environment, 6, 10, 15, 36, 41, 43, 58, 63, 64, 88, 104, 113, 116, 118, 119, 145, 155–157, 161, 171, 196, 208, 244 Epidermal Growth Factor (EGF), 177, 189 Epigenetic, 3, 7, 25, 43, 104, 114, 116, 128, 261 Epilepsy, 3–5, 42, 76, 103, 261 Event-related potentials (ERP), 12, 47–49, 51, 57, 60, 62, 63, 68, 71, 72, 73, 77, 79, 118, 163, 165, 167, 168 P300, 12, 48, 49, 51, 57, 66, 68, 71–74, 146, 167 Expressivity, 8

F Functional Magnetic Resonance Imaging (FMRI), 46, 50, 57–70, 78, 79, 88, 91–94, 96–98, 114, 117, 148, 163, 165, 166, 232, 239, 244, 253

G Gamma-amino-butyric acid (GABA), 76, 91, 97, 116, 125, 135, 172, 188, 215, 217 Gamma-Aminobutyric Acid type A receptor (GABAA), 49 Genetics, 3, 7, 41–44, 46, 48–51, 53, 54, 87, 90, 93, 95, 96, 98, 99, 118, 127, 129, 131, 173, 182, 190, 196, 206–208, 211, 212, 214, 215, 218, 223, 261 Genetic marker, 5, 92, 107, 244 Genome Wide Association (GWA), 87, 88, 135, 141 Genomics, 3, 10, 11, 87, 148 Glutamate, 13, 60, 72, 98, 114, 131, 135, 145, 151, 188, 215, 217 Gray Matter (GM), 11, 13, 45, 64, 92, 93, 97, 98, 133, 135

H Hallucinations, 53, 68, 73, 75, 89, 121, 127, 146, 179 Heritability, 9, 28, 42, 44, 46–50, 72–74, 96–98, 115, 117, 118, 120, 130, 132, 156, 160, 163, 170, 171, 178, 180, 183, 185–187, 190, 200, 215, 230, 234, 235, 260 Hippocampus, 46–48, 51, 63, 73, 74, 76–79, 98, 118, 129, 161, 166, 180, 245, 249 HOPA, 11 Hormone, 11, 13, 70, 254, 261 Hypofrontality, 97, 244 Hypothalamo-Pituitary-Adrenal axis (HPA axis), 230, 245

I Imaging genetics, 48, 87, 90, 93, 95, 96, 98, 99

L Learning, 50, 51, 58, 63, 119, 125–127, 129–136, 147, 148, 168, 177, 178, 180, 181, 185, 195, 198, 200–207, 218, 230 Lithium (Li), 108, 109, 199

Liquid chromatography, 3, 14 Locus heterogeneity, 135, 188

M Magnetoencephalography (MEG), 57, 62, 63, 65, 77, 90, 125, 132, 147, 148, 163, 168, 169 Magnetic resonance imaging (MRI), 31, 51, 57, 59–62, 64, 69, 70, 75, 88–92, 96–98, 125, 132, 146, 148, 239, 253 Magnetic resonance spectroscopy (MRS), 30, 32, 57, 60, 88, 90–92, 96, 98 Major depression, 3, 8, 11, 13, 24, 28, 31, 32, 34, 161, 206, 249 Mass spectrometry (MS), 10, 14, 103–105, 107, 108 Measurement and Treatment Research to Improve Cognition in (MATRICS), 57–60, 63, 65, 155, 160 Melatonin, 13 Memory, 11, 12, 28, 45–51, 53, 58, 64, 66–71, 73, 74, 77–79, 97, 104, 117, 118, 120, 125–127, 129–136, 143, 149, 151, 155–161, 164, 171, 177–181, 183, 185–187, 195, 197–204, 206, 207, 247 Metabolomics, 3, 8, 10, 11, 104 Methylenetetrahydrofolate reductase (MTHFR), 11 Minor physical anomaly (MPA), 11, 12, 239 Mismatch negativity (MMN), 12, 47, 49, 51, 57, 66, 71–74, 76, 114, 118 Mitochondria, 108 Mitochondrial DNA, 96, 98 Molar ratio, 12, 13 Monozygotic twins (MZ), 43, 125, 132, 133, 155, 158, 159, 195, 199–201, 204, 223 Mood disorders, 3–5, 11, 13, 104, 160, 206 Movement abnormalities, 12, 34, 167, 222, 239–254, 261 Multiple sclerosis, 3–5, 59, 64, 147 Myoclonic epilepsy, 42

N N-acetyl-aspartate (NAA), 88, 92, 98 Nerve growth factor (NGF), 12, 131 Neural network, 38, 45, 78, 165, 173, 180 Neuregulin–1(NRG1), 96–98, 132, 155, 161, 171, 177, 188, 189 Neurochemistry, 259 Neurocognitive deficit, 11, 12, 45, 57, 58, 132, 165, 177, 178, 24, 247 Neurodegenerative disorder, 14 Neurodevelopment, 41, 189 Neurodiagnosis, 141 Neuroimaging endophenotypes, 48, 89, 96, 99 Neuroinformatics, 141, 147 Neuron, 13, 49, 50, 60, 61, 64, 67, 76–79, 147, 150, 166, 168, 215, 242, 243, 245 Neuropathology, 14, 98 Neurophysiology, 41, 47, 49, 50, 53, 61, 178, 211, 215, 218, 242 Neurophysiological endophenotypes, 47, 48, 113, 117, 118, 178, 184, 188 Neuropsychiatric disorders, 3–6, 8, 11, 13–15, 42, 74, 87, 93, 98, 99, 103–109, 114, 177 Neuroscience, 3, 15, 53, 54, 71, 104, 125, 136, 142, 148 Nicotinic acetylcholine receptor (nAChR), 114, 117, 119 N-methyl-D-aspartate (NMDA), 49, 72, 76, 120, 125, 135, 145, 177, 189, 215

Index North American Prodromal Longitudinal Study (NAPLS), 239, 247 Nuclear magnetic resonance spectroscopy, 3, 14

O Obsessive-Compulsive Disorder (OCD), 6, 14, 31, 228, 230, 234 Obsessive-Compulsive Personality Disorder (OCPD), 227, 230, 233 Operational Criteria Checklist of Psychotic Illness (OPCRIT), 148

P P50 Sensory Abnormality, 11, 12, 114, 117, 119, 120 Panic disorder, 234 Parkinson’s disease, 3, 10, 13, 14, 45, 106, 107, 241–243, 253 Peripheral markers, 259 Personality, 6, 11, 13, 34, 51, 54, 76, 125, 182, 183, 185, 227, 228, 230, 239, 245, 248–250 Pharmacogenetics, 6, 7, 25 Pharmacogenomics, 6, 7, 15, 25, 69 Pharmacoproteomics, 6 Phenotype, 6, 9, 10, 30, 32–34, 41–44, 51–53, 78, 87, 89, 90, 92–98, 104, 114, 115, 118, 125, 127, 128, 131, 134, 136, 147, 151, 156, 161, 170, 177, 188–190, 195, 196, 200, 207, 208, 218, 227, 253, 262 Phenotypic variation, 41, 52, 200, 261 Plasticity, 59, 79 Polymorphism, 27, 37, 44, 46, 47, 49, 52, 87, 88, 96–98, 114, 115, 117, 121, 135, 151, 171, 177, 188, 253 Positive and Negative Syndrome Scale (PANSS), 114, 121 Positron emission tomography (PET), 57, 60, 66–68, 88, 90–92, 97, 98, 132, 163, 165, 167, 239, 243, 244, 253 Prefrontal cortex (PFC), 47, 57, 63, 66–69, 79, 97, 107, 118, 119, 135, 145, 146, 156, 165, 166, 168, 171, 179–181, 207, 231, 243 Prepulse Inhibition (PPI), 12, 34, 47, 49, 51, 114, 116, 118 Prepulse inhibition of startle, 47, 49, 118, 171 Prognostic markers, 6, 27 Proteomics, 3, 7, 8, 10, 13, 103–109, 259

R Regulator of G protein signaling, 4 (RGS4), 114, 117, 146, 171, 188 Rey Auditory Verbal Learning (RAVLT), 46, 50, 195, 201 Ribonucleic acid (RNA), 5, 23

S Selective serotonin reuptake inhibitor (SSRI), 23–25, 27, 30, 31 Seizures, 47 Serotonin (5-HT), 23, 24, 27, 91, 98, 171, 172, 188, 227, 230, 231 Serotonin receptor (5-HT2C and, 5-HT2A) 12, 72, 146, 230, 231 Scale for the Assessment of Positive (Psychotic) Symptoms (SAPS), 132 Scale for the Assessment of Negative (Psychotic) Symptoms (SANS), 132 Screening markers, 6 Schizophrenia, 3–6, 8, 11–14, 28–32, 34, 36, 41–54, 57–68, 70–79, 87–89, 95–99, 103, 104, 107, 108, 113–121,

273 125–136, 141–143, 145–152, 155–161, 163–173, 177–190, 195, 197, 201, 202, 204, 206–208, 211–218, 221–226, 239–248, 251–254, 259–261 Schizotypal Personality Disorder (SPD), 239, 246–251 Side effects, 10, 25, 104, 108, 109, 120, 121, 169, 179, 239, 241, 250, 253, 254 Single Nucleotide Polymorphisms (SNPs), 13, 87–89, 93–97, 134, 135, 177, 188, 189 Single photon emission computed tomography (SPECT), 57, 88, 91, 92, 97, 98 Smooth pursuit, 11–13, 34, 43 Social cognition, 58, 66, 164, 166, 170, 172, 173, 227, 233–235 State marker, 6, 15, 211–218, 222 Stratification markers, 6 Stress, 14, 32, 53, 63, 70, 158, 228, 230, 244–246, 248, 253, 254 Striatum, 47, 63, 118, 165, 166, 231, 242, 243, 247, 249 Stockings of Cambridge (SOC), 195, 201 Stroke, 64 Structured Clinical Interview for DSM Disorders (SCID), 125, 130–132, 134, 239, 248 Suicide, 13, 206 Suppression of P50, 11, 13, 47, 116, 118–120 Synapse, 43, 64, 76, 150 Symptoms, 3, 5, 7, 8, 15, 23–25, 28, 29, 31, 37, 41–46, 48, 52–54, 57, 59, 60, 62, 67, 68, 74, 75, 78, 89, 97, 103, 104, 106–109, 113–115, 119–121, 126–128, 132–134, 145–147, 150, 156, 157, 159, 164, 165, 169, 170, 172, 178–180, 196, 198, 199, 204, 206, 207, 211, 216–218, 234, 239, 240, 242, 245, 247, 248, 250–252, 254 Symptom dimensions, 208

T Thalamus, 12, 48, 63, 70, 118, 119, 180, 215, 231, 239, 242, 243 The long serotonin transporter promoter region (5-HTTLPR), 11 Tolerance, 53, 129, 144 Toxicity, 99, 199 Tower of London (TOL), 195, 201 Trait marker, 3, 6, 8, 15, 44, 164, 169, 171, 211–214, 216–218, 221–223, 225 Transcranial magnetic stimulation, 261 Translational approach, 4, 15, 212

V Verbal memory, 12, 45, 51, 58, 66, 126, 127, 129, 130, 132–134, 143, 179, 180, 186, 187, 198, 199, 202–204, 207 Variable number of tandem repeat (VNTR), 97, 98 Vascular dementia, 106 Visual scan path, 166, 170, 221, 223 Voxel based morphometry (VBM), 90

W Wisconsin Card Sorting Test (WCST), 96, 129, 146, 155, 160, 195, 197, 198, 201, 203, 205, 207, 227, 23, 233 Working Memory, 12, 45, 46, 48, 49, 51, 53, 58, 64, 66–70, 74, 78, 79, 97, 117, 118, 120, 125–127, 129–131, 133–136, 151, 155–161, 171, 177–180, 183, 187, 197, 200, 201, 203, 204